Available online at www.sciencedirect.com
Nurs Outlook 62 (2014) 138e156
www.nursingoutlook.org
Assessing the relationships between nurse work hours/ overtime and nurse and patient outcomes: Systematic literature review Sung-Heui Bae, PhD, MPH, RNa,*, Donna Fabry, DNP, CNS, RNb b
a School of Nursing, University of Texas at Austin, Austin, TX School of Nursing, University at Buffalo, State University of New York, Buffalo, NY
article info
abstract
Article history: Received 4 July 2013 Revised 15 October 2013 Accepted 30 October 2013
Background: The effects of work hours/overtime on nurse and patient outcomes
Keywords: Adverse nurse outcomes Adverse patient outcomes Nurse overtime Nurse work hours
and specific components of work hours (per shift and per week) and overtime on these effects have not been systematically examined. Purpose: The purpose of this review was to systematically evaluate the effect of nurse overtime and long work hours on nurse and patient outcomes. Methods: An online search of six electronic bibliographic databases was conducted for research published from 2000 to 2013. Discussion: Twenty-one nurse outcome measures and 19 patient outcome measures were found in relationships with work hours and overtime. A total of 67 relationships to nurse outcomes and 41 relationships to patient outcomes were examined. Conclusions: The findings of this review suggested that evidence supporting positive relationships between working long hours and adverse outcomes to the nurses is strong. However, to make a conclusion of the positive relationship between long work hours and adverse patient outcomes, more evidence is needed. Cite this article: Bae, S.-H., & Fabry, D. (2014, APRIL). Assessing the relationships between nurse work hours/overtime and nurse and patient outcomes: Systematic literature review. Nursing Outlook, 62(2), 138-156. http://dx.doi.org/10.1016/j.outlook.2013.10.009.
In order to provide continuous nursing care, 24-hour coverage goes beyond the typical 9:00 a.m. to 5:00 p.m., Monday through Friday work day (Trinkoff et al., 2011). Traditionally, three 8-hour shifts have been used (Josten, Ng-A-Tham, & Thierry, 2003). Globally, various work scheduling methods are used to cover the 24 hours a day, 7 days a week needs for patient care. Some scheduling methods exceed traditional 8 work hours per day (e.g., 12 hours). When nurses work overtime, they often work long periods of time per day as well as per week. For example, in 2008, an estimated
3,063,162 licensed registered nurses (RNs) lived in the United States (U.S. Department of Health and Human Services, 2010). According to the 2010 U.S. Census (U.S. Census Bureau, 2010), licensed RNs comprised 1% of the total population (308,745,538). Fifty-four percent of the respondents to the 2008 National Sample Survey of Registered Nurses worked more than 39 hours per week, resulting in more than half of RNs working at least 2,000 hours per year in their principal nursing positions (U.S. Department of Health and Human Services, 2010). This is approximately 200 hours per
* Corresponding author: Sung-Heui Bae, School of Nursing, University of Texas at Austin, 1710 Red River, Austin, TX 78701. E-mail address: [email protected] (S.-H. Bae). 0029-6554/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.outlook.2013.10.009
Nurs Outlook 62 (2014) 138e156
year more than the average American worker (Fleck, 2009). If we consider those nurses who have more than one position, the total number of hours could be more than this. Furthermore, nurses are more likely to work 12-hour shifts in the United States (American Nurses Association [ANA], 2009). All this implies nurses not only work long hours, but they also return to work more quickly, resulting in less sleep and rest between shifts. Long work hours can be defined by per shift as well as per week. In a similar concept, overtime can be measured by the contrast between scheduled work hours and actual work hours. Overtime can lead to long work hours. The culmination of working longer hours throughout a workweek leads to shorter periods of rest between shifts, which not only can affect nurses’ recovery time but also increase their exposure time to work stress and potential hazards (Centers for Disease Control and Prevention, 2002; Sonnentag & Zijlstra, 2006). To protect nurses from working such long hours, since 2000, states began to regulate nurses’ mandatory overtime and shift lengths with the minimum resting time between shifts (Bae, Brewer, & Korvner, 2012). The ANA supports the state legislation of such laws and is in pursuit of federal legislation (ANA, 2012a). Proponents of a ban on mandatory overtime expect that these state regulations improve not only nurses’ working conditions but also nurses’ job satisfaction and retention, which may result in compromised patient quality of care (Washington State Department of Labor and Industries, 2002). Evidence shows that nurses working longer shifts or overtime experienced fatigue and poor quality of sleep, which affected their vigilance, alertness, reaction time, and decision-making ability (Geiger-Brown, Tinkoff, & Rogers, 2011; Trinkoff et al., 2011). As a result, the likelihood of sustaining an injury dramatically increased. Previous studies found that long work hours per day and per week were associated with a higher incidence of musculoskeletal injuries and needlesticks in nurses (Ilhan, Durukan, Aras, Turkcuoglu, & Aygun, 2006; Trinkoff, Le, Geiger-Brown, Lipscomb, & Lang, 2006; Trinkoff, Le, Geigher-Brown, & Lipscomb, 2007). Working unplanned overtime was also associated with the occurrence of work-related injuries and workrelated illnesses (de Castro et al., 2010). Regarding patient outcomes, researchers found a link between long work hours and adverse patient outcomes. Stone et al. (2007) found a higher ratio of overtime was associated with occurrences of catheter-associated urinary tract infection and decubitus ulcers. The risk of making medical errors was three times higher when nurses worked shifts lasting of 12.5 hours or more (Rogers, Hwang, Scott, Aiken, & Dignes, 2004). The most recent study found long work hours per day during nurses’ typical work schedules were significantly related to patient mortality after controlling for hospital staffing levels and hospital characteristics (Trinkoff et al., 2011). Nurses working more than 40 hours per week also perceived an
139
increase in the number of medication errors, falls with injuries, and nosocomial infections (Olds & Clarke, 2010). Several reviews have been conducted in the health care sector to examine the effects of shift length on the quality of patient care and health care provider outcomes (Poissonnet & Veron, 2000; Fletcher et al, 2004; Estabrooks et al, 2009). Recently, Estabrooks et al. (2009) reviewed 12 studies to identify evidence on the effect of shift length (8-hour vs. 12-hour shifts) on the quality of patient care and health care provider outcomes. They found insufficient evidence to conclude the shift length has an impact on patient outcomes or specific health provider outcomes. However, these reviews were limited to the effects of the shift length on outcomes. We found no reviews of studies to date that rigorously and systematically assessed the evidence about the strength of the effect of nurse work hours/overtime and nurse and patient outcomes and the specific components of work hours (e.g., per shift and per week) and overtime (e.g., mandatory and voluntary overtime), which may have an adverse effect on nurse and patient outcomes. Because each study uses various measures of nurse work hours and overtime, accumulative evidence from those studies should be carefully evaluated to make a conclusion. Therefore, the current review conducted a systematic assessment of these empirical studies to draw specific evidence of the impact of nurse work hours and overtime practice on nurse and patient outcomes. Eventually this evidence can be used to develop guidelines on nursing work hours and overtime usage, and researchers, managers, and policy makers will have a better ability to assess and modify nurses’ working hours.
