ABBASS, IBRAHIM MS Pharm, RPh; HELTON, JEFFREY PhD; MHATRE, SHIVANI MS; SANSGIRY, SUJIT S. PhD
Author Affiliations: Fleming Centre for Healthcare Management (Mr Abbass) and School of Public Health, University of Texas (Mr Helton), and Division of Pharmacy Administration and Public Health, Department of Clinical Sciences and Administration, University of Houston, College of Pharmacy (Ms Mhatre and Dr Sansgiry), Houston, TX.
This study has no source of funding.
The authors have disclosed that they have no significant relationship with, or financial interest in, any commercial companies pertaining to this article.
Corresponding author: Sujit S. Sansgiry, PhD, Division of Pharmacy Administration and Public Health, Department of Clinical Sciences and Administration, College of Pharmacy, University of Houston, 1441 Moursund St, Houston, TX 77030 (email@example.com).
Since the advent of electronic health record (EHR) technology, managers may seek to understand the impact of EHR on nurse productivity and quality of care. The term EHR as used in this article defines the array of computer software applications commonly used to communicate orders for medical care, document pertinent facts regarding a patient’s medical history, and to disseminate results of diagnostic testing. It has been surmised that EHR technology could favorably affect healthcare operations in terms of efficiency and patient outcomes—both of which rely on nursing staff.1,2 This article explores the impact that EHR technology can have on nursing productivity.
One of the early studies that examined the impact of EHR on nurses’ time reported that using computers in hospitals was significantly associated with time saved for nurses that could be translated into financial savings.1 The time saved was attributed mainly to the flow of patient information between departments in the hospital. Minda and Brundage3 also reported significant time savings that were attributed to nurses’ use of a computerized system in the documentation processes in lieu of handwritten documentation. A randomized clinical trial reported reduction in nurses’ documentation time that equated to an increase in direct patient care (which includes nursing care activities such as administration of drugs, sampling for biochemistry, observation of patient, cleaning sheets, changing ventilator setting) after installing a computerized system in the ICU.4 Another study conducted in an ICU setting reported a significant reduction in documentation time (52 minutes in an 8-hour shift) after installing a more advanced documenting system that was suggestively associated, but not significantly (P = .085), with increase in direct patient care time.5 As computers evolved, additional capabilities were embedded into the computer system to provide a variety of functions to facilitate and improve care. Improved graphics were incorporated into the computer system to make them more user-friendly and enhance the flexibility in the design of the system to accommodate the workflow of nurses.5–7 The supplemental studies conducted to investigate the impact of these evolving systems on nursing time, medication errors, and quality measures did not find significant time savings or an increase in the time spent on direct patient care after using these computer systems, although other metrics (medication errors and quality measures) were significantly improved.8,9 Nevertheless, the results were not consistent across various hospital settings. Pierpont and Thilgen10 and Marasovic et al11 reported in two separate studies that nurses did not achieve significant time savings after the installation of a computerized system in an ICU. In summary, there is no conclusive evidence that EHR technologies improve efficiency in patient care delivery.
There are three concerns with the previous studies reported. First, as humans adapt to new technology, the time saved due to the implementation of these technology gets diverted to other resources. It is not clear if such diverted resources are used only to improve direct patient care. Even if time saved in one aspect of care were noted (eg, reduction of documentation time), that newly available time would not necessarily be translated into increased care provided to current patients or time allocation to new patients.3,5 Second, it is noteworthy to mention that previous studies were conducted in confined locations, usually specific hospital settings such as the ICU or other unit within the hospital. Third, there was variation between computer systems implemented at each hospital and their capabilities. It would be difficult to draw a decisive conclusion or make a general inference about the impact of EHR on nurses’ time and productivity in view of such variations. As a result, the underlying global impact of EHR on nursing time and productivity in US hospitals remains ambiguous and unexplored. A prior endeavor to arrive at a conclusion on the national impact of EHR on nurse productivity through extrapolating the results of some studies was criticized for methodological deficiencies.12,13 Thus, the aim of this study was to estimate the global impact of EHR on nurse productivity across US hospitals using national databases. The hypothesis examined here is that hospitals with higher levels of EHR utilization require fewer nurses to produce the same output than those with lower EHR usage levels.
