Practicum Project Plan

FEATURE ARTICLE Exploring the Relationship Between Patient Call-Light Use Rate and Nurse Call-Light Response Time in Acute Care Settings HUEY-MING TZENG, PhD, RN JANET L. LARSON, PhD, RN, FAAN Generally speaking, call-light use rate is associated with how often patients or family visitors have unmet needs and require assistance. 1It has been reported that the three most common reasons for the use of call-lights are requests for pain medication, personal assistance, and bathroom assistance. 2 In a study, about 6% of the patients indicated that their use of the call-light was for an urgent situation. About 15% (14.3%) indicated that their use of the call light was always related to their safety, and 39% indicated that their use of the call-light sometimes related to safety. 2 Both patients and staff report that a long call-light response is perceived as a risk factor for falling. 3,4 In other words, patient call-light usage and nurse responsiveness to call lights are two intertwined concepts that could affect patients’ safety during hospital stays and ultimately be targeted for intervention to decrease the rate of falls and improve patient safety. However, little is known about the relationship betwe en call-light usage and call-light response time.

The National Quality Forum (NQF) suggested ap- proaches for falls’ assessment by including, for example, process measures to quantify the level of staff adher- ence to organizational policy that represents effective falls prevention practices and patient-centered measures as evidence of the degree to which patients’ values and preferences is respected. 5In a recent survey, discharged 138 CIN: Computers, Informatics, Nursing &March 2011 CIN: Computers, Informatics, Nursing &Vol. 29, No. 3, 138–143 &Copyright B2011 Wolters Kluwer Health | Lippincott Williams & Wilkins Patient call-light usage and nurse responsiveness to call lights are two intertwined concepts that could affect patients’ safety during hospital stays.

Little is known about the relationship between call- light usage and call-light response time. Conse- quently, this exploratory study examined the relationship between the patient-initiated call-light use rate and the nursing staff’s average call-light response time in a Michigan community hospital.

It used hospital archived data retrieved from the call-light tracking system for the period from February 2007 through June 2008. Curve estima- tion regression and multiple regression analyses were conducted. The results showed that the call- light response time was not affected by the total nursing hours or RN hours. The nurse call-light response time was longer when the patient call- light use rate was higher and the average length of stay was shorter. It is likely that a shorter length of stay contributes to the nursing care activity level on the unit because it is associated with a higher frequency of patient admissions/discharges and treatment per patient-day. This suggests that the nursing care activity level on the unit and number of call-light alarms could affect nurse call-light re- sponse time, independently of the number of nurses available to respond. KEY WORDS Accidental falls &Call light &Hospitals &Inpatients & Nurses &Safety Author Affiliations: Department of Nursing, School of Health Profes-sions and Studies, The University of Michigan-Flint, Flint (Dr Tzeng)and Division of Acute, Critical and Long-term Care Programs(Dr Larson), School of Nursing, University of Michigan, Ann Arbor. This project was partially supported by grant R03HS018258 from the Agency for Healthcare Research and Quality. The content is solelythe responsibility of the authors and does not necessarily represent theofficial views of the Agency for Healthcare Research and Quality. Corresponding author: Huey-Ming Tzeng, PhD, RN, Department of Nursing, School of Health Professions and Studies, The Univer-sity of Michigan-Flint, 303 E. Kearsley Street, WS White Building,Flint, MI 48502-1950 ([email protected]). DOI: 10.1097/NCN.0b013e3181fc41d9 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1 older patients reported that a major safety concern dur- ing hospitalization was the lack of staff response to call lights. 4In another study, the majority of the nursing staff (81.6%) recognized that call lights were meaningful, and half of the staff members believed that call lights mattered to patient safety and required nursing staff at- tention. In contrast, 44% thought that answering call lights prevented them from performing critical aspects of the nursing role. Therefore, the importance of call-light response time is not uniformly recognized and priori- tized by staff. 6 In previous research that used call-light tracking data, 7,8 more calls for assistance were associated with longer call- light response times and fewer fall-related patient injuries per 1000 patient-days. Specifically, longer call-light re- sponse times were associated with fewer total falls and fewer fall-related patient injuries per 1000 patient-days.

