Practicum Project Plan

JONA Volume 43, Number 5, pp 302-307 CopyrightB2013 Wolters Kluwer Health | Lippincott Williams & Wilkins THE JOURNAL OF NURSING ADMINISTRATION Expanding Potential of Radiofrequency Nurse Call Systems to Measure Nursing Time in Patient Rooms Linda Fahey, DNP, RN Karen Dunn Lopez, PhD, RNJudith Storfjell, PhD, RN Gail Keenan, PhD, RN OBJECTIVE:The objective of this study was to deter- mine the utility and feasibility of using data from a nurse call system equipped with radiofrequency iden- tification data (RFID) to measure nursing time spent in patient rooms.

BACKGROUND:Increasing the amount of time nurses spend with hospitalized patients has become a focus after several studies demonstrating that nurses spend most of their time in nondirect care activities rather than delivering patient care. Measurement of nursing time spent in direct care often involves labor- intensive time and motion studies, making frequent or continuous monitoring impractical.

METHODS:Mixed methods were used for this de- scriptive study. We used 30 days of data from an RFID nursecallsystemcollectedon1unitinacommunity hospital to examine nurses time spent in patient rooms.

Descriptive statistics were applied to calculate this per- centage by role and shift. Data technologists were sur- veyed to assess how practical the access of data would be in a hospital setting for use in monitoring nursing time spent in patient rooms.

RESULTS:The system captured 7393 staff hours. Of that time, 7% did not reflect actual patient care time, so these were eliminated from further analysis. Theremaining 6880 hours represented 91% of expected worked time. RNs and nursing assistants spent 33% to 36% of their time in patient rooms, presumably providing direct care.

CONCLUSIONS:Radiofrequency identification data technology was found to provide feasible and accurate means for capturing and evaluating nursing time spent in patient rooms. Depending on the outcomes per unit, leaders should work with staff to maximize pa- tient care time.

Over the last decade, concerns about patient safety and, specifically, high mortality on medical-surgical units have spurred an intense focus on the quality of nursing care delivery. 1Recently, hospital readmission rates have been added to the list of concerns. Research supports that the amount of direct nursing care is a key factor in realizing positive patient outcomes. 2Increas- ing nursing direct care time has been supported in redesign efforts through theTransforming Care at the Bedsideprojects. 1Current methods of measur- ing direct care nursing time involve resource-intense time and motion studies or interrupt patient care to gather self-reported activities, making it difficult to measure performance and effort in this area. Simple cost-effective tools are needed for monitoring direct nursing time to support the testing of new models that will increase these value-added services.

Radiofrequency identification data (RFID), being deployed as part of many nurse call systems, could provide a solution to this measurement challenge. Al- though these automated systems cannot provide infor- mation about specific nursing activities being performed, they can provide information about nursing time spent in direct patient care areas. Greater than 90% of time spent in patient rooms involves the delivery 302JONA Vol. 43, No. 5 May 2013 Author Affiliations:Vice President and Chief Nursing Officer (Dr Fahey), Decatur Memorial Hospital, Illinois; Assistant Professor (Dr Dunn Lopez), College of Nursing, Department of Health Sciences, University of Illinois, Chicago; Senior Vice President and Chief Nursing Officer (Dr Storfjell), Loma Linda University Medical Center, California; Associate Professor and Director, Nursing Informatics Initiative (Dr Keenan), College of Nursing, Department of Health Sciences, University of Illinois, Chicago.

The authors declare no conflicts of interest.

Correspondence: Dr Fahey, Decatur Memorial Hospital, 2300 N Edward St, Decatur, IL 62526 ([email protected]).

DOI:10.1097/NNA.0b013e31828eebe1 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. of nursing care, and percentages of time in patients’; rooms closely match percentages of direct care mea- sured on previous studies. 3,4 Therefore, the time nurses spend in patient rooms can serve as a proxy measure- ment for direct care time. This study was designed to determine the utility and feasibility of using RFID to measure nursing time spent in patient rooms. For proj- ect purposes, key terms are defined in the Figure 1. Literature Review There have been a number of studies conducted to measure nursing time spent providing direct patient care. 3,4,6-20 Sampling techniques and observation methods vary. Twelve noted studies involved time and motion analysis with independent observers collecting data. 4,6-11,14-16,18,20 Direct observation, although subjective, has been typically used to document activities in time and motion studies. Focus groups were used in 4 of these cited studies in combination with direct observation or self-reported measurement to determine direct care percentages.