Study Design and Data Acquisition
A retrospective cross-sectional study, where hospitals were the unit of analysis, was conducted to measure the influence of EHR on nurse productivity. Two sources of data were obtained for years 2007 and 2008, the American Hospital Association (AHA) survey and the Centers for Medicare & Medicaid Services (CMS) data.
The AHA survey is one of the most widely used surveys in healthcare that annually gathers information from more than 6500 hospitals, representing around 98% of hospitals in the United States. It captures hospital demographics, services, finances, and utilizations. The AHA data were utilized to gather demographic information regarding hospitals that encompassed hospital size, adjusted patient-days, and the degree of EHR for years 2007 and 2008. The CMS data were utilized to calculate the case-mix index (CMI) for the same period. Both databases were linked via Medicare number, which is a unique unidentified patient-specific number. This study was considered under the exempt category of the institutional review board.
Nursing productivity in the hospital setting was examined through a production function where input was defined as the number of full-time equivalent (FTE) RNs who produced output defined as adjusted patient-days, calculated using the formula: adjusted patient-days = inpatient-days + [inpatient-days × (outpatient revenue / inpatient revenue)]. Adjusted patient-days were used as a measure for hospital productivity that accounted for inpatient and outpatient workload. Nurse contribution to output was a function of many factors such as CMI, hospital size, academic status, and the presence and type of obstetric care unit. The degree of EHR was determined through the status of four essential EHR components: order entry, clinical decision support, storing patient data, and results management. Depending on the existence and implementation status of these applications, hospitals were assigned scores from 0 to 2, where 0 meant that the application did not exist, 1 meant the application existed but was partially implemented, and 2 meant the application existed and it was fully implemented. An EHR index was created by summing up all the numeric values that indicated both the presence and implementation status of EHR applications, which ranged from 0 to 8, where 0 meant no EHR application existed in the hospital and upper-range 8 meant that all four EHR essential applications existed in the hospital and they were fully implemented. The index was further divided into a binary variable for analyses where hospitals were categorized into high (EHR index ≥5) and low (EHR index <5) EHR penetration. The CMI represented the average diagnosis-related group (DRG) relative weight for a particular hospital and was calculated by summing the DRG weights for all Medicare discharges and divided by the number of discharges.
It was assumed that hospitals would have a nursing staff according to their workload (adjusted patient-days). Furthermore, it was assumed that EHR implementation relieved nurses from documentation and administrative tasks and thus enabled them to perform more patient care tasks (which includes all nursing care activities associated with the patient, which is assigned to the nurse), after controlling for important confounders such as CMI, adjusted patient-days, EHR application, number of beds, type of obstetric care unit, and affiliation with medical school. It should be noted that “nurses” in this study refers to only RNs since they are concerned with both technological implementation and patient care.
Sampling and Statistical Analyses
The AHA survey for years 2007 and 2008 encompassed data from 6312 and 6407 hospitals, respectively. The CMI from the CMS data included data from 3947 hospitals in 2007 and 3750 hospitals in 2008. After merging AHA and CMS databases by year and accounting for missing data, the final sample for the two combined years consisted of 3368 hospitals.
Both FTE and adjusted patient-days were log transformed because of lack of normality. Hospitals with very small adjusted patient-days (<2390) or high/low level of CMI (3 SDs from the mean) (>2.3 or <0.5) were excluded from the analysis to make the study sample more representative of an average hospital with respect to workload and CMI. A random-effects model was used to analyze the data, where the FTE for RNs was the dependent variable and other covariates were the independent variables. A random-effects model is a method of analyzing panel data that assumes that the individual-specific effects are uncorrelated with the regressors (observable variables). All analyses were performed using STATA version 11 (StataCorp, College Station, TX) at a priori significance level of .05.
The AHA data demonstrated an overall high EHR implementation levels within hospitals (Figure 1). Almost two-thirds of the respondent hospitals in both years (63.9% in 2007 and 68.4% in 2008) had a high EHR index (≥5). Only a small proportion of the respondent hospitals (6.5% and 5.4% in 2007 and 2008, respectively) did not have any component application of EHR, while 16.6% and 20.5% of respondent hospitals in 2007 and 2008, respectively, had all components of EHR fully implemented. Also, there was an 8.6% increase in EHR implementation from 2007 (n= 3025) to 2008 (n = 3284).