Aim The aim of this review was to systematically evaluate the effect of nurse overtime and long work hours on nurse and patient outcomes. The research questions were how strong was the effect of nurse overtime and long work hours on nurse and patient adverse outcomes and which specific components of nurse overtime (e.g., mandatory and voluntary) and work hours (e.g., per shift, per week, and how many hours) were adversely related to nurse and patient outcomes.
Design From the preliminary assessment, we found the majority of studies were observational in nature. For the homogeneity of studies included in this review, only observational studies were included. Therefore, the recommendations for meta-analysis of observational
140
Nurs Outlook 62 (2014) 138e156
Figure 1 e Search and retrieval process.
studies regarding search strategy, definition of inclusion criteria, quality appraisal, and data abstraction and synthesis were used for this study (Stroup et al., 2000).
Search Strategy The search strategy for this systematic review was guided by a preliminary literature review that identified relevant searching terms; these included nurse (nursing) work hours, overtime, work schedule, and work shift. An online search of the electronic bibliographic databases CINAHL, Cochrane Database of Systematic Reviews, PubMed, PsycINFO, JSTOR, OVID, and Web of Science was conducted. Those databases were selected based on the preliminary literature review in this topic.
Inclusion/Exclusion Criteria Research articles written in English and published from January 2000 until March 2013 were included in
this review. In order to provide recent evidence of the effect of nurse overtime and long hours, we included those studies published from 2000. The studies were included if they were published in a peer-reviewed journal, consisted of nurses working in health care facilities as the study population, were reported as primary research, included the relationship of nurses’ work hours and overtime to nurse and/or patient outcomes as a measure of analysis (either one or both), had a quantitative observational study design, and nurse and/or patient outcomes were operationalized as the dependent variables and nurse work hours as independent variables.
Search Outcomes Using the searching terms, we found 2,836 articles from titles and keywords searches. We screened those articles using their titles for the presence of a study on nurse work hours and overtime and its association with either nurse or patient outcomes. If it was
141
Nurs Outlook 62 (2014) 138e156
difficult to distinguish the inclusion criteria by title alone, the abstract was reviewed. Using this screening method, a total of 144 articles were retrieved. After abstract screening, 37 articles were selected and fully read for further screening. Of these 37 articles, 13 articles were excluded because they did not measure nurse work hours (four articles), were not empirical studies (two articles), were not studies in nursing (one article), and were not published in peer-reviewed journals (six articles). The remaining 24 articles were then assessed for the methodological quality. Four articles (Han, Trinkoff, Storr, & Geiger-Brown, 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011) used the same data collected from one study, which is the Nurses’ Worklife and Health Study (NWHS). Other than that, each article represents each study (Figure 1).
2010), studies that scored less than 0.50 were rated as weak, studies that scored 0.50 to 0.75 were rated as moderate, and studies that scored greater than 0.75 were rated as strong. Seventeen articles had moderate methodological quality. Seven articles were assessed to be of weak methodological quality. However, because there were no shortcomings in measurement and analysis in those studies, we also included them for the review. Thus, all 24 articles were included for the review. Table 1 presents the summary score of each article, and Table 2 presents the summary of the quality assessment of studies included. Although those four articles (Han, Trinkoff, Storr, & Geiger-Brown, 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011) used the same data collected from the NWHS, each article used different study designs and data sets (e.g., longitudinal vs. cross-sectional). Thus, we reported quality scores for each article separately.
Quality Appraisal Data Extraction and Synthesis For the quality appraisal, the Quality Assessment and Validity Tool for Correlation Studies was used (Estabrooks et al., 2001; Estabrooks et al., 2003; Cummings et al., 2008). This instrument used 14 questions to evaluate the design, sample, measurement, and statistical analysis for quantitative observational studies with a total of 14 possible points. The total number of points scored was divided by 14, and the greater the number, the better the quality of the study. According to the adapted version of the quality assessment and validity tool criteria (Cummings et al.,
Figure 1 shows the search and retrieval process. The following data were extracted from the 24 articles in the final inclusion group: author, year of publication, country, participants/sample, setting, framework/ theoretical model, design, nurse work hours and overtime, measures of nursing outcomes, measures of patient outcomes, analysis, and significance of the associations with nurse work hours (Table 3). To clarify those four articles were from the same study (NWHS), we reported both the number of articles and the
Table 1 e Summary of Quality Scores Author
Year
Journal Abbreviation
Quality Score
Score Intervals
Barker & Nussbaum Berney & Needleman de Castro et al. Dorrian et al. Han et al. Ilhan et al. Ilhan et al. Josten et al. Lipscomb et al. Louie et al. Olds & Clarke Portela et al. Rajbhandary & Basu Rogers et al. Schluter et al. Scott et al. Scott et al. Stimpfel et al. Stone et al. Stone et al. Tanaka et al. Trinkoff et al. Trinkoff et al. Trinkoff et al.