On the surface, these relationships are not logical, and the relationship between patien t call-light usage and nurse call-light response time remain s unclear. In fact, the call- light is a vital patient communication link during hos- pitalization, and it is one of the few means by which cognitively intact patients can exercise meaningful control over their care. 1 As a result, in this study, we proposed that longer length of stay, greater call-light use rate, and fewer nursing staff andRNswouldleadtolongerstaffcall-lightresponsetime (see Figure 1 for the proposed model, which was tested in this article). This exploratory study used archived hos- pital data to understand the relationship between the patient- or family-initiated call-light use rate per patient- day and the nursing staff’s average call-light response time in adult acute inpatient care settings. Three control variables were proposed, which were average length of stay, total nursing hours per patient-day (HPPDs), and total RN HPPDs. The potential effects of these control variables were supported by previous research, 5,9 and they were available within the existing database. Average length of stay served as a proxy for the mean acuity level of the patients cared for in each study unit. Total nursing hours and RN HPPDs were staffing-related measures. The research question was: What is the relationship between the call-light use rate per patient-day and the nursing staff’s average call-light response time, after con- trolling for length of stay, total nursing HPPDs, and total RN HPPDs? The present study was designed to explore issues from a system/macro approach by analyzing ar- chived, inpatient care unit–level data. Therefore, variables included in this report were limited by the availability of the study hospital’s preexisting database as a study limitation. METHODS Design This exploratory study was conducted in five adult acute inpatient care units in a community hospital in Michigan.

The project was approved by the institutional review board of the study hospital. We used archived hospital data for the period between February 2007 and June 2008 (17 months) for a secondary data analysis. Units involved included two medical units, one surgical unit, one medical- surgical combined unit, and one acute rehabilitation unit.

The unit of analysis was the unit-month; 85 unit-month data points were available for analyses.

Data Source and Collection Call-Light Data The information recorded from the patient room call- light tracking system was used in this study. The study hospital uses the Responder IV system (Rauland, Chicago, IL). There are multiple call priorities in the Re- sponder IV system. Typical calls made from the pillow speaker or call cord at the bed are considered normal calls. The call cord at the bed is used when a patient is lying on the bed or sitting on a bedside chair or com- mode. A normal call may be answered by a staff mem- ber and then cancelled at its station of origin, that is, FIGURE 1. The proposed model that was tested in this article. CIN: Computers, Informatics, Nursing &March 2011 139 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1 the patient room. A normal call may also be answered and may or may not be cancelled at a console, which is usually located at the nurses’ station. The Responder IV system has the ability to record all call data, including call-light usage and response time in seconds. These data can be analyzed in vario us reports (eg, the average response time to a call placed from the call cord at the bed in the inpatient unit A, for a specific period).

The average response time is re corded as either ‘‘voice response’’ (call was answered and cancelled at the console) or ‘‘staff response’’ (call was answered and cancelled from the patient room). According t o conversations with the site coordinator of the study hospital, very few calls were answered at the console ( G5%, estimated). As stated in the policies of the study hospital, nursing staff must go into the room to answer the lights. When the staff members an- swered the lights, they would turn the light off in the room.

Staff members often canceled the lights as soon as they went into the room to let the other staff members know that someone was attending to the request for assistance.

However, in some situations (but not very common), the light can be cancelled after th e patient’s need has been met or the task has been completed. During busy hours, call lights may be answered by a unit clerk first before being relayed to nursing staff. Call lights were seldom cancelled by a unit clerk at the console located in the nurses’ station; in some situations, however, call lights may be cancelled by the responsible staff members at the console before going into the room to answer the call.

In this study, only the normal calls were included in the analysis. Those generated through the call-light button attached to the patient room wall when a pa- tient is lying on the bed or sitting on a bedside chair or commode were included in the analysis. The tracking system generates the total call-light use (number of calls) by unit and by month (by unit-month; from the first day of the month to the last day of the month). The number of recorded days is documented; failures to re- cord (eg, due to power outages) are expected to be rare.

This information was used to justify the counts of the total call-light use for the defined period.

The computation for the call-light use rate per patient- day by unit-month was as follows: (the counts of the total call-light use / the number of the covered days) the total number of days for the month / the total patient-days for the month.

For purposes of this research, the response time was defined as the time from initiation of a call to the can- cellation of the call at the patient’s room, and the average call-light response time (in seconds) was generated from the call-light tracking system. The average response time to normal calls that were cancelled from the patient rooms was used and reported as staff response. In other words, the calls cancelled at the console were excluded in the staff response report. The equation for the average call-light response time by unit-month is as follows: (sum of the call-light response time for the calls in seconds)/(total call-light use).

Average Length of Stay For the purposes of quality control, the average length of stay was routinely collected in the patient management database. Length of stay was calculated by unit-month. The computation was as follows: total patient-days / total discharges for a given period.