Table 1 provides the level of implementation status for EHR applications in hospitals for 2007 and 2008. The most frequent fully implemented component of EHR in the respondent hospitals was the result management applications (n = 1334 in 2007 and n = 1502 in 2008) followed by order-entry applications (1129 in 2007 and 1288 in 2008).
Table 2 provides the information on nurse FTE, adjusted patient-days, and CMI per hospital size. Most hospitals in the sample were categorized as small to medium bed size (≤200 beds) with a median RN FTE of 206 hours (mean, 358 [SD, 442.3]) and an average CMI of 1.4 (SD, 0.3) for years 2007 and 2008. The adjusted patient-days had a median of 64 340 days in 2007 and 65 547 days in 2008, while the median nurse FTE for the respondent hospitals in 2007 was 191 compared with 196 in 2008.
The study hypothesis was tested using the random-effects model and illustrated in Table 3. Independent variables in the model explained 87% of the overall variance of the dependent variable (log [RN FTE]). Hospitals with high EHR penetration were positively associated with more (log) RN FTE (coefficient = 0.2, P < .001), indicating that hospitals with higher EHR penetration utilize more RNs. Additionally, there was a statistically significant interaction between EHR index and CMI scores (coefficient = −0.1, P < .01), indicating that hospitals with higher CMI were more productive when using EHR compared with hospitals with lower CMI. The question of whether hospitals with higher workload (large adjusted patients-days) would benefit more or less from EHR application was examined through an interaction variable between EHR index and adjusted patient-days (log). The interaction variable was not statistically significant, and thus, we could not infer any relationship between EHR and hospitals productivity after adjusting for patient-days.
Although intuitively it is anticipated that as EHR use increases, the number of nurses employed by hospitals would decrease, this study appears to suggest that is not so. Instead, this work provides preliminary evidence that hospitals with lower EHR usage had fewer nurses employed compared with hospitals with higher EHR penetration. However, the EHR impact on nurse productivity appears moderated through CMI.
While the above findings appear to be counterintuitive, consideration of current literature provides some insight to this observation. Earlier studies reported that EHR use would relieve nurses from conducting logistical tasks.5 However, the time saved was not consistently utilized in either direct patient care activities or care for additional patients.8 Thus, unless the time released by higher EHR functionalities can be delegated to enhanced care for new patients, EHR may not create significant efficiency in reducing nurse FTE. Moreover, as more EHR applications are implemented in hospitals, more clinical nurses may be hired to oversee applications and to compensate for the temporary loss of productivity that occurs upon EHR implementation. Tips provided by the Healthcare Financial Management Association indicate that a large healthcare institution might need to contract with an additional 60 to 100 nurses during EHR implementation.14 It is unclear to what extent any differences in nurse productivity can be attributed to use of more EHR applications, taking into consideration that small hospitals may not demonstrate any difference in productivity. Thus, the premise that EHR may maximize nursing productivity cannot be supported based on findings from this study. Hospitals with high EHR penetration did not demonstrate more productivity as it was anticipated that higher EHR would lead to reduced number of nurses. Nevertheless, using EHR in hospitals with higher CMI could provide more efficiency compared with their counterparts with lower CMI. Hospital clinical managers considering implementation of new EHR technology would be well served by tempering expectations of improved nursing productivity as a means of generating returns on such investments. Implementing new technology in the absence of assessment of the work processes used with that technology could result in efficiency loss.15 Consequently, managers considering new EHR technology may at the same time need to evaluate underlying work processes for efficiency gains if the goal of an EHR implementation strategy includes improvement of labor efficiency. Since EHR usage tends to improve patient care documentation,16 perhaps that outcome and its contribution to better patient care and to improved demonstration of quality under future pay-for-performance systems would be better measures of the overall benefit of a hospital’s EHR implementation efforts. Thus, financial decision makers and healthcare providers could consider the use of EHR after reviewing their CMI and assessing its other advantages before implementation.
