The Clinical Question

Cochrane Database of Systematic Reviews Skin antisepsis for reducing central venous catheter-related infections (Review) Lai NM, Lai NA, O’Riordan E, Chaiyakunapruk N, Taylor JE, Tan K Lai NM, Lai NA, O’Riordan E, Chaiyakunapruk N, Taylor JE, Tan K.

Skin antisepsis for reducing central venous catheter-related i nfections.

Cochrane Database of Systematic Reviews 2016, Issue 7. Art. No.: CD010140.

DOI: 10.1002/14651858.CD010140.pub2.

www.cochranelibrary.com Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by Jo hn Wiley & Sons, Ltd. T A B L E O F C O N T E N T S1 HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 SUMMARY OF FINDINGS FOR THE MAIN COMPARISON . . . . . . . . . . . . . . . . . . .

7 BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8 OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8 METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13 RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Figure 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Figure 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Figure 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Figure 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 26 DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35 CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

66 DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Analysis 1.1. Comparison 1 Povidone-iodine (in aqueous solut ion) versus no skin antisepsis, Outcome 1 Catheter-related BSI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Analysis 1.2. Comparison 1 Povidone-iodine (in aqueous solut ion) versus no skin antisepsis, Outcome 2 Catheter colonisation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Analysis 2.1. Comparison 2 Chlorhexidine (in aqueous soluti on) versus no skin antisepsis, Outcome 1 Septicaemia. . 71 Analysis 2.2. Comparison 2 Chlorhexidine (in aqueous soluti on) versus no skin antisepsis, Outcome 2 Catheter colonisation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Analysis 2.3. Comparison 2 Chlorhexidine (in aqueous soluti on) versus no skin antisepsis, Outcome 3 Number of patients who required antibiotics during in-dwelling period of cathete r. . . . . . . . . . . . . . . . . . 72 Analysis 3.1. Comparison 3 Alcohol versus no skin antisepsis , Outcome 1 Catheter colonisation. . . . . . . . 72 Analysis 4.1. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 1 Catheter-related BSI. . . . . . . 73 Analysis 4.2. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 2 Catheter-related BSI per 1000 catheter- days. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 Analysis 4.3. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 3 All-cause mortality. . . . . . . 75 Analysis 4.4. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 4 Catheter colonisation. . . . . . 76 Analysis 4.5. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 5 Catheter colonisation per 1000 catheter- days. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Analysis 4.6. Comparison 4 Chlorhexidine versus povidone-i odine, Outcome 6 Insertion site infection. . . . . . 78 Analysis 5.1. Comparison 5 Chlorhexidine (in aqueous soluti on) versus alcohol, Outcome 1 Catheter-related BSI. . 78 Analysis 5.2. Comparison 5 Chlorhexidine (in aqueous soluti on) versus alcohol, Outcome 2 Catheter colonisation. . 79 Analysis 6.1. Comparison 6 Povidone-iodine (in aqueous solut ion) versus alcohol, Outcome 1 Catheter-related BSI. 79 Analysis 6.2. Comparison 6 Povidone-iodine (in aqueous solut ion) versus alcohol, Outcome 2 Catheter colonisation. 80 Analysis 7.1. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 1 Catheter-related BSI. . . . . . . 81 Analysis 7.2. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 2 Catheter-related BSI per 1000 catheter- days. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Analysis 7.3. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 3 Catheter colonisation. . . . . . . 82 Analysis 7.4. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 4 Catheter colonisation per 1000 catheter- days. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Analysis 7.5. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 5 Skin colonisation. . . . . . . . 83 Analysis 7.6. Comparison 7 Alcohol versus octenidine in alcoho l, Outcome 6 Adverse effects. . . . . . . . . 83 i Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 8.1. Comparison 8 Chlorhexidine (in alcohol) plus povidone-iodine (in aqueous solution) versus chlorhexidine (in alcohol), Outcome 1 Catheter colonisation. . . . . . . . . . . . . . . . . . . . . . . . 84 Analysis 8.2. Comparison 8 Chlorhexidine (in alcohol) plus po vidone-iodine (in aqueous solution) versus chlorhexidine (in alcohol), Outcome 2 Catheter colonisation per 1000 catheter-days . . . . . . . . . . . . . . . . 84 Analysis 9.1. Comparison 9 Chlorhexidine (in alcohol) plus po vidone-iodine (in aqueous solution) versus povidone-iodine (in aqueous solution), Outcome 1 Catheter colonisation. . . . . . . . . . . . . . . . . . . . 85 Analysis 9.2. Comparison 9 Chlorhexidine (in alcohol) plus po vidone-iodine (in aqueous solution) versus povidone-iodine (in aqueous solution), Outcome 2 Catheter colonisation per 1000 catheter-days. . . . . . . . . . . . 85 Analysis 10.1. Comparison 10 Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine aqueous 10% scrub, Outcome 1 Cathete r colonisation. . . . . . . . . . . 86 86 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

93 CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94 DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94 SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

95 DIFFERENCES BETWEEN PROTOCOL AND REVIEW . . . . . . . . . . . . . . . . . . . . .

96 INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. [Intervention Review] Skin antisepsis for reducing central venous catheter-related infections Nai Ming Lai 1 ,2 , Nai An Lai 3 , Elizabeth O’Riordan 4 , Nathorn Chaiyakunapruk 2 ,5 ,6 , Jacqueline E Taylor 7 , Kenneth Tan 8 1 School of Medicine, Taylor’s University, Subang Jaya, Malays ia.2 School of Pharmacy, Monash University Malaysia, Selangor, Ma laysia.

3 Intensive Care Unit, Queen Elizabeth II Jubilee Hospital, C oopers Plains, Australia.4 Faculty of Nursing and Midwifery, The University of Sydney and The Children’s Hospital at Westmead, Sydney, A ustralia.5 Center of Pharmaceutical Outcomes Research, Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences, Phitsanu lok, Thailand.6 School of Population Health, The University of Queensland, Brisbane, Australia. 7 Monash Newborn, Monash Medical Centre/Monash University, C layton, Australia.8 Department of Paediatrics, Monash University, Melbourne, Australia Contact address: Nai Ming Lai, School of Medicine, Taylor’s Uni versity, Subang Jaya, Malaysia. [email protected] , [email protected] .

Editorial group: Cochrane Wounds Group.

Publication status and date: New, published in Issue 7, 2016.

Citation: Lai NM, Lai NA, O’Riordan E, Chaiyakunapruk N, Taylor JE, Tan K. Skin antisepsis for reducing cen- tral venous catheter-related infections. Cochrane Database of Systematic Reviews 2016, Issue 7. Art. No.: CD010140. DOI:

10.1002/14651858.CD010140.pub2.

Copyright © 2016 The Cochrane Collaboration. Published by Jo hn Wiley & Sons, Ltd.

A B S T R A C T Background The central venous catheter (CVC) is a device used for many functio ns, including monitoring haemodynamic indicators and admin- istering intravenous medications, uids, blood products and parenteral nutrition. However, as a foreign object, it is susceptible to colonisation by micro-organisms, which may lead to catheter-rela ted blood stream infection (BSI) and in turn, increased mortality, morbidities and health care costs.

Objectives To assess the effects of skin antisepsis as part of CVC care for r educing catheter-related BSIs, catheter colonisation, and patient mortality and morbidities.

Search methods In May 2016 we searched: The Cochrane Wounds Specialised Regist er; The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library); Ovid MEDLINE (including In-Process & Othe r Non-Indexed Citations and Epub Ahead of Print); Ovid EMBASE and EBSCO CINAHL Plus. We also searched clinical trial re gistries for ongoing and unpublished studies. There were no restrictions with respect to language, date of publication or s tudy setting.

Selection criteria We included randomised controlled trials (RCTs) that assessed any type of skin antiseptic agent used either alone or in combination, compared with one or more other skin antiseptic agent(s), placeb o or no skin antisepsis in patients with a CVC in place.

Data collection and analysis Two authors independently assessed the studies for their el igibility, extracted data and assessed risk of bias. We expressed our results in terms of risk ratio (RR), absolute risk reduction (ARR) and numbe r need to treat for an additional bene cial outcome (NNTB) for dichotomous data, and mean difference (MD) for continuous data, with 95% con dence intervals (CIs). 1 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Main results Thirteen studies were eligible for inclusion, but only 12 studies contributed data, with a total of 3446 CVCs assessed. The total number of participants enrolled was unclear as some studies did not pr ovide such information. The participants were mainly adults a dmitted to intensive care units, haematology oncology units or general w ards. Most studies assessed skin antisepsis prior to insertion and regularly thereafter during the in-dwelling period of the CVC, ranging from every 24 h to every 72 h. The methodological quality of the included studies was mixed due to wide variation in their risk of bias. Most trials did not adequately blind the participants or personnel, and four of the 12 studies had a high risk of bias for incomplete out come data.

Three studies compared different antisepsis regimens with n o antisepsis. There was no clear evidence of a difference in all o utcomes examined, including catheter-related BSI, septicaemia, cathet er colonisation and number of patients who required systemic antibiotics for any of the three comparisons involving three different an tisepsis regimens (aqueous povidone-iodine, aqueous chlorhe xidine and alcohol compared with no skin antisepsis). However, there were great uncertainties in all estimates due to underpowered analyses and the overall very low quality of evidence presented.There wer e multiple head-to-head comparisons between different skin an tiseptic agents, with different combinations of active substance and ba se solutions. The most frequent comparison was chlorhexidine solution versus povidone-iodine solution (any base). There was very low quality evidence (downgraded for risk of bias and imprecision)that chlorhexidine may reduce catheter-related BSI compared with pov idone-iodine (RR of 0.64, 95% CI 0.41 to 0.99; ARR 2.30%, 95% CI 0.06 to 3.70%). This evidence came from four studies involving 1 436 catheters. None of the individual subgroup comparisons ofaqueous chlorhexidine versus aqueous povidone-iodine, alcoholic chlo rhexidine versus aqueous povidone-iodine and alcoholic chlor hexidine versus alcoholic povidone-iodine showed clear differences for catheter-related BSI or mortality (and were generally underpo wered).

Mortality was only reported in a single study.

There was very low quality evidence that skin antisepsis with chlorhexidine may also reduce catheter colonisation relative to povidone- iodine (RR of 0.68, 95% CI 0.56 to 0.84; ARR 8%, 95% CI 3% to 12%; ; ve studies, 1533 catheters, downgraded for risk of bias, indirectness and inconsistency).

Evaluations of other skin antiseptic agents were generally in single, small studies, many of which did not report the prim ary outcome of catheter-related BSI. Trials also poorly reported other ou tcomes, such as skin infections and adverse events.

Authors’ conclusions It is not clear whether cleaning the skin around CVC insertion s ites with antiseptic reduces catheter related blood stream infection compared with no skin cleansing. Skin cleansing with chlorhexid ine solution may reduce rates of CRBSI and catheter colonisation compared with cleaning with povidone iodine. These results ar e based on very low quality evidence, which means the true effect s may be very different. Moreover these results may be in uenced by the nature of the antiseptic solution (i.e. aqueous or alcohol-based).

Further RCTs are needed to assess the effectiveness and safet y of different skin antisepsis regimens in CVC care; these sho uld measure and report critical clinical outcomes such as sepsis, catheter-rela ted BSI and mortality.

P L A I N L A N G U A G E S U M M A R Y Skin antisepsis for reducing central venous catheter-rela ted infections Review Question We reviewed the evidence about whether using antiseptic trea tments on people’s skin helps reduce infections related to cent ral venous catheters (CVCs).

Background Central venous catheters (CVCs) are thin, exible tubes that a re inserted through the skin into a large vein, often in the arm or chest.

The tube can then be used to give uids, medicine and nutrition t o chronically and critically ill patients. However, CVCs pose a signi cant risk of infection by providing a way for micro-organi sms (germs) to spread into the body at the point where the cathet er is inserted. In order to try to reduce catheter-related infectio ns, healthcare staff frequently use antiseptic solutions toclean the skin around the catheter insertion site, both prior to insertion a nd whilst the catheter is in place. In this review, we summarise the evidence of the bene ts and harms of using antiseptics on the skin, and t he effects of different antiseptic solutions.

Search date 2 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. We searched multiple medical databases in May 2016.

Study characteristics In May 2016 we searched medical databases to nd randomised controlled trials looking at the use of skin antiseptics in peoplewith CVCs. We included 13 studies in this review, although only 12 s tudies contributed data for a total of 3446 CVCs. The study participants were mainly adults in intensive care units or other specialist hospital units. We reported our ndings in terms of the numbe r of catheters, as some studies did not provide the number of patients assess ed, and some patients had more than one CVC.One study was fund ed by a national research body, ve studies were funded in whole o r in part by at least a pharmaceutical company, and in the remaini ng seven studies funding sources were not stated.

Key results Three studies examined the effect of cleansing versus no cleans ing, and found no clear evidence of differences in blood infectio ns, infections in the catheter and need for antibiotics between pat ients who received cleansing compared to those who did not. Chlo rhexidine solution may reduce blood infections associated with the cathet er compared with povidone-iodine solution (reducing the infecti on rate from 64 cases per 1000 patients with a CVC with povidone io dine to 41 cases of infection per 1000 with chlorhexidine). This translates into the need to treat 44 people to avoid one addit ional bloodstream infection. Chlorhexidine solution may (com pared with povidone iodine solution) also reduce the presence of infectio us organisms within the catheter (reduced from 240 infected cathe ters per 1000 people to 189 infected catheters per 1000 people). It is unclear whether antiseptic skin cleansing in uences mortality rates as only one study reported this and although similar death rate s were observed with povidone iodine and chlorhexidine, smal l numbers mean a difference cannot be ruled out.

Quality of evidence The overall quality of evidence was poor due to aws in the way t he studies were designed, small study sizes, inconsistency ofthe results between the included studies and the nature of the outcomes rep orted. These aws have reduced our con dence in the results of th e studies. This means we cannot be certain whether cleaning the sk in around CVC insertion sites with antiseptic reduces catheter-related blood stream infection and other harmful effects, such as overa ll blood infections and mortality compared with no skin cleansi ng.

Cleansing with chlorhexidine solution may be more effective t han povidone iodine but the quality of the evidence was very lo w. 3 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N[ Explanation ] Chlorhe xidine compare d to povidone - iodine f or patie nts with a ce ntral ve nous cathe te r P atie nt or population: p a t i en t s w i t h a c en t ra l ven o u s c a t h et er Se ttings : h o s p i t a l i n p a t i en t s I nte rve ntion : c h l o rh exi d i n e Comparison : p o vi d o n e-i o d i n e Outcome s I llustrative comparative risk s* ( 9 5 % CI ) Re lative e f f e ct ( 9 5 % CI ) No. of P articipants( studie s) Quality of the e vide nce( GRADE) Assume d risk Corre sponding risk P ovidone - iodine Chlorhe xidine Ca t h et er-rel a t ed BSI - o ver- a l l c o m p a ri s o n b et w een c h l o rh exi d i n e a n d p o vi - d o n e-i o d i n e ( d u ri n g i n -p a t i en t s t a y) St u d y p o p u l a t i o n RR 0 . 6 4 ( 0 . 4 1 t o 0 . 9 9 ) 1 4 3 6( 4 RCTs ) ⊕ Very l o w b,c 6 4 p er 1 0 0 0 4 1 p er 1 0 0 0 ( 2 6 t o 6 3 ) M o d era t e a 4 6 p er 1 0 0 0 2 9 p er 1 0 0 0 ( 1 9 t o 4 5 ) Ca t h et er-rel a t ed BSI - s u b - g ro u p : c h l o rh exi d i n e i n a q u eo u s s o l u t i o n vers u s p o vi d o n e-i o d i n e i n a q u eo u s s o l u t i o n St u d y p o p u l a t i o n RR 0 . 6 4 ( 0 . 3 2 t o 1 . 2 8 ) 4 5 2( 2 RCTs ) ⊕ Very l o w c,d 8 6 p er 1 0 0 0 5 5 p er 1 0 0 0 ( 2 8 t o 1 1 0 ) M o d era t e 8 4 p er 1 0 0 0 5 4 p er 1 0 0 0( 2 7 t o 1 0 8 ) 4 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Ca t h et er-rel a t ed BSI - s u b - g ro u p : c h l o rh exi d i n e i n a l c o - h o l vers u s p o vi d o n e-i o d i n e i n a q u eo u s s o l u t i o n St u d y p o p u l a t i o n RR 0 . 7 7( 0 . 3 9 t o 1 . 5 3 ) 5 0 3( 2 RCTs ) ⊕ Very l o w c,d 7 0 p er 1 0 0 0 5 4 p er 1 0 0 0 ( 2 7 t o 1 0 8 ) M o d era t e 6 9 p er 1 0 0 0 5 3 p er 1 0 0 0 ( 2 7 t o 1 0 6 ) Ca t h et er-rel a t ed BSI - s u b - g ro u p : c h l o rh exi d i n e i n a l c o - h o l vers u s p o vi d o n e-i o d i n e i n a l c o h o l St u d y p o p u l a t i o n RR 0 . 4 ( 0 . 1 3 t o 1 . 2 4 ) 4 8 1( 1 RCT) ⊕⊕⊕ M o d era t e c 4 2 p er 1 0 0 0 1 7 p er 1 0 0 0 ( 5 t o 5 2 ) M o d era t e 4 2 p er 1 0 0 0 1 7 p er 1 0 0 0( 5 t o 5 2 ) Pri m a ry BSI o r c l i n i c a l s ep - s i s N o s t u d i es u n d er t h i s c o m p a ri s o n a s s es s ed t h i s o u t - c o m e. A l l -c a u s e m o rt a l - i t y - Ch l o rh exi d i n e i n a q u e- o u s s o l u t i o n vers u s p o vi - d o n e-i o d i n e i n a q u eo u s s o - l u t i o n Cl i n i c a l a s s es s m en t St u d y p o p u l a t i o n RR 1 . 1 5( 0 . 7 2 t o 1 . 8 3 ) 2 1 3( 1 RCT) ⊕⊕ l o w c,e 2 3 6 p er 1 0 0 0 2 7 1 p er 1 0 0 0 ( 1 7 0 t o 4 3 2 ) M o d era t e 2 3 6 p er 1 0 0 0 2 7 1 p er 1 0 0 0 ( 1 7 0 t o 4 3 2 ) A l l -c a u s e m o rt a l i t y - Ch l o rh exi d i n e i n a l c o h o l vers u s p o vi d o n e-i o - d i n e i n a q u eo u s s o l u t i o n St u d y p o p u l a t i o n RR 0 . 8 ( 0 . 4 8 t o 1 . 3 4 ) 2 2 2( 1 RCT) ⊕⊕ l o w c,e 5 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Cl i n i c a l a s s es s m en t 2 3 6 p er 1 0 0 0 1 8 9 p er 1 0 0 0( 1 1 3 t o 3 1 6 ) M o d era t e 2 3 6 p er 1 0 0 0 1 8 9 p er 1 0 0 0( 1 1 3 t o 3 1 6 ) M o rt a l i t y a t t ri b u t a b l e t h e CVC-rel a t ed i n f ec t i o n s .

N o s t u d i es u n d er t h i s c o m p a ri s o n a s s es s ed t h i s o u t - c o m e. * Th e b a s i s f o r t h e assume d risk( e. g . t h e m ed i a n c o n t ro l g ro u p ri s k a c ro s s s t u d i es ) i s p ro vi d ed i n f o o t n o t es . Th ecorre sponding risk( a n d i t s 9 5 % c o n f i d en c e i n t erva l ) i s b a s ed o n t h e a s s u m ed ri s k i n t h e c o m p a ri s o n g ro u p a n d t h e re lative e f f e cto f t h e i n t erven t i o n ( a n d i t s 9 5 % CI) .

BSI : b l o o d s t rea m i n f ec t i o n ; CI :Co n f i d en c e i n t erva l . GRA DE Wo rk i n g Gro u p g ra d es o f evi d en c e High quality: Fu rt h er res ea rc h i s very u n l i k el y t o c h a n g e o u r c o n f i d en c e i n t h e es t i m a t e o f ef f ec t .

M ode rate quality: Fu rt h er res ea rc h i s l i k el y t o h a ve a n i m p o rt a n t i m p a c t o n o u r c o n f i d en c e i n t h e es t i m a t e o f ef f ec t a n d m a y c h a n g e t h e es t i m a t e.

Low quality: Fu rt h er res ea rc h i s very l i k el y t o h a ve a n i m p o rt a n t i m p a c t o n o u r c o n f i d en c e i n t h e es t i m a t e o f ef f ec t a n d i s l i k el y t o c h a n g e t h e es t i m a t e.

V e ry low quality: We a re very u n c ert a i n a b o u t t h e es t i m a t e. a ’M o d era t e ri s k ’ w a s c a l c u l a t ed f ro m t h e m ed i a n c o n t ro l even t ra t e f o r ea c h o u t c o m e.

b Th ree o f t h e f o u r i n c l u d ed s t u d i es h a d u n c l ea r ri s k s o f b i a s i n a l l o c a t i o n c o n c ea l m en t , a n d a l l h a d h i g h ri s k s o f b i a s i n b l i n d i n g o f p a rt i c i p a n t s a n d p ers o n n el .

c Th e 9 5 % CI w a s w i d e.

d Th ere w a s a n o vera l l very s eri o u s c o n c ern o n ri s k o f b i a s t h a t res u l t ed i n d o w n g ra d i n g o f t w o l evel s : b o t h s t u d i es h a d u n c lea r ri s k o f b i a s u n d er a l l o c a t i o n c o n c ea l m en t a n d h i g h ri s k o f b i a s u n d er b l i n d i n g o f p a rt i c i p a n t s a n d p ers o n n el , a n d o n e s t ud y h a d s eri o u s u n i t o f a n a l ys i s i s s u e a s t h e o u t c o m e w a s rep o rt e d u s i n g c a t h et ers a s t h e u n i t , a n d t h e n u m b er o f c a t h et ers a n a l ys ed exc eed ed t h e n u m b er o f p a rt i c i p a n t s b y o ver 5 0 %, re f l ec t i n g t h a t f a c t t h a t s o m e p a t i en t s rec ei ved m u l t i p l e c a th et ers d u ri n g t h e s t u d y, w h i c h c o u l d h a ve s eri o u s l y a f f ec t ed t h e ef f ec t es t i m a t e.

e Th e s i n g l e s t u d y h a d u n c l ea r ri s k i n a l l o c a t i o n c o n c ea l m en t , h i g h ri s k i n b l i n d i n g o f p a t i en t s a n d p ers o n n el w h i c h m i g h t g i ve ri s e t o p erf o rm a n c e b i a s , w h i c h i n t u rn m i g h t a f f ec t t h e ri s k o f m o rt a l i t y, a s w el l a s h i g h ri s k o f a t t ri t i o n b i a s . 6 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. B A C K G R O U N D Please refer to Appendix 1 for a glossary of terms (lay de nitions in the context of this review only).

Description of the condition The concept of central venous catheterisation was rst introduce d in the early part of the last century by Bleichroder, Forssmann , Duffy and Authaniac, after Bleichroder reportedly inserted t he rst central venous catheter (CVC) in a human in 1905 ( Puri 2009 ). In the past four decades, the use of the CVC has become important in the management of many critically and chronically i ll patients. Insertion of a CVC provides secure vascular access for the administration of intravenous medications, uids, blood pr oducts and parenteral nutrition. It also serves as an essential cond uit for blood sampling, haemodynamic monitoring, renal replacemen t therapy and plasmapheresis.

It is estimated that 5 million CVCs are inserted every year in the United States and 200,000 each year in the UK ( Worthington 2005 ). One of the major problems associated with the use of CVCs is colonisation by micro-organisms that could result in loca l or systemic infection. Research has shown that infectious com- plications associated with CVCs cause signi cant morbidity and mortality, with considerable costs to the healthcare system ( CDC 2011 ;Cicalini 2004 ). In the USA, approximately 80,000 reported cases of CVC-associated blood stream infections (BSIs) occur in intensive care units (ICUs) every year; this number more than triples when considering the entire hospital system ( CDC 2011 ).

Although the exact mortality attributable to these BSIs rema ins unclear, reports have cited gures up to 35% ( CDC 2011 ). The associated cost incurred due to BSIs is considerable, including costs of additional medication, nursing time and increased len gth of hospital stay. The total annual cost of caring for patients w ith CVC-associated BSIs in the USA alone is estimated to range any- where from USD 296 million to USD 2.3 billion ( CDC 2011 ).

Micro-organisms colonise the CVCs and gain access to the blood stream of the patients via three main routes ( CDC 2011 ;Cicalini 2004 ;Pagani 2008 ):

• External surface of CVC through contaminated insertion site • Internal surface of CVC through contamination of catheter hubs, injection ports and lines; usually by the hands of healthcare workers or patients • Contaminated intravenous drugs, infusates and nutritiona l preparations.

For short-term CVCs, investigators have proposed colonisati on from the skin to the external surface of the CVCs as the major route of infection, while for long-term CVCs, the internal sur - face route becomes increasingly important, as the micro-organ- isms gain access to the internal surface as a result of contaminat ion from repeated handling of the CVCs ( Cicalini 2004 ). Description of the intervention A number of evidence-based guidelines have been developed in recent years aimed at reducing CVC-associated BSIs. Important measures recommended by two of the major guidelines include the following ( CDC 2011 ;Pratt 2007 ):

• Staff education • Quality assurance: systematically monitoring compliance to the established guidelines and evaluating issues relating to compliance • Hand hygiene • The use of aseptic technique during insertion and use of CVCs • Effective skin antisepsis at the insertion site • Maximum sterile barrier precautions (i.e. wearing sterile gloves, sterile gown, a cap and a mask and using a large sterile drape) • Use of subclavian vein as the preferred site of insertion rather than the internal jugular or femoral veins, as this ha s been shown to reduce infectious, mechanical and thrombotic complications ( Hamilton 2007 ) • The use of antimicrobial or antiseptic impregnated CVCs.

Effective skin antisepsis throughout the in-dwelling period of the catheter may prevent microbial contamination of the insertion site, thus delaying or reducing the risk of catheter colonisation and the subsequent development of infective complications. Given tha t insertion site contamination leads to colonisation on the ext ernal catheter surface and infection, one would expect skin antisepsi s to have some impact on reducing BSIs, especially with short-term CVCs.

Pioneering work by Pasteur, Semmelweis and Lister laid the f oun- dation for the practice of antisepsis in medicine ( Bankston 2005 ; Bynum 2008 ;Nuland 2003 ). Antisepsis is de ned as the preven- tion of infection by inhibiting the growth of causative micro-or - ganisms, while antiseptics are antimicrobial substances capab le of producing antisepsis ( Taber 2016 ). An ideal antiseptic agent would need to be immediately and persistently effective when appli ed to living tissues, including when a small amount of blood is pres ent, and to be effective against all pathogenic bacteria, viruses, fungi, protozoa, tubercle bacilli and bacterial spores ( Taber 2016 ). At the same time it should be non-toxic to living tissue, hypoallerg enic and safe to use repetitively on all parts of the body ( Edwards 2008 ; Hardin 1997 ). Human skin naturally has abundant microbiolog- ical ora which include resident (i.e. colonising) ora and tran- sient (i.e. contaminating or non-colonising) ora. Resident o ra tend to inhabit deeper layers of the skin and therefore are no t readily removed by the mechanical action of washing with soap and water. In contrast, transient ora are not consistently pr esent in most people and can usually be removed by mechanical action( Larson 1995 ;Ryan 2004 ). Both resident and transient ora are implicated in the pathogenesis of CVC-associated infections, t hus effective skin antisepsis may require not only mechanical remo val 7 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. but also the chemical killing and inhibition of both the resident and transient ora of the human skin ( Edwards 2008 ).

How the intervention might work There is a large number of antiseptic agents available and th ree are considered particularly important in skin antisepsis: chl orhexi- dine, iodine and alcohol. All three agents have a broad spectru m of activity against gram positive, gram negative, aerobic and a naero- bic bacteria, enveloped viruses such as human immunode ciency virus (HIV), herpes simplex virus (HSV) and cytomegalovirus (CMV), as well as fungi, although they differ in their effects ag ainst tubercle bacilli and bacterial spores. We summarise their chara c- teristics here: • Chlorhexidine, which is available mostly as chlorhexidine gluconate and less commonly as chlorhexidine acetate or hydrochloride ( Martindale 2016 ), exercises its antimicrobial action chie y by causing a disruption of microbial cell membranes. Its activity against tubercle bacilli and bacterial spores is limited ( Larson 1995 ;Russell 1986 ). Chlorhexidine gluconate has an intermediate onset of effect, which is reporte d to be minimally affected by organic materials such as blood, pu s or sputum. It also appears to cause relatively low level of ski n irritation and has little allergenic potential. However, i ts activity is pH dependent, and its effect is known to be compromised by many substances, including those used in natural soaps ( Larson 1995 ;Martindale 2016 ).

• Iodine and iodophors exert their antimicrobial effects through chemical destruction of the microbial cell wall and cellular contents. They are effective against tubercle bacilli a nd bacterial spores. They kill bacteria within seconds to minutes but are rapidly inactivated in the presence of organic materials s uch as blood, pus or sputum. There have been reports of frequent ski n irritation, allergic reactions and systemic toxicity in susce ptible individuals ( Edwards 2008 ;Hardin 1997 ;Larson 1995 ).

• Alcohols are available as either ethyl (ethanol), normal- propyl (n-propyl) or isopropyl alcohol for use as antiseptic agents. Alcohols derive their antimicrobial activity from denaturation of cellular proteins. They are effective agains t tubercle bacilli but less so against bacterial spores. Alcohols have a rapid onset of action, but they lose their antimicrobial effe cts very quickly. Importantly for this review, they are often combined with other agents such as chlorhexidine gluconate or iodine to achieve optimal antisepsis. Alcohols are also poor cleaning agents, and their use is usually not recommended when signi cant amounts of blood or dirt are present. There have be en reports of excessive skin drying and discomfort following application ( Larson 1995 ;Martindale 2016 ).

Other antiseptic agents include the following ( Larson 1995 ; Martindale 2016 ):

• Triclosan • Hexachlorophene • Chloroxylenol • Quarternary ammonium compounds such as cetrimide and benzalkonium chloride • Octenidine dihydrochloride • Phenolic or carbolic acid compounds • Hydrogen peroxide.

Why it is important to do this review A meta-analysis showed that using chlorhexidine gluconate for catheter site care reduced the risk of catheter-related BSIs by 49% when compared with povidone iodine ( Chaiyakunapruk 2002 ). However, the meta-analysis only evaluated chlorhexidine gluconate and povidone-iodine as skin antiseptics, and some st ud- ies within it assessed a combination of arterial catheters as w ell as central and peripheral venous catheters. Some uncertaintie s re- main regarding the best agent, or combination of agents, for u se as skin antisepsis for CVCs alone; the optimal interval betw een application of antiseptics as well as the best method for apply ing these agents. Examination of the latest National Healthcare Sa- fety Network report, which superseded the National Nosocomia l Infections Surveillance ( NNIS 2004 ), revealed that the CVC-as- sociated BSI rate in different ICUs in the USA ranges from 1.0 to 5.6 BSI per 1000 CVC-days ( Edwards 2008 ). These gures compare favourably with the previous NNIS gures of 2.7 to 7.4 BSI per 1000 CVC-days ( NNIS 2004 ). The observed improve- ment in CVC-associated BSI rate is probably multifactorial in n a- ture, but the recent educational and awareness campaigns about nosocomial infections and the implementation of infection con- trol measures in many hospitals in the USA may have played a ro le.