2011 2006 2010 2006 2011 2006 2008 2003 2002 2010 2010 2005 2010 2004 2012 2006 2007 2012 2006 2007 2010 2006 2007 2011
J Adv Nurs Policy Polit Nurs Pract Inter Nurs Rev Chronobiol Int J Nurs Adm J Adv Nurs J Adv Nurs J Adv Nurs Scand J Work Environ Health Qual Saf Health Care J Safety Res Rev Saude Publica Health Polit Health Aff Int J Nurs Stud Am J Crit Care Sleep Health Aff Med Care Med Care Ind Health Am J Ind Med Infect Control Hosp Epidemiol Nurs Res
0.36 0.64 0.50 0.43 0.57 0.36 0.43 0.58 0.57 0.50 0.50 0.36 0.64 0.57 0.43 0.57 0.50 0.50 0.57 0.57 0.43 0.57 0.64 0.64
Weak Moderate Moderate Weak Moderate Weak Weak Moderate Moderate Moderate Moderate Weak Moderate Moderate Weak Moderate Moderate Moderate Moderate Moderate Weak Moderate Moderate Moderate
Score intervals: 0.75 ¼ strong.
142
Nurs Outlook 62 (2014) 138e156
Table 2 e Summary of Quality Assessment of the Articles Included Number of Articles
Design Was the study prospective study (other than cross-sectional)? Was probability sampling used? Sample Was sample size justified? Was sample drawn from more than one site? Was anonymity protected? Was response rate more than 60%? Measurement Were independent variables measured reliably? Were independent variables measured using a valid instrument? Were outcome variables measured using a valid instrument? Were outcome variables measured by using other than nurse perception? Was a theoretical model/framework used for guidance? Statistical analysis If multiple effects studies, were correlations analyzed? Were covariates controlled? Were outliers managed?
No
Yes
16
8
14
10
24 6
0 18
0 16
24 8
0
24
0
24
0
24
17
7
19
5
23
1
5 23
19 1
number of studies throughout this review when they were needed.
2011). Donabedian’s model of structure, process, and outcome was used in two of the studies (Stone et al., 2006; Stone et al., 2007), and the effort recovery model (Josten et al., 2003), cycle of shortages in nursing supply model (Rajbhandary & Basu, 2010), and balance theory (Trinkoff et al., 2011) were each used in one study. Compared with experimental studies (e.g., randomization clinical trial), the observation study samples cannot be randomized because the investigator has no control over the subject’s treatment or working hours in these cases. Three articles from two studies used a longitudinal study design (Trinkoff et al., 2006; Scott, Rogers, Hwang, & Zhang, 2007; Trinkoff et al., 2007), and the remainder used a cross-sectional study design.
Nurse Work Hours and Overtime The four articles from the NWHS used the Standard Shiftwork Index (SSI), which has been used internationally to standardize self-report measures used in shift-work research (Han et al., 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011). Using the SSI, researchers measured comprehensive characteristics of nurse work schedules. Among those characteristics, the components of work hours (per day and per week), breaks, and overtime are relevant to this review. These shift work characteristics are equivalent with what other studies measured for work hours and overtime. Most other studies used daily work hours and weekly work hours. In terms of overtime, researchers defined the actual work hours being beyond the scheduled work hours. By integrating the findings of these studies, we used eight concepts regarding nurse work hours and overtime. The concepts include shift length (daily work hours), weekly work hours, the number of shifts per week, breaks, overtime (no specification), voluntary overtime, mandatory overtime, and on call.
Associations of Nurse Work Hours and Overtime with Nursing Outcomes
Results Twenty-four articles from 21 studies were analyzed, and their characteristics are summarized in Table 3. Of the final 21 studies, 11 were conducted in the United States; two were in Canada and Turkey; and one study each was in the Netherlands, Australia, Australia/New Zealand, Philippines, Brazil, and Japan. Although work scheduling methods differ between countries, overtime and long work hours beyond the traditional 8 work hours per day (e.g., 12 hours) are common characteristics of work schedules among nurses. Therefore, the inclusion of these international studies added value to this review. The studies were conducted in acute care hospitals, nursing homes, and other health care settings. Theoretical models were used in five of the studies (Josten et al., 2003; Stone et al., 2006; Stone et al., 2007; Rajbhandary & Basu, 2010; Trinkoff et al.,
Twenty-one different nurse outcome variables (needlestick, musculoskeletal disorder/injuries, back pain, injuries, illness, overweight/obesity, alcohol consumption, health complaints, absenteeism, burnout, job satisfaction/dissatisfaction, intent to stay/intent to leave, motor vehicle crash [MVC] or near MVC, drowsy driving, drowsiness at work, need for recovery, fatigue, varicose veins, physical discomfort and increased exertion, emotional disorders, and lack of time for housework) were reported in the 17 articles from 15 studies (Table 4).
Shift Lengths (Daily Work Hours) Thirteen articles reported the association between shift lengths and nurse outcomes. When the shift length increased, there was an increase in needlesticks (Ilhan et al., 2006; Trinkoff et al., 2007), musculoskeletal
Table 3 e Characteristics of the Included Studies Author, Year, Country
Sample, Setting
745 nurses, various settings
Berney & Needleman (2006), U.S.