12,13,17,19 One study used triggered self-reports with locations validated by RFID tracking. 3The amount of observation time in these studies ranged from 72 to 800 hours involving 1 to 30 observers. The average direct care percentage in the cited studies across a wide variety of practice set- tings was 31% (range, 15.9%-44%). 4,6-11,14,16,18,20 The noted time and motion studies involved sig- nificant human capital resources to collect data. There was no standard method for sampling, and no studies were identified where time was continuously moni- tored for all nurses over extended periods. Because nurses’ work has been found to be highly variable with multiple tasks over an entire shift often lasting less than 1 minute each, 9,10,21,22 limiting the sampling times may not provide an accurate picture of nurses’ time spent in direct patient care. Alternatively, ex- panding direct observation to larger segments of time is costly and time-consuming. The methods used to gather data for these studies may not be feasible for ongoing performance improvement efforts and may be time and cost prohibitive.

More recently, researchers have begun to use RFID to measure various aspects of nursing work. Measure- ment of staff response to patient call light requests and measurement of distance walked by nurses on medical surgical units have been accomplished in other studies supplemented by the use of RFID technology. 3,23 Sim- ulation of individual nursing workloads using RFID data combined with patient information from other sources has been used to test prototypes for pre- dicting individual nurse workloads. 24-26 Collectively, these studies demonstrate that RFID/nurse call systems can be used for purposes beyond measuring patient response times. By capturing nurse location, researchers Figure 1. Definitions for key terms.

JONA Vol. 43, No. 5 May 2013303 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. were able to examine various aspects of nursing work- flow including distance traveled, response to patient requests, and individual workload predictions. We proposed to add to these findings by examining the utility and feasibility of using data from a nurse call system in this study equipped with RFID to mea- sure nursing time spent in patient rooms.

Methods Mixed methods were used for this descriptive study with quantitative data extracted from an existing nurse call system to evaluate the utility for measuring nursing time spent in patient rooms. To evaluate feasibility of the application of RFID, qualitative data were col- lected through an open-ended survey about the prac- ticality of using the data and this approach.

Setting The setting was a medical unit at a 300-bed Midwestern community hospital with an average census of 33 pa- tients where a call system with RFID had been deployed for several years. The unit used a combination of care partners and team nursing astheir care delivery model with a staffing plan that consisted of approximately 60% RNs and 40% LPNs or nursing assistants.

Sample A convenience sample of 30 days of data generated from RFID sensors on the medical unit was used. Data included paired ingress and egress times for each event that sensors detected indicating staff movement on the unit. All staff on the unit was expected to wear trans- mission tags, and each patient room had a sensor to de- tect staff location. Initially, the data included 1 031 739 events, but 284 200 events remained after duplicates were removed. For the quantitative portion of the study, 2 specialists accessed and formatted the data.

Human Subjects This study was reviewed and granted exemption by both the university and hospital investigational re- view boards. Existing data with no staff identifiers were used for the quantitative analysis. For the qualitative survey, participation was voluntary, and no individ- ual identifiers were included in the results.

Data Preparation and Analysis Data were sorted by date and shift. Overlapping and a mix of 8- and 12-hour shifts necessitated a review by event to accomplish the shift assignments. Events that did not reflect actual patient care (discrepancies) were removed from any further calculations.

To determine if the system had adequately cap- tured all staff who worked, RFID data were com-pared with staff counts in administrative staffing data.

After determining how many staff were captured by the system,acalculationwascompletedtoestablishhow much of each individual’s time was captured. Ex- pected time was estimated by assigning 8.5 hours to each individual who worked a full shift and 4 hours to those who worked part of the shift. This was com- paredwiththetotaltimecapturedbyRFIDonthat shift for those individuals.

Two methods of analysis, a validated method and simplified method, were used to calculate percentage of time spent in the room. For the validated method (Figure 1), discrepancies were removed, and shifts were assigned manually to each healthcare worker. For the simplified method (Figure 1), shifts were not manually sorted, and discrepancies were not removed. Events were assigned based only on time. Total percentages of time in patient rooms as well as percentages by shift were calculated using both methods. Results The 284 200 events in the sample represented 7393 hours of staff time. Discrepancies constituted 7% of the time, leaving 6880 hours for further analysis (Table 1). When the RFID data were compared with administrative staffing data, 96% of the expected 1038 separate worked shifts were captured. In some cases, the RFID tags reflected different roles than staffing data (ie, RN in RFID, certified nursing assistant [CNA] on staffing sheet). For those staff detected by the sys- tem, 95% of expected worked hours were captured.