The impact of different skin antisepsis regimens in the prese nce of comprehensive infection control measures and lower baselin e BSI rates remains unclear. Furthermore, the availability of new studies using different skin antiseptic preparations and t he contin- uing emergence of drug resistant micro-organisms necessitates a systematic review to aid clinical decision-making and to highli ght future research needs ( O’Grady 2002 ;Parienti 2004 ;Pratt 2007 ).

O B J E C T I V E S To assess the effects of skin antisepsis around central venous catheter sites, on rates of catheter-related BSIs, catheter colo nisa- tion, and patient mortality and morbidities.

M E T H O D S Criteria for considering studies for this review 8 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Types of studies We included randomised controlled trials (RCTs) and cluster RCTs comparing one skin antiseptic regimen (a single agent or a combination of agents) with another regimen (a single agent or a combination of agents, placebo or no antisepsis). We ex- cluded cross-over studies due to the possible contaminating ef- fect of one intervention over another. We also excluded studies assessing CVCs for haemodialysis, as this is covered by anoth er Cochrane review ( McCann 2010 ).

Types of participants We included studies involving adults and children cared for in a hospital setting (in adult or paediatric wards or ICUs) with a ny underlying illness and a CVC inserted for any reason during t he study period. Studies that enrolled a patient more than once w ere acceptable provided that the enrolment took place in separate h os- pital admissions. We excluded studies conducted in neonatal se t- tings, for example in a neonatal intensive care unit (NICU), as t he types of catheters used, the insertion site and techniques, th e pos- sible complications as well as the risk factors for sepsis are di fferent compared with those in older children and adults ( Trieschmann 2007 ).

Types of interventions Intervention The use of any skin antiseptic regimen (a single agent or a combi - nation of agents) used for cleansing the skin around CVC inser - tion sites.

Comparisons A different skin antisepsis regimen (a single agent or a combin ation of agents), placebo or no skin antisepsis for CVC insertion sit es.

We required that the selection, insertion, use, maintenance a nd removal of CVCs in the intervention and comparison groups fol - lowed the standard protocol of the hospital setting in the stu dy.

The skin antisepsis regimen had to be the only systematic dif fer- ence between comparison groups (i.e., not catheter material or concurrent CVC-related antiseptic measures).

We accepted the duration of the studies as variously speci ed by the authors. We did not place any limit on the minimum and maximum duration of the follow-up period for each study.

Types of outcome measures Primary outcomes Number of patients with CVC-related blood stream infection (BSI) • Catheter-related BSI con rmed by laboratory • Primary BSI or clinical sepsis.

We present the criteria for the diagnosis of CVC-related BSI in Appendix 2 (Pagani 2008 ).

Mortality • All-cause mortality • Mortality attributable to CVC-related infections.

We included suitable studies using other de nitions of CVC-re - lated and associated infections, provided the authors justi ed their de nitions with valid sources.

Secondary outcomes • Number of patients with insertion site infection, either microbiologically documented (i.e. exudates at catheter insert ion site yield a micro-organism with or without concomitant BSI) or clinically documented (i.e. erythema or induration within 2 cm of the catheter insertion site in the absence of associated BSI a nd without accompanying purulence) ( Pagani 2008 ) • Number of patients with catheter colonisation, as de ned by the study authors using well-accepted de nitions such as a signi cant growth of micro-organism (more than 15 colony- forming units (CFU)) from the catheter tip, subcutaneous segment or catheter hub in the absence of clinical signs of infection ( Pagani 2008 ) • Number of drug-resistant organisms from cultures, including insertion site cultures, catheter cultures and blood cultures • Number of adverse events associated with the use of antiseptic agents, including skin irritation, contact dermat itis, systemic allergic reaction and anaphylaxis • Antibiotic usage during hospitalisation • Length of hospitalisation, either ICU stay or overall hospital stay • Cost of care, including cost of the antiseptic agent and the cost of treating any adverse effects • Quality of life, measured using validated tools.

Search methods for identi cation of studies Electronic searches We searched the following databases for relevant RCTs: • The Cochrane Wounds Specialised Register (searched 23 May 2016); • The Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library) (2016, Issue 4); 9 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. •Ovid MEDLINE (including In-Process & Other Non- Indexed Citations and Epub Ahead of Print) (1946 to 23 May 2016); • Ovid EMBASE (1974 to 23 May 2016); • EBSCO CINAHL Plus (1937 to 23 May 2016).

We used the search strategy in Appendix 3 to search the Cochrane Central Register of Controlled Trials (CENTRAL). We adapted this strategy for Ovid MEDLINE, Ovid EMBASE and EBSCO CINAHL Plus which can be found in Appendix 4 ,Appendix 5 and Appendix 6 , respectively. We combined the MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identi fying randomised trials in MEDLINE: sensitivity- and precision-ma x- imising version (2011 revision) ( Lefebvre 2011 ). We combined the EMBASE search with the Ovid EMBASE lter developed by the UK Cochrane Centre ( Lefebvre 2011 ). We combined the CINAHL searches with the trial lters developed by the Scottis h Intercollegiate Guidelines Network ( SIGN 2015 ).

We searched the following trial registries for details of ong oing clinical trials and unpublished studies.

• ClinicalTrials.gov ( http://www.clinicaltrials.gov/ ).

• WHO International Clinical Trials Registry Platform ( http://apps.who.int/trialsearch/Default.aspx ).

• EU Clinical Trials Register ( https:// www.clinicaltrialsregister.eu/ ).

Searching other resources We checked for further reports of eligible studies using the cit ation lists of papers identi ed by the above strategies. We also sca nned references lists of relevant Cochrane reviews and guidelines and contacted experts in the eld.

Data collection and analysis Selection of studies Two review authors (NML, EOR) independently assessed the rs t round of search results for potentially relevant studies. We retrieved in full those that appeared to meet the inclusion criteria, or w here this could not be determined, for further assessment. Two rev iew authors independently assessed the full papers retrieved, resolving any disagreement with input from a third review author (NC). We included the studies if they ful lled the criteria for inclusio n as outlined above and if the amount of information contained in t he article enabled the extraction of outcome data for meta-analysi s.

We screened publications for duplicate reports of the same tria l and contacted the trial authors for clari cation when necessary.

If we con rmed a duplicate publication, we identi ed a primary reference, but extracted unique data from all versions. Data extraction and management Two pairs of review authors (NAL and NML, PL and EOR) inde- pendently extracted and coded all data for each included study us - ing a pro forma designed speci cally for this review. Each pair w as responsible for half of the total number of included studies. We extracted the following information on each study: study desi gn, participants, setting, sample size, nature of intervention , compar- ison, outcomes, methods (unit of allocation and analysis) and r e- sults. We screened for duplicate entries of patients, where po ssible, by matching the initial number of patients recruited against t he total number along each step in the conduct of the study.

We found a discrepancy between the number of catheter and the number of patients in most studies. This was due to multiple catheters being inserted in some patients who were enrolled a fter each insertion. We were unable to limit our analysis to one cath eter per participant as none of the studies provided the data in thi s format.

We resolved any disagreement among the review authors by dis - cussion and formulation of a consensus acceptable to all members of the review team.

Assessment of risk of bias in included studies Two authors (NAL and NML) independently assessed each in- cluded study using the Cochrane tool for ’Risk of bias’ assessm ent ( Higgins 2011a ). This tool addresses six speci c domains.

1. Sequence generation 2. Allocation concealment 3. Blinding 4. Incomplete outcome data 5. Selective outcome reporting 6. Other issues (e.g. extreme baseline imbalance, design- speci c risks of bias such as recruitment in cluster for cluster- RCT, block randomisation of unblinded trials or fraud).

We present detailed criteria on which we based our judgement in Appendix 7 . We assessed blinding and completeness of outcome data for each outcome separately. We completed a ’Risk of bias’ table for each eligible study. We resolved any disagreement a mong the review authors by discussion to achieve a consensus. We pre- sented an overall assessment of the risk of bias using a ’Risk of bias summary gure’, which presented all of the judgement in a cross - tabulation of study by entry. This display of internal valid ity indi- cated the weight the reader may give to the results of each study .

In addition, we assessed whether trials followed a standard proto- col for all groups under study with regard to the insertion, us e, maintenance and removal of CVC, and regarding the concurrent use of other antiseptic measures such as antimicrobial impreg nated CVCs, antiseptic-soaked dressing and prophylactic antibioti cs. We referred to the study protocol, where available, for further details if necessary. We made relevant remarks in the corresponding ’R isk of bias’ table for each study if there were signi cant concerns in this aspect. 10 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Measures of treatment effect For dichotomous data, we used risk ratio (RR) to measure out- come estimates of the same scale. We estimated the number needed to treat for an additional bene cial outcome (NNTB) from the pooled risk difference (RD) using an online NNTB calculator ( http://nntonline.net/visualrx/ ). For continuous data, we pooled measures at a similar time point using the mean difference (MD).

Two studies reported the measure of variance as a standard err or (SE) or 95% con dence intervals (CI) ( Humar 2000 ;Dettenkofer 2010 ). We obtained standard deviations (SD) for the above-men- tioned studies from the SE using the formula SD = SE x square root of the number of participants, and from the 95% CI using th e formula SD = square root of the number of participants x (upper limit or CI −lower limit of CI)/3.92.

Unit of analysis issues One potential unit of analysis issue that we had anticipated w as the issue that arose as a result of the studies using catheters , rather than patients, as the unit of analysis in catheter-related out comes such as catheter-related BSI and catheter colonisation. Ideally , if the study performed randomisation and analysis based on the par- ticipants, and each participant had only one catheter evaluated , adjustment for clustering would not have been necessary. How- ever, if a study included multiple catheters per patient and cle arly stated so, we would have assessed whether the authors had und er- taken statistical adjustment to account for the effects of cluste ring by using appropriate analysis models such as the ’generalise d esti- mating equation’ (GEE) model ( Higgins 2011b ). If investigators had made adjustments for clustering, we would have combined th e study with other studies in the meta-analysis. If they had not , or if it was unclear whether there were adjustments made, we woul d have assessed the number of catheters as well as participants i n the study. If the studies had also reported the number of particip ants with events and the total number analysed, we would have only reported the outcomes using the participants, rather than cath eters as the unit of analysis. However, if the study did not provide par- ticipant-level data, we would not have been able to avoid the un it of analysis issues. We would have acknowledged this as a major limitation of the review in our discussion and undertaken sen si- tivity analysis to assess the pooled results after excluding studies with no adjustments for clustering.

However, in this review, none of the included studies provide d participant-level data for catheter-speci c outcomes. As a resul t, we could not adjust for the unit of analysis issue, nor could we per form sensitivity analysis to assess the results with and without studies with unadjusted unit of analysis issues. We have acknowledge d this in our discussion, as planned.

Another possible unit of analysis issue that could have arise n was the effects of clustering that arose in cluster-RCTs in which ran- domisation was performed at the unit, rather than the partici pant level. However, we did not include any cluster-RCTs in this revi ew.Had we identi ed an eligible cluster-RCT (e.g. trial in which the assignment to intervention or control group was made at the le vel of the unit or ward rather than the individual), we would have addressed the possible unit of analysis issues as follows.

First, we would have assessed whether the authors had made ad - justments for the effects of clustering to account for non-inde- pendence among the participants by using appropriate analysi s models such as the ’generalised estimating equation’ (GEE) mo del ( Higgins 2011b ).

If investigators did not make adjustments for the effects of cl us- tering, we would have performed adjustment by multiplying t he SEs of the nal effect estimates by the square root of the ’desi gn effect’, represented by the formula ’1 + ( m− 1) x ICC’, where m is the average cluster size (number of participants per cluster) and ICC is the intracluster correlation. We would have determined the average cluster size mby dividing the total number of participants by the total number of clusters. We would have used an assumed ICC of 0.10, which has been proposed to be a realistic general estimate based on previous similar studies ( Campbell 2001 ). We would also have combined the adjusted nal effect estimates fr om each trial with their SEs in our meta-analysis using the generi c inverse-variance methods, as stated in the Cochrane Handbook for Systematic Reviews of Interventions ( Higgins 2011b ).

If it were impossible to nd out whether trialists made adjus tments on the effect of clustering, we would still have included the stu dies concerned in our meta-analysis using the effect estimates repor ted by the authors, and performed sensitivity analyses to asses s how excluding those studies would affect the overall pooled estim ates.

Dealing with missing data We assessed whether there was a high attrition rate and wheth er an intention-to-treat analysis was performed. To assess whethe r the dropout rate was important, we inspected the absolute attrit ion rate and the attrition rate in relation to the event rates for the inter- vention and the comparison groups. If the absolute dropout ra te was 20% or more, we judged the study to be at high risk of bias due to incomplete outcome data. If the dropout rate was lower than 20%, we used a ’worst-case-scenario’ method for the primary outcomes ( Guyatt 1993 ). For instance, for an unfavourable out- come such as catheter-related BSI or mortality, if the results of a trial favoured the intervention group, we assumed all drop outs from the intervention group to have developed the outcome, an d all dropouts from the comparison group to have not developed t he outcome. We then analysed to see if such an assumption changed the direction of the results (e.g. from favouring the interven tion group to favouring the comparison group). If so, we considered the dropout rate to be signi cant. We made the reverse assump- tion when a trial favoured the comparison group, or when the outcomes examined were favourable, such as survival or treatm ent success. 11 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Assessment of heterogeneity We assessed all the included studies in terms of their clinical and methodological characteristics. 1. Baseline characteristics of the participants 2. Clinical settings of the studies (e.g. intensive care units, oncology wards, renal units) 3. Co-interventions 4. Methodological quality (as detailed in the ’Risk of bias’ assessment, for example studies at high risk of bias are de n ed as studies with unclear or no allocation concealment, and studies where participants, caregivers or investigators are not blin ded, or where blinding is unclear) 5. Nature of intervention (comparison between one skin antiseptic regimen and placebo as opposed to comparison of two active regimens) 6. Outcome assessment and unit of analysis.

We visually inspected the forest plots for any evidence of hete ro- geneity of treatment effects. We used the I 2 statistic ( Higgins 2003 ) to measure inconsistency in the results, with a value of 50% or greater indicating moderate to substantial statistical hete rogeneity.

We found signi cant statistical heterogeneity in one analysi s ( Analysis 4.4 ) and provided a plausible explanation the possible reason for heterogeneity in the form of risk of attrition bia s in some included studies. We decided to still provide the pooled estim ate for this analysis and separated the studies based on the risk of attrition bias in our pre-speci ed sensitivity analysis.

Assessment of reporting biases We planned to screen for publication bias in our review using a funnel plot if there were more than 10 studies included in the analysis. If publication bias was implied by a signi cant asym metry of the funnel plot, we would have included a statement in our results with a corresponding note of caution in our discussion. We did not generate any funnel plot in this review as there were f ewer than 10 studies included in the analysis across all the comparis ons and outcomes.

Data synthesis We used Review Manager software to perform meta-analysis of t he included studies ( RevMan 2014 ). We used a xed-effect model for most of our analyses, as there was no substantial clinical and s tatis- tical heterogeneity. For the outcomes with substantial clinica l and statistical heterogeneity that was not satisfactorily expla ined or re- duced by subgroup analyses, we used a random-effects model that took into account between-study variability within the analys is and lessened the possibility of spurious inferences of signi cance compared to the xed-effect model. We used the Mantel-Haen- szel method to analyse all the dichotomous outcomes, as we an- ticipated relatively frequent events for most of our outcomes . For continuous outcomes, we employed the inverse variance methods using the effect measure of mean differences. In our assessmen tof the effects of missing data, we compared our adjusted analys is using the best- and worst-case scenarios to the completer analys is as reported by the study authors.

When there were more than two arms evaluated in a study, for ex- ample, aqueous chlorhexidine versus alcoholic chlorhexidine ver- sus aqueous povidone-iodine, we set up separate pairwise com- parisons as subgroups under the major comparison of chlorhex- idine versus povidone-iodine, as follows: aqueous chlorhexi dine versus aqueous povidone-iodine; and alcoholic chlorhexidine ver- sus aqueous povidone-iodine. In so doing, we halved the total number of participants and events in the povidone-iodine grou p to avoid double-counting.

Had we identi ed studies that assessed cost-effectiveness, we planned to provide only a narrative review of their ndings a nd not directly compare costs in studies using different units of m ea- surement, due to the complexity of analysing cost-effectivenes s if different price-years were used.

Subgroup analysis and investigation of heterogeneity In this review, we created subgroups of comparisons based on the solution used, for example, a subgroup for chlorhexidine in aqueous solution versus povidone iodine in aqueous solutio n, and another subgroup for chlorhexidine in alcohol versus povidon e- iodine in aqueous solution.

Had data been available, we would have carried out the followi ng subgroup analyses: 1. Short term CVCs (less than 10 days) versus longer term CVCs (10 days or more) 2. CVCs with antimicrobial modi cations (antimicrobial impregnation, cuffs, hubs) versus CVCs with no antimicrobial modi cations 3. Studies undertaken in paediatric patients versus adult patients 4. Studies undertaken in different patient populations wit h different levels of care (intensive care patients, oncology pat ients, renal patients and patients in general medical or surgical war ds) 5. Studies undertaken with co-interventions (e.g. sepsis prevention bundle) versus studies done without co-intervent ions 6. Studies that used rigorous criteria (e.g. as outlined in Pagani 2008 ) for determining catheter-related infections versus studies that used more liberal criteria.

Sensitivity analysis We performed the following sensitivity analyses. 1. Best- and worst-case scenarios to assess the impact of missing data, as described in the section ’ Dealing with missing data ’.

2. Including and excluding studies with unclear and high risks of selection bias, namely, studies with unclear or high risk fo r random sequence generation, allocation concealment or both. 12 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Had suf cient data been available, we would have performed ad- ditional sensitivity analyses to include and exclude studies with methodological issues other than selection bias, such as a lack o f blinding to the participants, caregivers or investigators, o r where blinding was unclear.

’Summary of ndings’ table We created a ’Summary of ndings’ table, which displayed seven major outcomes in our review, using the web-based GRADEpro software ( http://gdt.guidelinedevelopment.org ) ( Schünemann 2011a ). We used the eight GRADE considerations (study limita- tions, consistency of effect, imprecision, indirectness and pub li- cation bias, large effect, plausible confounding and dose resp onse relationship) to assess the overall quality of the body of ev idence ( Schünemann 2011b ). In generating the ’Summary of ndings’ table, we interpreted the median control group event rate for the outcome as ’moderate risk’.

R E S U L T S Description of studies Results of the search We identi ed 609 records from the initial search of the Cochrane Wounds Group Specialised Register, CENTRAL, MEDLINE, EMBASE and CINAHL. We performed additional searches from relevant published studies and identi ed two further studi es that appeared to be relevant. After removing duplicates, there we re 574 records. Of these, 107 articles appeared to be relevant after w e in- spected the titles. We evaluated the abstracts and if necessary , the full text of the articles, excluding 84 of the 107 records, includ ing one duplicate publication of another excluded study. Of the re- maining 23 articles, one was an ongoing study, and we could not fully assess six as we are still awaiting their full texts or f urther in- formation from the authors. Ultimately, 16 articles describi ng 13 studies were available and met our inclusion criteria. Among t hese 16 articles, three were additional publications relating to t hree in- cluded studies. The ow diagram of the studies from the initia l search to the meta-analysis is shown in Figure 1 . We describe all the included studies in the Characteristics of included studies table and note the reasons for excluding the others in the Characteristics of excluded studies table. 13 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Figure 1. Study ow diagram. 14 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Included studies We included 13 RCTs, conducted in eight countries, including the USA (four studies), France (two studies), and Canada, Germany, Iran, Japan, Spain, Switzerland and Finland (1 study each). Ten trials were single centre RCTs and three were multicentre RCTs( Dettenkofer 2010 ;Humar 2000 ;Yasuda 2013 ) The number of patients recruited ranged from 50 (with 50 CVCs) in Sadowski 1988 to 420 (with 998 CVCs) in Vallés 2008 .Mimoz 1996 , Mimoz 2007 and Yasuda 2013 did not report the number of participants. Prager 1984 recruited children (n = 3) in addition to adults (in this case, n = 159), while Sadowski 1988 recruited children and adolescent from 10 weeks to 15 years of age. All studies included participants of both sexes.

Six studies recruited patients from the medical/surgical ICUs ( Maki 1991 ;Mimoz 1996 ;Mimoz 2007 ;Vallés 2008 ;Tuominen 1981 ;Yasuda 2013 ), two studies recruited patients who were ei- ther pre- or post-cardiac surgery ( Levy 1988 ;Yousefshahi 2013 ), one study enrolled patients from a burns unit ( Sadowski 1988 ), one from haematology and surgical units ( Dettenkofer 2010 ) and the remaining three studies were conducted hospital-wide, whi ch included intensive-care and non intensive-care patients ( Humar 2000 ;Langgartner 2004 ;Prager 1984 ). The average duration of catheterisation, where reported, varied from 2 to 21.1 days (r ange 1 to > 30 days).

There were ten basic comparisons between two or three arms in the included studies, with subgroups based on type of solutio n in two comparisons.

• Comparison 1: povidone-iodine (in aqueous solution) versus no skin antisepsis ( Prager 1984 ).

• Comparison 2: chlorhexidine (in aqueous solution) versus no skin antisepsis ( Tuominen 1981 ).

• Comparison 3: alcohol versus no skin antisepsis ( Sadowski 1988 ).

• Comparison 4: chlorhexidine versus povidone-iodine ( Humar 2000 ;Maki 1991 ;Mimoz 2007 ;Vallés 2008 ;Yasuda 2013 ). The speci c subgroups for this comparison are listed below based on the different preparations of chlorhexidine a nd/ or povidone-iodine:

◦Chlorhexidine in aqueous solution versus povidone- iodine in aqueous solution ( Maki 1991 ;Vallés 2008 ).

◦ Chlorhexidine in alcohol versus povidone-iodine in aqueous solution ( Humar 2000 ;Vallés 2008 ).

◦ Chlorhexidine in alcohol versus povidone-iodine in alcohol ( Mimoz 2007 ).

◦ Chlorhexidine in alcohol versus povidone-iodine (base solution unknown) ( Yasuda 2013 ).

Among the studies included in this comparison, two ( Vallés 2008 ; Yasuda 2013 ) carried out three-arm comparison. Vallés 2008 com- pared 2% chlorhexidine in aqueous solution (group 1), 0.5% chlorhexidine in alcohol (group 2) and 10% povidone-iodine in aqueous solution (group 3), while Yasuda 2013 compared 1% chlorhexidine in alcohol (group 1), 0.5% chlorhexidine in alcohol (group 2) and 10% povidone-iodine (base solution unknown).

Because the authors of Yasuda 2013 did not specify the base so- lution for the povidone-iodine group, we could not include this study in any subgroup in our meta-analysis. • Comparison 5: chlorhexidine (aqueous) versus alcohol ( Maki 1991 ).

• Comparison 6: povidone-iodine versus alcohol.

◦Povidone-iodine in aqueous solution versus alcohol ( Maki 1991 ).

◦ Povidone-iodine-impregnated adherent lm versus alcohol ( Levy 1988 ).

• Comparison 7: alcohol versus octenidine in alcohol ( Dettenkofer 2010 ).

• Comparison 8: chlorhexidine (in alcohol) plus povidone- iodine (in aqueous solution) versus chlorhexidine in alcohol ( Langgartner 2004 ).

• Comparison 9: chlorhexidine (in alcohol) plus povidone- iodine (in aqueous solution) versus povidone-iodine (in aqueo us solution) ( Langgartner 2004 ).

• Comparison 10: Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine 10% aqueous scrub ( Yousefshahi 2013 ).

In terms of the timing of intervention, most studies assesse d skin antisepsis prior to insertion and regularly thereafter dur ing the in- dwelling period of the catheters, ranging from every 24 h to ev ery 72 h. Three studies evaluated the skin antisepsis intervent ion only prior to catheter insertion ( Levy 1988 ;Yasuda 2013 ;Yousefshahi 2013 ), and one study examined skin antisepsis prior to removal of the catheters ( Sadowski 1988 ).Maki 1991 and Mimoz 1996 evaluated central venous as well as arterial catheters, altho ugh only Maki 1991 provided a separate report of patients receiving CVCs for the outcomes of catheter-related BSI and catheter colonisati on, while only Mimoz 1996 provided CVC-speci c reports for both outcomes per 1000 catheter-days.

The concentration of chlorhexidine-based solution used in the studies ranged from 0.05% to 2%, with three studies using a com - bination of chlorhexidine plus alcohol. The concentration of po vi- done-iodine was 10% in all studies except Mimoz 2007 , which used 5% povidone-iodine together with 70% ethanol. All of the studies that evaluated alcohol used 70% isopropyl alcohol exce pt Dettenkofer 2010 , which used a combination of 45% 2-propanol or 74% ethanol with 10% 2-propanol.

In terms of concomitant CVC-related infection control measures, six studies clearly described the use of maximal sterile barri er 15 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. precaution ( Dettenkofer 2010 ;Humar 2000 ;Langgartner 2004 ; Mimoz 1996 ;Mimoz 2007 ;Vallés 2008 ), three studies described part of the maximal sterile precaution (such as the use of steril e gloves, gown or dressing) without explicitly mentioning max imal sterile precaution ( Levy 1988 ;Maki 1991 ;Yousefshahi 2013 ), and four studies did not provide any clear description ( Prager 1984 ; Sadowski 1988 ;Tuominen 1981 ;Yasuda 2013 ).

The included studies assessed almost exclusively two major ou t- comes, namely, catheter colonisation or equivalent (all 13 stud - ies) and catheter-related BSI or equivalent (8 studies). The oth er outcomes assessed were sepsis, skin colonisation, insertion site in- fection, number of patients who required antibiotics during t he period of catheter use and adverse effects (only evaluated in on e study). Only one study reported mortality ( Vallés 2008 ), and no study reported cost of care or quality of life.

Control group risk of infection varied from 6.0% to 32.0% for catheter colonisation, and from 4.1% to 9.8% for catheter-relat ed BSI.

Of the eight studies that evaluated the primary outcome of catheter-related BSI, all except Yasuda 2013 clearly de ned this outcome in line with our de nitions, detailed in Appendix 2 . The exact wording varied among the studies, but the de nitions in - volved a positive blood culture in the presence of catheter with clinical evidence of sepsis, improvement of the clinical signs fo l- lowing removal of the catheters or both. One study ( Yousefshahi 2013 ) used the Centers for Disease Control and Prevention (CDC) de nitions of catheter-related BSI ( CDC 2011 ), which were also consistent with the de nitions adopted in this review. M ost studies used previously validated laboratory methods to pe rform catheter and blood cultures, adopting microbiological de niti ons for colonisation and bloodstream infection that were consiste nt with published literature in the evaluation of catheter-rela ted infec- tions, including the use of molecular subtyping. In Yasuda 2013 , the published abstract did not contain the de nition of cathete r- related BSI.

All studies reported catheter-related outcomes such as catheter - related BSI and catheter colonisation using the catheter as the unit of analysis. Ten of the 13 included studies provided the numbe r of participants alongside the number of catheters, although non e pro- vided separate reports of the catheter-related outcomes using par- ticipants as the unit of analysis. The number of catheters match ed the number of participants in six studies ( Dettenkofer 2010 ; Levy 1988 ;Humar 2000 ;Maki 1991 ;Sadowski 1988 ;Yousefshahi 2013 ); in three studies, the number of catheters exceeded the num- ber of participants: by 10% in Prager 1984 , 18% in Langgartner 2004 and 50% in Vallés 2008 . In Tuominen 1981 , there were fewer catheters analysed than participants enrolled, with no reason provided.

We did not incorporate the outcome data of Yasuda 2013 into our meta-analysis, as it was published only as an abstract and d id not state the base solution used (either aqueous or alcohol) fo r the povidone-iodine group. We are awaiting further information fromthe authors.

In terms of funding source, one study ( Dettenkofer 2010 ) received funding from a national research agency, ve studies ( Humar 2000 ; Maki 1991 ;Mimoz 1996 ;Mimoz 2007 ;Prager 1984 ) were funded in whole or in part by a pharmaceutical company, and in the remaining seven studies ( Langgartner 2004 ;Levy 1988 ;Sadowski 1988 ;Tuominen 1981 ;Vallés 2008 ;Yasuda 2013 ;Yousefshahi 2013 ), the sources of funding were not stated.

Excluded studies We excluded a total of 83 articles based on one or more of the following reasons. 1. Study design or article type (54 studies): the studies were either retrospective or prospective cohort studies, cross-over study, before-and-after intervention studies, prospective n on- randomised intervention studies, meta-analyses, economic analyses with no original trial data, in vitro experiments, studies with research questions or outcomes that did not match our review, commentaries or an abstract of an included study, excluded study or a study awaiting classi cation.

2. Population (17 studies): the participants in the studies wer e either neonates, people undergoing haemodialysis or all pa tients in ICU, not only those with CVCs in place. 3. Intervention (25 studies): the studies either assessed antimicrobial-impregnated dressing or cerebral ventricular catheter. 4. Insuf cient information (four studies): the studies either reported combined outcome data for arterial, venous or Swan Gantz catheters (or a combination of these), with no separate reporting for venous catheter and little possibility of conta cting the authors for further information, or they reported outcom e data that were unsuitable for meta-analysis.

Among the excluded articles, three articles were merged with ot her articles as their secondary references on the basis of duplicati on of information as stated under reason number 1 above, includi ng two included studies ( Maki 1991 ;Mimoz 1996 ) and one excluded study ( Garland 2009b ).

A description of each study is available in the ’ Characteristics of excluded studies ’ table.

Risk of bias in included studies There was a wide variation in the risk of bias of the included studies. Overall, there was approximately a one-third split in the domains that were judged to be low risk, unclear risk and high r isk.

There was at least one high-risk domain in each of the included studies. All studies were judged to be at high risk for blindi ng of participants, except Dettenkofer 2010 (low risk) and Yousefshahi 2013 (unclear risk). Yasuda 2013 had unclear risks of bias in all domains, as there was insuf cient information in the publish ed abstract. The proportions of included studies with low, high a nd unclear risks of bias in each domain is illustrated in Figure 2 , and 16 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. the risk of bias judgment of each included study in each domain is depicted in Figure 3 . Additionally, we have provided a detailed description of the risk of bias of each study in the ’ Characteristics of included studies ’ table. We summarise our risk of bias assessments for each domain below.