161 hospitals
de Castro et al. (2010), Philippine
655 registered nurses working in various settings
Dorrian et al. (2006), Australia
23 nurses, hospitals
Han et al. (2011), 2,103 nurses, U.S. various settings
Design
Not specified Crosssectional
Nurse Work Hours and Overtime Working40 hours/ week
Measures of Nurse Outcomes Physical and mental fatigue, workload,
Measures of Patient Outcomes
Analysis
Significance of the Associations with Nurse Work Hours ( p .05)
NA
Paired two Longer shift lengths and increased sided t-tests; correlation e hours per week Spearmans associated with higher levels of physical and total fatigue and acute and chronic fatigue, not mental fatigue Not specified Longitudinal Overtime (% of NA UTI, UGI bleed, Negative Increased overtime related to decreases overtime hours) pneumonia, shock, binomial in mortality cardiac arrest, regression sepsis, and failure to rescue NA Multivariate Frequency of Not specified CrossWork hours, shift Work-related injury mandatory/ logistic sectional length, frequency in past year, work regression unplanned of mandatory/ related illness in overtime per month unplanned the past year, associated with overtime, overtime missed work for work related hours/month more than 2 days in injuries, work the past year, back related illness, and pain missing two or more days of work Logistic Sleep duration Medical errors Not specified CrossOvertime (scheduled Sleep length and regression shorter on work sectional vs. actual) quality, fatigue (frequency, type, near errors, days than days off; levels, caffeine observed errors) sleep shorter when intake, alcohol, error or near error prescription recorded; sleep medication duration was a significant predictor of error occurrence Not specified CrossLong work hours, Overweight/obesity NA Logistic Long work hours per sectional weekly burden, regression shift related to required on call/ increase in overtime, lack of overweight/obesity rest
Nurs Outlook 62 (2014) 138e156
Barker & Nussbaum (2011), U.S.
Theoretical Model
(continued on next page)
143
Author, Year, Country
Sample, Setting
Theoretical Model
Design
Nurse Work Hours and Overtime
Measures of Nurse Outcomes
Measures of Patient Outcomes
Not specified Crosssectional
Daily working hours
Sharps and needlestick injuries during professional life, within the last year
NA
Ilhan et al. 418 nurses (2008), Turkey working in hospitals
Not specified Crosssectional
Weekly working hours, daily work duration
NA
Josten et al. (2003), Netherlands
134 nurses, nursing home
Effort Recovery Model
8 hours vs. 9 hours
Burnout (Maslach Burnout Inventory) e emotional exhaustion (EE), depersonalization (DP) and personal accomplishment (PA) Fatigue and need for recovery scales, health complaint scale, self -performance rating, effort rating scale
Lipscomb et al. (2002), U.S.
1,163 nurses, no data for setting
Not specified Crosssectional
Louie et al. 1 ICU, 15 beds, 36 (2010), Canada patients, 55 hypoglycemic events
Not specified Crosssectional
Crosssectional
NA
NA Job control, Working>8 hours/ psychological day,>40 hours/ week, 6 years 7 days demands, physical demands, per musculoskeletal week,>50 hours/ disorders week, 12 hours/ day,44 hours/ Diseases and week and>28 hours disorders (varicose veins, per week of house work hypertension, depression, strain, injury, headaches, skin allergies, pulmonary/GI disorders, sleep complaints) Working 8-hour shift Absenteeism vs. 12-hour shift
8.5 hours vs. 12.5 hours, overtime
Measures of Patient Outcomes
NA
Logistic regression
NA
Negative binomial regression
Sleep/wake patterns, Medical errors, mood, and caffeine near misses intake
Logistic regression
Nurs Outlook 62 (2014) 138e156
Olds & Clarke (2010), U.S.
Nurse Work Hours and Overtime
Among LPN, working 12 hours related to greater absenteeism compared with those working 8 hrs Errors increased with 12.5 hours shift and working overtime (continued on next page)
145
Author, Year, Country
Sample, Setting
Theoretical Model
Design
4,419 nurses & midwives, various settings
Not specified Crosssectional
Scott et al. (2006), U.S
502 nurses, hospitals
Not specified Crosssectional
Scott et al. (2007), U.S.
895 nurses; 11,334 work shifts, hospital staff nurses 22,275 nurses; 396 hospitals, hospitals
Stimpfel et al. (2012), U.S.
Stone et al. (2006), U.S.
13 hours per dissatisfaction, shift intent to leave
13 hospitals, 99 Donabedian’s Crossnursing units, 805 theory sectional RNs in New York City
8-hour vs. 12-hour shift
Incident reports Satisfaction, (medication events, emotional falls, decubitus exhaustion, ulcers), riskdepersonalization, adjusted patient personal safety indicators accomplishment, (decubitus ulcer, satisfied with failure to rescue), current scheduling, no intention to stay, nurse perceptions of quality missed shifts, injury, turnover, staffing and costs, absenteeism and costs
Analysis
Significance of the Associations with Nurse Work Hours ( p .05)
Logistic regression
Working 40 hours or more per week related to increase in alcohol consumption Longer shift duration increased risk of errors
Logistic regression
GEE, logistics regression models Logistic and linear regression
GEE
Working>8.5-8 hours 8 hours, 9e11 hours, 12 hours Hours/day 8 hours 8 hours vs. 9 hours 8 hours vs. 12 hours 8 hours vs. 12 hours 8 hours vs. >8 hours 8 hours vs. 12 hours 8e9, 10e11, 12e13, >13 hours per shift 8e9, 10e11, 12e13, >13 hours per shift 8 hours vs. 12 hours 9 hours vs. 12 hours 8 hours vs. 12 hours 8e9, 10e11, 12e13, >13 hours per shift 8.5 hours, >8.5e8.5e8.5e12 hours Hours/day B. Weekly work hours Hours/week >40 hours 40 hours, >40 hours Hours/week 40 hours, >40 hours Weekly work duration 40 hours, >40 hours 40 hours, >40 hours >84 hours per week 40 hours, >40 hours 40 hours, >40 hours 41e80 hours per week >60 hours per week >44 hours per week >44 hours per week 12.5 hours 8.5 hours < vs. 8.5e12.5 8.5 hours < vs. > 12.5 hours >13 hours 8e9, 10e11, 12e13, >13 hours per shift B. Weekly work hours 40 hours, >40 hours 40 hours, >40 hours 40 hours, >40 hours >40 hours >50 hours >40 hours C. Breaks
Nurs Outlook 62 (2014) 138e156
www.nursingoutlook.org
Assessing the relationships between nurse work hours/ overtime and nurse and patient outcomes: Systematic literature review Sung-Heui Bae, PhD, MPH, RNa,*, Donna Fabry, DNP, CNS, RNb b
a School of Nursing, University of Texas at Austin, Austin, TX School of Nursing, University at Buffalo, State University of New York, Buffalo, NY
article info
abstract
Article history: Received 4 July 2013 Revised 15 October 2013 Accepted 30 October 2013
Background: The effects of work hours/overtime on nurse and patient outcomes
Keywords: Adverse nurse outcomes Adverse patient outcomes Nurse overtime Nurse work hours
and specific components of work hours (per shift and per week) and overtime on these effects have not been systematically examined. Purpose: The purpose of this review was to systematically evaluate the effect of nurse overtime and long work hours on nurse and patient outcomes. Methods: An online search of six electronic bibliographic databases was conducted for research published from 2000 to 2013. Discussion: Twenty-one nurse outcome measures and 19 patient outcome measures were found in relationships with work hours and overtime. A total of 67 relationships to nurse outcomes and 41 relationships to patient outcomes were examined. Conclusions: The findings of this review suggested that evidence supporting positive relationships between working long hours and adverse outcomes to the nurses is strong. However, to make a conclusion of the positive relationship between long work hours and adverse patient outcomes, more evidence is needed. Cite this article: Bae, S.-H., & Fabry, D. (2014, APRIL). Assessing the relationships between nurse work hours/overtime and nurse and patient outcomes: Systematic literature review. Nursing Outlook, 62(2), 138-156. http://dx.doi.org/10.1016/j.outlook.2013.10.009.