The Total Nursing HPPDs and RN HPPDs The variable of the total nursing HPPDs was defined as the number of productive hours worked by nursing staff with direct care responsibilities per patient-day and the RN HPPDs as the hours worked by RNs. The National Database of Nursing Quality Indicators quarterly reports were obtained from the study hospital. The total nursing HPPDs and RN HPPDs were abstracted from the reports.

The quarterly data were then entered into each corre- sponding unit-month.

Data Analyses Abstracted data were entered and processed using SPSS 16.0 in Windows (SPSS Inc, Chicago, IL). Descriptive analyses were conducted for all study variables. Pearson correlation analyses were conducted to understand the relationships among study variables. Curve estimation regression models were used to investigate the relation- ships between the call-light response time (as the de- pendent variable) and the predictor variables, including the call-light use rate, length of stay, total nursing HPPDs, and RN HPPDs. Each relationship between the dependent ( Y) and independent ( X) variables was tested using four curve estimation regression models, includ- ing a linear model whose equation is Y=b0+( b1* X) (the values were modeled as a linear function of the in- dependent variable), a logarithmic model whose equa- tion is Y =b0+( b1 * ln( X)), an inverse model whose equation is Y =b0+( b1/ X), and a quadratic model whose equation is Y=b0+( b1* X) + (b2 * X2).10 Note that these four models were chosen based on the preliminary examination of the scatterplots between the dependent variable and the independent variables. The anal- ysis of variance for each curve estimation model was con- ducted, and the best curve estimation regression model for each independent variable was determined based on the significance of Fvalues. The independent variables were then transformed accordingly as needed. All three control variables were first entered into the initial multiple regression model at the same time, and then the call-light use rate was entered into the final model. Standardized 140 CIN: Computers, Informatics, Nursing &March 2011 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1 regression coefficients were used for interpretation pur- poses; "was set at .05. RESULTS Descriptive characteristics of the study variables are pre- sented in Table 1. The average c all-light response time was 3 minutes 2 seconds (SD, 66.11 seconds; range, 87–367 seconds). For exploratory purposes, Pearson correlation analyses were conducted among study variables (Table 2).

The data fit an inverse model [ Y =b0+( b1/ X)] for the average call-light use rate ( R2=0.06, F=4.93, P= .03, b0 = 206.06, b1= j85.24), a linear model [ Y=b0+ (b1* X)] for length of stay ( R2= 0.20, F= 20.55, PG.01, b0 = 235.53, b1= j9.56), a linear model for total nurs- ing HPPDs ( R2= 0.10, F=9.00, PG.01, b0= j4.62, b1 = 22.89), and the linear model for total RN HPPDs (R2=0.06, F=5.48, P= .02, b0 = 39.04, b1=30.05). The monthly call-light use rat edataweretransformed. ThemultipleregressionanalysesareshowninTable 3.

The initial multiple regression model (including only three control variables of total nursing HPPDs, total RN HPPDs, and length of stay) explains 20.4% of the variance of the average call-light response time, where length of stay was a significant predictor of the average call- light response time. The final regression model included all three control variables and the call-light use rate per patient-day (in- verse). This model explains 26.9% of the variance of the average call-light response ti me. The average call-light re- sponse time was higher, when the average length of stay was shorter and the call-light use rate was higher. Note that the relationship ( "coefficient) is negative because the inverse value was used for the c all-light use rate. In com- parison with the initial model, the final model was statisti- cally improved (change in R2= 0.071, F=8.20, PG.01). DISCUSSION The major findings were that st aff call-light response time was longer when the length of stay was shorter and when the call-light use rate was high er. In addition, the call-light response time was not affected by the number of nurses and number of RNs present (Figure 2). This result is con- sistent with previous research, 7,8 where higher patient call-light use rate was associ ated with longer staff response time to call lights.

The observed relationship between length of stay and call-light response time was not consistent with our pro- posed model. We initially thought that a longer length of stay represented a higher level of patient disease acuity and that this would have a negative effect on call-light response time. But the opposite was true, although a previous study concluded that length of stay rises proportionally with pa- tient disease acuity. 11 It is likely that a shorter length of stay contributes to a higher le vel of nursing care activities on the unit because it is associated with a higher frequency of patient admissions and discharges, procedures, treat- ment, and patient education p er patient-day. The staff members in such units are busier, and that might result in a longer call-light response time. Alternatively, it is pos- sible that a shorter average length of stay implies a health- ier patient population, and the staff members in such units may perceive that that patie nts’ needs are not so urgent, and that might result in a longer call-light response time.