Figure 2. Risk of bias graph: review authors’ judgements abo ut each risk of bias item presented as percentages across all included studies. 17 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Figure 3. Risk of bias summary: review authors’ judgements about each risk of bias item for each included study. 18 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Allocation For random sequence generation, we judged 6 of the 13 in- cluded studies to have low risk of bias ( Dettenkofer 2010 ;Humar 2000 ;Mimoz 1996 ;Mimoz 2007 ;Tuominen 1981 ;Vallés 2008 ).

For allocation concealment, three studies had low risk of bias ( Dettenkofer 2010 ;Langgartner 2004 ;Mimoz 2007 ). In these studies, the authors clearly stated the method of sequence gen - eration, which involved some form of random number scheme, mostly by computers. There were also clear statements in the ’Methods’ that reassured the readers of the independence bet ween sequence generation and allocation. Two studies were judged t o be at high risk in sequence generation as well as allocation con- cealment, as they allocated participants either using an alter nate sequence or based on their hospital registration numbers ( Prager 1984 ;Yousefshahi 2013 ). There was an unclear risk of bias in one or both domains for 8 of the 13 included studies due to insuf - cient information provided in the articles.

Blinding All of the studies except Dettenkofer 2010 ,Yasuda 2013 and Yousefshahi 2013 had a high risk of bias with regard to blinding of participants. Maki 1991 ,Mimoz 1996 and Mimoz 2007 clearly stated that they did not blind participants, while other stud ies did not specify. However, blinding was considered very unlike ly in these studies because they compared either a skin antisepsi s regimen against no regimen, one skin antisepsis solution ag ainst another with a different appearance, or a skin antisepsis reg imen against a different and clearly distinguishable infection con trol measure with no documented attempt to mask the participants.

Eight studies did not report blinding of outcome assessors ( Humar 2000 ;Langgartner 2004 ;Levy 1988 ;Prager 1984 ;Sadowski 1988 ; Tuominen 1981 ;Yasuda 2013 ;Yousefshahi 2013 ), while the other ve did not make any clear statements one way or the other ( Dettenkofer 2010 ;Maki 1991 ;Mimoz 1996 ;Mimoz 2007 ;Vallés 2008 ). Although investigators objectively measured the outcome of catheter colonisation, catheter-related BSI required some d egree of clinical judgment, which might have been affected by lack of blinding.

Incomplete outcome data We judged studies to have a high risk of attrition bias for the following three reasons, alone or in combination: 1. High absolute attrition rates ( ≥20% attrition) or an attrition rate that was higher than the event rates in the cont rol group 2. Vulnerability of the pooled estimates to best- and worst- case scenarios using the dropouts in the assigned groups 3. Marked imbalance in the attrition rates between the assigned groups.

Four studies had high risk of bias in this domain either becaus e they had more than 20% withdrawals ( Dettenkofer 2010 ;Humar 2000 ;Langgartner 2004 ) or because their results changed signi - cantly with best- and worst-case scenarios ( Vallés 2008 ). Six stud- ies had low risk of bias ( Levy 1988 ;Maki 1991 ;Mimoz 2007 ; Prager 1984 ;Sadowski 1988 ;Yousefshahi 2013 ), and the infor- mation on withdrawal was not suf cient in the remaining three studies ( Mimoz 1996 ;Tuominen 1981 ;Yasuda 2013 ).

Selective reporting Nine studies had low risk of reporting bias ( Dettenkofer 2010 ; Humar 2000 ;Maki 1991 ;Mimoz 1996 ;Mimoz 2007 ;Prager 1984 ;Sadowski 1988 ;Tuominen 1981 ;Yousefshahi 2013 ), and three studies carried a high risk ( Langgartner 2004 ;Levy 1988 ; Sadowski 1988 ). The three studies that were judged to have high risk of reporting bias did not report key outcomes that would b e expected in such types of studies, such as catheter-related BSI, clinical sepsis or mortality.

Other potential sources of bias We screened for other potential sources of bias including extre me baseline imbalance, block randomisation of unblinded trials , unit of analysis issues and any evidence of fraud. As blinding was h ighly unlikely in most included studies, the use of block randomisat ion posed an additional risk of bias due to the possibility of dis rupting the integrity of the random sequence with educated guess on the likely allocation of the future participants ( Higgins 2011a ). Two studies ( Humar 2000 ;Vallés 2008 ) were judged to have high risk under ’other potential sources of bias’ as they used block rand omi- sation, and the authors did not state whether they used varyi ng block sizes in either trial.

Unit of analysis issues were a particular concern in three studi es ( Langgartner 2004 ;Prager 1984 ;Vallés 2008 ), in which the num- ber of catheters analysed exceeded the total number of particip ants.

This meant that some participants had multiple catheters anal ysed in the study as the authors of the three studies did not limit o ne catheter per participants in the analyses. The results might h ave been affected as the outcomes data from multiple catheters from the same participants were most likely not independent from e ach other. A more detailed description of the risk of bias of the tr ials is provided in ’ Assessment of risk of bias in included studies ’.

Effects of interventions 19 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. See: Summary of ndings for the main comparison Chlorhexidine compared to povidone-iodine in reducing cathet er related infections In this review, we assessed outcomes for a total of 3446 cathete rs in our meta-analysis of 12 studies. The total number of participa nts was unclear as some studies did not report this detail. Overal l, we carried out 10 comparisons, with variations related to the b ase solution in comparisons 4 and 6.

• Comparison 1: povidone-iodine (in aqueous solution) versus no skin antisepsis ( Prager 1984 ).

• Comparison 2: chlorhexidine (in aqueous solution) versus no skin antisepsis ( Tuominen 1981 ).

• Comparison 3: alcohol versus no skin antisepsis ( Sadowski 1988 ).

• Comparison 4: chlorhexidine versus povidone-iodine.

◦Chlorhexidine in aqueous solution versus povidone- iodine in aqueous solution ( Maki 1991 ;Vallés 2008 ).

◦ Chlorhexidine in alcohol versus povidone-iodine in aqueous solution ( Humar 2000 ;Vallés 2008 ).

◦ Chlorhexidine in alcohol versus povidone-iodine in alcohol ( Mimoz 2007 ).

• Comparison 5: chlorhexidine (in aqueous solution) versus alcohol ( Maki 1991 ).

• Comparison 6: povidone-iodine versus alcohol.

◦Povidone-iodine in aqueous solution versus alcohol ( Maki 1991 ).

◦ Povidone-iodine-impregnated adherent lm versus alcohol ( Levy 1988 ).

• Comparison 7: alcohol versus octenidine in alcohol ( Dettenkofer 2010 ).

• Comparison 8: chlorhexidine (in alcohol) plus povidone- iodine (in aqueous solution) versus chlorhexidine in alcohol ( Langgartner 2004 ).

• Comparison 9: chlorhexidine (in alcohol) plus povidone- iodine (in aqueous solution) versus povidone-iodine (in aqueo us solution) ( Langgartner 2004 ).

• Comparison 10: Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine 10% aqueous scrub ( Yousefshahi 2013 ).

Below, we report on our outcomes of interest in order of the com- parisons that examined them.

Primary outcomes Catheter-related BSI Comparison 1: aqueous povidone iodine versus no skin antisepsis (1 RCT, 179 catheters) Prager 1984 was the only study that compared povidone iodine in aqueous solution versus with no skin antisepsis (dry dress ing).

There was no clear evidence of a difference in the rate of catheter - related BSI (RR 0.99, 95% CI 0.37 to 2.61; 179 catheters; Analysis 1.1 ). The estimate is very uncertain as the comparison was un- derpowered to detect important differences in the outcome. The quality of evidence for this outcome was rated as very low due to very serious risk of bias issues (random sequence generatio n, allocation concealment, non-blinding of participants and unit o f analysis issue) as well as imprecision.

Comparisons 2: aqueous chlorhexidine versus no skin anti- sepsis and comparison 3: alcohol versus no skin antisepsis No study reported this outcome for these comparisons.

Comparison 4: chlorhexidine versus povidone-iodine (4 RCTs, 1436 catheters) Overall, chlorhexidine (any solution) was associated with a lo wer rate of catheter-related BSI than povidone-iodine (any solutio n) (absolute risk reduction (ARR) of 2.30%, 95% con dence interval (CI) 0.06% to 3.70%; risk ratio (RR) 0.64, 95% CI 0.41 to 0.99; NNTB 44, 95% CI 27 to 1563; four studies, 1436 catheters, I 2 = 0%; Analysis 4.1 ;Figure 4 ). This evidence was very low quality, downgraded for imprecision (one level) and risks of bias (two le v- els) in allocation concealment, blinding of participants and un it of analysis issues under “other sources of bias”. Analyses of sub- groups according to the base solution used showed no clear dif- ferences between chlorhexidine and povidone-iodine in the rat es of catheter-related BSI: chlorhexidine in aqueous solution ve rsus povidone-iodine in aqueous solution (RR 0.64, 95% CI 0.32 to 1.28, 2 studies, 452 catheters, I 2 = 15%), chlorhexidine in alcohol versus povidone-iodine in aqueous solution (RR 0.77, 95% CI 0.39 to 1.53; 2 studies, 503 catheters, I 2 = 0%), chlorhexidine in alcohol versus povidone-iodine in alcohol (RR 0.40, 95% CI 0.13 to 1.24; 1 study, 481 catheters). The small number of trials in each subgroup means that the comparisons were underpowered, and the results are uncertain. We considered the evidence from the data to be of very low overall quality (downgraded for impr e- cision (one level) and risks of bias (two levels) in allocation con - cealment, blinding of participants and unit of analysis issue s. We have highlighted the results for these outcomes from the over all comparison of chlorhexidine versus povidone-iodine as well as the three subgroup comparisons in our Summary of ndings for the main comparison . 20 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Figure 4. Forest plot of comparison: 1 Chlorhexidine versuspovidone-iodine, outcome: 1.1 Catheter- related BSI. For the outcome of catheter-related BSI per 1000 catheter-days, chlorhexidine was associated with an apparent lower BSI rate co m- pared with povidone-iodine (RR 0.53, 95% CI 0.30 to 0.94; 4 studies, 1450 catheters, I 2 = 0%; Analysis 4.2 ). Analyses of sub- groups according to the base solution used found evidence of a possible difference between chlorhexidine in alcohol versus p ovi- done-iodine in aqueous solution (RR 0.49, 95% CI 0.25 to 0.95; 3 studies, 661 catheters, I 2 = 31%), but relative effects were un- clear for the other base solutions in comparison (chlorhexidine in aqueous solution versus povidone-iodine in aqueous solutio n (RR 0.82, 95% CI 0.23 to 2.93; 1 study, 308 catheters), and chlorhex- idine in alcohol versus povidone-iodine in alcohol (RR 0.41, 95% CI 0.06 to 2.92; 1 study, 481 catheters). All subgroup compar- isons were underpowered and the overall quality of evidence f or this outcome was very low due to very serious risk of bias issue s (non-blinding of participants, incomplete outcome data and unit of analysis issues).

Comparison 5: aqueous chlorhexidine versus alcohol (1 RCT, 99 catheters) A single small study compared chlorhexidine in aqueous soluti on with alcohol ( Maki 1991 ) and found no clear difference in the absolute rate of catheter-related BSI between the alcohol-base d solution and the chlorhexidine-based solution (RR 0.24, 95% CI 0.02 to 2.54; 99 catheters; Analysis 5.1 ). The comparison was underpowered and the quality of evidence for this outcome was low due to risk of bias of the study (non-blinding) and imprecisi on.

Comparison 6: aqueous povidone-iodine versus alcohol (1 RCT, 109 catheters) Maki 1991 , the only study that compared povidone-iodine in aqueous solution with alcohol did not nd a clear difference in t he rate of catheter-related BSI between the two groups (RR 1.04, 95 % CI 0.24 to 5.08; 109 catheters; Analysis 6.1 ). The comparison was underpowered and the quality of evidence for this outcome was low due to risk of bias issue (non-blinding of the participants) and imprecision.

Comparison 7: alcohol versus octenidine in alcohol (1 RCT, 387 catheters) Dettenkofer 2010 was the only study to compare alcohol ver- sus octenidine in alcohol, and found no clear difference between groups in the absolute rate of catheter-related BSI (RR 2.01, 95 % CI 0.88 to 4.59; 387 catheters; Analysis 7.1 ) or catheter-related BSI per 1000 catheter-days (RR 2.18, 95% CI 0.54 to 8.77; 387 catheters; Analysis 7.2 ). The comparison was underpowered and the quality of evidence for both outcomes was low due to risk of bias issue (incomplete outcome data) and imprecision.

Septicaemia (whether or not CVC-related) Comparison 2: chlorhexidine versus no skin antisepsis (1 RCT , 136 participants) The only study that reported the outcome of septicaemia (irre- spective of its relationship with CVC) was Tuominen 1981 , which compared chlorhexidine with no skin antisepsis. This study of 136 participants compared the use of 0.05% chlorhexidine in aqueou s solution with no skin antisepsis and found no clear difference in the rate of septicaemia between the two groups, but the result was 21 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. inconclusive due to imprecision (RR 2.91, 95% CI 0.31 to 27.31; Analysis 2.1 ). The quality of evidence for this outcome was low due to risk of bias issue (non-blinding of participants) and imp re- cision, as stated above.

Mortality (all-cause or CVC-related) Comparison 4: chlorhexidine versus povidone-iodine (1 RCT, 329 participants analysed, 106 participants in povidone-io- dine group were included in both subgroup comparisons be- low) A single study ( Vallés 2008 ) reported mortality. The study divided the participants into three groups: chlorhexidine in aqueous so- lution, chlorhexidine in alcohol and povidone-iodine in aqueo ussolution. Analyses according to subgroups showed no clear dif- ferences in the rates of mortality between chlorhexidine in aq ue- ous solution and povidone-iodine in aqueous solution (RR 1.15 , 95% CI 0.72 to 1.83; 213 participants) ( Analysis 4.3 ), or between chlorhexidine in alcohol and povidone-iodine in aqueous solu- tion (RR 0.80, 95% CI 0.48 to 1.34; 222 participants) ( Analysis 4.3 )(Figure 5 ). However, the comparison was underpowered to detect important differences in the outcome, and the quality of evidence for both analyses was low due to a combination of risk o f bias issues and imprecision in the outcome estimates ( Summary of ndings for the main comparison ). Consequently true differences in the mortality associated with use of chlorhexidine or povid one iodine cannot be ruled out.

Figure 5. Forest plot of comparison: 1 Chlorhexidine versus povidone-iodine, outcome: 1.3 All-cause mortality. Secondary outcomes Catheter colonisation Comparison 1: aqueous povidone-iodine versus no skin antisepsis (1 RCT, 179 catheters) Based on Prager 1984 , the only study in this underpowered com- parison, it is unclear whether there is any difference in the ef fect on catheter colonisation of aqueous povidone iodine and no ski n antisepsis (RR 0.93, 95% CI 0.53 to 1.60; 179 catheters; Analysis 1.2 ). There was very low quality evidence due to serious risk of bias (random sequence generation, allocation concealment, non- blinding of participants and unit of analysis issue) and indi rectness of the outcome. Comparison 2: aqueous chlorhexidine versus no skin antisepsis (1 RCT, 124 catheters) Based on Tuominen 1981 , the only study to compare chlorhex- idine in aqueous solution with no skin antisepsis, there was no clear difference in the rate of catheter colonisation and theref ore uncertainty as to their relative effects remains (RR 1.26, 95% C I 0.61 to 2.59; 124 catheters; Analysis 2.2 ). The quality of evidence was very low due to risk of bias (non-blinding of participants), indirectness of the outcome and imprecise estimate from an un- derpowered analysis.

Comparison 3: alcohol versus no skin antisepsis (1 RCT, 50 catheters) Based on a single study in this underpowered analysis ( Sadowski 1988 ), it remains unclear whether there is a difference between cleansing the skin with alcohol and no skin antisepsis prior to catheter removal (RR 0.75, 95% CI 0.30 to 1.85; 50 catheters; 22 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 3.1 ). The quality of evidence was very low due to risk of bias (non-blinding of the participants), indirectness and impre ci- sion.

Comparison 4: chlorhexidine versus povidone-iodine (5 RCTs, 1533 catheters) Pooled analysis of ve studies that compared chlorhexidine wi th povidone iodine showed an overall reduction in the risk of cath eter colonisation with chlorhexidine (RR 0.68, 95% CI 0.56 to 0.84; ARR 8%, 95% CI 3 to 12%; NNTB 13, 95% CI 9 to 34; 5 stud- ies, 1533 catheters, I 2 = 55%; Analysis 4.4 ;Figure 6 ). Analysing subgroups according to the solution, there appeared to be redu c- tions in rates of catheter colonisation favouring chlorhexidi ne in the following comparisons:

Figure 6. Forest plot of comparison: 1 Chlorhexidine versus povidone-iodine, outcome: 1.4 Catheter colonisation. • Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution (RR 0.60, 95% CI 0.40 to 0.91; 2 studies, 442 catheters, I 2 = 56%).

• Chlorhexidine in alcohol versus povidone-iodine in alcohol (RR 0.52, 95% CI 0.34 to 0.80; 1 study, 481 catheters).

However, the rate of catheter colonisation between chlorhexid ine in alcohol versus povidone-iodine in aqueous solution appear ed to be similar (RR 0.86, 95% CI 0.64 to 1.14; 3 studies, 600 catheter s, I 2 = 58%).

There was moderate heterogeneity present for the overall po oledanalysis, as indicated by the I 2 of 55%. The extent of heterogeneity remained even with the studies separated into subgroups accor ding to the solution used, as shown above. We investigated other p ossi- ble sources of heterogeneity by exploring factors that were pr esent in the population, intervention, comparison, outcome de nit ions and risk of bias among the included studies. We noted that al- though there were some differences in the characteristics of the included studies in terms of population (surgical versus cardia c versus general ICUs) and intervention (different concentrati ons of chlorhexidine used, duration of catheterisation and the concur rent 23 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. use of other antiseptic substances alongside chlorhexidine-based solution), these differences did not plausibly explain the de gree of heterogeneity, as separating the studies into subgroups acco rding to these factors did not reduce the degree of heterogeneity.

However, we identi ed one plausible source of heterogeneity un- der the risk of bias criterion. We found that only two out of ve in- cluded studies ( Maki 1991 ;Mimoz 1996 ) had low risk of attrition bias, while the other three were at high risk of bias in this do main.

The two studies with low risk of attrition bias showed signi cant bene ts of chlorhexidine compared with povidone-iodine, whil st the remaining studies showed no signi cant difference betwee n the two groups. Grouping studies with low risk and high risk o f attrition bias separately reduced the I 2 statistic to 0% and 41%, respectively.

We undertook best- and worst-case scenarios to determine the im - pact of missing data from these three studies and found that th e overall pooled analysis was substantially altered, with th e best-case scenario moving the direction of the pooled estimate to signi - cantly and substantially favour the chlorhexidine group, and the worst-case scenario moving the pooled estimate to signi cantly favour the povidone-iodine group (see ’ Sensitivity analysis ’ for de- tails).

Having identi ed a plausible explanation for the observed h etero- geneity, we still decided to combine all ve studies under thre e different subgroups according to the type of solution used (eit her aqueous or alcohol). Taking all considerations, the overall qu ality of evidence for this outcome was very low, as there were very ser i- ous concerns regarding risk of bias (non-blinding of participant s, incomplete outcome data and unit of analysis issue), indirectne ss of the outcome and inconsistency among the study results.

Comparison 5: aqueous chlorhexidine versus alcohol (1 RCT, 99 catheters) According to a single study ( Maki 1991 ), it remains unclear whether there is a difference in the rates of catheter colonisat ion between chlorhexidine in aqueous solution and alcohol (RR 0.38 , 95% CI 0.11 to 1.33; 99 catheters; Analysis 5.2 ), but the compar- ison was underpowered. The quality of evidence for this outcom e was very low due to risk of bias (non-blinding of participants), indirectness and imprecision.

Comparison 6: aqueous povidone-iodine versus alcohol (3 RCTs, 169 catheters) It is unclear whether there is a difference in the rates of cathet er colonisation between patients who received CVC cleansing with povidone-iodine and those who receive cleansing with alcohol, either overall (RR 1.76, 95% CI 0.76 to 4.09; 2 studies, 169 catheters, I 2 = 43%), or in subgroups comparing povidone-iodine in aqueous solution versus alcohol (RR 1.25, 95% CI 0.49 to 3.14 ; 1 study, 109 catheters) or povidone-iodine-impregnated adher ent lm versus alcohol (RR 9.00, 95% CI 0.51 to 160.17; 1 study, 60 catheters; Analysis 6.2 ). The comparisons were underpowered, and the overall quality of evidence for this outcome was very lo w due to risk of bias (non-blinding of participants), indirectness of the outcome and imprecision.

Comparison 7: alcohol versus octenidine in alcohol (1 RCT, 322 catheters) Dettenkofer 2010 , the only study to compare alcohol versus octenidine in alcohol, showed that alcohol alone is probably as - sociated with a higher rate of catheter colonisation compared to octenidine (RR 2.26, 95% CI 1.22 to 4.21; 322 catheters; Analysis 7.3 ). However, there appeared to be no clear difference between the two groups in terms of catheter colonisation per 1000 cathet er- days (RR 2.23, 95% CI 0.79 to 6.29; 322 catheters; Analysis 7.4 ).

The quality of evidence for both outcomes was low, due to con- cerns in risk of bias (non-blinding of participants) and indirect ness of the outcomes.

Comparison 8: chlorhexidine in alcohol plus povidone- iodine in aqueous solution versus chlorhexidine in alcohol (1 RCT, 88 catheters) In an underpowered analysis from a single study ( Langgartner 2004 ), a combination of chlorhexidine plus povidone-iodine ap- peared to be associated with lower rate of catheter colonisatio n (RR 0.19, 95% CI 0.04 to 0.81; 88 catheters; Analysis 8.1 ) as well as catheter colonisation per 1000 catheter-days (RR 0.19, 95% CI 0.06 to 0.59; 88 catheters; Analysis 8.2 ) compared with chlorhex- idine alone, although the effects were uncertain due to the ver y low quality of evidence, which was reduced by risk of bias (non- blinding of participants, incomplete outcome data, unit of ana ly- sis issue), indirectness and imprecision.

Comparison 9: chlorhexidine in alcohol plus povidone- iodine in aqueous solution versus povidone-iodine in aqueous solution (1 RCT, 95 catheters) In another single-study, underpowered analysis based on Langgartner 2004 , there appeared to be lower rate of catheter colonisation (RR 0.15, 95% CI 0.04 to 0.62; 95 catheters; Analysis 9.1 ) as well as catheter colonisation per 1000 catheter-days (RR 0.17, 95% CI 0.05 to 0.52; 95 catheters; Analysis 9.2 ) using a com- bination of chlorhexidine and povidone-iodine compared with u s- ing povidone-iodine alone, but the effects were very uncertain due to the very low quality of evidence, which was reduced by risk of bias (non-blinding of participants, incomplete outcome data, un it of analysis issue), indirectness and imprecision. 24 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Comparison 10: Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine 10% aqueous scrub (1 RCT, 249 catheters) From the single study in this underpowered comparison ( Yousefshahi 2013 ), it is uncertain whether there is any clear differ- ence between the two groups in the rate of catheter colonisation (RR 1.08, 95% CI 0.68 to 1.72; 249 catheters; Analysis 10.1 ) due to the very low quality of evidence, which was reduced by risk of bias (random sequence generation, allocation concealment), indi - rectness and imprecision.

Insertion site infection Comparison 4: Chlorhexidine versus povidone-iodine (1 RCT, 242 catheters) Based on the result of a single study ( Humar 2000 ) in an under- powered analysis, it is uncertain whether there is any clear di ffer- ence between chlorhexidine (in alcohol) and povidone-iodine (in aqueous solution) with regard to insertion site infection, a s the quality of evidence was very low due to risk of bias (non-blind- ing of the participants, incomplete outcome data), indirectness and imprecision. The authors reported this outcome as the mean CFU count (MD −2.80, 95% CI −9.10 to 3.50; 242 catheters; Analysis 4.6 ).

Skin colonisation Comparison 7: Alcohol versus octenidine in alcohol (1 RCT, 365 catheters) Based on the results of Dettenkofer 2010 , using alcohol alone probably resulted in higher mean CFU compared with octeni- dine in alcohol (MD 79.00 CFUs, 95% CI 32.76 to 125.24; 365 catheters; Analysis 7.5 ). The quality of evidence was moderate as it was reduced by imprecision of the effect estimates from an un- derpowered analysis.

Adverse effects Comparison 7: Alcohol versus octenidine in alcohol (1 RCT, 398 participants) A single study, Dettenkofer 2010 , reported the rates of various ad- verse effects on the skin, the de nitions of which appeared to o ver- lap. For example, the authors reported “skin irritation”, “ burning”, “skin irritation and burning”, “itching”, “skin lesions”, “ burning and skin lesions”, “itching and skin irritation” as the outcom esunder adverse effects. To avoid duplication, we included only t he most commonly reported adverse effect, namely, skin irritati on.

For this outcome, there was moderate quality evidence showing no clear difference between in adverse effect rates between pat ients whose CVC sites were cleansed with alcohol and those who were cleansed with octenidine in alcohol (RR 0.85, 95% CI 0.60 to 1.20; 398 participants; Analysis 7.6 ). The quality of evidence was reduced by imprecision of the effect estimates from an underpow - ered analysis.

Number of patients who were on antibiotics during the period of catheter use Comparison 2: Chlorhexidine in aqueous solution versus no skin antisepsis ( 1 RCT, 136 participants) The only study that evaluated this outcome, Tuominen 1981 found no clear difference between the two groups with regard to the number of patients who required antibiotics during the pe riod of catheter use (RR 0.84, 95% CI 0.55 to 1.27; 136 participants; Analysis 2.3 ). The quality of evidence was low due to risk of bias (non-blinding of participants) and imprecision from an under- powered analysis.

Number of drug-resistant organisms from culture, length of hospitalisation, cost of care and quality of life No studies in any comparison assessed these outcomes.

Subgroup analyses Other than separating the subgroups according to the type of solution used in comparisons 4 and 6, we did not perform any additional subgroup analyses as speci ed in our ’ Methods ’ because the data in each study were not presented separately for vario us potential subgroups (for short term versus longer term CVCs, for paediatric versus adult patients and for patients in ICU ver sus those in other wards). Likewise, there was only a single study includ ed in many comparisons, and all studies in the meta-analysis used diagnostic criteria for catheter-related infections that were in line with our pre-speci ed criteria (see ’ Included studies ’).

Sensitivity analysis We performed sensitivity analyses on the most commonly asses sed outcomes, namely, catheter-related BSI (primary outcome) and catheter colonisation (secondary outcome) to evaluate the impact of excluding some studies based our prede ned criteria (unclear or no allocation concealment (selection bias) and signi cant dropou t rates (attrition bias)). We assessed the impact of missing data i n studies with high dropout rates using the best- and worst-case scenarios. We did not carry out a sensitivity analysis according to 25 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. the criterion of blinding because we considered that all but one study were at high risk in this domain.

Catheter-related BSI Comparison 4: Chlorhexidine versus povidone-iodine Selection bias: None of the four studies included were at low risk of bias for both random sequence generation and allocatio n concealment. Maki 1991 was at unclear risk for both items while the other three studies were at low risk for at least one of the items. We decided to perform the sensitivity analysis by exclu ding Maki 1991 . While the point estimate changed only slightly, the con dence interval expanded to cross the line of no effect, shift ing the result to become non-signi cant (before exclusion: RR 0.64, 95% CI 0.41 to 0.99; after exclusion: RR 0.68, 95% CI 0.43 to 1.08).

Attrition bias : Two studies had a high risk of attrition bias ( Humar 2000 ,Vallés 2008 ). We conducted best- and worst-case scenarios by assuming the outcome for the patients with missing data as described in the Dealing with missing data section. The direction of the pooled estimate differed markedly between the best- a nd worst-case scenarios as well as from the actual results reported , namely the ’completer analysis’ (best-case scenario: RR 0.35, 95 % CI 0.24 to 0.50, I 2 : 0%; worst-case scenario: RR 1.47, 95% CI 1.01 to 2.14, I 2 : 64%; actual results reported: RR 0.64, 95% CI 0.41 to 0.99, I 2 : 0%).

There was only a single study included for all the other compar- isons.

Catheter colonisation Comparison 4: Chlorhexidine versus povidone-iodine Selection bias : One study ( Maki 1991 ) had unclear risk of bias in both random sequence generation and allocation concealment, whilst the other studies were at low risk of bias for at least o ne item. The exclusion of Maki 1991 did not result in a substantial change in the pooled estimates (before exclusion: RR 0.68, 95% CI 0.56 to 0.84; after exclusion: RR 0.72, 95% CI 0.58 to 0.88).

Attrition bias : Three of the ve studies ( Humar 2000 ; Langgartner 2004 ;Vallés 2008 ) included in this comparison had high or unclear risk of attrition bias. We conducted best- and worst-case scenarios. With the best-case scenario, the pooled est i- mate showed substantial reduction in the risk of catheter colon i- sation favouring the chlorhexidine group (RR 0.56, 95% CI 0.47 to 0.68, I 2 : 73%), and with the worst-case scenario, there was no signi cant difference between the two groups (RR 0.90, 95% CI 0.74 to 1.09, I 2 : 72%). Results from both the best- and worst- case scenarios differed markedly with the actual results repor ted, namely the ’completer analysis’ (RR 0.68, 95% CI 0.56 to 0.84, I 2 : 55%).

There were insuf cient studies in all the other comparisons to enable a meaningful sensitivity analysis.

D I S C U S S I O N Summary of main results This review identi ed a wide variety of skin antisepsis regi mens that comprised different combinations of an active substance (su ch as chlorhexidine) and base solution (such as aqueous or alcoholi c solution). However, a limited number of studies (and sometime s just one) examined each regimen. Based on very limited eviden ce, there were no clear differences between various skin antiseps is reg- imens for our primary outcome of catheter-related BSI, althoug h for the overall comparison between chlorhexidine and povidon e- iodine, there appeared to be a reduction in catheter-related BS I associated with chlorhexidine. Notably, two studies conducted in the 1980s, one comparing povidone-iodine in aqueous solution with no skin antisepsis and the other comparing chlorhexidine in aqueous solution with no skin antisepsis, found no differen ce in the rates of BSI between the intervention group and the contro l group ( Prager 1984 ;Tuominen 1981 ). However, these were small studies with some methodological issues, and the evidence the y provide is very inconclusive.

Based on a single study ( Vallés 2008 ), there were similar rates of mortality between chlorhexidine-based solution and povid one- iodine based solution. However, the analyses were underpow ered for any clear conclusion to be drawn with regards to this outcome.

In the outcome of catheter colonisation, some differences exist ed between different skin antisepsis regimens, with regimens contain- ing chlorhexidine appearing to be more effective than regimen s containing povidone-iodine in reducing risk.