In order to provide continuous nursing care, 24-hour coverage goes beyond the typical 9:00 a.m. to 5:00 p.m., Monday through Friday work day (Trinkoff et al., 2011). Traditionally, three 8-hour shifts have been used (Josten, Ng-A-Tham, & Thierry, 2003). Globally, various work scheduling methods are used to cover the 24 hours a day, 7 days a week needs for patient care. Some scheduling methods exceed traditional 8 work hours per day (e.g., 12 hours). When nurses work overtime, they often work long periods of time per day as well as per week. For example, in 2008, an estimated
3,063,162 licensed registered nurses (RNs) lived in the United States (U.S. Department of Health and Human Services, 2010). According to the 2010 U.S. Census (U.S. Census Bureau, 2010), licensed RNs comprised 1% of the total population (308,745,538). Fifty-four percent of the respondents to the 2008 National Sample Survey of Registered Nurses worked more than 39 hours per week, resulting in more than half of RNs working at least 2,000 hours per year in their principal nursing positions (U.S. Department of Health and Human Services, 2010). This is approximately 200 hours per
* Corresponding author: Sung-Heui Bae, School of Nursing, University of Texas at Austin, 1710 Red River, Austin, TX 78701. E-mail address: [email protected] (S.-H. Bae). 0029-6554/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.outlook.2013.10.009
Nurs Outlook 62 (2014) 138e156
year more than the average American worker (Fleck, 2009). If we consider those nurses who have more than one position, the total number of hours could be more than this. Furthermore, nurses are more likely to work 12-hour shifts in the United States (American Nurses Association [ANA], 2009). All this implies nurses not only work long hours, but they also return to work more quickly, resulting in less sleep and rest between shifts. Long work hours can be defined by per shift as well as per week. In a similar concept, overtime can be measured by the contrast between scheduled work hours and actual work hours. Overtime can lead to long work hours. The culmination of working longer hours throughout a workweek leads to shorter periods of rest between shifts, which not only can affect nurses’ recovery time but also increase their exposure time to work stress and potential hazards (Centers for Disease Control and Prevention, 2002; Sonnentag & Zijlstra, 2006). To protect nurses from working such long hours, since 2000, states began to regulate nurses’ mandatory overtime and shift lengths with the minimum resting time between shifts (Bae, Brewer, & Korvner, 2012). The ANA supports the state legislation of such laws and is in pursuit of federal legislation (ANA, 2012a). Proponents of a ban on mandatory overtime expect that these state regulations improve not only nurses’ working conditions but also nurses’ job satisfaction and retention, which may result in compromised patient quality of care (Washington State Department of Labor and Industries, 2002). Evidence shows that nurses working longer shifts or overtime experienced fatigue and poor quality of sleep, which affected their vigilance, alertness, reaction time, and decision-making ability (Geiger-Brown, Tinkoff, & Rogers, 2011; Trinkoff et al., 2011). As a result, the likelihood of sustaining an injury dramatically increased. Previous studies found that long work hours per day and per week were associated with a higher incidence of musculoskeletal injuries and needlesticks in nurses (Ilhan, Durukan, Aras, Turkcuoglu, & Aygun, 2006; Trinkoff, Le, Geiger-Brown, Lipscomb, & Lang, 2006; Trinkoff, Le, Geigher-Brown, & Lipscomb, 2007). Working unplanned overtime was also associated with the occurrence of work-related injuries and workrelated illnesses (de Castro et al., 2010). Regarding patient outcomes, researchers found a link between long work hours and adverse patient outcomes. Stone et al. (2007) found a higher ratio of overtime was associated with occurrences of catheter-associated urinary tract infection and decubitus ulcers. The risk of making medical errors was three times higher when nurses worked shifts lasting of 12.5 hours or more (Rogers, Hwang, Scott, Aiken, & Dignes, 2004). The most recent study found long work hours per day during nurses’ typical work schedules were significantly related to patient mortality after controlling for hospital staffing levels and hospital characteristics (Trinkoff et al., 2011). Nurses working more than 40 hours per week also perceived an
139
increase in the number of medication errors, falls with injuries, and nosocomial infections (Olds & Clarke, 2010). Several reviews have been conducted in the health care sector to examine the effects of shift length on the quality of patient care and health care provider outcomes (Poissonnet & Veron, 2000; Fletcher et al, 2004; Estabrooks et al, 2009). Recently, Estabrooks et al. (2009) reviewed 12 studies to identify evidence on the effect of shift length (8-hour vs. 12-hour shifts) on the quality of patient care and health care provider outcomes. They found insufficient evidence to conclude the shift length has an impact on patient outcomes or specific health provider outcomes. However, these reviews were limited to the effects of the shift length on outcomes. We found no reviews of studies to date that rigorously and systematically assessed the evidence about the strength of the effect of nurse work hours/overtime and nurse and patient outcomes and the specific components of work hours (e.g., per shift and per week) and overtime (e.g., mandatory and voluntary overtime), which may have an adverse effect on nurse and patient outcomes. Because each study uses various measures of nurse work hours and overtime, accumulative evidence from those studies should be carefully evaluated to make a conclusion. Therefore, the current review conducted a systematic assessment of these empirical studies to draw specific evidence of the impact of nurse work hours and overtime practice on nurse and patient outcomes. Eventually this evidence can be used to develop guidelines on nursing work hours and overtime usage, and researchers, managers, and policy makers will have a better ability to assess and modify nurses’ working hours.