However, little research has substantiated the relation- ship between length of stay and patient disease acuity, considering the impact of health insurance in the United Ta b le 1 Descriptive Characteristics of the Study Variables (n = 85 unit-months) Variable Mean SD Range Call-light response time, seconds 181.99 66.11 87–367 Call-light use rate per patient-day 4.59 1.86 0.95–7.77 Length of stay, d 5.60 3.08 3.51–14.35 Total nursing HPPDs 8.15 0.90 5.87–9.71 Total RN HPPDs 4.76 0.55 3.27–5.82 Ta b le 2 Summary of the Pearson Correlation Analyses (n = 85 unit-months) Variable/Correlation Coefficient Call-Light Response Time Call-Light Use Rate (Inverse) Length of Stay Total Nursing HPPDs Total RN HPPDs Call-light response time, s — Call-light use rate per patient-day (inverse) j0.24 a — Length of stay, d j0.56 b j0.12 — Total nursing HPPDs 0.31 b j0.12 j0.60 b — Total RN HPPDs 0.25 a 0.05 j0.62 b 0.91 b — aPG.05. bPG.01. CIN: Computers, Informatics, Nursing &March 2011 141 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1 States as an example (eg, Medicare and Medicaid as government-sponsored health insurance programs, Blue Cross Blue Shield of Michigan as a nonprofit insurance corporation). A recent study compared two neurosur- gery patient groups, those receiving the Massachusetts’ health insurance mandate (MassHealth) and those re- ceiving Commonwealth Care as a subsidized insurance program. In comparison with Commonwealth Care pa- tients, the length of stay for MassHealth patients was significantly longer, although disease acuity was signifi- cantly lower than the average. 12 Thus, more research is needed to further verify whether a shorter length of stay indeed implies a higher level of nursing care activities or a lower level of patient disease acuity in acute inpatient care settings.

Although the multiple regression analysis did not find total nursing HPPDs and total RN HPPDs as significant predictors of staff call-light response time (Table 3), the correlation analyses showed that when total nursing HPPDs and total RN HPPDs were higher, staff call-light response time was longer (Table 2). This can be explained by the significant relationsh ip between total nursing hours and length of stay; it is likely that the colinearity between length of stay and total nursing and RN HPPDs was the reason why total nursing HPPDs and total RN HPPDs were not significant predictors of staff’s response time to call lights in the multiple regr ession models. In addition, under the assumption that a shorter length of stay implies a higher level of nursing care activities, when the average length of stay was shorter, a unit manager could have scheduled more nursing and RN HPPDs to accommodate the increased workload. This is another possible explana- tion for why the call-light response time was not affected by the number of nurses and number of RNs present. Ta b l e 3 Summary of the Multiple Regression Models a: Predicting Average Nurse Call-Light Response Time by Total Nursing HPPDs, Total RN HPPDs, Length of Stay, and Call-Light Use Rate Per Patient-Day (n = 85 unit-months) Model Summary R2 Adjusted R2 Change in R2 FValue for the Change (df =1/ df =2) Pof the FValue Model 1 0.233 0.204 0.233 8.193 (3/81) .001 a Model 2 0.304 0.269 0.071 8.199 (1/80) .005 a Model/Predictors Nonstandard Coefficient: B(SE) Standard Coefficient: " tP 95% CI for B Model 1 Constant 187.31 (77.54) 2.42 .02 b 33.04/341.59 Total nursing HPPDs 32.18 (16.93) .44 1.90 .06 j1.51/65.87 Total RN HPPDs j45.55 (28.52) j.38 j1.60 .11 j102.29/11.19 Length of stay j9.12 (2.48) j.43 j3.68 .001 a j14.05/ j4.18 Model 2 Constant 259.25 (78.44) 3.31 .001 a 103.15/415.35 Total nursing HPPDs 11.83 (17.72) .16 0.67 .51 j23.42/47.09 Total RN HPPDs j17.71 (29.01) j.15 j0.61 .54 j75.44/40.01 Length of stay j10.60 (2.43) j.49 j4.36 .001 a j15.44/ j5.76 Call-light use rate d j106.76 (37.28) j.30 j2.86 .001 a j180.96/ j32.56 aPredictors for model 1: length of stay, total nursing HPPDs, and total RN HPPDs; predictors for model 2: length of stay, total nursing HPPDs, total RN HP PDs, and call-light use rate (inverse).bPG.01. cPG.05. dBased on the finding of the curve estimation regression models, call-light use rate data were transformed, and the inverse data were used in the regres sion models.