One trial showed that octenidine in alcohol appeared to be more ef- fective than alcohol alone in reducing catheter colonisation. Th ree separate studies that compared chlorhexidine, povidone-iodi ne and alcohol-based solution, respectively, with no skin antise psis did not nd any clear difference in the rates of catheter coloni- sation between the intervention group and the control group, al- though the amount of evidence based on these studies is insuf - cient to draw any clear conclusion. Analysis based on very small number of studies and catheters suggested that a combination o f chlorhexidine and povidone-iodine appeared to be more effecti ve than either agent alone in reducing catheter colonisation. Sin gle- study analyses showed that there were no clear differences in t he rates of insertion site infection, skin colonisation or adver se events between different skin antisepsis regimens examined. Over all, the results of this meta-analysis need to be interpreted with caut ion, as the majority of the included studies were not suf ciently powe red 26 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. to detect a clear difference in the outcomes, and some signi cant results came from small, methodologically awed studies, as men- tioned above.

Overall completeness and applicability of evidence We identi ed 13 studies that matched our selection criteria in terms of population, intervention, comparison and outcomes, and data were unavailable for analysis in 1 out of 13 studies. A to tal of 3446 catheters were assessed. The studies took place in Euro pe, the USA and Asia, from 1981 to 2013, in settings where CVCs are commonly used, such as the ICUs and haematology and oncology units. However, there are certain limitations in the complete ness of this review. For example, among the participants, children were grossly underrepresented, and most of the included studies d id not adequately asses some of the key prespeci ed outcomes of th is review, including primary BSIs, mortality, adverse effects a nd costs.

Furthermore, we were unable to undertake most of the subgrou p analyses because there were insuf cient data.

Quality of the evidence Overall, the quality the evidence for the majority of outcomes assessed was very low to moderate due to the small number of studies included in each comparison and variable risk of bias of the included studies. The strongest evidence comes from the overal l comparison between antiseptic solutions containing chlorhex idine against antiseptic solutions containing povidone-iodine, f or which there were ve studies. However, all comparisons in this revi ew suffered from a lack of power in the analysis, as evidenced by th e small number of trials and catheters in each comparison. The lack of power in the analysis has seriously affected our con dence in interpreting the results in general, as we were unable to det ermine whether non-statistically signi cant results were indicative of true (null) effects or of insuf cient data for detecting differences. Also, in the case of a statistically signi cant difference, an analysi s with a small number of trials and catheters lessens the reliabilit y of the results due to concerns about the effects of small studies in exacerbating the impact of biases ( Sterne 2011 ).

A second major limitation in the quality of the evidence gather ed was the risk of attrition bias, as four studies had high risks and three had unclear risks. In studies with high risk of attrition bias , the pooled results varied substantially between the best- and w orst-case scenarios and from the actual results reported, and this preclu ded us from drawing a rm conclusion on the results of the outcomes concerned. Besides, a lack of blinding of the participants in mos t studies, as well as the unit of analysis issues in some studie s in which multiple catheters in the same participants were analyse d as separate units has further affected the overall methodolo gical rigour of the included studies, and in turn the quality of evid ence.Overall, the body of evidence gathered in this review did not a llow us to reach a robust conclusion regarding the effectiveness and safety of various skin antisepsis regimes in reducing CVC-rel ated infections (see Summary of ndings for the main comparison for the outcome data under the major comparisons in this review).

Potential biases in the review process We performed a comprehensive search of multiple databases wit h independent screening, selection and assessment of eligible studies.

However, we were unable to obtain all relevant data; ve stud ies are awaiting assessment, as there were dif culties obtaining fu ll texts, and another one is an ongoing study. Many of the excluded studi es assessed a combination of arterial and venous catheters but di d not report outcome data separately for CVC, which prevented us from including a larger body of potentially relevant informa tion.

We are currently waiting for authors of the studies concerned to provide us relevant data for our future updates.

There are some unresolved unit of analysis issues in this rev iew:

for catheter-speci c outcomes such as catheter-related BSI and catheter colonisation, we reported the results in the same way as the original studies, using catheters rather than participan ts as the unit of analysis. As a result, the review included multiple cat heters in the same participants. Our failure to adjust for this unit o f analysis issue might have affected the results.

Agreements and disagreements with other studies or reviews The ndings of this review are broadly in line with two other r e- views on this topic ( Adams 2007 ;Chaiyakunapruk 2002 ), which concluded that antiseptic solutions containing chlorhexidine are more effective than those containing povidone-iodine in reduci ng catheter colonisation and catheter-related BSI. Of the two revi ews, Adams 2007 was a narrative review and Chaiyakunapruk 2002 was a systematic review that evaluated chlorhexidine against po vidone- iodine in all vascular catheters, including arterial and centra l and peripheral venous catheters. Chaiyakunapruk 2002 showed that antiseptic solutions containing chlorhexidine reduced cathet er-re- lated BSI on average by 49% compared with povidone-iodine, although there was a great degree of uncertainty on the magni- tude of its bene t, as re ected by a wide con dence interval (RR 0.51, CI 0.27 to 0.97; 8 RCTs including three that evaluated on ly CVCs, 4143 arterial and venous catheters combined including 1493 CVCs). Our review, which is focused only on central venous catheters, included two more trials but a slightly smaller num ber of CVCs, and we showed that a solution containing chlorhexidin e reduced catheter-related BSI by an average of 36% (relative ef- fect) compared with povidone-iodine (RR 0.64, CI 0.41 to 0.99; 5 studies, 1436 CVCs). 27 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. A U T H O R S ’ C O N C L U S I O N S Implications for practice Very low quality evidence suggests that antiseptic solutions containing chlorhexidine may reduce catheter colonisation and catheter-related BSI compared with antiseptic solutions conta in- ing povidone-iodine. It is unclear whether skin cleansing for C VCs with any solution is bene cial compared with no skin antisepsi s. It is still unclear whether skin antisepsis as part of CVC care red uces overall sepsis and mortality. While the evidence gathered in t his review does not change the current recommendations that favour the use of chlorhexidine-containing solution for skin antisep sis in CVC care, uncertainties remain on its value in improving patie nt mortality and morbidity.

Implications for research Further trials in skin antisepsis in patients with a CVC are w ar- ranted. This review highlights the paucity of high-quality re search answering questions on whether skin antisepsis in patients with CVC reduces overall rates of sepsis and mortality. Furthermo re, the evolving patterns of hospital-associated infections, accom pa- nying progress in infection control measures and microbiologi - cal diagnostic techniques have resulted in changing effectiven ess of various interventions employed. Future trials should in cludethe two key outcomes, overall rate of sepsis and mortality, al ong- side catheter-speci c outcomes such as catheter-related BSI and catheter colonisation, with a clear description of the settings , par- ticipants and concurrent infection control measures to enable an evaluation of the results in relation to these factors. If pos sible, investigators should blind participants and personnel, or a t the very least outcome assessors, with measures in place (such as tra in- ing of care personnel on handling the catheters for the purpose of research and the implementation of a standard protocol with regards to the handling of study catheters during and after of ce hours) to reduce loss of data.

A C K N O W L E D G E M E N T S We are grateful for the contribution of peer reviewers Joan We bster, Susan O’Meara, Judith Tanner, Ankur Barua, Clifford Richard - son, Mark Rodgers, Jane Nadel, Marian Brady, Gill Worthy, Vi c- toria Steelman and Dayanithee Chetty for kindly spending ti me to comment on our draft protocol and review and suggesting im- provements. We thank Meggan Harris for copyediting the revie w.

We also acknowledge the contribution by Dr Rachel Wel Lynn Ooi in assisting Nai Ming Lai in screening through the search results from CENTRAL to identify potentially relevant arti cles.

R E F E R E N C E S References to studies included in this review Dettenkofer 2010 {published data only} Dettenkofer M, Wilson C, Gratwohl A, Schmoor C, Bertz H, Frei R, et al. Skin disinfection with octenidine dihydrochloride for central venous catheter site care:

a double-blind, randomized, controlled trial. Clinical Microbiology and Infection 2010; Vol. 6, issue 16:600–6.

Humar 2000 {published data only} Humar A, Ostromecki A, Direnfeld J, Marshall JC, Lazar N, Houston PC, et al. Prospective randomized trial of 10% povidone-iodine versus 0.5% tincture of chlorhexidine as cutaneous antisepsis for prevention of central venous catheter infection. Clinical Infectious Diseases 2000; Vol.

31, issue 4:1001–7.

Langgartner 2004 {published data only} Langgartner J, Linde HJ, Lehn N, Reng M, Schölmerich J, Glück T. Combined skin disinfection with chlorhexidine/ propanol and aqueous povidone-iodine reduces bacterial colonisation of central venous catheters. Intensive Care Medicine 2004; Vol. 30, issue 6:1081–8.

Levy 1988 {published data only} Levy JH, Nagle DM, Curling PE, Waller JL, Kopel M, Tobia V. Contamination reduction during central venous catheterization. Critical Care Medicine 1988; Vol. 16, issue 2:165–7.

Maki 1991 {published data only} ∗ Maki DG, Ringer M, Alvarado CJ. Prospective randomised trial of povidone-iodine, alcohol, and chlorhexidine for prevention of infection associated with central venous and arterial catheters. The Lancet 1991; Vol. 338, issue 8763:

339–43.

Maki DG, Ringer M, Alvarado CJ. Prospective randomized trial of povidone-iodine, alcohol, and chlorhexidine for prevention of infection associated with central venous and arterial catheters. CINA: Of cial Journal of the Canadian Intravenous Nurses Association 1993;9(1):10–15.

Mimoz 1996 {published data only} ∗ Mimoz O, Pieroni L, Lawrence C, Edouard A, Costa Y, Samii K, et al. Prospective, randomized trial of two antiseptic solutions for prevention of central venous or arterial catheter colonization and infection in intensive care unit patients. Critical Care Medicine 1996;24(11):1818–23.

Mimoz O, Pieroni L, Lawrence C, Edouard A, Samii K.

Prospective trial of povidone-iodine (PI) and chlorhexidine (CH) for prevention of catheter-related sepsis (CRS).

Proceedings of the 34th Interscience Conference on Antimicrobial Agents and Chemotherapy; 1994 Oct 4-7; Orlando (FL). 1994. 28 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 2007{published data only} Mimoz O, Villeminey S, Ragot S, Dahyot-Fizelier C, Laksiri L, Petitpas F, et al. Chlorhexidine-based antiseptic solutio n vs alcohol-based povidone-iodine for central venous catheter care. Archives of Internal Medicine 2007; Vol. 167, issue 19:2066–72.

Prager 1984 {published data only} Prager RL, Silva J. Colonization of central venous catheters.

Southern Medical Journal 1984; Vol. 77, issue 4:458–61.

Sadowski 1988 {published data only} Sadowski DA, Harrell DA, Maley MP, Warden GD. The value of culturing central-line catheter tips in burn patients.

The Journal of Burn Care and Rehabilitation 1988;9(1):

66–8.

Tuominen 1981 {published data only} Tuominen M, Valtonen VV, Nikki P. The effect of local antiseptic, chlorhexidine, in preventing infection from central venous catheterization: a clinical study. Annals of Clinical Research 1981; Vol. 13, issue 6:425–8.

Vallés 2008 {published data only} Vallés J, Fernández I, Alcaraz D, Chacón E, Cazorla A, Canals M, et al. Prospective randomized trial of 3 antiseptic solutions for prevention of catheter colonization in an intensive care unit for adult patients. Infection Control and Hospital Epidemiology 2008; Vol. 29, issue 9:847–53.

Yasuda 2013 {published data only} ∗ Yasuda H, Sanui M, Fujitani S. Comparison of three cutaneous antiseptic solutions for the prevention of catheter colonization. Proceedings of the 43rd Critical Care Congress of the Society of Critical Care Medicine, SCCM 2014; 2014 Jan 9-13; San Francisco. Critical Care Medicine 2013; 41(12 Suppl 1):A195.

Yasuda H, Sanui M, Komuro T, Hatakeyama J, Matsukubo S, Kawano S, et al. Comparison of three cutaneous antiseptic solutions for the prevention of catheter colonization in an ICU for adult patients: A multicenter prospective randomized controlled trial. Critical Care2015; 19 (Suppl 1):73. [DOI: 10.1186/cc14153] Yousefshahi 2013 {published data only} Yousefshahi F, Azimpour K, Boroumand MA, Naja M, Barkhordari K, Vaezi M, et al. Can a new antiseptic agent reduce the bacterial colonization rate of central venous lines in post-cardiac surgery patients?. Journal of Tehran University Heart Center 2013;8(2):70–5.

References to studies excluded from this review Almeida 2009 {published data only} Almeida M, Ferreira A, Reis P, Alves V, Dias C, Granja C.

Reducing catheter-related bloodstream infections (CRBSI) in the ICU with an evidence-based intervention. Intensive Care Medicine 2009;35(Suppl 1):S271.

Apisarnthanarak 2010 {published data only} Apisarnthanarak A, Thongphubeth K, Yuekyen C, Warren DK, Fraser VJ. Effectiveness of a catheter-associated bloodstream infection bundle in a Thai tertiary care center: a 3-year study.

American Journal of Infection Control 2010; 38 (6):449–55.

Assadian 2004 {published data only} Assadian O. Skin antiseptic in reducing the risk of central venous catheter-related infections. Critical Care Medicine 2004; 32(3):887–8.

Astle 2005 {published data only} Astle CM, Jensen L. A trial of ExSept for hemodialysis central venous catheters. Nephrology Nursing Journal 2005; 32 (5):517–25.

Balamongkhon 2007 {published data only} Balamongkhon B, Thamlikitkul V. Implementation of chlorhexidine gluconate for central venous catheter site care at Siriraj Hospital, Bangkok, Thailand. American Journal of Infection Control 2007;35(9):585–8.

Bilir 2009 {published data only} Bilir A, Yelken B, Erkan A. Chlorhexidine, octenidine or povidone iodine for catheter related infections: a randomised controlled trial. Critical Care 2009; Vol. 13, issue Suppl 1:S79.

Borghesi 2011 {published data only} Borghesi A, Tzialla C, Decembrino L, Manzoni P, Stronati M. New possibilities of prevention of infection in the newborn. Journal of Maternal-Fetal and Neonatal Medicine 2011; 24(Suppl 2):28–30.

Bowling 2010 {published data only} Bowling G, Leykum L. Decreasing the rate of contaminated blood cultures. Proceedings of 2010 Annual Meeting of the Society of Hospital Medicine; 2010 Apr 8-11; Washington, DC. Journal of Hospital Medicine 2010;5(Suppl 1):85.

Camins 2010 {published data only} Camins BC, Richmond AM, Dyer KL, Zimmerman HN, Coyne DW, Rothstein M, et al. A crossover intervention trial evaluating the ef cacy of a chlorhexidine-impregnated sponge in reducing catheter-related bloodstream infections among patients undergoing hemodialysis. Infection Control and Hospital Epidemiology 2010;31(11):1118–23.

Carrer 2005 {published data only} Carrer S, Bocchi A, Bortolotti M, Braga N, Gilli G, Candini M, et al. Effect of different sterile barrier precautions and central venous catheter dressing on the skin colonization around the insertion site. Minerva Anestesiologica2005;71 (5):197–206.

Casey 2003 {published data only} Casey AL, Worthington T, Lambert PA, Quinn D, Faroqui MH, Elliott TS. A randomized, prospective clinical trial to assess the potential infection risk associated with the PosiFlow needleless connector. Journal of Hospital Infection 2003; Vol. 54, issue 4:288–93.

Casey 2007 {published data only} Casey AL, Burnell S, Whinn H, Worthington T, Faroqui MH, Elliott TS. A prospective clinical trial to evaluate the microbial barrier of a needleless connector. The Journal of Hospital Infection 2007;65(3):212–8. 29 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Cepkova 2006{published data only} Cepkova M, Matthay MA. Reducing risk in the ICU:

vascular catheter-related infections. Infections in Medicine 2006; 23(4):141–52.

Chaiyakunapruk 2003 {published data only} Chaiyakunapruk N, Veenstra DL, Lipsky BA, Sullivan SD, Saint S. Vascular catheter site care: the clinical and economic bene ts of chlorhexidine gluconate compared with povidone iodine. Clinical Infectious Diseases 2003;37 (6):764–71.

Crawford 2004 {published data only} Crawford AG, Fuhr JP, Rao B. Cost-bene t analysis of chlorhexidine gluconate dressing in the prevention of catheter-related bloodstream infections. Infection Control and Hospital Epidemiology 2004;25(8):668–74.

Daghistani 1996 {published data only} Daghistani D, Horn M, Rodriguez Z, Schoenike S, Toledano S. Prevention of indwelling central venous catheter sepsis. Medical and Pediatric Oncology 1996;26(6):405–8.

Darouiche 2007 {published data only} Darouiche RO. Preventing catheter-related infectious complications. Journal of Supportive Oncology 2007;5(2):

70–1.

Darouiche 2008 {published data only} Darouiche RO. Prevention of infections associated with vascular catheters. International Journal of Arti cial Organs 2008; 31(9):810–9.

Dean 2011 {published data only} Dean R, Dillworth J, Phillips M. Assessment of daily bathing protocols: a comparison of chlorhexidine solution and chlorhexidine impregnated cloths. Proceedings of Critical Care Congress; 2012 Feb 4-8; Houston, TX.

Critical Care Medicine 2011;39(Suppl 12):142.

Dettenkofer 2002 {published data only} Dettenkofer M, Jonas D, Wiechmann C, Rossner R, Frank U, Zentner J, et al. Effect of skin disinfection with octenidine dihydrochloride on insertion site colonization of intravascular catheters. Infection 2002; Vol. 30, issue 5:

282–5.

Eggimann 2010 {published data only} Eggimann P, Joseph C, Thevenin MJ, Voirol P, Bellini C, Pagani JL, et al. Impact of chlorhexidine-impregnated sponges on catheter-related infections rate. Intensive Care Medicine 2010;36(Suppl 2):S128.

Eyberg 2008 {published data only} Eyberg CI, Pyrek J. A controlled randomized prospective comparative pilot study to evaluate the ease of use of a transparent chlorhexidine gluconate gel dressing versus a chlorhexidine gluconate disk in healthy volunteers. The Journal of the Association for Vascular Access 2008;13(3):

112–7.

Freiberger 1992 {published data only} Freiberger D, Bryant J, Marino B. The effects of different central venous line dressing changes on bacterial growth in a pediatric oncology population. Journal of Pediatric Oncology Nursing 1992; Vol. 9, issue 3:3–7. Fukunaga 2004 {published data only} Fukunaga A, Naritaka H, Fukaya R, Tabuse M, Nakamura T. Povidone-iodine ointment and gauze dressings associated with reduced catheter-related infection in seriously ill neurosurgical patients. Infection Control and Hospital Epidemiology 2004; Vol. 25, issue 8:696–8.

Garcia 2010 {published data only} Garcia J, Island E, McLaughlin G, Langshaw A, Socarra M, Lalanne J, et al. Prevention of central venous catheter infections using chlorhexidine: Scrub the Hub campaign in transplant patients. American Journal of Transplantation 2010; 10(2):422–3.

Garcia-Teresa 2007 {published data only} Garcia-Teresa MA, Casado-Flores J, Delgado-Dominguez MA, Roqueta-Mas J, Cambra-Lasaosa F, Concha-Torre A, et al. Infectious complications of percutaneous central venous catheterization in pediatric patients: a Spanish multicenter study. Intensive Care Medicine 2007;33(3):466–76.

Garcia-Vazquez 2011 {published data only} Garcia-Vazquez E, Murcia-Paya J, Canteras M, Gomez J.

In uence of a hygiene promotion programme on infection control in an intensive-care unit. Clinical Microbiology and Infection 2011;17(6):894–900.

Garland 1996 {published data only} Garland JS, Alex CP, Mueller CD, Cisler-Kahill LA.

Local reactions to a chlorhexidine gluconate-impregnated antimicrobial dressing in very low birth weight infants. The Pediatric Infectious Disease Journal 1996;15(10):912–4.

Garland 2001 {published data only} Garland JS, Alex CP, Mueller CD, Otten D, Shivpuri C, Harris MC, et al. A randomized trial comparing povidone- iodine to a chlorhexidine gluconate-impregnated dressing for prevention of central venous catheter infections in neonates. Pediatrics2001;107(6):1431–6.

Garland 2009a {published data only} Garland JS, Alex CP, Uhing MR, Peterside IE, Rentz A, Harris MC. Pilot trial to compare tolerance of chlorhexidine gluconate to povidone-iodine antisepsis for central venous catheter placement in neonates. Journal of Perinatology 2009; 29(12):808–13.

Garland 2009b {published data only} ∗ Garland JS, Harris MC, Uhing MR, Alex CP, Peterside I, Rentz A. Randomized pilot trial to assess safety of 2% chlorhexidine gluconate (CHG) in neonates with percutaneously placed central venous catheters (PICC).

Proceedings of the Pediatric Academic Societies Annual Meeting; 2009 May 2-5; Baltimore, MD. 2009.

Jeffery S, Garland JS, Harris MC, Uhing MR, Alex CP, Peterside I, Rentz A. Randomized pilot trial to assess safety of 2% chlorhexidine gluconate (CHG) in neonates with percutaneously placed central venous catheters (PICC).

Proceedings of the Pediatric Academic Societies Annual Meeting; 2009 May 2-5; Baltimore, MD. Baltimore:

Pediatric Academic Societies, 2009. 30 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Gilad 2006{published data only} Gilad J, Borer A. Prevention of catheter-related bloodstream infections in the neonatal intensive care setting. Expert Review of Anti-Infective Therapy 2006;4(5):861–73.

Girard 2012 {published data only} Girard R, Comby C, Jacques D. Alcoholic povidone- iodine or chlorhexidine-based antiseptic for the prevention of central venous catheter-related infections: in-use comparison. Journal of Infection and Public Health 2012;5 (1):35–42.

Gnass 2004 {published data only} Gnass SA, Barboza L, Bilicich D, Angeloro P, Treiyer W, Grenovero S, et al. Prevention of central venous catheter- related bloodstream infections using non-technologic strategies. Infection Control and Hospital Epidemiology 2004; 25 (8):675–7.

Gunst 2011 {published data only} Gunst M, Matsushima K, Vanek S, Gunst R, Sha S, Frankel H. Peripherally inserted central catheters may lower the incidence of catheter-related blood stream infections in patients in surgical intensive care units. Surgical Infections 2011; 12(4):279–82.

Habibzadeh 2013 {published data only} Habibzadeh F, Yousefshahi F, Rouhipour N. Unethical conduct of underpowered clinical trials. Journal of Tehran University Heart Center 2013;8(4):213–4.

Hachem 2002 {published data only} Hachem R, Raad I. Prevention and management of long- term catheter related infections in cancer patients. Cancer Investigation 2002;20(7-8):1105–13.

Halpin 1991 {published data only} Halpin DP, O’Byrne P, McEntee G, Hennessy TP, Stephens RB. Effect of a betadine connection shield on central venous catheter sepsis. Nutrition1991;7(1):33–4.

Hanazaki 1999 {published data only} Hanazaki K, Shingu K, Adachi W, Miyazaki T, Amano J. Chlorhexidine dressing for reduction in microbial colonization of the skin with central venous catheters: a prospective randomized controlled trial. The Journal of Hospital Infection 1999;42(2):165–8.

Hill 1990 {published data only} Hill RL, Fisher AP, Ware RJ, Wilson S, Casewell MW.

Mupirocin for the reduction of colonization of internal jugular cannulae: a randomized controlled trial. The Journal of Hospital Infection 1990;15(4):311–21.

Huang 2006 {published data only} Huang SS, Yokoe DS, Hinrichsen VL, Spurchise LS, Datta R, Miroshnik I, et al. Impact of routine intensive care unit surveillance cultures and resultant barrier precautions on hospital-wide methicillin-resistant Staphylococcus aureus bacteremia. Clinical Infectious Diseases 2006;43(8):971–8.

Hutchinson 1990 {published data only} Hutchinson SK, Waskerwitz M, Martin K, Faubion W, Revesz S. Nonocclusive, clean permanent right atrial catheter dressing change procedures compared with occlusive, sterile permanent right atrial catheter dressing change procedures in children with cancer.

Journal of Pediatric Oncology Nursing 1990;7(2):71.

Ishikawa 2010 {published data only} Ishikawa Y, Kiyama T, Haga Y, Ishikawa M, Takeuchi H, Kimura O, et al. Maximal sterile barrier precautions do not reduce catheter-related bloodstream infections in general surgery units: a multi-institutional randomized controlled trial. Annals of Surgery 2010;251(4):620–3.

Ishizuka 2009 {published data only} Ishizuka M, Nagata H, Takagi K, Kubota K. Comparison of 0.05% chlorhexidine and 10% povidone-iodine as cutaneous disinfectant for prevention of central venous catheter-related bloodstream infection: a comparative stud y.

European Surgical Research 2009;43(3):286–90.

Johnson 2005 {published data only} Johnson DW, Van Eps C, Mudge DW, Wiggins KJ, Armstrong K, Hawley CM, et al. Randomized, controlled trial of topical exit-site application of honey (Medihoney) versus mupirocin for the prevention of catheter-associated infections in hemodialysis patients. Journal of the American Society of Nephrology 2005;16(5):1456–62.

Khattak 2010 {published data only} Khattak AZ, Ross R, Ngo T, Shoemaker CT. A randomized controlled evaluation of absorption of silver with the use of silver alginate (Algidex) patches in very low birth weight (VLBW) infants with central lines. Journal of Perinatology 2010; 30(5):337–42.

Khouli 2009 {published data only} Khouli HI, Jahnes K, Mathew J, Gohil A, Shapiro J, Rose K, et al. Medical residents’ performance in maximum barrier precautions during central venous catheter placement:

effect of simulation-based training. Chest2009;136(4):12S.

Krein 2007 {published data only} Krein SL, Hofer TP, Kowalski CP, Olmsted RN, Kauffman CA, Forman JH, et al. Use of central venous catheter- related bloodstream infection prevention practices by US hospitals. Mayo Clinic Proceedings 2007;82(6):672–8.

Kruse 1999 {published data only} Kruse JA. Chlorhexidina gluconate solution prevented catheter colonization and infection [La solucion de gluconato de clorhexidina previene la colonizacion e infeccion del cateter]. Enfermedades Infecciosas y Microbiologia Clinica 1999;19(1):37–8.

Kulkarni 2013 {published data only} Kulkarni AP, Awode M. A prospective randomised trial to compare the ef cacy of povidone-iodine 10% and chlorhexidine 2% for skin disinfection. Indian Journal of Anaesthesia 2013;57(3):270–5.

Lange 1997 {published data only} Lange BJ, Weiman M, Feuer EJ, Jakobowski D, Bilodeau J, Stallings VA, et al. Impact of changes in catheter management of infectious complications among children with central venous catheters. Infection Control and Hospital Epidemiology 1997;18(5):326–32. 31 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Le Corre 2003{published data only} Le Corre I, Delorme M, Cournoyer S. A prospective, randomized trial comparing a transparent dressing and a dry gauze on the exit site of long term central venous catheters of hemodialysis patients. Journal of Vascular Access2003;4 (2):56–61.

Legras 1997 {published data only} Legras A, Cattier B, Dequin PF, Boulain T, Perrotin D.

Prospective randomised trial for prevention of vascular- catheter infection: Alcohol chlorhexidine versus povidone- iodine. [Etude prospective randomisée pour la prévention des infections liées aux cathéters: chlorhexidine alcoolique contre polyvidone iodée]. Réanimation Urgences 1997; Vol. 6, issue 1:5–11.

Levy 2005 {published data only} Levy I, Katz J, Solter E, Samra Z, Vidne B, Birk E, et al. Chlorhexidine-impregnated dressing for prevention of colonization of central venous catheters in infants and children: a randomized controlled study. The Pediatric Infectious Disease Journal 2005;24(8):676–9.

Madeo 1998 {published data only} Madeo M, Martin CR, Turner C, Kirkby V, Thompson DR. A randomized trial comparing Arglaes (a transparent dressing containing silver ions) to Tegaderm (a transparent polyurethane dressing) for dressing peripheral arterial catheters and central vascular catheters. Intensive and Critical Care Nursing 1998;14(4):187–91.

Mahieu 2001 {published data only} Mahieu LM, De Dooy JJ, Lenaerts AE, Leven MM, De Muynck AO. Catheter manipulations and the risk of catheter-associated bloodstream infection in neonatal intensive care unit patients. Journal of Hospital Infection 2001; 48(1):20–6.

Maki 1981 {published data only} Maki DG, Band JD. A comparative study of polyantibiotic and iodophor ointments in prevention of vascular catheter- related infection. The American Journal of Medicine 1981;3:

739–44.

Maki 1992 {published data only} Maki D. Choosing a disinfectant for central catheters.

Nursing Times 1992;12:54–5.

McCann 2016 {published data only} McCann M, Fitzpatrick F, Mellotte G, Clarke M. Is 2% chlorhexidine gluconate in 70% isopropyl alcohol more effective at preventing central venous catheter- related infections than routinely used chlorhexidine gluconate solutions: a pilot multicenter randomized trial (ISRCTN2657745)?. American Journal of Infection Control 2016:Epub ahead of print.

Montecalvo 2012 {published data only} Montecalvo MA, McKenna D, Yarrish R, Mack L, Maguire G, Haas J, et al. Chlorhexidine bathing to reduce central venous catheter-associated bloodstream infection: impact and sustainability. American Journal of Medicine 2012;125 (5):505–11. Munoz-Price 2009 {published data only} Munoz-Price LS, Hota B, Stemer A, Weinstein RA.

Prevention of bloodstream infections by use of daily chlorhexidine baths for patients at a long-term acute care hospital. Infection Control and Hospital Epidemiology 2009; 30 (11):1031–5.

Munoz-Price 2012 {published data only} Munoz-Price LS, Dezfulian C, Wyckoff M, Lenchus JD, Rosalsky M, Birnbach DJ, et al. Effectiveness of stepwise interventions targeted to decrease central catheter-associat ed bloodstream infections. Critical Care Medicine2012;40(5):

1464–9.

Nikoletti 1999 {published data only} Nikoletti S, Leslie G, Gandossi S, Coombs G, Wilson R.

A prospective, randomized, controlled trial comparing transparent polyurethane and hydrocolloid dressings for central venous catheters. American Journal of Infection Control 1999;27(6):488–96.

Noto 2014 {published data only} Noto M, Domenico H, Talbot T, Byrne D, Wheeler A.

Healthcare-associated infections and chlorhexidine bathing:

a pragmatic cluster-randomized trial. Proceedings of the 2015 Critical Care Congress 2015; 2015 Jan 17-21; Phoenix, AZ. Critical Care Medicine 2014;42(12 Suppl 1):

A1478–9.