Aim The aim of this review was to systematically evaluate the effect of nurse overtime and long work hours on nurse and patient outcomes. The research questions were how strong was the effect of nurse overtime and long work hours on nurse and patient adverse outcomes and which specific components of nurse overtime (e.g., mandatory and voluntary) and work hours (e.g., per shift, per week, and how many hours) were adversely related to nurse and patient outcomes.
Design From the preliminary assessment, we found the majority of studies were observational in nature. For the homogeneity of studies included in this review, only observational studies were included. Therefore, the recommendations for meta-analysis of observational
140
Nurs Outlook 62 (2014) 138e156
Figure 1 e Search and retrieval process.
studies regarding search strategy, definition of inclusion criteria, quality appraisal, and data abstraction and synthesis were used for this study (Stroup et al., 2000).
Search Strategy The search strategy for this systematic review was guided by a preliminary literature review that identified relevant searching terms; these included nurse (nursing) work hours, overtime, work schedule, and work shift. An online search of the electronic bibliographic databases CINAHL, Cochrane Database of Systematic Reviews, PubMed, PsycINFO, JSTOR, OVID, and Web of Science was conducted. Those databases were selected based on the preliminary literature review in this topic.
Inclusion/Exclusion Criteria Research articles written in English and published from January 2000 until March 2013 were included in
this review. In order to provide recent evidence of the effect of nurse overtime and long hours, we included those studies published from 2000. The studies were included if they were published in a peer-reviewed journal, consisted of nurses working in health care facilities as the study population, were reported as primary research, included the relationship of nurses’ work hours and overtime to nurse and/or patient outcomes as a measure of analysis (either one or both), had a quantitative observational study design, and nurse and/or patient outcomes were operationalized as the dependent variables and nurse work hours as independent variables.
Search Outcomes Using the searching terms, we found 2,836 articles from titles and keywords searches. We screened those articles using their titles for the presence of a study on nurse work hours and overtime and its association with either nurse or patient outcomes. If it was
141
Nurs Outlook 62 (2014) 138e156
difficult to distinguish the inclusion criteria by title alone, the abstract was reviewed. Using this screening method, a total of 144 articles were retrieved. After abstract screening, 37 articles were selected and fully read for further screening. Of these 37 articles, 13 articles were excluded because they did not measure nurse work hours (four articles), were not empirical studies (two articles), were not studies in nursing (one article), and were not published in peer-reviewed journals (six articles). The remaining 24 articles were then assessed for the methodological quality. Four articles (Han, Trinkoff, Storr, & Geiger-Brown, 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011) used the same data collected from one study, which is the Nurses’ Worklife and Health Study (NWHS). Other than that, each article represents each study (Figure 1).
2010), studies that scored less than 0.50 were rated as weak, studies that scored 0.50 to 0.75 were rated as moderate, and studies that scored greater than 0.75 were rated as strong. Seventeen articles had moderate methodological quality. Seven articles were assessed to be of weak methodological quality. However, because there were no shortcomings in measurement and analysis in those studies, we also included them for the review. Thus, all 24 articles were included for the review. Table 1 presents the summary score of each article, and Table 2 presents the summary of the quality assessment of studies included. Although those four articles (Han, Trinkoff, Storr, & Geiger-Brown, 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011) used the same data collected from the NWHS, each article used different study designs and data sets (e.g., longitudinal vs. cross-sectional). Thus, we reported quality scores for each article separately.
Quality Appraisal Data Extraction and Synthesis For the quality appraisal, the Quality Assessment and Validity Tool for Correlation Studies was used (Estabrooks et al., 2001; Estabrooks et al., 2003; Cummings et al., 2008). This instrument used 14 questions to evaluate the design, sample, measurement, and statistical analysis for quantitative observational studies with a total of 14 possible points. The total number of points scored was divided by 14, and the greater the number, the better the quality of the study. According to the adapted version of the quality assessment and validity tool criteria (Cummings et al.,
Figure 1 shows the search and retrieval process. The following data were extracted from the 24 articles in the final inclusion group: author, year of publication, country, participants/sample, setting, framework/ theoretical model, design, nurse work hours and overtime, measures of nursing outcomes, measures of patient outcomes, analysis, and significance of the associations with nurse work hours (Table 3). To clarify those four articles were from the same study (NWHS), we reported both the number of articles and the
Table 1 e Summary of Quality Scores Author
Year
Journal Abbreviation
Quality Score
Score Intervals
Barker & Nussbaum Berney & Needleman de Castro et al. Dorrian et al. Han et al. Ilhan et al. Ilhan et al. Josten et al. Lipscomb et al. Louie et al. Olds & Clarke Portela et al. Rajbhandary & Basu Rogers et al. Schluter et al. Scott et al. Scott et al. Stimpfel et al. Stone et al. Stone et al. Tanaka et al. Trinkoff et al. Trinkoff et al. Trinkoff et al.