This study was limited by its lack of control and information on the impact of EHR on the quality of care. Cross-sectional studies lack ability to establish causal relationships. Validation of these results using a longitudinal study would provide more conclusive results. Discretion should be exercised when generalizing these figures to all US hospitals as only those that reported EHR implementation were considered, and hospitals with no or low EHR implementation might be reluctant to report their EHR status in the AHA survey. The study did not account for the impact of any other computer system existing in the hospital on nurse productivity.17 Furthermore, the study did not evaluate or control the impact of EHR on productivity of other healthcare personal. Future studies should be conducted to estimate the long-term financial impact of EHR on productivity with various outputs.
Hospitals with higher EHR penetration have more nurses employed compared with hospitals with low EHR penetration. This difference decreased for hospitals with higher case-mix index values. The role and implementation of EHR in hospitals need to be considered by justifying the long-term financial savings and the improvement in productivity of healthcare personnel.
1. Hendrickson G, Kovner CT. Effects of computers on nursing resource use. Do computers save nurses time? Comput Nurs. 1990; 8 (1): 16–22.
2. Jha A, DesRoches C, Kralovec P, Joshi M. A progress report on electronic health records in U.S. hospitals. Health Aff. 2009; 360 (16): 1628–1638.
3. Minda S, Brundage DJ. Time differences in handwritten and computer documentation of nursing assessment. Comput Nurs. 1994; 12 (6): 277–279.
4. Bosman RJ, Rood E, Oudemans-van Straaten HM, Van der Spoel JI, Wester JP, Zandstra DF. Intensive care information system reduces documentation time of the nurses after cardiothoracic surgery. Intensive Care Med. 2003; 29 (1): 83–90.
5. Wong DH, Gallegos Y, Weinger MB, Clack S, Slagle J, Anderson CT. Changes in intensive care unit nurse task activity after installation of a third-generation intensive care unit information system. Crit Care Med. 2003; 31 (10): 2488–2494.
6. Garrido T, Jamieson L, Zhou Y, Wiesenthal A, Liang L. Effect of electronic health records in ambulatory care: retrospective, serial, cross sectional study. BMJ. 2005; 330 (7491): 581.
7. Kritz S, Brown LS Jr, Chu M, John-Hull C, Madray C, Zavala R, Louie B. Electronic medical record system at an opioid agonist treatment programme: study design, pre-implementation results and post-implementation trends. J Eval Clin Pract. 2011. doi: 10.1111/j.1365-2753.2011.01664.x.
8. Menke JA, Broner CW, Campbell DY, McKissick MY, Edwards-Beckett JA. Computerized clinical documentation system in the pediatric intensive care unit. BMC Med Inform Decis Mak. 2001; 1: 3.
9. Abbass I, Mhatre S, Sansgiry S, Tipton J, Frost C. Impact and determinants of commercial computerized prescriber order entry on the medication administration process. Hosp Pharm. 2011; 46 (5): 341–348.
10. Pierpont GL, Thilgen D. Effects of computerized charting on nursing activity in intensive care. Crit Care Med. 1995; 23 (6): 1067–1073.
11. Marasovic C, Kenney C, Elliott D, Sindhusake D. A comparison of nursing activities associated with manual and automated documentation in an Australian intensive care unit. Comput Nurs. 1997; 15 (4): 205–211.
15. Brant-Lucich K. Principles of process redesign. In Langabeer J, ed. Performance Improvement in Hospitals and Health Systems. Chicago, IL: Healthcare Information Management Systems Society; 2009: 65–79.
16. Gunningberg L, Fogelberg-Dahm M, Ehrenberg A. Improved quality and comprehensiveness in nursing documentation of pressure ulcers after implementing an electronic health record in hospital care. J Clin Nurs. 2009; 18 (11): 1157–1564.
17. Dwibedi N, Sansgiry SS, Frost CP, et al.. Effect of bar-code-assisted medication administration on nurses’ activities in an intensive care unit: a time-motion study. Am J Health Syst Pharm. 2011; 68 (11): 1026–1031.
© 2012 Lippincott Williams & Wilkins, Inc.