Parienti 2004 {published data only} Parienti JJ, Du Cheyron D, Ramakers M, Malbruny B, Leclercq R, Le Coutour X, et al. Alcoholic povidone- iodine to prevent central venous catheter colonization: A randomized unit-crossover study. Critical Care Medicine 2004; 32(3):708–13.

Peterson 2011 {published data only} Peterson K. Central venous catheter injection cap disinfection: chlorhexidine versus 70% alcohol. Journal of Pediatric Nursing 2011;26(2):e6.

Raad 1994 {published data only} Raad II, Hohn DC, Gilbreath BJ, Suleiman N, Hill LA, Bruso PA, et al. Prevention of central venous catheter-related infections by using maximal sterile barrier precautions during insertion. Infection Control and Hospital Epidemiology 1994;15(4 Pt 1):231–8.

Render 2006 {published data only} Render ML, Brungs S, Kotagal U, Nicholson M, Burns P, Ellis D, et al. Evidence-based practice to reduce central line infections. Joint Commission Journal on Quality and Patient Safety 2006;32(5):253–60.

Rezaei 2009 {published data only} Rezaei J, Esfandiari Kh, Tavakoli H, Sadooghi M, Hasibi M, Behzadi M. Evaluation of mupirocin ointment in control of central venous catheter related infections: a randomized clinical trial. Tehran University Medical Journal 2009;67(6):

428–34.

Richardson 2006 {published data only} Richardson D. Literature review [Alcoholic povidone–iodine to prevent central venous catheter colonization: a 32 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. randomized unit–crossover study].Journal of the Association for Vascular Access 2006;11(2):76.

Rickard 2004 {published data only} Rickard CM, Lipman J, Courtney M, Siversen R, Daley P. Routine changing of intravenous administration sets does not reduce colonization or infection in central venous catheters. Infection Control and Hospital Epidemiology 2004; 25 (8):650–5.

Rijnders 2003 {published data only} Rijnders BJ, Van Wijngaerden E, Wilmer A, Peetermans WE. Use of full sterile barrier precautions during insertion of arterial catheters: a randomized trial. Clinical Infectious Diseases 2003;36(6):743–8.

Rubinson 2004 {published data only} Rubinson L, Diette GB. Best practices for insertion of central venous catheters in intensive-care units to prevent catheter-related bloodstream infections. Journal of Laboratory and Clinical Medicine 2004;143(1):5–13.

Rupp 2008 {published data only} Rupp ME, Fitzgerald T, Puumala S, Anderson JR, Craig R, Iwen PC, et al. Prospective, controlled, cross-over trial of alcohol-based hand gel in critical care units. Infection Control and Hospital Epidemiology 2008;29(1):8–15.

Ruschulte 2009 {published data only} Ruschulte H, Franke M, Gastmeier P, Zenz S, Mahr KH, Buchholz S, et al. Prevention of central venous catheter related infections with chlorhexidine gluconate impregnated wound dressings: a randomized controlled trial. Annals of Hematology 2009; Vol. 88, issue 3:267–72.

Schwebel 2012 {published data only} Schwebel C, Lucet JC, Vesin A, Arrault X, Calvino- Gunther S, Bouadma L, et al. Economic evaluation of chlorhexidine-impregnated sponges for preventing catheter- related infections in critically ill adults in the Dressing Study. Critical Care Medicine 2012;40(1):11–17.

Sheehan 1993 {published data only} Sheehan G, Leicht K, O’Brien M, Taylor G, Rennie R. Chlorhexidine versus povidone-iodine as cutaneous antisepsis for prevention of vascular-catheter infection.

Proceedings of the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy; 1993 Oct 17-20; New Orleans, LA. 1993:a1616.

Spiegler 2010 {published data only} Spiegler P. Comparing central line and arterial line infections. Clinical Pulmonary Medicine 2010;17(5):248–9.

Swan 2014 {published data only} Swan J, Bui L, Pham V, Shirkey B, Graviss E, Hai S, et al. RCT Of chlorhexidine vs. soap & water bathing for prevention of hospital-acquired infections in SICU.

Proceedings of the 2015 Critical Care Congress; 2015 Jan 17-21; Phoenix, AZ. Critical Care Medicine 2014;42(12 Suppl 1):A1369–70.

Tietz 2005 {published data only} Tietz A, Frei R, Dangel M, Bolliger D, Passweg JR, Gratwohl A, et al. Octenidine hydrochloride for the care of central venous catheter insertion sites in severely immunocompromised patients.

Infection Control and Hospital Epidemiology 2005;26(8):703–7.

Van Esch 2002 {published data only} Van Esch J. Chlorhexidine reduced catheter tip colonisation more than 10% povidone-iodine in critically ill neonates.

Evidence-Based Nursing 2002;5(3):73.

Zingg 2008 {published data only} Zingg W, Cartier-Fassler V, Walder B. Central venous catheter-associated infections. Best Practice and Research:

Clinical Anaesthesiology 2008;22(3):407–21.

Zingg 2009 {published data only} Zingg W, Imhof A, Maggiorini M, Stocker R, Keller E, Ruef C. Impact of a prevention strategy targeting hand hygiene and catheter care on the incidence of catheter- related bloodstream infections. Critical Care Medicine2009; 37 (7):2167–73.

References to studies awaiting assessment Atahan 2012 {published data only} Atahan K, Cokmez A, Bekoglu M, Durak E, Tavusbay C, Tarcan E. The effect of antiseptic solution in central venous catheter care. Bratislavske Lekarske Listy 2012;113 (9):548–51.

BIlir 2013 {published data only} Bilir A, Yelken B, Erkan A. Chlorhexidine, octenidine or povidone iodine for catheter related infections: a randomised controlled trial. Journal of Research in Medical Sciences 2013;18(6):510–2.

Giles 2002 {published data only} Giles Y, Aksoy M, Tezelman S. What really affects the incidence of central venous catheter-related infections for short-term catheterization?. Acta Chirurgica Belgica2002; 102 (4):256–8.

Knasinski 2000 {unpublished data only} Knasinski V, Maki DG. A prospective, randomized, controlled trial of 1% chlorhexidine 75% alcohol vs. 10% povidone iodine for cutaneous disinfection and follow- up site care with central venous and arterial catheters.

Proceedings of the National Association of Vascular Access Network Conference; 2000 September 6-10; San Diego, CA. San Diego, 2000.

Mimoz 2015 {published data only} Mimoz O, Lucet JC, Kerforne T, Pascal J, Souweine B, Goudet V, et al. Skin antisepsis with chlorhexidine- alcohol versus povidone iodine-alcohol, with and without skin scrubbing, for prevention of intravascular-catheter- related infection (CLEAN): an open-label, multicentre, randomised, controlled, two-by-two factorial trial.

Lancet 2015;386(10008):2069–2077. [DOI: 10.1016/ S0140-6736] Yamamoto 2014 {published data only} Yamamoto N, Kimura H, Misao H, Matsumoto H, Imafuku Y, Watanabe A, et al. Ef cacy of 1.0% chlorhexidine- gluconate ethanol compared with 10% povidone-iodine 33 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. for long-term central venous catheter care in hematology departments: a prospective study.American Journal of Infection Control 2014;42(5):574–6.

References to ongoing studies Goudet 2013 {published data only} Goudet V, Timsit JF, Lucet JC, Lepape A, Balayn D, Seguin S, et al. Comparison of four skin preparation strategies to prevent catheter-related infection in intensive care unit (CLEAN trial): a study protocol for a randomized controlled trial. Trials2013;14:114.

Additional references Adams 2007 Adams D, Elliot TS. Skin antiseptics used prior to intravascular catheter insertion. British Journal of Nursing 2007; 16(5):278–80.

Bankston 2005 Bankston J. Joseph Lister and the History of Antiseptics .

Hockessin: Mitchell Lane Publishers, 2005.

Bynum 2008 Bynum W. The History of Medicine: A Very Short Introduction . New York: Oxford University Press, 2008.

Campbell 2001 Campbell MK, Mollison J, Grimshaw JM. Cluster trials in implementation research: estimation of intracluster correlation coef cients and sample size. Statistics in Medicine 2001; 20(3):391–9.

CDC 2011 Centers for Disease Control and Prevention (CDC).

Guidelines for the prevention of intravascular catheter- related infections, 2011. http://www.cdc.gov/hicpac/BSI/ BSI-guidelines-2011.html (accessed 4 June 2015).

Chaiyakunapruk 2002 Chaiyakunapruk N, Veenstra DL, Lipsky BA, Saint S.

Chlorhexidine compared with povidone-iodine solution for vascular catheter-site care: a meta-analysis. Annals of Internal Medicine 2002;136(11):792–801.

Cicalini 2004 Cicalini S, Palmieri F, Petrosillo N. Clinical review: new technologies for prevention of intravascular catheter-relat ed infections. Critical Care 2004;8(3):157–62.

Edwards 2008 Edwards JR, Peterson KD, Andrus ML, Dudeck MA, Pollock DA, Horan TC. National Healthcare Safety Network (NHSN) Report, data summary for 2006 through 2007, issued November 2008. American Journal of Infection Control 2008;36(9):609–26.

Guyatt 1993 Guyatt GH, Sackett DL, Cook DJ, Evidence-Based Medicine Working Group. Users’ guides to the medical literature II. How to use an article about therapy or prevention A. Are the results of the study valid?. JAMA 1993; 270(21):2598–601. Hamilton 2007 Hamilton HC, Foxcroft DR. Central venous access sites for the prevention of venous thrombosis, stenosis and infection in patients requiring long-term intravenous therapy. Cochrane Database of Systematic Reviews 2007, Issue 3. [DOI: 10.1002/14651858.CD004084.pub2] Hardin 1997 Hardin W, Nichols R. Handwashing and patient skin preparation. In: Malangoni MA editor(s). Critical Issues in Operating Room Management . Philadelphia: Lippincott- Raven, 1997:133–49.

Higgins 2003 Higgins JPT, Thompson SG, Deeks JJ, Altman DG.

Measuring inconsistency in meta-analyses. British Medical Journal 2003;327:557–60.

Higgins 2011a Higgins JPT, Altman DG, Sterne JAC. Chapter 8: Assessing risk of bias in included studies. In: Higgins JPT, Green S (editors). Cochrane Handbook of Systematic Reviews of Interventions. Version 5.1.0 [updated March 2011].

The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

Higgins 2011b Higgins JPT, Deeks J, Altman DG. Chapter 16:

Special topics in statistics. In: Higgins JPT, Green S (editors). Cochrane Handbook of Systematic Reviews of Interventions. Version 5.1.0 [updated March 2011].

The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

Larson 1995 Larson EL. APIC guideline for handwashing and hand antisepsis in health care settings. American Journal of Infection Control 1995;23(4):251–69.

Lefebvre 2011 Lefebvre C, Manheimer E, Glanville J. Chapter 6: Searching for studies. In: Higgins JPT, Green S (editors). Cochrane Handbook of Systematic Reviews of Interventions. Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

Martindale 2016 Bray eld, A (editor). Martindale: the complete drug reference. www.medicinescomplete.com/mc/martindale/ current/ (accessed 6 July 2016).

McCann 2010 McCann M, Moore ZEH. Interventions for preventing infectious complications in haemodialysis patients with central venous catheters. Cochrane Database of Systematic Reviews 2010, Issue 1. [DOI: 10.1002/ 14651858.CD006894.pub2] NNIS 2004 US Department of Health and Human Services. National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2004, issued October 2004. American Journal of Infection Control 2004; 32(8):470–85. 34 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Nuland 2003Nuland SB. The Doctors’ Plague: Germs, Childbed Fever and the Strange Story of Ignaz Semmelweis . New York: WW Norton, 2003.

O’Grady 2002 O’Grady NP, Alexander M, Dellinger EP, Gerberding JL, Heard SO, Maki DG, et al. Guidelines for the prevention of intravascular catheter-related infections. Centers for Disea se Control and Prevention. Morbidity and Mortality Weekly Report (MMWR) 2002; Vol. 51, issue RR–10:1–29.

Pagani 2008 Pagani JL, Eggimann P. Management of catheter-related infection. Expert Review of Anti-infective Therapy 2008;6(1):

31–7.

Pratt 2007 Pratt RJ, Pellowe CM, Wilson JA, Loveday HP, Harper PJ, Jones SR, et al. epic2: National evidence-based guidelines for preventing healthcare-associated infections in NHS hospitals in England. Journal of Hospital Infection 2007;65 (Suppl 1):S1–64.

Puri 2009 Puri N, Puri V, Dellinger RP. History of technology in the intensive care unit. Critical Care Clinics 2009;25(1):185- 200, ix.

RevMan 2014 [Computer program] The Nordic Cochrane Centre, The Cochrane Collaboration.

Review Manager (RevMan). Version 5.3. Copenhagen:

The Nordic Cochrane Centre, The Cochrane Collaboration, 2014.

Russell 1986 Russell AD. Chlorhexidine: antibacterial action and bacterial resistance. Infection1986;14(5):212–5.

Ryan 2004 Ryan KJ, Ray G (editors). Sherris Medical Microbiology . 4th Edition. New York: McGraw-Hill, 2004.

Schünemann 2011a Schünemann HJ, Oxman AD, Higgins JPT, Vist GE, Glasziou P, Guyatt GH. Chapter 11: Presenting results and ’Summary of ndings’ tables. In: Higgins JPT, Green S (editors). Cochrane Handbook of Systematic Reviews of Interventions. Version 5.1.0 [updated March 2011].

The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org. Schünemann 2011b Schünemann HJ, Oxman AD, Vist GE, Higgins JPT, Deeks JJ, Glasziou P, et al. Chapter 12: Interpreting results and drawing conclusions. In: Higgins JPT, Green S (editors). Cochrane Handbook of Systematic Reviews of Interventions. Version 5.1.0 [updated March 2011].

The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

SIGN 2015 Scottish Intercollegiate Guidelines Network (SIGN). Search lters. http:www.sign.ac.uk/methodology/ lters.html# random (accessed 4 June 2015).

Sterne 2011 Sterne JAC, Egger M, Moher D (editors). Chapter 10:

Addressing reporting biases. In: Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Intervention .Version 5.1.0 [updated March 2011].

The Cochrane Collaboration, 2011. Available from www.cochrane-handbook.org.

Taber 2016 Venes D (editor). Taber’s Cyclopedic Medical Dictionary Online. http://www.tabers.com/tabersonline/view/Tabers- Dictionary/738695/0/antiseptic?q=antiseptic&ti=0 (accessed 6 July 2016).

Trieschmann 2007 Trieschmann U, Cate UT, Sreeram N. Central venous catheters in children and neonates - what is important?.

Images in Paediatric Cardiology 2007;9(4):1–8. [PUBMED:

22368674] Worthington 2005 Worthington T, Elliott TS. Diagnosis of central venous catheter related infection in adult patients. Journal of Infection 2005;51(4):267–80.

References to other published versions of this review Lai 2012 Lai NA, Lai NM, O’Riordan E, Chaiyakunapruk N, Taylor JE, Tan K. Skin antisepsis during catheter insertion for reducing central venous catheter related infections.

Cochrane Database of Systematic Reviews 2012, Issue 10.

[DOI: 10.1002/14651858.CD010140] ∗ Indicates the major publication for the study 35 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. C H A R A C T E R I S T I C S O F S T U D I E S Characteristics of included studies[ordered by study ID] Dettenkofer 2010 Methods Multicentre RCT (Switzerland) Study period: May 2002 to June 2005 Setting: 2 haematology units and 1 surgical unit in 2 universi ty hospitals Participants Adult patients who required a CVC.

Number of participants: 400 Number of catheters; 400 Age: median age of 59 years (25% quartile of 48 to 70 years) Sex: 66% male overall Interventions 2-arm comparison of skin antisepsis prior to catheter insertion 1. Intervention A: 0.1% octenidine with 30% I-propanol plus 45 % 2-propanol.

2. Intervention B: 74% ethanol with 10% 2-propanol. Outcomes • Catheter colonisation • Skin colonisation • Catheter-related BSI • Adverse events Outcomes assessed at various points during in-patient stay. Notes The unit of analysis was patient, and it appeared that 1 cathet er per patient was analysed although this was not stated explicitly Funding source: the study was funded partly by the Swiss Natio nal Science Foundation Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Methods’, ’Randomisation and interven- tions’: “The randomisation code was pro- duced by the independent Centre for Clin- ical Studies using computerised random number generator… used a strati cation factor and block randomisation with ran- domly varying block length” Allocation concealment (selection bias) Low risk ’Methods’, ’Randomisation and interven- tions’: As above, and “The randomisation was realised using closed envelopes, ensur- ing that the sequence was concealed before patients entered the trial.” 36 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Dettenkofer 2010 (Continued) Blinding of participants and personnel (performance bias) All outcomes Low risk ’Methods’, ’Randomisation and interven- tions’: “The patients, staff administering the intervention, the microbiology lab were all blinded to the assignment.” Blinding of outcome assessment (detection bias) All outcomes Low risk ’Methods’, ’Randomisation and interven- tions’: “The patients, staff administering the intervention, the microbiology lab were all blinded to the assignment” Incomplete outcome data (attrition bias)All outcomes High risk Discussion, paragraph 2: “20% of the catheters were not cultured, however they were equally distributed”. The absolute rate of post randomisation exclusion was high for the outcome of catheter colonisa- tion. However, the authors appeared to fol- low the intention-to-treat principle as they analysed the patients for whom the data was available in the originally assigned group Selective reporting (reporting bias) Low risk Authors reported all 4 major outcomes as stated in the ’Methods’, namely, catheter colonisation, skin colonisation, catheter- related BSI and adverse effects in suf cient detail in the ’Results’ Other bias Low risk None identi ed Humar 2000 Methods Multicentre RCT (Canada) Study period: Period of study not speci ed but authors stated that study conducted over 1 year (paragraph 1, results) Setting: hospital-wide Participants ’Patients and methods’, ’Patients’: “All patients > 18 year s of age who had CVCs inserted for any purpose were eligible for inclusion in the study, prov ided the treating physician felt the inserted catheter would be present for a minimum of 72 hours.” Number of participants: 242 Number of catheters; 374 Age: mean of 58.3 years +/- range of 16.8 years (chlorhexidine g roup ) and 62.2 years +/- range of 16.0 years (povidone-iodine group) Sex: 78% male in chlorhexidine group and 72% male in povidone-i odine group Interventions Comparison of 2 active agents for initial and subsequent cutaneous antisepsis for catheter care 1. Intervention A: 10% povidone-iodine.

2. Intervention B: 0.5% tincture of chlorhexidine. 37 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Humar 2000 (Continued) Outcomes assessed at various points during in-patient stay. Outcomes • Catheter-related BSI (de nite and probable) • Catheter colonisation • Insertion site infection Notes Funding source: the study was funded by Physicians Services Inco rporated (North York, Ontario, Canada) and Medi-Flex (Overland Park, KS). Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Methods’, ’Study design’: Randomisation was achieved “by the use of blinded block randomisation schedule” Allocation concealment (selection bias) Unclear risk Although the authors stated that the block randomisation schedule was “blinded”, there was no further information provided on how treatment assignment was allocated using the random sequence generated at the time of enrolment Blinding of participants and personnel (performance bias) All outcomes High risk The authors did not report whether blind- ing was achieved; blinding for clinical out- come assessment was highly unlikely be- cause the antiseptic solutions used differed in appearance Blinding of outcome assessment (detection bias)All outcomes Unclear risk Blinding for microbiological outcome as- sessment was unclear as this was not stated in the paper Incomplete outcome data (attrition bias) All outcomes High risk For the outcomes of catheter-related BSI and catheter colonisation, trialists only analysed 180 out of 242 patients that were initially enrolled (74%). The authors stated that 62 catheters were not analysed because the catheter tips were not available for cul- ture, the underlying reasons of which were not provided For the outcome of insertion site (“exit site”) infection which was not dependent on catheter culture, trialists included all 242 patients in the analysis The authors appeared to follow the inten- tion-to-treat principle as they analysed the 38 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Humar 2000 (Continued) patients for whom the data was available in the originally assigned group Selective reporting (reporting bias) Low risk Authors reported all the outcomes stated in the ’Methods’ with suf cient detail in the ’Results’ Other bias High risk The study employed a block randomisation schedule with high likelihood that blind- ing of participants and personnel could not be achieved. This posed a risk to the integrity of the random sequence which would be vulnerable to disruption follow- ing educated guesses by those involved in the study on the likely assigned group of the future participants Langgartner 2004 Methods Single-centre RCT (Germany) Study period: May 1999 to August 2002.

Setting: Inpatient hospital wards and ICUs Participants ’Materials and methods’: ”Adult inpatients scheduled for el ective CVC placement during normal working hours were eligible for participation in the s tudy. Patients from normal wards as well as from the intensive care units were included. Pa tients known to be allergic to iodine or chlorhexidine were excluded as were all patients w ho needed a CVC placed under emergency conditions. No underlying disease was de ned as an exclusion criteria.

“ Number of participants: 119 Number of catheters: 200 (140 analysed) Age: mean age ranged from 50.5 to 56.6 years (SD ranged from 14. 8 to 17.2 years) (reported separately according to three groups) Sex: overall 60.7% male. Interventions Skin disinfection prior to catheter insertion and daily durin g the change of dressings with 1 of the 3 regimens 1. Intervention A: povidone-iodine 10% aqueous solution.

2. Intervention B: propanol 70%/chlorhexidine 0.5%.

3. Intervention C: propanol 70%/chlorhexidine 0.5% followe d by PVP-iodine 10%.

Outcomes assessed at various points during in-patient stay. Outcomes Catheter colonisation Notes Funding source: not stated. Risk of bias 39 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Langgartner 2004 (Continued) Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear risk Not reported Allocation concealment (selection bias) Low risk ’Materials and methods’: ”Sealed and num- bered envelopes contained the randomisa- tion code together with the instructions for skin disinfection and forms for the docu- mentation of the procedure.“ Blinding of participants and personnel (performance bias) All outcomes High risk Blinding of patients and carers not re- ported, although blinding appeared very unlikely because the number of antiseptic solution used for each group and their ap- pearances were different Blinding of outcome assessment (detection bias)All outcomes Unclear risk It was not stated whether the personnel tak- ing the swabs and the interpreter of the mi- crobiological tests were blinded to the allo- cation Incomplete outcome data (attrition bias)All outcomes High risk ’Materials and methods’: ”In addition to the 140 catheters evaluated, 60 more catheters had been included but had to be excluded from analysis: in 5 cases, pa- tients had died with the catheter in place, in 38 cases microbiological analysis of the catheter tip had not been performed and 17 catheters were lost during follow-up (e.

g. the patient was taken to a different clinic with the CVC in place).” In total, 200 catheters were recruited but only 140 were evaluated, which represented an overall dropout rate of 30%. It was un- clear why trialists did not perform microbi- ological analyses in the 38 catheters as men- tioned However, the authors appeared to fol- low the intention-to-treat principle as they analysed the patients for whom the data was available in the originally assigned group Selective reporting (reporting bias) High risk The only outcome stated in the ’Meth- ods’ and reported was catheter coloni- sation. Some important outcomes such as catheter-related blood stream infection, clinical sepsis and mortality were not re- ported 40 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Langgartner 2004 (Continued) Other bias High risk There was a unit of analysis issue in which the number of catheters analysed exceeded the number of participants by nearly 18%, and the outcome was reported using catheters as the units Levy 1988 Methods Single-centre RCT (USA) Study period: not reported Setting: no clear description of the study setting except that t he study was conducted on “patients undergoing coronary artery surgery” Participants ’Patients and methods’: “60 patients scheduled for coronary a rtery surgery were studied during right internal jugular vein cannulation for PA cathete r insertion.” Number of participants: 60 Number of catheters;60 Age: not reported Sex: not reported Interventions Comparison of 2 skin preparation regimes before insertion o f CVC 1. Intervention A: 1 minute-cleaning with 70% isopropyl alcoho l followed by draping with a sterile non-absorbent sheet with an iodophor-i mpregnated adherent lm placed over the aperture.

2. Intervention B: povidone-iodine swabs followed by drapin g with a sterile non- absorbent sheet. Outcomes • Catheter colonisation • Bacterial contamination of surgical gloves Outcomes assessed at various points during in-patient stay. Notes Funding source: not stated. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear risk ’Patients and methods’: “Patients were as- signed randomly assigned to one of two groups.” There was no further information, includ- ing on random sequence generation Allocation concealment (selection bias) Unclear risk There was no information in the paper to enable an assessment on whether random sequence generation was independent from allocation 41 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Levy 1988 (Continued) Blinding of participants and personnel (performance bias) All outcomes High risk Although the authors did not explicitly say, blinding of the patient and personnel was highly unlikely because the 2 skin antisep- sis regimes differed in the way of admin- istration (1 using a liquid solution and an additional adherent lm and the other us- ing a swab without an adherent lm) Blinding of outcome assessment (detection bias) All outcomes Unclear risk Blinding for microbiological outcome as- sessment not reported Incomplete outcome data (attrition bias)All outcomes Low risk Authors analysed all 60 participants ini- tially enrolled and seemed to follow the in- tention-to-treat principle Selective reporting (reporting bias) High risk Authors reported both major outcomes named in the ’Methods’, catheter coloni- sation and positive glove culture, in suf - cient detail in the ’Results’. However, they did not include major patient-related out- comes such as catheter-related BSI, sepsis or mortality Other bias Low risk None identi ed Maki 1991 Methods Single-centre RCT (USA) Study period: 1986-1987.

Setting: surgical ICU Participants All adult patients over 18 years old Number of participants:176 Number of catheters;176 Age: mean age ranged from 51 to 53 years (SD of 19 in all three gro ups) Sex: not reported. Interventions Skin antisepsis prior to CVC insertion and every 48 h thereaf ter using 1 of 3 antiseptic solutions 1. Intervention A: 10% povidone-iodine.

2. Intervention B: 70% isopropyl alcohol.

3. Intervention C: 2% chlorhexidine gluconate. Outcomes • “Catheter-related infections” (catheter colonisation) • “Catheter-related bacteraemia” (catheter-related BSI) Outcomes assessed at various points during in-patient stay. 42 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Maki 1991 (Continued) Notes Although not clearly stated, it appeared that each patient had only 1 catheter included in the study, as Table 1 in the article suggested. Authors stud ied both venous and arterial catheters and reported outcome data separately Funding source: partly funded by Stuart Corporation (ICI, Ltd ) of Wilmington, Delaware Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear risk ’Materials and methods’, ’Procedures for insertion and care of catheters’: “At the time of insertion, each catheter was randomised to one of three antiseptic solutions . . .” There was no description of random se- quence generation Allocation concealment (selection bias) Unclear risk Not reported Blinding of participants and personnel (performance bias) All outcomes High risk ’Materials and methods’, ’Source of clinical data’: “Although it was not possible for the users or the research nurses to be blinded to the antiseptic agent used . . .” Blinding of outcome assessment (detection bias)All outcomes Low risk ’Materials and methods’, ’Source of clinical data’: “[T]he research microbiologist who processed all cultures had no knowledge of the antiseptic group to which the catheter had been assigned” Incomplete outcome data (attrition bias)All outcomes Low risk It appeared that there were no with- drawals, as the number of catheters anal- ysed matched the number of catheters en- rolled initially. The authors appeared to follow the intention-to-treat principle by analysing the catheters in the originally as- signed groups Selective reporting (reporting bias) Low risk Authors reported both major outcomes of catheter colonisation and catheter-related BSI as stated in the ’Methods’ in suf cient detail in the ’Results’. An additional out- come of adverse event was reported, al- though this was reported as an overall per- centage without separating venous from ar- terial catheters Other bias Low risk None identi ed 43 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 1996 Methods Single-centre RCT (France) Study period: 1 July 1992 to 31 October 1993 Setting: surgical-trauma ICU Participants Consecutive patients aged 18 years and above who were scheduled to receive a non- tunnelled central venous catheter, an arterial catheter or bot h Number of participants: not reported Number of catheters; 158 Age: mean age from 51 to 54 years (SD 18 to 19)(reported separatel y in two groups) Sex: not reported Interventions Comparison of the following 2 skin antiseptic regimens prio r to catheter insertion and every 48 h post insertion 1. Intervention A: 0.25% chlorhexidine+ 0.025% benzalkoniu m 2. Intervention B: 10% povidone-iodine Outcomes • Catheter colonisation • Catheter-related BSI Outcomes assessed at various points during in-patient stay. Notes Trialists studied both arterial catheters and CVCs. They did not report data separately for CVC and arterial catheters except for the outcomes of cathete r colonisation per 1000 catheter-days and catheter-related sepsis per 1000 catheter-day s Funding source: funded in part by Les Laboratoires Nicholas, G aillard, France Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Materials and methods’, ’Randomisation procedure’: ”Each patient requiring at least one catheter was randomly allocated to one of two groups by drawing envelopes from an urn.“ Allocation concealment (selection bias) Unclear risk ’Materials and methods’, ’Randomisation procedure’: ”“Each patient requiring at least one catheter was randomly allocated to one of two groups by drawing envelopes from an urn.” It was unclear who drew the envelopes and when. It was also unclear whether the en- velops were sealed and opaque. If the en- velop was drawn by the investigator in- volved in the enrolment, there was a high risk of violating allocation concealment, for example, by redrawing 44 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 1996 (Continued) Blinding of participants and personnel (performance bias) All outcomes High risk ’Materials and methods’, ’Blood cultures’:

“Although it was not possible for the re- search team to be blinded to the antisep- tic agents used, the research microbiologist who processed all cultures had no knowl- edge of the antiseptic group to which the catheter had been assigned.” Blinding of outcome assessment (detection bias) All outcomes Low risk ’Materials and methods’, ’Blood cultures’:

“Although it was not possible for the re- search team to be blinded to the antisep- tic agents used, the research microbiologist who processed all cultures had no knowl- edge of the antiseptic group to which the catheter had been assigned.” Incomplete outcome data (attrition bias)All outcomes Unclear risk There was no information on post ran- domisation withdrawals, nor any descrip- tion on the use of intention-to-treat analy- sis Selective reporting (reporting bias) Low risk Authors reported the major outcomes stated in the ’Methods’, namely catheter colonisation and catheter related sepsis, in suf cient details in the ’Results’. The authors provided separate data for CVCs and arterial catheters for the outcomes of catheter colonisation per 1000 catheter- days and catheter-related sepsis per 1000 catheter-days Other bias Low risk None identi ed Mimoz 2007 Methods Single-centre RCT (France) Study period: 14 May 2004 to 29 June 2006 Setting: surgical ICU Participants Adult inpatients Number of participants: not reported Number of catheters; 538 Age: mean age 57-58 years (SD 18-19) (reported separately in two g roups) Sex: 67.4% men in chlorhexidine group and 75.7% men in povidon e-iodine group Interventions Skin antisepsis using the following 2 regimens prior to CVC insertion and thereafter every 72 h 1. Intervention A: chlorhexidine gluconate, 0.025% benzalko nium chloride. 45 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 2007 (Continued) 2. Intervention B: 4% benzylic alcohol combined versus 5% povi done-iodine in 70% ethanol. Outcomes • Catheter colonisation • Catheter-related BSI Outcomes assessed at various points during in-patient stay. Notes Funding source: this study was supported by Centre Hospitali er et Universitaire de Poitiers and unrestricted grants from Bayer HealthCare and V iatris Pharmaceuticals Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Methods’, ’Randomisation’: “The ran- domisation sequences were generated by computer and conveyed to the investiga- tors by means of sealed envelopes, 1 for each catheter, with instructions to select en- velopes in numerical order.” Allocation concealment (selection bias) Low risk ’Methods’, ’Randomisation’: “The ran- domisation sequences were generated by computer and conveyed to the investiga- tors by means of sealed envelopes, 1 for each catheter, with instructions to select en- velopes in numerical order.” Blinding of participants and personnel (performance bias) All outcomes High risk ’Methods’, ’Randomisation’: “Although it was not possible for the nurses and attend- ing physicians to be blinded to the antisep- tic agent used because of different colours of the 2 solutions (brown for the povidone- iodine and colourless for the chlorhexidine- based solution), the microbiologists who processed all of the cultures and the re- search team who reviewed the outcomes were unaware of the type of antiseptic so- lution used.” Blinding of outcome assessment (detection bias)All outcomes Low risk ’Methods’, ’Randomisation’: “Although it was not possible for the nurses and attend- ing physicians to be blinded to the antisep- tic agent used because of different colours of the 2 solutions (brown for the povidone- iodine and colourless for the chlorhexidine- based solution), the microbiologists who processed all of the cultures and the re- search team who reviewed the outcomes 46 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 2007 (Continued) were unaware of the type of antiseptic so- lution used.” Incomplete outcome data (attrition bias) All outcomes Low risk There was 11% withdrawal, with a similar number of catheters excluded from anal- ysis from the 2 groups. The authors have clearly stated the reasons for withdrawal and appeared to follow the intention-to- treat principle by analysing the available pa- tient data in the originally assigned groups Selective reporting (reporting bias) Low risk Authors reported the 2 major outcomes stated in the ’Methods’, namely, catheter colonisation and catheter-related BSI, in suf cient details in the ’Results’ Other bias Low risk None identi ed Prager 1984 Methods Single-centre RCT (USA) Study period: not reported Setting: hospital departments of General Surgery (123), Medi cine (20), Thoracic Surgery (19), Neurosurgery (8), Obstretrics and Gynaecology (3), Paediatri cs (3) and others (3) Participants All hospital inpatients who required a CVC Number of participants: 159 adults, 3 children Number of catheters; 179 Age: not reported Sex: not reported Interventions Skin antisepsis applied daily after CVC insertion.