When numbers of staff detected and the amount of time detected per staff are combined and compared with the expected hours from administrative staffing sheets, 91% of expected nursing worked time was captured by the RFID system. The percentage time in patient rooms was 34% for all staff and was consistent for all roles except LPNs (Table 2). The percentage for Table 1.Discrepancy Events by Category for 30-Day Study Period Events n Hours Discrepancies only (1) Overlap at shift change 238 6 (2) Mislabeled case manager 698 58 (3) Periods detected where staff not providing care227 9 (4) Idle tags 864 439 Total 2027 512 Total Discrepancies + nondiscrepancies 284 200 7393 Discrepancies/total 1% 7% 304JONA Vol. 43, No. 5 May 2013 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. LPNs was higher, but their hours represented only 5% of the sample. The range for total percentages across shifts was 33% to 36% for day, evening, and night shifts and 34% to 35% for weekends versus weekdays.

When percentages for the 30-day period were calcu- lated using the simplified method, there was a 2% to 4% decrease in time spent in patient rooms.

Two data technologists involved in the extraction of RFID data for this study reported on the feasibility of routinely extracting these data. The technologists described the effort as more mental than physical, requiring 6 to 10 hours each to complete their por- tions of the data preparation. The temporal demand was similar to tasks they routinely performed. Most of the time involved initial learning and setup of the process, and staff expected time would be reduced significantly for subsequent access and analysis. Specific skills required to access these data included Microsoft SQL server experience, ability to write SQL code, and ability to create flat ASCII files. The primary barrier identified was the requirement to manually access data.

Discussion This study was designed to determine if RFID asso- ciated with nurse call systems could be used to mea- sure the percentage of time that nurses spent in patient rooms. The review of data demonstrated that the sys- tem captured 91% of worked time. The high capturerate exceeds the sampling rate generally used for pre- vious studies, so it should be adequate for measurement of time in patient rooms. 21 Although the capture rate was high, accuracy also needs to be considered. The original data downloaded from the server contained a large number of duplicates caused by a file configuration in the RFID server. This manual process was laborious, so work with the RFID vendor in establishing a file configuration that can be accessed without duplicates would be important for continuous monitoring. Of the other discrepancies, idle tags left in detectable areas when they were not being worn by staff accounted for 86% of the dis- crepancy time. To effectively use these data for mon- itoring time in patient rooms, staff needs to place their RFID tags in locations that are not detectable when inactive. Data screens eliminating events where no movement of staff has occurred for 1 hour could ensure compliance. There were also discrepancies in the number of people and roles compared with staffing sheets. Float staff did not always wear tags, and in some cases, staff borrowed a tag that did not match their job status. This resulted in a 4% decrease in total staff captured compared with staffing sheets. This could be prevented with a process to assign tags accurately for staff floating to the unit with periodic comparisons to administrative staffing data to ensure compliance.

After discrepancies were removed, the percentage of time in patient rooms was easily calculated using Table 2.Time in Patient Rooms aby Role and Time Period Time Period During 30-d Study Role Total Hours on Unit Hours in Patient Rooms % Time in Patient Rooms All 30 d RN 4445 1526 34 LPN 362 136 37 CNA 2074 710 34 Total 6880 2372 34 Weekdays RN 3267 1128 35 LPN 233 87 37 CNA 1485 509 34 Total 4985 1724 35 Weekends RN 1178 397 34 LPN 128 49 38 CNA 589 202 34 Total 1896 648 34 Day shifts RN 1658 585 35 LPN 0 0 0 CNA 1202 452 38 Total 2860 1038 36 Evening shifts RN 1702 576 34 LPN 151 68 45 CNA 515 150 29 Total 2368 793 34 Night shifts RN 1085 364 34 LPN 211 68 32 CNA 357 109 30 Total 1653 541 33 aPercentage of time in room = hours in patient rooms / total hours on unit. JONA Vol. 43, No. 5 May 2013305 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Microsoft Excel, a readily available spreadsheet program.

The percentage of time spent in patient rooms for this study of 34% was higher than the 31% average for direct care previously cited, but near the middle of the range (15%-44%) found in those studies. All nurs- ing time spent in the room is not necessarily direct care time, which could account for this difference. Calcu- lating percentages for single shifts was rarely included in previous studies and proved challenging for this study as well. Because of overlapping and varying shift lengths, each event was manually assigned to a shift based on the times that the employee had worked.