2011 2006 2010 2006 2011 2006 2008 2003 2002 2010 2010 2005 2010 2004 2012 2006 2007 2012 2006 2007 2010 2006 2007 2011
J Adv Nurs Policy Polit Nurs Pract Inter Nurs Rev Chronobiol Int J Nurs Adm J Adv Nurs J Adv Nurs J Adv Nurs Scand J Work Environ Health Qual Saf Health Care J Safety Res Rev Saude Publica Health Polit Health Aff Int J Nurs Stud Am J Crit Care Sleep Health Aff Med Care Med Care Ind Health Am J Ind Med Infect Control Hosp Epidemiol Nurs Res
0.36 0.64 0.50 0.43 0.57 0.36 0.43 0.58 0.57 0.50 0.50 0.36 0.64 0.57 0.43 0.57 0.50 0.50 0.57 0.57 0.43 0.57 0.64 0.64
Weak Moderate Moderate Weak Moderate Weak Weak Moderate Moderate Moderate Moderate Weak Moderate Moderate Weak Moderate Moderate Moderate Moderate Moderate Weak Moderate Moderate Moderate
Score intervals: 0.75 ¼ strong.
142
Nurs Outlook 62 (2014) 138e156
Table 2 e Summary of Quality Assessment of the Articles Included Number of Articles
Design Was the study prospective study (other than cross-sectional)? Was probability sampling used? Sample Was sample size justified? Was sample drawn from more than one site? Was anonymity protected? Was response rate more than 60%? Measurement Were independent variables measured reliably? Were independent variables measured using a valid instrument? Were outcome variables measured using a valid instrument? Were outcome variables measured by using other than nurse perception? Was a theoretical model/framework used for guidance? Statistical analysis If multiple effects studies, were correlations analyzed? Were covariates controlled? Were outliers managed?
No
Yes
16
8
14
10
24 6
0 18
0 16
24 8
0
24
0
24
0
24
17
7
19
5
23
1
5 23
19 1
number of studies throughout this review when they were needed.
2011). Donabedian’s model of structure, process, and outcome was used in two of the studies (Stone et al., 2006; Stone et al., 2007), and the effort recovery model (Josten et al., 2003), cycle of shortages in nursing supply model (Rajbhandary & Basu, 2010), and balance theory (Trinkoff et al., 2011) were each used in one study. Compared with experimental studies (e.g., randomization clinical trial), the observation study samples cannot be randomized because the investigator has no control over the subject’s treatment or working hours in these cases. Three articles from two studies used a longitudinal study design (Trinkoff et al., 2006; Scott, Rogers, Hwang, & Zhang, 2007; Trinkoff et al., 2007), and the remainder used a cross-sectional study design.
Nurse Work Hours and Overtime The four articles from the NWHS used the Standard Shiftwork Index (SSI), which has been used internationally to standardize self-report measures used in shift-work research (Han et al., 2011; Trinkoff et al., 2006; Trinkoff et al., 2007; Trinkoff et al., 2011). Using the SSI, researchers measured comprehensive characteristics of nurse work schedules. Among those characteristics, the components of work hours (per day and per week), breaks, and overtime are relevant to this review. These shift work characteristics are equivalent with what other studies measured for work hours and overtime. Most other studies used daily work hours and weekly work hours. In terms of overtime, researchers defined the actual work hours being beyond the scheduled work hours. By integrating the findings of these studies, we used eight concepts regarding nurse work hours and overtime. The concepts include shift length (daily work hours), weekly work hours, the number of shifts per week, breaks, overtime (no specification), voluntary overtime, mandatory overtime, and on call.
Associations of Nurse Work Hours and Overtime with Nursing Outcomes
Results Twenty-four articles from 21 studies were analyzed, and their characteristics are summarized in Table 3. Of the final 21 studies, 11 were conducted in the United States; two were in Canada and Turkey; and one study each was in the Netherlands, Australia, Australia/New Zealand, Philippines, Brazil, and Japan. Although work scheduling methods differ between countries, overtime and long work hours beyond the traditional 8 work hours per day (e.g., 12 hours) are common characteristics of work schedules among nurses. Therefore, the inclusion of these international studies added value to this review. The studies were conducted in acute care hospitals, nursing homes, and other health care settings. Theoretical models were used in five of the studies (Josten et al., 2003; Stone et al., 2006; Stone et al., 2007; Rajbhandary & Basu, 2010; Trinkoff et al.,
Twenty-one different nurse outcome variables (needlestick, musculoskeletal disorder/injuries, back pain, injuries, illness, overweight/obesity, alcohol consumption, health complaints, absenteeism, burnout, job satisfaction/dissatisfaction, intent to stay/intent to leave, motor vehicle crash [MVC] or near MVC, drowsy driving, drowsiness at work, need for recovery, fatigue, varicose veins, physical discomfort and increased exertion, emotional disorders, and lack of time for housework) were reported in the 17 articles from 15 studies (Table 4).
Shift Lengths (Daily Work Hours) Thirteen articles reported the association between shift lengths and nurse outcomes. When the shift length increased, there was an increase in needlesticks (Ilhan et al., 2006; Trinkoff et al., 2007), musculoskeletal
Table 3 e Characteristics of the Included Studies Author, Year, Country
Sample, Setting
745 nurses, various settings
Berney & Needleman (2006), U.S.