1. Intervention A: povidone-iodine for skin antisepsis 2. Control: no skin antisepsis Outcomes • Catheter colonisation • “Catheter-related septicaemia” (catheter-related BSI) Outcomes assessed at various points during in-patient stay. Notes Funding source: supported in part by the Purdue Frederick Comp any, Wilmington, Delaware Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) High risk The exact method of sequence generated was not described. However, in the ’Meth- 47 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Prager 1984 (Continued) ods’, the authors stated that patients were randomised according to hospital registra- tion number, suggesting that they used al- ternation, instead of true randomisation Allocation concealment (selection bias) High risk As above Blinding of participants and personnel (performance bias) All outcomes High risk Although the authors did not explicitly say, it was unlikely that the participants and the care providers were blinded, as the study assessed skin antisepsis versus no skin anti- sepsis Blinding of outcome assessment (detection bias)All outcomes Unclear risk Blinding of microbiological outcome asses- sor not reported Incomplete outcome data (attrition bias)All outcomes Low risk Although the authors did not describe any withdrawals, it appeared that all catheters that were initially enrolled were analysed in the originally assigned groups Selective reporting (reporting bias) Low risk Authors reported the major outcomes of catheter colonisation and catheter-related BSI as stated in the ’Methods’ in suf cient detail in the ’Results’ The authors also reported an additional outcome of skin erythema. However, this was reported as an overall percentage of pa- tients with colonised catheters, not accord- ing to the allocated groups, and so it did not allow data extraction for meta-analysis.

Nevertheless, this did not affect our judg- ment on the overall risk of reporting bias in any major way Other bias High risk There was a unit of analysis issue in which the number of catheters analysed exceeded the number of participants by nearly 10%, and the outcomes were reported using catheters as the units 48 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Sadowski 1988 Methods Single-centre RCT (USA) Study period: November 1982 to December 1985 Setting: surgical ICU Participants Adult burn patients with a CVC in place Number of participants: 50 Number of catheters; 50 Age: mean age of 5.4 years (10 weeks to 15 years) Sex: 68% male Interventions Skin antisepsis prior to catheter removal:1. Intervention: 70% isopropyl alcohol applied for 3 minutes prior to catheter removal 2. Control: no skin antisepsis Outcomes • Catheter colonisation • Positive blood culture (not reported according to group allocati on) Outcomes assessed at various points during in-patient stay. Notes Funding source: not stated. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear risk ’Materials and methods’: Patients were “randomly assigned to one of two groups”.

Method of random sequence generation not described Allocation concealment (selection bias) Unclear risk Not reported Blinding of participants and personnel (performance bias) All outcomes High risk Although not stated in the article, blinding appeared highly unlikely because the inter- vention involved an additional measure in catheter site care Blinding of outcome assessment (detection bias)All outcomes Unclear risk Blinding of microbiological outcome asses- sor not reported Incomplete outcome data (attrition bias)All outcomes Low risk Although not clearly stated, it appeared that all 50 patients were analysed in their originally assigned groups as the tabulated results suggest Selective reporting (reporting bias) High risk There were 2 major outcomes reported, namely, catheter colonisation (positive catheter tip culture) and positive blood cul- 49 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Sadowski 1988 (Continued) ture. However, the data from positive blood culture was unsuitable to be included in the meta-analysis as it was reported only as an overall gure and not according to the al- located groups Other bias Low risk None identi ed Tuominen 1981 Methods Single-centre RCT (Finland) Study period:not reported.

Setting: ICU Participants Adult inpatients admitted to ICU who required a CVC. No exclus ion criteria stated Number of participants:136 Number of catheters; 136 (124 analysed) Age: not reported Sex: not reported Interventions Skin antisepsis applied prior to CVC insertion and regularl y thereafter 1. Intervention A: chlorhexidine 0.05% added to the sterile g auze and applied at the CVC insertion site twice daily 2. Intervention B: sterile gauze application without chlorhe xidine Outcomes •Septicaemia • Catheter colonisation • Adverse effects • Number of patients on antibiotics during the in-dwelling time of the catheters Outcomes assessed at various points during in-patient stay. Notes No adverse effects were recorded in either group, so we do not in clude the data in our analysis Funding source: not stated. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Patients and methods’: The patients were “randomly allocated to one of two groups” Allocation concealment (selection bias) Unclear risk Not adequately described Blinding of participants and personnel (performance bias) All outcomes High risk Not stated in the paper, but blinding appears unlikely as the trial involved a comparison between the application of chlorhexidine-soaked gauze versus a dry 50 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Tuominen 1981 (Continued) sterile gauze Blinding of outcome assessment (detection bias) All outcomes Unclear risk Blinding of microbiological outcome asses- sor not reported Incomplete outcome data (attrition bias) All outcomes Unclear risk The authors did not provide information on the initial number of patients and catheters recruited or the eventual number analysed Selective reporting (reporting bias) Low risk The outcomes were not de ned in the ’Methods’. However, authors reported all major outcomes, including septicaemia, catheter colonisation and adverse effects, in suf cient detail Other bias Low risk None identi ed Vallés 2008 Methods Single-centre RCT (Spain) Study period: 1 Jan 2005 to 3 June 2006 Setting: adult medical-surgical ICU in a university hospital Participants Patients requiring a CVC Number of participants: 420 Number of catheters; 998 (631 analysed) Age: mean age from 60 to 61 years (SD 16-17) (reported separately in three groups) Sex: not reported. Interventions 3-arm comparison of the following skin antiseptic regimens ap plied prior to CVC in- sertion and every 72 h thereafter 1. Intervention A: 10% aqueous povidone-iodine solution.

2. Intervention B: 0.5% alcoholic chlorhexidine-gluconate sol ution.

3. Intervention C: 2% aqueous chlorhexidine-gluconate soluti on. Outcomes •Catheter colonisation • “Catheter related sepsis” (catheter-related BSI) • “Catheter related bacteraemia” • Mortality was not speci ed as an outcome in the methods, but mor tality gures were reported in the “Patient characteristics” table.

Outcomes assessed at various points during in-patient stay. Notes Funding source: not stated. Risk of bias 51 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Vallés 2008 (Continued) Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Low risk ’Methods’, ’Study design’: The random se- quence was generated by “[b]y use of a blinded block randomisation schedule” Allocation concealment (selection bias) Unclear risk Not adequately reported Blinding of participants and personnel (performance bias) All outcomes High risk Although not stated by the authors, blind- ing to patients and caregivers appeared highly unlikely, as the antiseptic solutions used differed in appearance Blinding of outcome assessment (detection bias)All outcomes Low risk ’Methods’, ’Bacteriologic methods’:

“The microbiologists who performed the catheter-tip cultures had no knowledge of the antiseptic group to which the catheter had been assigned.” Incomplete outcome data (attrition bias)All outcomes High risk Authors excluded from analysis 367/998 (36.7%) of the catheters initially ran- domised for various reasons ( Figure 1 of the paper). They excluded 279 catheters post enrolment because they did not meet the inclusion criteria. However, among these excluded catheters, the reason given for 179 of them was that they were “not cultured”.

It was unclear what the underlying reasons were for failure to obtain culture in these catheters, and whether the excluded data here were missing at random Trialists excluded 88 further catheters be- cause they were inserted beyond 72 h af- ter discharge from ICU. These 88 catheters were evenly distributed among the 3 as- signed groups (61 between the 2 chlorhex- idine groups and 27 in the povidone-io- dine group). However, following the con- struction of the best- and worst-case scenar- ios using the dropouts, the direction of the effect estimates swung from signi cantly favouring the chlorhexidine group (best- case scenario for chlorhexidine group) to signi cantly favouring the povidone-iodine group (worst-case scenario for chlorhexi- dine group). It was unclear whether the au- thors followed the intention-to-treat prin- ciple by analysing all available data accord- 52 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Vallés 2008 (Continued) ing to the originally assigned groups, as there was no mention of participants who crossed over to the other group We accorded the study high risk in this domain due to the high absolute dropout rate including the 179 catheters that were not adequately accounted for, as mentioned above, and the vulnerability of the result estimates to best- and worst-case scenarios Selective reporting (reporting bias) Low risk Authors reported all 3 outcomes stated in the ’Methods’, namely, catheter coloni- sation, catheter-related BSI (“catheter-re- lated sepsis”) and catheter-related bacter- aemia in suf cient detail in the ’Results’ In addition, they also reported the impor- tant outcome of mortality in the “Patient characteristics” table. although this was not a pre-speci ed outcome in the methods. Other bias High risk The study employed a block randomisation schedule with high likelihood that blind- ing of participants and personnel were not achieved. This posed a risk to the integrity of the random sequence, which would be vulnerable to disruption following edu- cated guesses by those involved in the study on the likely assigned group of the future participants There was a serious unit of analysis issue in which the number of catheters analysed ex- ceeded the number of participants by over 50%, and the major outcomes were re- ported using catheters as the units Yasuda 2013 Methods Multicentre RCT (Japan) Study period: March 2014 (not further details provided) Setting: 23 Japanese ICUs Participants ’Participants’: “Patients over 18 years of age undergoing CV C and AC placement for more than 72 hours” Number of participants:not reported Number of catheters; 137 Age: not reported Sex: not reported 53 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Yasuda 2013 (Continued) Interventions 3-arm comparison for skin antisepsis prior to catheter inserti on 1. Intervention A: 1% chlorhexidine gluconate (CHG) in alcohol.

2. Intervention B: 0.5% CHG in alcohol.

3. Intervention C: 10% povidone iodine (base solution unknow n). Outcomes •Catheter colonisation • Catheter-related BSI Outcomes assessed at various points during in-patient stay. Notes For this review, we combined the data for 1% CHG and 0.5% CHG as t here was no signi cant difference in the results between the 2 groups This was an interim analysis of the full study and was publish ed in abstract form Funding source: not stated. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) Unclear risk Not mentioned in the published abstract Allocation concealment (selection bias) Unclear risk Not mentioned in the published abstract Blinding of participants and personnel (performance bias) All outcomes Unclear risk Not mentioned in the published abstract Blinding of outcome assessment (detection bias) All outcomes Unclear risk Not mentioned in the published abstract Incomplete outcome data (attrition bias)All outcomes Unclear risk Not mentioned in the published abstract Selective reporting (reporting bias) Unclear risk Not mentioned in the published abstract Other bias Unclear risk Insuf cient information in the published abstract to assess the risks of bias Yousefshahi 2013 Methods Single-centre RCT (Iran) Study period: not reported.

Setting: cardiac-surgical ICU Participants Adult patients admitted to ICU after cardiac surgery Number of participants: 249 Number of catheters; 249 54 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Yousefshahi 2013 (Continued) Age: mean age of 57 and 60 years (range 51 to 68) (reported separa tely in two groups) Sex: 76.1% and 76.5% male (reported separately in two groups) Interventions Skin antisepsis prior to CVC insertion. 1. Intervention A: Sanosil (which consisted of hydrogen peroxi de and silver).

2. Intervention B: pure water (as adjunct to chlorhexidine 2% ba th plus povidone- iodine 10% scrub). Outcomes • Catheter colonisation • Sepsis Outcomes assessed at various points during in-patient stay. Notes The number of CVCs evaluated matched the number of participant s Funding source: not stated. Risk of bias Bias Authors’ judgement Support for judgement Random sequence generation (selection bias) High risk From the authors’ description, it appeared that an alternate sequence was used follow- ing an initial coin toss to determine the daily order of the grouping ’Methods’: “[A]ll the patients were sep- arated into the intervention and control groups based on simple randomisation and entry sequence to the pre-operation room.

Each day, a simple coin randomisation technique was used to determine the group for the rst patient and the spraying of pure water or Sanosil 2% on the catheter loca- tion (from the upper chest to the mandible) . Subsequently, odd and even numbers were used to determine the group of the other patients.” Allocation concealment (selection bias) High risk From the authors’ description, it appeared that an alternate sequence was used follow- ing an initial coin toss to determine the daily order of the grouping ’Methods’: “[A]ll the patients were sep- arated into the intervention and control groups based on simple randomisation and entry sequence to the pre-operation room.

Each day, a simple coin randomisation technique was used to determine the group for the rst patient and the spraying of pure water or Sanosil 2% on the catheter loca- 55 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Yousefshahi 2013 (Continued) tion (from the upper chest to the mandible) . Subsequently, odd and even numbers were used to determine the group of the other patients.” Blinding of participants and personnel (performance bias) All outcomes Unclear risk From the authors’ description, it appeared that the patients and the person who re- moved the catheters to send for culture were blinded (see below). However, the au- thors did not state whether the nurse who sprayed the study substance was blinded to the study materials ’Methods’: “Both spray bottles were similar in shape and cover. Sanosil does not have any colour or smell and is similar to water, and the patients were blinded to the study.

” ’Methods’: “Each day, two trained ICU nurses, blinded to the group type of the patients, collected the tips of ve removed catheters aseptically...” Blinding of outcome assessment (detection bias)All outcomes Unclear risk Not reported Incomplete outcome data (attrition bias)All outcomes Low risk It appeared that all patients recruited ini- tially had their CVCs analysed Selective reporting (reporting bias) Low risk Authors reported the 2 key outcomes spec- i ed in the ’Methods’, namely, catheter colonisation and sepsis, in the ’Results’. As no patient in either group developed sep- sis, we did not include this outcome in our meta-analysis Other bias Low risk None identi ed AC : arterial catheter; BSI: bloodstream infection; CHG: chlorhexidine-gluconate; CVC: central venous catheter; ICU: intensive care unit; PA: pulmonary artery; RCT: randomised controlled trial. 56 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Characteristics of excluded studies[ordered by study ID] Study Reason for exclusion Almeida 2009 Before-and-after study. Basis of exclusion: design Apisarnthanarak 2010 Quasi-experimental before-and-after study. Basis of exclusio n: design Assadian 2004 A commentary to Parienti 2004 . Basis of exclusion: article type Astle 2005 An RCT that assessed ExSept versus chlorhexidine for patient s with haemodialysis catheters. Basis of exclu- sion: population Balamongkhon 2007 Non-randomised trial that assessed CVC site care using 2% chlor hexidine gluconate versus povidone-iodine.

Basis of exclusion: design Bilir 2009 This is a conference abstract of an study awaiting classi cation ( BIlir 2013 ) Borghesi 2011 A review article on infection control strategies for the newbor n. Basis of exclusion: article type and population Bowling 2010 A before-and-after study that assessed a multifaceted program me in decreasing blood culture contamination.

Basis of exclusion: study design Camins 2010 Cross-over study that assessed chlorhexidine-impregnated fo am dressing for prevention of catheter-related BSI in patients undergoing haemodialysis. Basis of exclusio n: design, population and intervention Carrer 2005 RCT that compared maximal sterile barrier (consisting of mask, cap, sterile gloves, gown, large drape) versus control precautions (mask, cap, sterile gloves, small drape) an d transparent polyurethane lm versus gauze dressing for reduction of CVC-related infections. Basis of excl usion: intervention Casey 2003 A single-centre RCT (UK) that compared the PosiFlow needleless co nnector against the standard luer cap attached to the CVCs for adult patients admitted for cardiac surgery. The authors used a factorial design which enabled the concurrent 3-arm comparison of 3 differe nt skin antiseptic solutions (0.5% chlorhexidine/alcohol, 70% isopropyl alcohol and 10% povidon e-iodine) applied prior to the insertion of the catheters. However, the major outcome assessed was “stopco ck entry port microbial contamination” rather than catheter colonisation, and this is not part of the p respeci ed outcomes in our review. Basis of exclusion: study design (design of the outcome) Casey 2007 RCT that compared a needless connector set (Clearlink Y-type exte nsion set) against standard 3-way stop- cocks with caps for reducing CVC related infections. Basis of exclu sion: intervention Cepkova 2006 A review article on reducing catheter-related infections in the ICU. Basis of exclusion: article type Chaiyakunapruk 2003 Cost-effectiveness analysis on chlorhexidine gluconate versus povidone-iodine for catheter site care. Basis of exclusion: article type Crawford 2004 Cost-bene t analysis of chlorhexidine gluconate dressing in r educing catheter-related infections. Basis of exclusion: article type 57 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. (Continued) Daghistani 1996 RCT that assessed antibiotic ush for CVCs in children with can cer. Basis of exclusion: intervention Darouiche 2007 A review article on strategies to prevent catheter-related infections. Basis of exclusion: article type Darouiche 2008 A review article on strategies to prevent catheter-related infections. Basis of exclusion: article type Dean 2011 A cross-over study that compared the use of chlorhexidine soluti on against chlorhexidine-impregnated cloth for CVC care. Basis of exclusion: study design Dettenkofer 2002 A quasi-randomised trial in which patients were assigned on an alternate basis to either octenidine-based skin antiseptic solution versus propanol-based solution. Addit ionally, the results were presented in 25th centile, median and 75th centile of quantitative skin culture (in CFU/24 cm2 ) which does not allow extraction for meta-analysis. Basis of exclusion: study design and data repo rting Eggimann 2010 A prospective non-randomised study that assessed catheter-rel ated infections following the introduction of various infection control strategies. Basis of exclusion: stu dy design Eyberg 2008 RCT that assessed chlorhexidine gluconate gel dressing versu s chlorhexidine gluconate disk in reducing CVC-related infections. Basis of exclusion: intervention Freiberger 1992 A quasi-experimental study comparing 2 skin antisepsis regim ens (chlorhexidine and povidone-iodine) and 2 types of dressing (Tegaderm and standard gauze) in a 4-arm com parison of different combinations. The authors only reported the results in F or X 2 values along with the P values, without reporting the raw dat a, which precluded data extraction for meta-analysis. Basis of excl usion: study design and data reporting Fukunaga 2004 A non-randomised study with historical cohort that assessed po vidone-iodine ointment in addition to dressing in reducing CVC-related infections. Basis of exclusio n: study design Garcia 2010 A non-randomised study that assessed the effect of chlorhexidi ne scrub of the CVC hub during each access in reducing CVC-related infections. Basis of exclusion: study d esign Garcia-Teresa 2007 A multicentre observational study that evaluated CVC-related infections in children. Basis of exclusion:

study design Garcia-Vazquez 2011 A before-and-after study that evaluated the effect of a hand hyg iene promotion programme in reducing infections in an ICU. Basis of exclusion: study design Garland 1996 An RCT that assessed the local reaction to a chlorhexidine glucon ate-impregnated antimicrobial dressing in very low birth weight infants. Basis of exclusion: populat ion and intervention Garland 2001 An RCT that compared chlorhexidine gluconate-impregnated dres sing with povidone-iodine skin scrub for prevention of CVC-related infections in neonates. Basis of ex clusion: population Garland 2009a An RCT that compared chlorhexidine gluconate with povidone-iod ine as skin antisepsis prior to CVC placement in neonates. Basis of exclusion: population 58 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. (Continued) Garland 2009b An RCT that assessed the safety of chlorhexidine gluconate in n eonates with percutaneously inserted central venous catheters. Basis of exclusion: population Gilad 2006 A review article on prevention of catheter-related BSI in the ne onatal intensive care setting. Basis of exclusion:

article type Girard 2012 A longitudinal cohort study that compared two CVC cleaning prot ocols (containing alcohol-based povi- done-iodine solution (Betadine alcolique) and chlorhexidine-b ased antiseptic (Biseptine), respectively) ad- ministered in different periods. Basis of exclusion: study d esign Gnass 2004 A prospective, non-randomised study that evaluated the effect of multiple infection control strategies in reducing catheter-related infections. Basis of exclusion: stud y design Gunst 2011 A non-randomised trial that compared antiseptic-impregnated C VC with peripherally-inserted central line in reducing catheter-related infections. Basis of exclusion: st udy design and intervention Habibzadeh 2013 A commentary on an included study ( Yousefshahi 2013 ) Hachem 2002 A review article on prevention of catheter-related infection in long-term catheters. Basis of exclusion: article type Halpin 1991 An RCT that evaluated the effect of povidone-iodine connection s hield that is incorporated in the catheter hub in reducing CVC-related infections. Basis of exclusion: int ervention Hanazaki 1999 An RCT that assessed the effect of chlorhexidine dressing in reducing catheter colonisation. Basis of exclu- sion: intervention Hill 1990 An RCT that assessed the effect of mupirocin ointment on colonis ation rate of internal jugular vein catheters.

Basis of exclusion: intervention Huang 2006 A retrospective study that assessed the effect of multiple inf ection control measures on the rates of methicillin- resistant Staphylococcus aureus infection in an adult ICU. Basi s of exclusion: study design Hutchinson 1990 An RCT that assessed occlusive versus non-occlusive right atrial catheter dressing change procedures in children with cancer. Basis of exclusion: intervention Ishikawa 2010 An RCT comparing maximal sterile barrier precaution versus st andard sterile barrier precaution measures during CVC insertion in reducing CVC-related infections. Basi s of exclusion: intervention Ishizuka 2009 A non-randomised trial that compared the use of chlorhexidine v ersus povidone-iodine for CVC site skin disinfection in 2 separate cohorts of patients. Basis of exclus ion: study design Johnson 2005 An RCT that compared honey versus mupirocin applied at the cathe ter exit site for preventing catheter- related infections in patients undergoing haemodialysis. B asis of exclusion: population and intervention Khattak 2010 An RCT that evaluated the absorption of silver in very low bir thweight infants who received silver alginate- impregnated central venous catheter. Basis of exclusion: popu lation and intervention 59 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. (Continued) Khouli 2009 A conference abstract that reports the impact of simulation trai ning on residents’ performance in adhering to maximum sterile barrier precaution during CVC insertion. Basis of exclusion: research question and design Krein 2007 A national survey on measures to reduce catheter-related BSI. B asis of exclusion: study design Kruse 1999 This is a commentary on an included study ( Mimoz 1996 ). Basis of exclusion: article type Kulkarni 2013 An RCT that compared the use of 10% povidone-iodine versus 2% chl orhexidine for skin disinfection prior to insertion of epidural or central venous catheters. The stud y combined both epidural and CVCs is the outcome reporting with no separate data for CVC, and more impo rtantly, the outcome of skin colonisation was assessed based on a skin swab that was taken immediately a fter the application of the skin antiseptic agent, which did not t in with our question of whether the appl ication of skin antiseptic agent reduces catheter-related infection during the period of catheter use. E xcluded on th basis of research question and design Lange 1997 A non-randomised trial that assessed a multifaceted strategy in CVC management in reducing catheter- related infection in children with chronic illness. Basis of ex clusion: study design Le Corre 2003 An RCT comparing transparent dressing versus a dry gauze appl ied at the exit site of the catheter on haemodialysis patients. Basis of exclusion: population and intervention Legras 1997 An RCT comparing alcohol-chlorhexidine against povidone-iodin e for skin antisepsis for intravascular catheters. The study evaluated a mixture of venous, arterial and Swan Gantz catheters with no separate outcome reporting for venous catheters. There were no contact de tails provided in the paper to request for separate data for venous catheters. Basis of exclusion: insuf cient information Levy 2005 An RCT that assessed the effectiveness of chlorhexidine glucon ate-impregnated dressing in reducing catheter-related infections in children. Basis of exclusion: in tervention Madeo 1998 An RCT comparing 2 different dressings for arterial and venou s catheters in reducing catheter-related infections. Basis of exclusion: intervention Mahieu 2001 A prospective cohort study that evaluated the effect of catheter manipulation on catheter-related BSI in neonates. Basis of exclusion: study design, population and i ntervention Maki 1981 A commentary on disinfectant for vascular catheters. Basis of ex clusion: article type Maki 1992 An RCT comparing different antibiotic ointments for preventing catheter-related infection. Basis of exclu- sion: intervention McCann 2016 A pilot RCT involving in 3 Irish outpatient hemodialysis uni ts compared 2% chlorhexidine gluconate (CHG) in 70% isopropyl alcohol with CHG solutions for central ve nous catheter exit site antisepsis. Basis of exclusion: population Montecalvo 2012 A prospective cohort study that evaluated the rates of catheter -related BSI over 3 study periods: pre- intervention (phase 1), in which all patients were bathed with s oap and water or non-medicated washcloths; active intervention (phase 2), in which patients were bathed wit h 2% chlorhexidine gluconate cloths with 60 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. (Continued) the number of baths administered and skin tolerability asse ssed; and post-intervention (phase 3), in which chlorhexidine bathing continued but without oversight by res earch personnel. Basis of exclusion: study design Munoz-Price 2009 A non-randomised study that evaluated a step-wise infection con trol approach in reducing catheter-related infection. Basis of exclusion: study design, intervention Munoz-Price 2012 A non-randomised study that evaluated the use of daily chlorhe xidine bath in reducing catheter-related infection. Basis of exclusion: study design Nikoletti 1999 An RCT comparing transparent polyurethane and hydrocolloid d ressings for CVC in reducing catheter- related infection. Basis of exclusion: intervention Noto 2014 A cluster-RCT that assessed the effects of daily chlorhexidine b athing on the rates of healthcare associated infection in general for all ICU patients, not speci c to patie nts with CVC in place. Basis of exclusion:

population Parienti 2004 A cluster-randomised cross-over study that assessed the effecti veness of alcoholic povidone-iodine in pre- venting catheter-related infection. Basis of exclusion: study design Peterson 2011 An evidence-based summary on the effectiveness of chlorhexidin e versus 70% alcohol for CVC injection cap disinfection. Basis of exclusion: article type Raad 1994 An RCT that assessed the effectiveness of maximal sterile pre caution during CVC insertion in reducing catheter-related infection. Basis of exclusion: intervention Render 2006 A cluster-randomised trial that assessed the effectiveness of 2 multifaceted infection control projects in reducing central line infections. Basis of exclusion: study des ign Rezaei 2009 An RCT that assessed the effectiveness of mupirocin ointment i n reducing catheter-related infection. Basis of exclusion: intervention Richardson 2006 A commentary on Parienti 2004 . Basis of exclusion: article type Rickard 2004 An RCT that assessed the effectiveness of changing intravenou s administration set for reducing catheter- related infection. Basis of exclusion: intervention Rijnders 2003 An RCT that assessed the use of full sterile barrier precautio n in reducing catheter-related infection. Basis of exclusion: intervention Rubinson 2004 A review article on measures to reduce catheter-related infectio n during insertion of CVC. Basis of exclusion:

article type Rupp 2008 A non-randomised, comparative, cross-over trial that evaluate d the effectiveness of alcohol-based hand gel in reducing hospital-acquired infections. Basis of exclusion: r esearch question, study design Ruschulte 2009 An RCT that assessed the effectiveness of chlorhexidine-impre gnated wound dressing in reducing CVC- related infection in patients undergoing chemotherapy. Basi s of exclusion: intervention 61 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. (Continued) Schwebel 2012 An economic analysis on chlorhexidine-impregnated sponges fo r reducing catheter-related infection. Basis of exclusion: article type Sheehan 1993 An article identi ed through a related review paper in the for m of a conference abstract. The text of the conference abstract could not be traced after contacting the author of the review article. We were unable to locate the contact details of the authors of this conference pap er to request for further information. The conference abstract did not appear to be published subsequentl y in full. Basis of exclusion: insuf cient information Spiegler 2010 A review article comparing central venous line and arterial lin e infections. Basis of exclusion: article type Swan 2014 A cluster-RCT that compared chlorhexidine bathing versus soap a nd water bathing in decreasing the rates of healthcare associated infection for all patients in ICUs, an d not only patients with a CVC in place. Basis of exclusion: population Tietz 2005 A prospective observational study that assessed the effectiv eness of octenidine hydrochloride for CVC site care in patients receiving bone marrow transplant. Basis of ex clusion: study design Van Esch 2002 An evidence-based summary that examined the role of chlorhexid ine versus povidone-iodine antisepsis for reducing catheter-related infection in neonates. Basis of exclu sion: article type Zingg 2008 An overview on catheter-related BSI. Basis of exclusion: articl e type Zingg 2009 A before-and-after study that assessed the effectiveness of an educational programme on promoting hand hygiene measures in reducing catheter-related BSI. Basis of ex clusion: study design BSI: bloodstream infection; CFU: colony-forming units; CVC: central venous catheter; RCT: randomised controlled trial.