For example, events occurring during change of shift between 3:00 and 3:30 PM could be attributable to either day or evening shift. Manual sorting was time- consuming and may not be practical in a hospital set- ting. Designation of usual assigned shifts in the system for each individual might facilitate the shift sort in a less manual manner but may not be necessary to effectively use the data. When percentages were calculated using the simplified method with minimal data sorting (du- plicate removal only), total percentages decreased by 2%. Day and evening shift times calculated using stan- dard shift ranges, ignoring overlapping shifts and dis- crepancies, varied by less than 2%. For example, using the detailed analysis, evening staff spent 34% of their time in patient rooms, but only 32% when all events were counted between 3:00 and 10:59 PM including those involving day staff at change of shift. Nights reflected a 4% decrease using the simple analysis be- cause of idle tag discrepancies. If that single cause of discrepancy is eliminated, time using the simplified method would be very close to times determined in the validated method. The simplified method eliminated approximately 20 hours of work.

With the use of the simplified method, the time required for data analysis can be minimized contrib- uting positively to the feasibility of using RFID for routine measurement of nursing time in patient rooms.

The specialists who accessed data believed that the skills and time required to do so were similar to other tasks that they performed. The specialists saw manual access as a barrier to continuous use but did acknowl- edge that processes would be streamlined after initial setup. Although this access is possible if hospital re- sources include experts with the appropriate skills, the data may be more easily provided in a reporting format by vendors who market these systems to hospitals.

Nurse leaders need to understand where the data reside in their systems so they can direct requests for data to the appropriate parties. Although the RFID and call system are often purchased together, they can be sup- plied by different vendors with RFID data residing on a separate server. Knowing whom to ask and estab- lishing what information is needed are important stepsto gain access to usable RFID data. To ensure accurate measurement, staff need to wear tags that correctly identify their roles, and idle tags need to be stored away from sensors. Nurse leaders need to ensure that data used for analysis do not contain duplicates and are appropriately segmented into shifts for a partic- ular unit. The data should be provided in a format and at a frequency that can be easily used to monitor nursing practice at the unit level, and the use of these data needs to occur in a nonthreatening, professional practice culture where dialogue about nursing work and redesign at all levels are an accepted practice in performance improvement. Practice and Research Implications With routine access to RFID data, percentage of nurse time in the patient rooms can be used to measure the effectiveness of practice changes designed to increase the amount of time nurses spend providing direct care.

For the medical unit in this study, a 10% increase in time at the bedside would meanthateachpatientre- ceived an additional hour of direct nursing care per day. These hours could be used for patient education, pain management, patient and family support, or dis- charge planning among other direct care activities. The cost for adding 1 hour of nursing care per patient to that unit would normally exceed $250,000 (estimated at $21 per hour for 12 000 hours), a cost that could be decreased or eliminated if processes can be redesigned to change how nurses spend their time.

Access to this rich source of data provides mul- tiple research and practice improvement opportuni- ties. Information about patterns of nursing work by individual nurse or role, frequency of in-room visits, and time spent in all areas of the unit can contribute to the design of innovative care models with improved workflow and well-designed physical environments.

The use of RFID data alone is limited because it does not provide information about actual care provided or quality outcomes. However, if this information is combined with electronic health record data, times required for different types of patients and time re- quired to achieve positive patient outcomes may be- come more evident. For example, is there an optimal amount of time that nurses need to spend with end-of- life patients to achieve effective pain control? If used in conjunction with interoperable nursing care data coded with standardized terminologies across hospitals, bench- marking efficient use of nursing resources could go far beyond simply looking at nurse-to-patient ratios or other traditional measures.

The future potential for use of RFID data in under- standing nursing practice is tremendous. Many hospi- tals are just beginning their journey into the benefits 306JONA Vol. 43, No. 5 May 2013 Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. of RFID technology, whereas others have invested in call systems with RFID but have not yet accessed the full potential for measuring nursing work or response times. If nursing leaders are involved in the selection of this technology for their institutions and in activeconversation with RFID vendors about data potential and reporting capabilities, they can create an environ- ment where important information about nursing work is readily and efficiently available to the staff who design and deliver care.

References 1. Rutherford P, Lee B, Greiner A.Transforming Care at the Bedside. IHI Innovation Series White Paper. Boston, MA:

Institute for Healthcare Improvement; 2004. www.IHI.org.

Accessed October 18, 2012.

2. Institute of Medicine.The Future of Nursing: Leading Change, Advancing Health. Washington, DC: The National Academies Press; 2011.