FIGURE 2. The tested model. 142 CIN: Computers, Informatics, Nursing &March 2011 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1 Limitations The scope of this study is limited to five adult acute in- patient care units in a community hospital located in the Midwestern region of the United States, reducing the abil- ity to generalize the findings. The call-light tracking system adopted in the study hospital cannot distinguish whether the call light is turned off before or after the reason for the call light was resolved and whether nursing staff effectively resolved the reason for the call light. In addition, this study investigated only the information recorded from the pa- tient room call-light system. The call-light system adopted by this study hospital does not have the capability to tell nursing staff the urgency level of each call light. The nurse must go into a patient’s room to determine the urgency level. The information recorded by the call-light systems in the bathrooms was not the focus of this study and should be investigated in future research. In addition, nursing staff cannot physically attend to more than one patient’s call at the same time. Consequently, we assume that the efficiency of the teamwork among a patient’s responsible nurse and unlicensed assisted personnel may contribute to the response time to each call light. Whether the call-light button is placed at a convenient location for the patient may also contribute to the patient call-light use rate per patient-day. In addition, nursing staff’s priorities among their assigned tasks may be a critical factor determining their call-light response time. These issues, not addressed in this article, are study limitations. CONCLUSION This study suggests that the levels of patient disease acuity and nursing care activities on the unit (as measured by length of stay) and number of c all-light alarms could affect nurse call-light response time, independently of the number of nurses available to respond. As for the practical im- plication for nursing execu tives and unit managers, when benchmarking staff call-light response time, it is essential to consider an inpatient care unit’s average length of stay and patient call-light use rate. As for future research di- rections, multihospital studies are needed to demonstrate whether this study’s conclusion holds across hospitals (eg, ownership, teaching status, bed size), types of units (eg, step- down, acute, subacute units), and different patient popu- lations (eg, percentage of cognitively intact vs impaired patients). It is also importa nt to understand whether nursing staff’s perceived levels of ca ll-light fatigue and teamwork among nursing colleagues are significant determinants of staff call-light response time. REFERENCES 1. Deitrick L, Bokovoy J, Stern G, Panik A. Dance of the call bells: usingethnography to evaluate patient satisfaction with quality of care.JNursCareQual . 2006;12(4):316–324. 2. Tzeng HM. Perspectives of patient and families about the nature ofand reasons for call light usage and staff response time. Medsurg Nurs . 2011. In Press. 3. Tzeng HM, Yin CY. The extrinsic risk factors for inpatient falls inhospital patient rooms. J Nurs Care Qual . 2008;23(3):234–242. 4. Tzeng HM, Yin CY. Perspectives of recently discharged patientson hospital fall-prevention programs. J Nurs Care Qual . 2009;24(1): 42–49.5. National Quality Forum (NQF). Safe ty practice 33: falls prevention. In: Safe Practices for Better Healthc are—2009 Update: A Consensus Report . Washington, DC: National Quality Forum; 2009:319–324. 6. Tzeng HM. Perspectives of staff nurses toward patient- and family-initiated call light usage and response time to call lights. Appl Nurs Res . 2011. In Press. DOI: 10.1016/j.apnr.2009.03.003. 7. Tzeng HM, Yin CY. Do call light use and response time correlatedwith inpatient falls and inpatient dissatisfaction? J Nurs Care Qual . 2009;24(3):232–242.8. Tzeng HM, Yin CY. Relationship between call light use and responsetime and inpatient falls in acute care settings. J Clin Nurs . 2009;18(23): 3333–3341.9. Currie L. Fall and injury prevention. In: Patient Safety and Quality: An Evidence-Based Handbook for Nurse (AHRQ Publication 08-0043). Hughes RG, ed. Rockville, MD: Agency for HealthcareResearch and Quality, US Department of Health and Human Ser-vices; 2008:195–250. 10. SPSS Inc. Curve estimation. In: SPSS Base 16.0 User’s Guide . Chicago, IL: SPSS Inc; 2007:455–459. 11. Mahmoud NN, Turpin RS, Yang G, Saunders WB. Impact of surgical site infections on length of stay and costs in selected colorectal pro-cedures. Surg Infect (Larchmt) . 2009;10(6):539–544. 12. Burrows AM, Moser RP, Weaver JP, Litwin DE, Pilitsis JG. Massachusetts health insurance mandate: effects on neurosurgicalpractice. J Neurosurg . 2010;112(1):202–207. CIN: Computers, Informatics, Nursing &March 2011 143 Copyright @ 201 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 1