161 hospitals
de Castro et al. (2010), Philippine
655 registered nurses working in various settings
Dorrian et al. (2006), Australia
23 nurses, hospitals
Han et al. (2011), 2,103 nurses, U.S. various settings
Design
Not specified Crosssectional
Nurse Work Hours and Overtime Working40 hours/ week
Measures of Nurse Outcomes Physical and mental fatigue, workload,
Measures of Patient Outcomes
Analysis
Significance of the Associations with Nurse Work Hours ( p .05)
NA
Paired two Longer shift lengths and increased sided t-tests; correlation e hours per week Spearmans associated with higher levels of physical and total fatigue and acute and chronic fatigue, not mental fatigue Not specified Longitudinal Overtime (% of NA UTI, UGI bleed, Negative Increased overtime related to decreases overtime hours) pneumonia, shock, binomial in mortality cardiac arrest, regression sepsis, and failure to rescue NA Multivariate Frequency of Not specified CrossWork hours, shift Work-related injury mandatory/ logistic sectional length, frequency in past year, work regression unplanned of mandatory/ related illness in overtime per month unplanned the past year, associated with overtime, overtime missed work for work related hours/month more than 2 days in injuries, work the past year, back related illness, and pain missing two or more days of work Logistic Sleep duration Medical errors Not specified CrossOvertime (scheduled Sleep length and regression shorter on work sectional vs. actual) quality, fatigue (frequency, type, near errors, days than days off; levels, caffeine observed errors) sleep shorter when intake, alcohol, error or near error prescription recorded; sleep medication duration was a significant predictor of error occurrence Not specified CrossLong work hours, Overweight/obesity NA Logistic Long work hours per sectional weekly burden, regression shift related to required on call/ increase in overtime, lack of overweight/obesity rest
Nurs Outlook 62 (2014) 138e156
Barker & Nussbaum (2011), U.S.
Theoretical Model
(continued on next page)
143
Author, Year, Country
Sample, Setting
Theoretical Model
Design
Nurse Work Hours and Overtime
Measures of Nurse Outcomes
Measures of Patient Outcomes
Not specified Crosssectional
Daily working hours
Sharps and needlestick injuries during professional life, within the last year
NA
Ilhan et al. 418 nurses (2008), Turkey working in hospitals
Not specified Crosssectional
Weekly working hours, daily work duration
NA
Josten et al. (2003), Netherlands
134 nurses, nursing home
Effort Recovery Model
8 hours vs. 9 hours
Burnout (Maslach Burnout Inventory) e emotional exhaustion (EE), depersonalization (DP) and personal accomplishment (PA) Fatigue and need for recovery scales, health complaint scale, self -performance rating, effort rating scale
Lipscomb et al. (2002), U.S.
1,163 nurses, no data for setting
Not specified Crosssectional
Louie et al. 1 ICU, 15 beds, 36 (2010), Canada patients, 55 hypoglycemic events
Not specified Crosssectional
Crosssectional
NA
NA Job control, Working>8 hours/ psychological day,>40 hours/ week, 6 years 7 days demands, physical demands, per musculoskeletal week,>50 hours/ disorders week, 12 hours/ day,44 hours/ Diseases and week and>28 hours disorders (varicose veins, per week of house work hypertension, depression, strain, injury, headaches, skin allergies, pulmonary/GI disorders, sleep complaints) Working 8-hour shift Absenteeism vs. 12-hour shift
8.5 hours vs. 12.5 hours, overtime
Measures of Patient Outcomes
NA
Logistic regression
NA
Negative binomial regression
Sleep/wake patterns, Medical errors, mood, and caffeine near misses intake
Logistic regression
Nurs Outlook 62 (2014) 138e156
Olds & Clarke (2010), U.S.
Nurse Work Hours and Overtime
Among LPN, working 12 hours related to greater absenteeism compared with those working 8 hrs Errors increased with 12.5 hours shift and working overtime (continued on next page)
145
Author, Year, Country
Sample, Setting
Theoretical Model
Design
4,419 nurses & midwives, various settings
Not specified Crosssectional
Scott et al. (2006), U.S
502 nurses, hospitals
Not specified Crosssectional
Scott et al. (2007), U.S.
895 nurses; 11,334 work shifts, hospital staff nurses 22,275 nurses; 396 hospitals, hospitals
Stimpfel et al. (2012), U.S.
Stone et al. (2006), U.S.
13 hours per dissatisfaction, shift intent to leave
13 hospitals, 99 Donabedian’s Crossnursing units, 805 theory sectional RNs in New York City
8-hour vs. 12-hour shift
Incident reports Satisfaction, (medication events, emotional falls, decubitus exhaustion, ulcers), riskdepersonalization, adjusted patient personal safety indicators accomplishment, (decubitus ulcer, satisfied with failure to rescue), current scheduling, no intention to stay, nurse perceptions of quality missed shifts, injury, turnover, staffing and costs, absenteeism and costs
Analysis
Significance of the Associations with Nurse Work Hours ( p .05)
Logistic regression
Working 40 hours or more per week related to increase in alcohol consumption Longer shift duration increased risk of errors
Logistic regression
GEE, logistics regression models Logistic and linear regression
GEE
Working>8.5-8 hours 8 hours, 9e11 hours, 12 hours Hours/day 8 hours 8 hours vs. 9 hours 8 hours vs. 12 hours 8 hours vs. 12 hours 8 hours vs. >8 hours 8 hours vs. 12 hours 8e9, 10e11, 12e13, >13 hours per shift 8e9, 10e11, 12e13, >13 hours per shift 8 hours vs. 12 hours 9 hours vs. 12 hours 8 hours vs. 12 hours 8e9, 10e11, 12e13, >13 hours per shift 8.5 hours, >8.5e8.5e8.5e12 hours Hours/day B. Weekly work hours Hours/week >40 hours 40 hours, >40 hours Hours/week 40 hours, >40 hours Weekly work duration 40 hours, >40 hours 40 hours, >40 hours >84 hours per week 40 hours, >40 hours 40 hours, >40 hours 41e80 hours per week >60 hours per week >44 hours per week >44 hours per week 12.5 hours 8.5 hours < vs. 8.5e12.5 8.5 hours < vs. > 12.5 hours >13 hours 8e9, 10e11, 12e13, >13 hours per shift B. Weekly work hours 40 hours, >40 hours 40 hours, >40 hours 40 hours, >40 hours >40 hours >50 hours >40 hours C. Breaks