Characteristics of studies awaiting assessment [ordered by study ID] Atahan 2012 Methods RCT (Czech Republic) Participants Adult surgical patients who required a CVC Interventions CVC insertion site disinfection with 10% povidone-iodine (Pov iiodeks) versus Savlosol (15% cetrimide, 1.5% chlorhexidine-gluconate, ethanol) Outcomes Catheter colonisation and catheter-related BSI Notes - 62 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. BIlir 2013 Methods RCT (Turkey) Participants Adult ICU patients who required a CVC Interventions 3-arm comparison: skin antisepsis using 4% chlorhexidine gluconate (n = 19), 10% povidone iodine (n = 19) or octenidine hydrochlorodine (n = 19) Outcomes Catheter colonisation and catheter-related BSI (“catheter-rela ted sepsis”), determined using “standard microbiological methods” (’Materials and methods’) Notes The study evaluated a mixture of venous and arterial catheter s with no separate analysis for venous catheters. This appears to be a conference abstract. We are awaiting further inf ormation from the authors Giles 2002 Methods RCT Participants Surgical patients who required a CVC Interventions Transparent occlusive dressing versus daily CVC site care withpovidone-iodine 10% solution Outcomes Catheter colonisation and catheter-related sepsis Notes Awaiting full text Knasinski 2000 Methods RCT Participants Unclear Interventions 1% chlorhexidine plus 75% alcohol versus 10% povidone iodine f or cutaneous disinfection and follow-up site care with central venous and arterial catheters Outcomes Catheter colonisation and catheter-related BSI Notes This title was identi ed as a conference abstract from an earlie r meta-analysis on a similar topic. There is no further information at this stage other than the title. The author of the meta-analysis paper with the title could not locate the abstract paper, and the study appeared not to be subsequen tly published in full. The study included both venous and arterial catheters, and it was unclear whether a separate o utcome report for venous catheters would be available.

We are awaiting the response of the study author for further i nformation 63 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Mimoz 2015 Methods Open-label multi-centre RCT with a two-by-two factorial design Participants Adults (age >/=18 years) admitted to one of 11 French intensive-care units and requiring at least one of central- venous, haemodialysis, or arterial catheters Interventions All intravascular catheters prepared with 2% chlorhexidine-70 % isopropyl alcohol (chlorhexidine-alcohol) or 5% povidone iodine-69% ethanol (povidone iodine-alcohol), with or without scrubbing of the skin with detergent before antiseptic application Outcomes “catheter-related infections”, catheter colonisation, advers e effects Notes Awaiting full-text report for speci c information on central v enous catheters Yamamoto 2014 Methods A comparative study (it is unclear from the abstract whether it is an RCT) Participants Haematology patients (age range unclear) Interventions 1% chlorhexidine-gluconate ethanol versus 10% povidone-iodine for skin antisepsis of CVC sites Outcomes Catheter-related BSI, catheter colonisation Notes Awaiting full text from the authors Characteristics of ongoing studies[ordered by study ID] Goudet 2013 Trial name or title Comparison of four skin preparation strategies to prevent ca theter-related infection in intensive care unit (CLEAN trial): a study protocol for a randomized controlled tria l Methods “A prospective multicenter, 2 × 2 factorial, randomized-controlled, assessor-blind trial” Participants Setting: 11 intensive care units in 6 French hospitals. Participants: All adult patients aged over 18 years requiring the insertion of 1 or more of the following: periph eral arterial catheter, non-tunnelled central venous catheter, haemodialysis catheter and arterial pulmonary cath eter Interventions Patients are allocated to 1 of the 4 skin preparation strategies: 2% chlorhexidine/70% isopropyl alcohol or 5% povidone iodine/69% ethanol, with and without prior skin scrubbing Outcomes Catheter-related BSI, catheter colonisation, cutaneous tolerance, length of hospitalisation, mortality and cost Starting date October 2012, lasting approximately 14 months Contact information Corresponding author: Olivier Mimoz [email protected] 64 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Goudet 2013 (Continued) Notes Clinicaltrials.gov number NCT01629550 . Protocol published in Trials, 2013:14: 114 65 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. D A T A A N D A N A L Y S E S Comparison 1. Povidone-iodine (in aqueous solution) versus no skin antisepsis Outcome or subgroup titleNo. of studies No. of participants Statistical method Effect size 1 Catheter-related BSI 1 179 Risk Ratio (M-H, Fixed, 95% CI) 0.99 [0.37, 2.61] 2 Catheter colonisation 1 179 Risk Ratio (M-H, Fixed, 95% CI) 0.93 [0.53, 1.60] Comparison 2. Chlorhexidine (in aqueous solution) versus no skin antis epsis Outcome or subgroup titleNo. of studies No. of participants Statistical method Effect size 1 Septicaemia 1 136 Risk Ratio (M-H, Fixed, 95% CI) 2.91 [0.31, 27.31] 2 Catheter colonisation 1 124 Risk Ratio (M-H, Fixed, 95% CI) 1.26 [0.61, 2.59] 3 Number of patients who required antibiotics during in-dwelling period of catheter 1 136 Risk Ratio (M-H, Fixed, 95% CI) 0.84 [0.55, 1.27] Comparison 3. Alcohol versus no skin antisepsis Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter colonisation 1 50 Risk Ratio (M-H, Fixed, 95% CI) 0.75 [0.30, 1.85] Comparison 4. Chlorhexidine versus povidone-iodine Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter-related BSI 4 1436 Risk Ratio (M-H, Fixed, 95% CI) 0.64 [0.41, 0.99] 1.1 Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution 2 452 Risk Ratio (M-H, Fixed, 95% CI) 0.64 [0.32, 1.28] 1.2 Chlorhexidine in alcohol versus povidone-iodine in aqueous solution 2 503 Risk Ratio (M-H, Fixed, 95% CI) 0.77 [0.39, 1.53] 66 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. 1.3 Chlorhexidine in alcohol versus povidone-iodine in alcohol 1 481 Risk Ratio (M-H, Fixed, 95% CI) 0.40 [0.13, 1.24] 2 Catheter-related BSI per 1000 catheter-days 4 1450 Risk Ratio (Fixed, 95% CI) 0.53 [0.30, 0.94] 2.1 Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution 1 308 Risk Ratio (Fixed, 95% CI) 0.82 [0.23, 2.93] 2.2 Chlorhexidine in alcohol versus povidone-iodine in aqueous solution 3 661 Risk Ratio (Fixed, 95% CI) 0.49 [0.25, 0.95] 2.3 Chlorhexidine in alcohol versus povidone-iodine in alcohol 1 481 Risk Ratio (Fixed, 95% CI) 0.41 [0.06, 2.92] 3 All-cause mortality 1 Risk Ratio (M-H, Fixed, 95% CI) Subtotals only 3.1 Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution 1 213 Risk Ratio (M-H, Fixed, 95% CI) 1.15 [0.72, 1.83] 3.2 Chlorhexidine in alcohol versus povidone-iodine in aqueous solution 1 222 Risk Ratio (M-H, Fixed, 95% CI) 0.80 [0.48, 1.34] 4 Catheter colonisation 5 1533 Risk Difference (M-H, Fixed, 95% CI) -0.08 [-0.12, -0.03] 4.1 Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution 2 452 Risk Difference (M-H, Fixed, 95% CI) -0.09 [-0.17, -0.02] 4.2 Chlorhexidine in alcohol versus povidone-iodine in aqueous solution 3 600 Risk Difference (M-H, Fixed, 95% CI) -0.04 [-0.11, 0.03] 4.3 Chlorhexidine in alcohol versus povidone-iodine in alcohol 1 481 Risk Difference (M-H, Fixed, 95% CI) -0.11 [-0.17, -0.04] 5 Catheter colonisation per 1000 catheter-days 5 1547 Risk Ratio (Fixed, 95% CI) 0.64 [0.50, 0.81] 5.1 Chlorhexidine in aqueous solution versus povidone-iodine in aqueous solution 1 308 Risk Ratio (Fixed, 95% CI) 0.69 [0.40, 1.20] 5.2 Chlorhexidine in alcohol versus povidone-iodine in aqueous solution 4 758 Risk Ratio (Fixed, 95% CI) 0.64 [0.48, 0.85] 5.3 Chlorhexidine in alcohol versus povidone-iodine in alcohol 1 481 Risk Ratio (Fixed, 95% CI) 0.53 [0.24, 1.17] 6 Insertion site infection 1 242 Mean Difference (IV, Fixed, 95% CI) -2.80 [-9.10, 3.50] 67 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Comparison 5. Chlorhexidine (in aqueous solution) versus alcohol Outcome or subgroup titleNo. of studies No. of participants Statistical method Effect size 1 Catheter-related BSI 1 99 Risk Ratio (M-H, Fixed, 95% CI) 0.24 [0.02, 2.54] 2 Catheter colonisation 1 99 Risk Ratio (M-H, Fixed, 95% CI) 0.38 [0.11, 1.33] Comparison 6. Povidone-iodine (in aqueous solution) versus alcohol Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter-related BSI 1 109 Risk Ratio (M-H, Fixed, 95% CI) 1.04 [0.21, 5.08] 2 Catheter colonisation 2 169 Risk Ratio (M-H, Fixed, 95% CI) 1.76 [0.76, 4.09] 2.1 Povidone-iodine in aqueous solution versus alcohol 1 109 Risk Ratio (M-H, Fixed, 95% CI) 1.25 [0.49, 3.14] 2.2 Povidone-iodine-impregnated adherent lm versus alcohol 1 60 Risk Ratio (M-H, Fixed, 95% CI) 9.0 [0.51, 160.17] Comparison 7. Alcohol versus octenidine in alcohol Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter-related BSI 1 387 Risk Ratio (M-H, Fixed, 95% CI) 2.01 [0.88, 4.59] 2 Catheter-related BSI per 1000 catheter-days 1 387 Risk Ratio (Fixed, 95% CI) 2.18 [0.54, 8.77] 3 Catheter colonisation 1 322 Risk Ratio (M-H, Fixed, 95% CI) 2.26 [1.22, 4.21] 4 Catheter colonisation per 1000catheter-days 1 322 Risk Ratio (Fixed, 95% CI) 2.23 [0.79, 6.29] 5 Skin colonisation 1 365 Mean Difference (IV, Fixed, 95% CI) 79.00 [32.76, 125. 24] 6 Adverse effects 1 398 Risk Ratio (M-H, Fixed, 95% CI) 0.85 [0.60, 1.20] 68 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Comparison 8. Chlorhexidine (in alcohol) plus povidone-iodine (in aqueous solution) versus chlorhexidine (in alcohol) Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter colonisation 1 88 Risk Ratio (M-H, Fixed, 95% CI) 0.19 [0.04, 0.81] 2 Catheter colonisation per 1000 catheter-days 1 88 Risk Ratio (Fixed, 95% CI) 0.19 [0.06, 0.59] Comparison 9. Chlorhexidine (in alcohol) plus povidone-iodine (in aqu eous solution) versus povidone-iodine (in aqueous solution) Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter colonisation 1 95 Risk Ratio (M-H, Fixed, 95% CI) 0.15 [0.04, 0.62] 2 Catheter colonisation per 1000 catheter-days 1 95 Risk Ratio (Fixed, 95% CI) 0.17 [0.05, 0.52] Comparison 10. Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine aqueous 10% scrub Outcome or subgroup title No. of studies No. of participants Statistical method Effect size 1 Catheter colonisation 1 249 Risk Ratio (M-H, Fixed, 95% CI) 1.08 [0.68, 1.72] 69 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 1.1. Comparison 1 Povidone-iodine (in aqueous solution) versus no skin antisepsis, Outcome 1 Catheter-related BSI.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 1 Povidone-iodine (in aqueous solution) versu s no skin antisepsis Outcome: 1 Catheter-related BSI Study or subgroup Povidone iodine No skin antisepsis Risk Ra tio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Prager 1984 7/84 8/95 100.0 % 0.99 [ 0.37, 2.61 ] Total (95% CI) 84 95 100.0 % 0.99 [ 0.37, 2.61 ] Total events: 7 (Povidone iodine), 8 (No skin antisepsis) Heterogeneity: not applicable Test for overall effect: Z = 0.02 (P = 0.98) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours povidone iodine Favours control Analysis 1.2. Comparison 1 Povidone-iodine (in aqueous sol ution) versus no skin antisepsis, Outcome 2 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 1 Povidone-iodine (in aqueous solution) versu s no skin antisepsis Outcome: 2 Catheter colonisation Study or subgroup Povidone iodine No skin antisepsis Risk Ra tio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Prager 1984 18/84 22/95 100.0 % 0.93 [ 0.53, 1.60 ] Total (95% CI) 84 95 100.0 % 0.93 [ 0.53, 1.60 ] Total events: 18 (Povidone iodine), 22 (No skin antisepsis) Heterogeneity: not applicable Test for overall effect: Z = 0.28 (P = 0.78) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours povidone iodine Favours control 70 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 2.1. Comparison 2 Chlorhexidine (in aqueous solution) versus no skin antisepsis, Outcome 1 Septicaemia.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 2 Chlorhexidine (in aqueous solution) versus n o skin antisepsis Outcome: 1 Septicaemia Study or subgroup Chlorhexidine Control Risk Ratio Weight R isk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Tuominen 1981 3/69 1/67 100.0 % 2.91 [ 0.31, 27.31 ] Total (95% CI) 69 67 100.0 % 2.91 [ 0.31, 27.31 ] Total events: 3 (Chlorhexidine), 1 (Control) Heterogeneity: not applicable Test for overall effect: Z = 0.94 (P = 0.35) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours chlorhexidine Favours control Analysis 2.2. Comparison 2 Chlorhexidine (in aqueous solut ion) versus no skin antisepsis, Outcome 2 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 2 Chlorhexidine (in aqueous solution) versus n o skin antisepsis Outcome: 2 Catheter colonisation Study or subgroup Chlorhexidine Control Risk Ratio Weight R isk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Tuominen 1981 13/60 11/64 100.0 % 1.26 [ 0.61, 2.59 ] Total (95% CI) 60 64 100.0 % 1.26 [ 0.61, 2.59 ] Total events: 13 (Chlorhexidine), 11 (Control) Heterogeneity: not applicable Test for overall effect: Z = 0.63 (P = 0.53) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours chlorhexidine Favours control 71 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 2.3. Comparison 2 Chlorhexidine (in aqueous solution) versus no skin antisepsis, Outcome 3 Number of patients who required antibiotics during in-dwel ling period of catheter.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 2 Chlorhexidine (in aqueous solution) versus n o skin antisepsis Outcome: 3 Number of patients who required antibiotics duri ng in-dwelling period of catheter Study or subgroup Chlorhexidine Control Risk Ratio Weight R isk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Tuominen 1981 25/69 29/67 100.0 % 0.84 [ 0.55, 1.27 ] Total (95% CI) 69 67 100.0 % 0.84 [ 0.55, 1.27 ] Total events: 25 (Chlorhexidine), 29 (Control) Heterogeneity: not applicable Test for overall effect: Z = 0.84 (P = 0.40) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours chlorhexidine Favours control Analysis 3.1. Comparison 3 Alcohol versus no skin antisepsi s, Outcome 1 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 3 Alcohol versus no skin antisepsis Outcome: 1 Catheter colonisation Study or subgroup Alcohol-based No skin antisepsis Risk Rat io Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Sadowski 1988 6/25 8/25 100.0 % 0.75 [ 0.30, 1.85 ] Total (95% CI) 25 25 100.0 % 0.75 [ 0.30, 1.85 ] Total events: 6 (Alcohol-based), 8 (No skin antisepsis) Heterogeneity: not applicable Test for overall effect: Z = 0.63 (P = 0.53) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol-based Favours no antisepsis 72 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.1. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 1 Catheter-related BSI.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 1 Catheter-related BSI Study or subgroup Chlorhexidine-based Povidone- iodine- based Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI 1 Chlorhexidine in aqueous solution versus povidone-iodin e in aqueous solution Maki 1991 1/67 5/77 10.3 % 0.23 [ 0.03, 1.92 ] Vall s 2008 17/211 10/97 30.4 % 0.78 [ 0.37, 1.64 ] Subtotal (95% CI) 278 174 40.7 % 0.64 [ 0.32, 1.28 ] Total events: 18 (Chlorhexidine-based), 15 (Povidone-iod ine-based) Heterogeneity: Chi 2 = 1.17, df = 1 (P = 0.28); I 2 =15% Test for overall effect: Z = 1.26 (P = 0.21) 2 Chlorhexidine in alcohol versus povidone-iodine in aqueo us solution Humar 2000 4/92 4/88 9.1 % 0.96 [ 0.25, 3.71 ] Vall s 2008 15/226 9/97 27.9 % 0.72 [ 0.32, 1.58 ] Subtotal (95% CI) 318 185 37.0 % 0.77 [ 0.39, 1.53 ] Total events: 19 (Chlorhexidine-based), 13 (Povidone-iod ine-based) Heterogeneity: Chi 2 = 0.13, df = 1 (P = 0.72); I 2 =0.0% Test for overall effect: Z = 0.73 (P = 0.46) 3 Chlorhexidine in alcohol versus povidone-iodine in alcoh ol Mimoz 2007 4/242 10/239 22.3 % 0.40 [ 0.13, 1.24 ] Subtotal (95% CI) 242 239 22.3 % 0.40 [ 0.13, 1.24 ] Total events: 4 (Chlorhexidine-based), 10 (Povidone-iodi ne-based) Heterogeneity: not applicable Test for overall effect: Z = 1.59 (P = 0.11) Total (95% CI) 838 598 100.0 % 0.64 [ 0.41, 0.99 ] Total events: 41 (Chlorhexidine-based), 38 (Povidone-iod ine-based) Heterogeneity: Chi 2 = 2.28, df = 4 (P = 0.69); I 2 =0.0% Test for overall effect: Z = 1.99 (P = 0.046) Test for subgroup differences: Chi 2 = 0.98, df = 2 (P = 0.61), I 2 =0.0% 0.01 0.1 1 10 100 Favours chlorhexidine Favours povidone-iodine 73 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.2. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 2 Catheter-related BSI per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 2 Catheter-related BSI per 1000 catheter-days Study or subgroup Chlorhexidine-based Povidone- iodine- based log [Risk Ratio] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI 1 Chlorhexidine in aqueous solution versus povidone-iodin e in aqueous solution Vall s 2008 211 97 -0.2 (0.65) 20.1 % 0.82 [ 0.23, 2.93 ] Subtotal (95% CI) 211 97 20.1 % 0.82 [ 0.23, 2.93 ] Heterogeneity: not applicable Test for overall effect: Z = 0.31 (P = 0.76) 2 Chlorhexidine in alcohol versus povidone-iodine in aqueo us solution Humar 2000 92 88 -0.12 (0.7) 17.3 % 0.89 [ 0.22, 3.50 ] Mimoz 1996 87 71 -1.34 (0.5) 34.0 % 0.26 [ 0.10, 0.70 ] Vall s 2008 226 97 -0.2 (0.65) 20.1 % 0.82 [ 0.23, 2.93 ] Subtotal (95% CI) 405 256 71.4 % 0.49 [ 0.25, 0.95 ] Heterogeneity: Chi 2 = 2.91, df = 2 (P = 0.23); I 2 =31% Test for overall effect: Z = 2.10 (P = 0.036) 3 Chlorhexidine in alcohol versus povidone-iodine in alcoh ol Mimoz 2007 242 239 -0.89 (1) 8.5 % 0.41 [ 0.06, 2.92 ] Subtotal (95% CI) 242 239 8.5 % 0.41 [ 0.06, 2.92 ] Heterogeneity: not applicable Test for overall effect: Z = 0.89 (P = 0.37) Total (95% CI) 858 592 100.0 % 0.53 [ 0.30, 0.94 ] Heterogeneity: Chi 2 = 3.49, df = 4 (P = 0.48); I 2 =0.0% Test for overall effect: Z = 2.17 (P = 0.030) Test for subgroup differences: Chi 2 = 0.58, df = 2 (P = 0.75), I 2 =0.0% 0.01 0.1 1 10 100 Favours chlorhexidine Favours povidone-iodine 74 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.3. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 3 All-cause mortality.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 3 All-cause mor tality Study or subgroup Chlorhexidine-based Povidone- iodine-based Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI 1 Chlorhexidine in aqueous solution versus povidone-iodin e in aqueous solution Vall s 2008 29/107 25/106 100.0 % 1.15 [ 0.72, 1.83 ] Subtotal (95% CI) 107 106 100.0 % 1.15 [ 0.72, 1.83 ] Total events: 29 (Chlorhexidine-based), 25 (Povidone-iod ine-based) Heterogeneity: not applicable Test for overall effect: Z = 0.59 (P = 0.56) 2 Chlorhexidine in alcohol versus povidone-iodine in aqueo us solution Vall s 2008 22/116 25/106 100.0 % 0.80 [ 0.48, 1.34 ] Subtotal (95% CI) 116 106 100.0 % 0.80 [ 0.48, 1.34 ] Total events: 22 (Chlorhexidine-based), 25 (Povidone-iod ine-based) Heterogeneity: not applicable Test for overall effect: Z = 0.84 (P = 0.40) Test for subgroup differences: Chi 2 = 1.04, df = 1 (P = 0.31), I 2 =3% 0.01 0.1 1 10 100 Favours chlorhexidine Favours povidone-iodine 75 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.4. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 4 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 4 Catheter colonisation Study or subgroup Chlorhexidine-based Povidone- iodine- based Risk Difference Weight Risk Difference n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI 1 Chlorhexidine in aqueous solution versus povidone-iodin e in aqueous solution Maki 1991 4/67 15/77 10.0 % -0.14 [ -0.24, -0.03 ] Vall s 2008 38/211 24/97 18.5 % -0.07 [ -0.17, 0.03 ] Subtotal (95% CI) 278 174 28.5 % -0.09 [ -0.17, -0.02 ] Total events: 42 (Chlorhexidine-based), 39 (Povidone-iod ine-based) Heterogeneity: Chi 2 = 0.89, df = 1 (P = 0.35); I 2 =0.0% Test for overall effect: Z = 2.38 (P = 0.017) 2 Chlorhexidine in alcohol versus povidone-iodine in aqueo us solution Humar 2000 31/92 24/88 12.5 % 0.06 [ -0.07, 0.20 ] Langgar tner 2004 11/45 16/52 6.7 % -0.06 [ -0.24, 0.11 ] Vall s 2008 34/226 24/97 18.9 % -0.10 [ -0.19, 0.00 ] Subtotal (95% CI) 363 237 38.1 % -0.04 [ -0.11, 0.03 ] Total events: 76 (Chlorhexidine-based), 64 (Povidone-iod ine-based) Heterogeneity: Chi 2 = 3.71, df = 2 (P = 0.16); I 2 =46% Test for overall effect: Z = 1.03 (P = 0.30) 3 Chlorhexidine in alcohol versus povidone-iodine in alcoh ol Mimoz 2007 28/242 53/239 33.4 % -0.11 [ -0.17, -0.04 ] Subtotal (95% CI) 242 239 33.4 % -0.11 [ -0.17, -0.04 ] Total events: 28 (Chlorhexidine-based), 53 (Povidone-iod ine-based) Heterogeneity: not applicable Test for overall effect: Z = 3.13 (P = 0.0017) Total (95% CI) 883 650 100.0 % -0.08 [ -0.12, -0.03 ] Total events: 146 (Chlorhexidine-based), 156 (Povidone-i odine-based) Heterogeneity: Chi 2 = 6.43, df = 5 (P = 0.27); I 2 =22% Test for overall effect: Z = 3.60 (P = 0.00032) Test for subgroup differences: Chi 2 = 1.96, df = 2 (P = 0.38), I 2 =0.0% -1 -0.5 0 0.5 1 Favours chlorhexidine Favours povidone-iodine 76 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.5. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 5 Catheter colonisation per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 5 Catheter colonisation per 1000 catheter-days Study or subgroup Chlorhexidine-based Povidone- iodine- based log [Risk Ratio] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI 1 Chlorhexidine in aqueous solution versus povidone-iodin e in aqueous solution Vall s 2008 211 97 -0.37 (0.28) 19.4 % 0.69 [ 0.40, 1.20 ] Subtotal (95% CI) 211 97 19.4 % 0.69 [ 0.40, 1.20 ] Heterogeneity: not applicable Test for overall effect: Z = 1.32 (P = 0.19) 2 Chlorhexidine in alcohol versus povidone-iodine in aqueo us solution Humar 2000 92 88 -0.3 (0.23) 28.7 % 0.74 [ 0.47, 1.16 ] Langgar tner 2004 45 52 -0.13 (0.32) 14.8 % 0.88 [ 0.47, 1.64 ] Mimoz 1996 87 71 -1.35 (0.4) 9.5 % 0.26 [ 0.12, 0.57 ] Vall s 2008 226 97 -0.47 (0.29) 18.1 % 0.63 [ 0.35, 1.10 ] Subtotal (95% CI) 450 308 71.1 % 0.64 [ 0.48, 0.85 ] Heterogeneity: Chi 2 = 6.49, df = 3 (P = 0.09); I 2 =54% Test for overall effect: Z = 3.06 (P = 0.0022) 3 Chlorhexidine in alcohol versus povidone-iodine in alcoh ol Mimoz 2007 242 239 -0.63 (0.4) 9.5 % 0.53 [ 0.24, 1.17 ] Subtotal (95% CI) 242 239 9.5 % 0.53 [ 0.24, 1.17 ] Heterogeneity: not applicable Test for overall effect: Z = 1.58 (P = 0.12) Total (95% CI) 903 644 100.0 % 0.64 [ 0.50, 0.81 ] Heterogeneity: Chi 2 = 6.78, df = 5 (P = 0.24); I 2 =26% Test for overall effect: Z = 3.65 (P = 0.00026) Test for subgroup differences: Chi 2 = 0.28, df = 2 (P = 0.87), I 2 =0.0% 0.01 0.1 1 10 100 Favours chlorhexidine Favours povidone-iodine 77 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 4.6. Comparison 4 Chlorhexidine versus povidone-iodine, Outcome 6 Insertion site infection.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 4 Chlorhexidine versus povidone-iodine Outcome: 6 Inser tion site infection Study or subgroup Chlorhexidine-based Povidone- iodine- based Mean Difference Weight Mean Difference N Mean(SD)[mean CFU] N Mean(SD)[mean CFU] IV,Fixed,95% CI IV,Fixed,95% CI Humar 2000 125 3.1 (21.2) 117 5.9 (28.1) 100.0 % -2.80 [ -9.10, 3.50 ] Total (95% CI) 125 117 100.0 % -2.80 [ -9.10, 3.50 ] Heterogeneity: not applicable Test for overall effect: Z = 0.87 (P = 0.38) Test for subgroup differences: Not applicable -100 -50 0 50 100 Favours chlorhexidine Favours povidone iodine Analysis 5.1. Comparison 5 Chlorhexidine (in aqueous solut ion) versus alcohol, Outcome 1 Catheter- related BSI.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 5 Chlorhexidine (in aqueous solution) versus a lcohol Outcome: 1 Catheter-related BSI Study or subgroup Chlorhexidine-based Alcohol based Risk R atio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Maki 1991 1/67 2/32 100.0 % 0.24 [ 0.02, 2.54 ] Total (95% CI) 67 32 100.0 % 0.24 [ 0.02, 2.54 ] Total events: 1 (Chlorhexidine-based), 2 (Alcohol based) Heterogeneity: not applicable Test for overall effect: Z = 1.19 (P = 0.23) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours chlorhexidine Favours povidone-iodine 78 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 5.2. Comparison 5 Chlorhexidine (in aqueous solution) versus alcohol, Outcome 2 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 5 Chlorhexidine (in aqueous solution) versus a lcohol Outcome: 2 Catheter colonisation Study or subgroup Chlorhexidine-based Alcohol-based Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Maki 1991 4/67 5/32 100.0 % 0.38 [ 0.11, 1.33 ] Total (95% CI) 67 32 100.0 % 0.38 [ 0.11, 1.33 ] Total events: 4 (Chlorhexidine-based), 5 (Alcohol-based) Heterogeneity: not applicable Test for overall effect: Z = 1.51 (P = 0.13) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol Favours chlorhexidine Analysis 6.1. Comparison 6 Povidone-iodine (in aqueous sol ution) versus alcohol, Outcome 1 Catheter- related BSI.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 6 Povidone-iodine (in aqueous solution) versu s alcohol Outcome: 1 Catheter-related BSI Study or subgroup Povidone- iodine-based Alcohol-based Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Maki 1991 5/77 2/32 100.0 % 1.04 [ 0.21, 5.08 ] Total (95% CI) 77 32 100.0 % 1.04 [ 0.21, 5.08 ] Total events: 5 (Povidone-iodine-based), 2 (Alcohol-base d) Heterogeneity: not applicable Test for overall effect: Z = 0.05 (P = 0.96) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Alcohol Povidone-iodine 79 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 6.2. Comparison 6 Povidone-iodine (in aqueous solution) versus alcohol, Outcome 2 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 6 Povidone-iodine (in aqueous solution) versu s alcohol Outcome: 2 Catheter colonisation Study or subgroup Povidone- iodine-based Alcohol-based Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI 1 Povidone-iodine in aqueous solution versus alcohol Maki 1991 15/77 5/32 93.4 % 1.25 [ 0.49, 3.14 ] Subtotal (95% CI) 77 32 93.4 % 1.25 [ 0.49, 3.14 ] Total events: 15 (Povidone-iodine-based), 5 (Alcohol-bas ed) Heterogeneity: not applicable Test for overall effect: Z = 0.47 (P = 0.64) 2 Povidone-iodine-impregnated adherent lm versus alcoho l Levy 1988 4/30 0/30 6.6 % 9.00 [ 0.51, 160.17 ] Subtotal (95% CI) 30 30 6.6 % 9.00 [ 0.51, 160.17 ] Total events: 4 (Povidone-iodine-based), 0 (Alcohol-base d) Heterogeneity: not applicable Test for overall effect: Z = 1.50 (P = 0.13) Total (95% CI) 107 62 100.0 % 1.76 [ 0.76, 4.09 ] Total events: 19 (Povidone-iodine-based), 5 (Alcohol-bas ed) Heterogeneity: Chi 2 = 1.77, df = 1 (P = 0.18); I 2 =43% Test for overall effect: Z = 1.31 (P = 0.19) Test for subgroup differences: Chi 2 = 1.64, df = 1 (P = 0.20), I 2 =39% 0.01 0.1 1 10 100 Favours alcohol Favours povidone-iodine 80 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 7.1. Comparison 7 Alcohol versus octenidine in alcohol, Outcome 1 Catheter-related BSI.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 1 Catheter-related BSI Study or subgroup Alcohol-based Octenidine Risk Ratio Weig ht Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Dettenkofer 2010 16/193 8/194 100.0 % 2.01 [ 0.88, 4.59 ] Total (95% CI) 193 194 100.0 % 2.01 [ 0.88, 4.59 ] Total events: 16 (Alcohol-based), 8 (Octenidine) Heterogeneity: not applicable Test for overall effect: Z = 1.66 (P = 0.097) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol-based Favours octenidine Analysis 7.2. Comparison 7 Alcohol versus octenidine in alc ohol, Outcome 2 Catheter-related BSI per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 2 Catheter-related BSI per 1000 catheter-days Study or subgroup Alcohol-based Octenidine log [Risk Ratio ] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI Dettenkofer 2010 193 194 0.78 (0.71) 100.0 % 2.18 [ 0.54, 8.77 ] Total (95% CI) 193 194 100.0 % 2.18 [ 0.54, 8.77 ] Heterogeneity: not applicable Test for overall effect: Z = 1.10 (P = 0.27) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol Favours octenidine 81 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 7.3. Comparison 7 Alcohol versus octenidine in alcohol, Outcome 3 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 3 Catheter colonisation Study or subgroup Alcohol-based Octenidine Risk Ratio Weig ht Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Dettenkofer 2010 28/157 13/165 100.0 % 2.26 [ 1.22, 4.21 ] Total (95% CI) 157 165 100.0 % 2.26 [ 1.22, 4.21 ] Total events: 28 (Alcohol-based), 13 (Octenidine) Heterogeneity: not applicable Test for overall effect: Z = 2.58 (P = 0.0099) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol-based Favours octenidine Analysis 7.4. Comparison 7 Alcohol versus octenidine in alc ohol, Outcome 4 Catheter colonisation per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 4 Catheter colonisation per 1000 catheter-days Study or subgroup Alcohol-based Octenidine log [Risk Ratio ] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI Dettenkofer 2010 157 165 0.8 (0.53) 100.0 % 2.23 [ 0.79, 6.29 ] Total (95% CI) 157 165 100.0 % 2.23 [ 0.79, 6.29 ] Heterogeneity: not applicable Test for overall effect: Z = 1.51 (P = 0.13) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol Favours octenidine 82 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 7.5. Comparison 7 Alcohol versus octenidine in alcohol, Outcome 5 Skin colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 5 Skin colonisation Study or subgroup Alcohol-based Octenidine Mean Difference Weight Mean Difference N Mean(SD)[mean CFU] N Mean(SD)[mean CFU] IV,Fixed,95% CI IV,Fixed,95% CI Dettenkofer 2010 178 100 (308.4) 187 21 (64.5) 100.0 % 79.00 [ 32.76, 125.24 ] Total (95% CI) 178 187 100.0 % 79.00 [ 32.76, 125.24 ] Heterogeneity: not applicable Test for overall effect: Z = 3.35 (P = 0.00081) Test for subgroup differences: Not applicable -200 -100 0 100 200 Favours alcohol Favours octenidine Analysis 7.6. Comparison 7 Alcohol versus octenidine in alc ohol, Outcome 6 Adverse effects.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 7 Alcohol versus octenidine in alcohol Outcome: 6 Adverse effects Study or subgroup Alcohol-based Octenidine Risk Ratio Weig ht Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Dettenkofer 2010 45/197 54/201 100.0 % 0.85 [ 0.60, 1.20 ] Total (95% CI) 197 201 100.0 % 0.85 [ 0.60, 1.20 ] Total events: 45 (Alcohol-based), 54 (Octenidine) Heterogeneity: not applicable Test for overall effect: Z = 0.93 (P = 0.35) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours alcohol Favours octenidine 83 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 8.1. Comparison 8 Chlorhexidine (in alcohol) pluspovidone-iodine (in aqueous solution) versus chlorhexidine (in alcohol), Outcome 1 Catheter colonisati on.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 8 Chlorhexidine (in alcohol) plus povidone-io dine (in aqueous solution) versus chlorhexidine (in alcohol) Outcome: 1 Catheter colonisation Study or subgroup CG and PI combined Chlorhexidine Risk Rati o Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Langgar tner 2004 2/43 11/45 100.0 % 0.19 [ 0.04, 0.81 ] Total (95% CI) 43 45 100.0 % 0.19 [ 0.04, 0.81 ] Total events: 2 (CG and PI combined), 11 (Chlorhexidine) Heterogeneity: not applicable Test for overall effect: Z = 2.25 (P = 0.025) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours CG-PI combination Favours chlorhexidine Analysis 8.2. Comparison 8 Chlorhexidine (in alcohol) plus povidone-iodine (in aqueous solution) versus chlorhexidine (in alcohol), Outcome 2 Catheter colonisati on per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 8 Chlorhexidine (in alcohol) plus povidone-io dine (in aqueous solution) versus chlorhexidine (in alcohol) Outcome: 2 Catheter colonisation per 1000 catheter-days Study or subgroup CG and PI combined Chlorhexidine log [Risk Ratio] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI Langgar tner 2004 43 45 -1.66 (0.58) 100.0 % 0.19 [ 0.06, 0.59 ] Total (95% CI) 43 45 100.0 % 0.19 [ 0.06, 0.59 ] Heterogeneity: not applicable Test for overall effect: Z = 2.86 (P = 0.0042) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours combined Favours Chlorhexidine 84 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 9.1. Comparison 9 Chlorhexidine (in alcohol) pluspovidone-iodine (in aqueous solution) versus povidone-iodine (in aqueous solution), Outcome 1 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 9 Chlorhexidine (in alcohol) plus povidone-io dine (in aqueous solution) versus povidone-iodine (in aqueous solution) Outcome: 1 Catheter colonisation Study or subgroup CG and PI combined Povidone-iodine Risk Ra tio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Langgar tner 2004 2/43 16/52 100.0 % 0.15 [ 0.04, 0.62 ] Total (95% CI) 43 52 100.0 % 0.15 [ 0.04, 0.62 ] Total events: 2 (CG and PI combined), 16 (Povidone-iodine) Heterogeneity: not applicable Test for overall effect: Z = 2.62 (P = 0.0088) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours CG-PI combination Favours povidone-iodine Analysis 9.2. Comparison 9 Chlorhexidine (in alcohol) plus povidone-iodine (in aqueous solution) versus povidone-iodine (in aqueous solution), Outcome 2 Catheter colonisation per 1000 catheter-days.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 9 Chlorhexidine (in alcohol) plus povidone-io dine (in aqueous solution) versus povidone-iodine (in aqueous solution) Outcome: 2 Catheter colonisation per 1000 catheter-days Study or subgroup CG and PI combined Povidone-iodine log [Ri sk Ratio] Risk Ratio Weight Risk Ratio N N (SE) IV,Fixed,95% CI IV,Fixed,95% CI Langgar tner 2004 43 52 -1.79 (0.58) 100.0 % 0.17 [ 0.05, 0.52 ] Total (95% CI) 43 52 100.0 % 0.17 [ 0.05, 0.52 ] Heterogeneity: not applicable Test for overall effect: Z = 3.09 (P = 0.0020) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours combined Favours povidone-iodine 85 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Analysis 10.1. Comparison 10 Sanosil (hydrogen peroxide and silver) versus water as adjunct to chlorhexidine 2% aqueous bath plus povidone-iodine aqueou s 10% scrub, Outcome 1 Catheter colonisation.