3. Hendrich A, Chow M, Skierczynski B, Zhenqiang L. A 36- hospital time and motion study: how do medical-surgical nurses spend their time?Perm J. 2008;12(3):25-34. http:// www.ncbi.nlm.nih.gov/pmc/articles/PMC3037121/. Accessed October 7, 2012.

4. Upenieks V, Kotlerman J, Akhavan J, Esser J, Ngo M. As- sessing nursing staffing ratios.Policy Polit Nurs Pract. 2007; 8(1):7-19.

5. Nielsen J.Usability Engineering. San Diego, CA: Academic Press; 1993.

6. Banner L, Olney CM. Automated clinical documentation: does it allow nurses more time for patient care?Comput Inform Nurs. 2009;27(2):75-81.

7. Barnett GV. A new way to measure nursing: computer timing of nursing time and support of laboring patients.Comput Inform Nurs. 2008;26(4):199-206.

8. Capuano T, Bokovoy J, Halkins D, Hitchings K. Work flow analysis: eliminating nonYvalue-added work.JNursAdm. 2004; 34(5):246-256.

9. Cornell P, Riordan M, Herrin-Griffith D. Transforming nurs- ing workflow, part 2: the impact of technology on nurse ac- tivities.JNursAdm. 2010;40(10):432-439.

10. Cornell P, Riordan M, Townsend-Gervis M, Mobley R.

Barriers to critical thinking workflow interruptions and task switching among nurses.JNursAdm. 2011;41(10):407-414.

11. Desjardins F, Cardinal L, Belzile E, McCusker J. Reorga- nizing nursing work on surgical units: a time-and-motion study.

Can J Nurs Leadersh. 2008;21(3):26-38.

12. FitzGerald M, Pearson A, Walsh K, Long L, Heinrich N.

Patterns of nursing: a review of nursing in a large metropolitan hospital.J Clin Nurs. 2003;2(3):326-332.

13. Harrison L, Nixon G. Nursing activity in general intensive care.J Clin Nurs. 2002;11(2):158-167.14. Herdman TH, Burgess LPA, Ebright PR, et al. Impact of continuous vigilance monitoring on nursing workflow.J Nurs Adm. 2009;39(3):123-129.

15. Korst LM, Eusebio-Angeja AC, Chamorro T, Aydin CE, Gregory KD. Nursing documentation time during implemen- tation of an electronic medical record.JNursAdm. 2003; 33(1):24-30.

16. Lee T, Mills ME, Lu M. Impact of a nursing information sys- tem on practice patterns in Taiwan.Comput Inform Nurs.

2008;26(4):207-214.

17. Myny D, Van Goubergen D, Limre V, Gobert M, Verhaeghe S, Defloor T. Determination of standard times of nursing ac- tivities based on a Nursing Minimum Dataset.JAdvNurs.

2010;66(1):92-102.

18. Poon EG, Keohane CA, Bane A, et al. Impact of barcode medication administration technology on how nurses spend their time providing patient care.J Nurs Adm. 2008;38(12):541-549.

19. Storfjell JL, Omoike O, Ohlson S. The balancing act: patient care time versus cost.J Nurs Adm. 2008;38(5):244-249.

20. Upenieks VV, Akhavan J, Kottlerman J, Esser J, Ngo MJ.

Value-addedcare:anewwayofassessingnursingstaffingratios and workload variability.JNursAdm. 2007;37(5):243-252.

21. Cornell P, Herrin-Griffith D, Keim C, et al. Transforming nursing workflow, part 1.JNursAdm. 2010;40(9):366-373.

22. McGillis Hall L, Pedersen C, Fairley L. Losing the moment:

understanding interruptions to nurses’ work.JNursAdm. 2010; 40(4):169-176.

23. Lucero RJ, Haomiao J, Stone P.Information technology, nurse staffing, and patient needs.Nurs Econ. 2011;29(4):189-194.

24. Baker R, Tindell S, Behan D, Turpin P, Rosenberger J, Punnakitikashem P. Phase I: creating an electronic prototype to generate equitable hospital nurse-to-patient assignments.

Comput Inform Nurs. 2010;28(1):57-62.

25. Sundaramoorthi D, Chen V, Rosenberger J, Kim S, Buckley- Behan D. A data-integrated simulation model to evaluate nurse- patient assignments.Health Care Manag Sci. 2009;12(3):

252-268.

26. Sundaramoorthi D, Rosenberger JM, Buckley-Behan D, Kim S. Evidence-based nursing. Simulation and patient assign- ment: unearthing patterns.Nurs Manage. 2010;41(11):14-16.