Review: Skin antisepsis for reducing central venous cathet er-related infections Comparison: 10 Sanosil (hydrogen peroxide and silver) vers us water as adjunct to chlorhexidine 2% aqueous bath plus pov idone-iodine aqueous 10% scrub Outcome: 1 Catheter colonisation Study or subgroup Sanosil Water Risk Ratio Weight Risk Ratio n/N n/N M-H,Fixed,95% CI M-H,Fixed,95% CI Yousefshahi 2013 26/113 29/136 100.0 % 1.08 [ 0.68, 1.72 ] Total (95% CI) 113 136 100.0 % 1.08 [ 0.68, 1.72 ] Total events: 26 (Sanosil), 29 (Water) Heterogeneity: not applicable Test for overall effect: Z = 0.32 (P = 0.75) Test for subgroup differences: Not applicable 0.01 0.1 1 10 100 Favours Sanosil Favours water A P P E N D I C E S Appendix 1. Glossary of terms (lay de nitions in the context of thi s review only) Colonisation : occupation by bacteria or other micro-organisms in a speci c body part or a device in the body without causing infection Erythema : redness Induration : a term usually used to describe the hardening of a small area o f the skin Infusates : liquid that is being infused through a device, such as a line, f rom the source (such as the uid bag) to the patient Nosocomial infection : also known as ahospital-acquired infection orHAI , an infection whose development is favoured by a hospital environment, such as one acquired by a patient during a hospita l visit or one developed among hospital staff. Such infectionsinclude fungal and bacterial infections and are aggravated by the redu ced resistance of individual patients.

Pathogenesis : the chain of events leading to the appearance of a disease or a m edical problem, described scienti cally in detail Placebo : a simulated or ’sham’ treatment that is designed to be indis tinguishable from the actual treatment in all aspects except fo r the active component tested Plasmapheresis : a medical procedure in which a person’s blood is channeled out of his body to a special ’ ltering machine’ and then returned to the body after the removal of the unwanted substa nce. It is used to treat a variety of medical problems in which unw anted substances, usually in the form of harmful antibodies, are pr oduced Purulence : the state where pus appears at or around a lesion such as a woun d Regimen : a systematic plan of single or multiple measures designed t o improve the health of a patient Single agent : the use of only one antiseptic agent A combination of agents : the use of more than one antiseptic agent together Transient ora : bacteria that occupy a speci c place in the body or a device for a sho rt-term period Subclavian vein : large blood vessels on each the side of the neck; commonly used a s a site for inserting a central venous catheter. 86 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Appendix 2. De nitions of infections linked to vascular access Table 1. De nitions of infections linked to vascular access (Pagani 2008) Type of infection Criteria Catheter colonisation A signi cant growth of a micro-organism (> 15 CFU) from the cathete r tip, subcutaneous segment or catheter hub in the absence of clinical signs of infecti on Exit-site/insertion site infection Microbiologically documented: exudates at catheter exit site yield a micro-organism with or without concomitant bloodstream infection. Clinically documented: erythema or induration within 2 cm of the catheter insertion site in the a bsence of associated bloodstream infection and without concomitant purulence Positive blood culture Micro-organism, potentially pathogenic, cultured from one or m ore blood culture Bloodstream infection Positive blood culture with a clinical sepsis (see below) Primary bloodstream infection Laboratory-con rmed bloodstream infection or clinical sepsis occurring without docu- mented infection Secondary bloodstream infection Laboratory-con rmed bloodstream infection secondary to anoth er documented infec- tion Clinical sepsis Requires one of the following with no other recognised cause: f ever (> 38° C), hypoten- sion (SBP < 90 mmHg), oliguria (< 20 ml/h); and all of the following : blood culture not performed or no organism detected in blood, no apparent in fection at another body site and clinical response to therapy following catheter remov al or change Catheter-associated bloodstream infection Primary bloodstream infection or clinical sepsis in the presence of an intravascular device Catheter-related bloodstream infection Laboratory-con rmed bloodstream infection in the presence of a n intravascular access:

at least 1 positive blood culture obtained from a peripheral v ein, clinical manifestation of infection and no apparent source of the bloodstream infectio n except the vascular access, and with 1 of the microbiological methods: a positive res ult of semi-quantitative (> 15 CFUs per catheter segment) or quantitative culture (> 10 3 CFU/catheter segment) with the same organism, paired quantitative blood cultures w ith a > 5:1 ratio device versus peripheral, differential time to positivity (blood cu lture obtained from a CVC is positive at least 2 h earlier than a peripheral blood culture) CFU : colony-forming units; CVC: central venous catheter; S BP: systolic blood pressure. 87 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Appendix 3. The Cochrane Central Register of Controlled Trials (CENTRAL) search strategy #1 MeSH descriptor: [Catheterization, Central Venous] explo de all trees #2 central next venous next catheter*:ti,ab,kw #3 central next venous next line*:ti,ab,kw #4 {or #1-#3} #5 MeSH descriptor: [Antisepsis] explode all trees #6 antisepsis:ti,ab,kw #7 MeSH descriptor: [Hand Hygiene] explode all trees #8 (handwash* or hand wash* or “hand hygiene”):ti,ab,kw #9 aseptic next technique*:ti,ab,kw #10 barrier next precaution*:ti,ab,kw #11 MeSH descriptor: [Anti-Infective Agents, Local] explode all trees #12 MeSH descriptor: [Chlorhexidine] explode all trees #13 MeSH descriptor: [Iodine] explode all trees #14 MeSH descriptor: [Povidone] explode all trees #15 MeSH descriptor: [Triclosan] explode all trees #16 MeSH descriptor: [Hexachlorophene] explode all trees #17 MeSH descriptor: [Cetrimonium Compounds] explode all tre es #18 MeSH descriptor: [Phenol] explode all trees #19 MeSH descriptor: [Hydrogen Peroxide] explode all trees #20 MeSH descriptor: [Alcohols] explode all trees #21 MeSH descriptor: [Soaps] explode all trees #22 (iodine* or povidone* or chlorhexidine or betadine or tricl osan or hexachlorophene or chloroxylenol or cetrimide or benza lkonium or benzylkonium or octenidine or phenol* or carbolic or “hydro gen peroxide” or alcohol or alcohols or antiseptic* or soap*):ti,ab,kw #23 skin near/3 disinfect*:ti,ab,kw #24 {or #5-#23} #25 {and #4, #24} in Trials Appendix 4. Ovid MEDLINE search strategy 1 exp Catheterization, Central Venous/ 2 central venous catheter*.tw.

3 central venous line*.tw.

4 or/1-3 5 exp Antisepsis/ 6 antisepsis.tw.

7 exp Hand Hygiene/ 8 (handwash* or hand wash* or hand hygiene).tw.

9 aseptic technique*.tw.

10 barrier precaution*.tw.

11 exp Anti-Infective Agents, Local/ 12 exp Chlorhexidine/ 13 exp Iodine/ 14 exp Povidone/ 15 exp Triclosan/ 16 exp Hexachlorophene/ 17 exp Cetrimonium Compounds/ 18 exp Phenol/ 19 exp Hydrogen Peroxide/ 20 exp Alcohols/ 21 exp Soaps/ 88 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. 22 (iodine* or povidone* or chlorhexidine or betadine or triclosan or hexachlorophene or chloroxylenol or cetrimide or benzal konium or benzylkonium or octenidine or phenol* or carbolic or hydrog en peroxide or alcohol or alcohols or antiseptic* or soap*).tw.

23 (skin adj3 disinfect*).tw.

24 or/5-23 25 4 and 24 26 randomized controlled trial.pt.

27 controlled clinical trial.pt.

28 randomi?ed.ab.

29 placebo.ab.

30 clinical trials as topic.sh.

31 randomly.ab.

32 trial.ti.

33 or/26-32 34 exp animals/ not humans.sh.

35 33 not 34 36 and/25,35 Appendix 5. Ovid EMBASE search strategy 1 exp central venous catheter/ 2 central venous catheter*.tw.

3 central venous line*.tw.

4 or/1-3 5 exp antisepsis/ 6 antisepsis.tw.

7 exp hand washing/ 8 (handwash* or hand wash* or hand hygiene).tw.

9 aseptic technique*.tw.

10 barrier precaution*.tw.

11 exp topical antiinfective agent/ 12 exp chlorhexidine/ 13 exp iodine/ 14 exp povidone/ 15 exp povidone iodine/ 16 exp triclosan/ 17 exp hexachlorophene/ 18 exp cetrimide/ 19 exp benzalkonium/ 20 exp octenidine/ 21 exp phenol/ 22 exp hydrogen peroxide/ 23 exp alcohol/ 24 exp soap/ 25 (iodine* or povidone* or chlorhexidine or betadine or triclo san or hexachlorophene or chloroxylenol or cetrimide or benzal konium or benzylkonium or octenidine or phenol* or carbolic or hydrog en peroxide or alcohol or alcohols or antiseptic* or soap*).tw.

26 (skin adj3 disinfect*).tw.

27 or/5-26 28 4 and 27 29 Randomized controlled trials/ 30 Single-Blind Method/ 31 Double-Blind Method/ 32 Crossover Procedure/ 89 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. 33 (random$ or factorial$ or crossover$ or cross over$ or cross-over$ or placebo$ or assign$ or allocat$ or volunteer$).ti,ab.

34 (doubl$ adj blind$).ti,ab.

35 (singl$ adj blind$).ti,ab.

36 or/29-35 37 exp animals/ or exp invertebrate/ or animal experiment/ o r animal model/ or animal tissue/ or animal cell/ or nonhuman/ 38 human/ or human cell/ 39 and/37-38 40 37 not 39 41 36 not 40 42 28 and 41 Appendix 6. EBSCO CINAHL Plus search strategy S1 (MH “Central Venous Catheters+”) S2 (MH “Catheterization, Central Venous+”) S3 TI central venous catheter* or AB central venous catheter* S4 TI central venous line* or AB central venous line* S5 S1 or S2 or S3 or S4 S6 TI antisepsis or AB antisepsis S7 (MH “Handwashing+”) S8 TI ( handwash* or hand wash* or hand hygiene ) or AB ( handwas h* or hand wash* or hand hygiene ) S9 TI aseptic technique* or AB aseptic technique* S10 TI barrier precaution* or AB barrier precaution* S11 (MH “Antiinfective Agents, Local+”) S12 (MH “Chlorhexidine”) S13 (MH “Iodine”) S14 (MH “Povidone-Iodine”) S15 (MH “Hexachlorophene”) S16 (MH “Benzalkonium Compounds”) S17 (MH “Phenols”) S18 (MH “Hydrogen Peroxide”) S19 (MH “Alcohols+”) S20 (MH “Soaps”) S21 TI iodine* or povidone* or chlorhexidine or betadine or tr iclosan or hexachlorophene or chloroxylenol or cetrimide or ben zalkonium or benzylkonium or octenidine or phenol* or carbolic or hydrog en peroxide or alcohol or alcohols or antiseptic* or soap* S22 AB iodine* or povidone* or chlorhexidine or betadine or tr iclosan or hexachlorophene or chloroxylenol or cetrimide or ben zalkonium or benzylkonium or octenidine or phenol* or carbolic or hydrog en peroxide or alcohol or alcohols or antiseptic* or soap* S23 TI skin N3 disinfect* or AB skin N3 disinfect* S24 S6 or S7 or S8 or S9 or S10 or S11 or S12 or S13 or S14 or S15 or S1 6 or S17 or S18 or S19 or S20 or S21 or S22 or S23 S25 S5 and S24 S26 MH “Clinical Trials+” S27 PT Clinical trial S28 TI clinic* N1 trial* or AB clinic* N1 trial* S29 TI ( singl* or doubl* or trebl* or tripl* ) and TI ( blind* or mask* ) S30 AB ( singl* or doubl* or trebl* or tripl* ) and AB ( blind* or mask* ) S31 TI randomi?ed control* trial* or AB randomi?ed control* tr ial* S32 MH “Random Assignment” S33 TI random* allocat* or AB random* allocat* S34 MH “Placebos” S35 TI placebo* or AB placebo* S36 MH “Quantitative Studies” S37 S26 or S27 or S28 or S29 or S30 or S31 or S32 or S33 or S34 or S35 or S36 90 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. S38 S25 and S37 Appendix 7. Risk of bias criteria 1. Was the allocation sequence randomly generated?

Low risk of bias The investigators describe a random component in the sequence generation process such as: referring to a random number table;using a computer random number generator; coin tossing; shuf ing car ds or envelopes; throwing dice; drawing of lots.

High risk of bias The investigators describe a non-random component in the seque nce generation process. Usually, the description would involve some systematic, non-random approach, for example: sequence genera ted by odd or even date of birth; sequence generated by some rul e based on date (or day) of admission; sequence generated by some rule based on hospital or clinic record number.

Unclear Insuf cient information about the sequence generation proces s to permit judgement of low or high risk of bias.

2. Was the treatment allocation adequately concealed?

Low risk of bias Participants and investigators enrolling participants could not foresee assignment because one of the following, or an equ ivalent method, was used to conceal allocation: central allocation (includ ing telephone, web-based and pharmacy-controlled randomisation); sequentially numbered drug containers of identical appearan ce; sequentially numbered, opaque, sealed envelopes.

High risk of bias Participants or investigators enrolling participants could p ossibly foresee assignments and thus introduce selection bia s, such as allocation based on: using an open random allocation schedule (e.g. a list o f random numbers); assignment envelopes used without appropriate safeguards (e.g. envelopes were unsealed or non-opaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.

Unclear Insuf cient information to permit judgement of low or high ri sk of bias. This is usually the case if the method of concealment i s not described or not described in suf cient detail to allow a de nit e judgement, for example if the use of assignment envelopes is described, but it remains unclear whether envelopes were sequentially n umbered, opaque and sealed.

3. Blinding - was knowledge of the allocated interventions a dequately prevented during the study?

Low risk of bias Any one of the following: • No blinding, but the review authors judge that the outcome and the outcome measurement are not likely to be in uenced by lack of blinding. 91 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. •Blinding of participants and key study personnel ensured, an d unlikely that the blinding could have been broken.

• Either participants or some key study personnel were not blin ded, but outcome assessment was blinded and the non-blinding o f others unlikely to introduce bias.

High risk of bias Any one of the following: • No blinding or incomplete blinding, and the outcome or outcome m easurement is likely to be in uenced by lack of blinding.

• Blinding of key study participants and personnel attempted, but likely that the blinding could have been broken.

• Either participants or some key study personnel were not blin ded, and the non-blinding of others likely to introduce bias.

Unclear Any one of the following: • Insuf cient information to permit judgement of low or high ri sk of bias.

• The study did not address this outcome.

4. Were incomplete outcome data adequately addressed?

Low risk of bias Any one of the following: • No missing outcome data.

• Reasons for missing outcome data unlikely to be related to tru e outcome (for survival data, censoring unlikely to be introduci ng bias).

• Missing outcome data balanced in numbers across intervention g roups, with similar reasons for missing data across groups.

• For dichotomous outcome data, the proportion of missing outcom es compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate .

• For continuous outcome data, plausible effect size (difference i n means or standardised difference in means) among missing outcomes not enough to have a clinically relevant impact on obser ved effect size.

• Missing data have been imputed using appropriate methods.

High risk of bias Either of the following: • Reason for missing outcome data likely to be related to true ou tcome, with either imbalance in numbers or reasons for missing data across intervention groups. • For dichotomous outcome data, the proportion of missing outcom es compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate. • For continuous outcome data, plausible effect size (difference i n means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in observed e ffect size.

• ‘As-treated’ analysis done with substantial departure of th e intervention received from that assigned at randomisation .

• Potentially inappropriate application of simple imputatio n.

Unclear Any one of the following: • Insuf cient reporting of attrition/exclusions to permit jud gement of low or high risk of bias (e.g. number randomised not s tated, no reasons for missing data provided). • The study did not address this outcome. 92 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. 5. Are reports of the study free of suggestion of selective outcome reporting?

Low risk of bias Either of the following: • The study protocol is available and all of the study’s prespeci ed (primary and secondary) outcomes that are of interest in the review have been reported in the prespeci ed way.

• The study protocol is not available but it is clear that the publ ished reports include all expected outcomes, including those th at were prespeci ed (convincing text of this nature may be uncommon).

High risk of bias Any one of the following: • Not all of the study’s prespeci ed primary outcomes have been r eported.

• One or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not prespeci ed. • One or more reported primary outcomes were not prespeci ed (unl ess clear justi cation for their reporting is provided, such as an unexpected adverse effect).

• One or more outcomes of interest in the review are reported inco mpletely so that they cannot be entered in a meta-analysis.

• The study report fails to include results for a key outcome that would be expected to have been reported for such a study.

Unclear Insuf cient information to permit judgement of low or high ri sk of bias. It is likely that the majority of studies will fallinto this category.

6. Other sources of potential bias Low risk of bias The study appears to be free of other sources of bias.

High risk of bias There is at least one important risk of bias. For example, the study:

• had a potential source of bias related to the speci c study desi gn used; • had extreme baseline imbalance; • has been claimed to have been fraudulent; or • had some other problem.

Unclear There may be a risk of bias, but there is either: • insuf cient information to assess whether an important risk of bias exists; or • insuf cient rationale or evidence that an identi ed problem w ill introduce bias. 93 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. C O N T R I B U T I O N S O F A U T H O R S Nai Ming Lai conceived the review question, coordinated and developed the review, performed the CENTRAL search, screened and selected the studies, entered the data, performed the analys es, developed the ’Summary of ndings’ tables, drafted the r esults, discussion, conclusions and abstract, edited the review, made an intellectu al contribution to the draft review writing, approved the nal version prior to submission and is guarantor for the review.

Nai An Lai made an intellectual contribution to the review writ ing and approved the nal version prior to submission.

Elizabeth O’Riordan made an intellectual contribution to the review writing and approved the nal version prior to submission.

Nathorn Chaiyakunapruk assisted in searching, provided som e full-text articles, made an intellectual contribution to the review writing and approved the nal version prior to submission.

Kenneth Tan participated in study selection, data entry and cro ss-checking, made an intellectual contribution to the review wri ting and approved the nal version prior to submission.

Jacqueline Taylor participated in study selection, data entry and cross-checking, edited the review draft and approved the na l version prior to submission.

Contributions of editorial base:

Nicky Cullum (Editor): edited the protocol and the review; advis ed on methodology, interpretation and review content; approved the nal review prior to submission.

Sally Bell-Syer and Gill Rizzello (Managing Editors): coordina ted the editorial process. Advised on methodology, interpretation and content. Edited the review.

Ruth Foxlee designed the search strategy, Amanda Briant and R eetu Child ran the searches and edited the search methods sectio n.

Denise Mitchell: assisted in searching and provided full-text articles D E C L A R A T I O N S O F I N T E R E S T Nai Ming Lai: none known.

Nai An Lai: none known.

Elizabeth O’Riordan: none known.

Nathorn Chaiyakunapruk: none known.

Jacqueline Taylor: none known.

Kenneth Tan: none known.

S O U R C E S O F S U P P O R T 94 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. Internal sources• No sources of support supplied External sources • This project was supported by the National Institute for Heal th Research, via Cochrane Infrastructure funding to Cochrane Wounds. The views and opinions expressed herein are those of the authors and do not necessarily re ect those of the Systematic Reviews Programme, NIHR, NHS or the Department of Health, UK .

D I F F E R E N C E S B E T W E E N P R O T O C O L A N D R E V I E W 1. We have amended the title of the review by omitting the phra se “during catheter insertion”. This was considered appropriate as all of our included studies examined skin antisepsis through out the in-dwelling period of the catheters with or without incl uding the period of insertion, and keeping the phrase “during catheter insertion” would be misleading. We have revised the text of o ur review from Background through to the Methods where appropriate to r e ect the change.

2. Under ’ Why it is important to do this review ’, we changed the original statements “However, in some studi es within the meta-analysis, a combination of antiseptics were used; for example, chlorhexi dine gluconate was sometimes evaluated in combination with al cohol.

There remain some uncertainties regarding the best agent or co mbination of agents to be used for skin antisepsis” to the following:

“However, the meta-analysis only evaluated chlorhexidine gl uconate and povidone-iodine as skin antiseptics, and some stud ies within it assessed a combination of arterial catheters as well as centr al and peripheral venous catheters. Some uncertainties remai n regarding the best agent, or combination of agents, for use as skin antis epsis for CVCs alone . . .”. This was because in this review, the studies included also used a combination of agents, and there were no st udies that assessed chlorhexidine gluconate separately, so the original statements did not justify the need for this review. Instead , the new statements more clearly re ect the differences betwee n this review and the earlier review mentioned.

3. Under ’ Types of studies ’, we added the following statements to further de ne the scop e of our selection of studies: “We excluded cross-over studies due to the possible contaminating effect of o ne intervention over another. We also excluded studies assessing CVCs for haemodialysis, as this is covered by another Cochrane revi ew ( McCann 2010 ).” 4. Under ’ Selection of studies ’, we omitted the reference to unpublished studies because we d id not nd any unpublished study in our search of the trials registries.

5. Under Electronic searches , we updated the CENTRAL and MEDLINE search strategies in line with the updated indexing terms in each database.

6. Under ’ Data extraction and management ’, we have re-written paragraph 2 to the following to re ect wha t was actually done in the review.

“We found a discrepancy between the number of catheter and the nu mber of patients in most studies, and this was due to multiple catheters being enrolled in some patients. However, we were u nable to limit our analysis to one catheter per participant as none of the individual studies provided the adjusted results based on o ne catheter per participant.” 7. We have added the section ’Unit of analysis issues’ to descri be how we would handle cluster-RCTs.

8. Under ’ Dealing with missing data ’, we revised our statement to include the absolute dropout ra te in our consideration in assessing the risk of attrition bias, as a number of included studies had very high absolute dropout rates. Our revised statements are shown below:

“To assess whether the dropout rate was signi cant, we inspect ed the absolute dropout rate and the dropout rate in relationto the event rates for the intervention and the comparison groups. If the a bsolute dropout rate was 20% or more, we judged the study to be at high risk for incomplete outcome data. If the dropout rate was lower than 20%, we used a ’worst-case-scenario’ method . . .” 9. Under ’ Assessment of heterogeneity ’, we revised the statement to re ect what was actually done in t he review, as follows:

“We found signi cant statistical heterogeneity in one analys is ( Analysis 4.4 ) and provided a plausible explanation the possible reason for heterogeneity in the form of risk of attrition bias in som e included studies. We decided to still provided the pooled est imate for this analysis and separated the studies based on the risk of attri tion bias in our pre-speci ed sensitivity analysis.” 95 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd. 10. Under ’ Sensitivity analysis ’, we re-wrote the section as follows to re ect the information that we gathered in the review and removed any mention of intention-to-treat analysis:

“We performed the following sensitivity analyses.

1. Best- and worst-case scenarios to assess the impact of missing data, as described in the section ’ Dealing with missing data ’.

2. Including and excluding studies with unclear and high risks o f selection bias, namely, studies with unclear or high risk for random sequence generation, allocation concealment or both.

Had suf cient data been available, we would have performed ad ditional sensitivity analyses to include and exclude studieswith methodological issues other than selection bias, such as a lack o f blinding to the participants, care givers or investigators,or where blinding was unclear.” 11. Under ’ Subgroup analysis and investigation of heterogeneity ’, we added the following statement to describe the separatio n of comparisons into subgroups based on the solution used, in res ponse to the referees’ comments in our draft review:

“In this review, we created subgroups of comparisons based on t he solution used, for example, a subgroup for chlorhexidine in aqueous solution versus povidone iodine in aqueous solution, and an other subgroup for chlorhexidine in alcohol versus povidone-i odine in aqueous solution.” I N D E X T E R M S Medical Subject Headings (MeSH) Anti-Infective Agents, Local [ ∗ therapeutic use]; Antisepsis [ ∗ methods]; Catheter-Related Infections [ ∗ prevention & control]; Central Venous Catheters [ ∗ adverse effects; microbiology]; Chlorhexidine [therapeutic u se]; Ethanol [therapeutic use]; Povidone-Iodine [ther- apeutic use]; Randomized Controlled Trials as Topic; Skin [ ∗ microbiology] MeSH check words Adult; Humans 96 Skin antisepsis for reducing central venous catheter-rela ted infections (Review) Copyright © 2016 The Cochrane Collaboration. Published by J ohn Wiley & Sons, Ltd.