Estimates of injury risks for healthcare personnel working night shifts and long hours
- Correspondence to Dr Allard E Dembe, Center for Health Outcomes, Policy and Evaluation Studies, The Ohio State University College of Public Health, 472 Cunz Hall, 1841 Neil Ave, Columbus, OH 43210, USA;
- Accepted 23 February 2009
Background: Evidence suggests that working long hours or unconventional shifts (night, evening and rotating shifts) can induce fatigue and stress in healthcare employees that might jeopardise quality of care and patient safety.
Methods: This study is based on a retrospective analysis of 13 years of occupational data from the National Longitudinal Survey of Youth, covering nearly 11 000 American workers. During the study period, 545 injuries were reported by employees in healthcare professions. Cox proportional hazard analyses were used to calculate adjusted hazard ratios comparing the risk of a job-related injury among healthcare workers in various types of demanding schedules to employees working conventional schedules. The analyses were stratified to estimate risks within different occupational classifications and care settings.
Results: The greatest injury risks to healthcare workers were in schedules involving overtime or at least 60 h per week. Interestingly, an elevated risk of injury was not observed for healthcare employees working 12 or more hours per day or for those in night, evening or rotating shifts. Among employees working overtime and long-hour (>60 h per week) schedules, those at medical provider offices had a significantly higher risk of injury (HR 2.86) than at hospitals, rehabilitation clinics or long-term care facilities. Support personnel, including aids, attendants, technicians, therapists and dieticians, faced a higher risk of injury than did physicians and nurses.
Conclusion: Healthcare managers responsible for quality improvement and patient safety programmes should be aware of the possibility for worker fatigue and injury in particular scheduling arrangements.
The duration and scheduling of work shifts is known to affect the performance of healthcare personnel. For instance, evidence indicates that the risk of errors is increased and vigilance diminished among nurses working long-hour shifts.12 Studies have found that physician trainees who work extended hours (eg >80 h per week, or 16 h consecutively) have an increased likelihood of fatigue-related errors and attention problems.34567 Nurses in rotating shifts report greater levels of stress and fatigue compared with those in non-rotating shifts.891011
Long working hours and certain types of shift-work schedules have been linked with an increased risk for automobile accidents and job injuries among healthcare workers. The risk for healthcare workers to be in an auto crash or a job-related accident increases in relation to the number of hours worked on a shift.12 In one study, 95.5% of nurses reported experiencing an automobile crash or near-miss during the past year when driving home after working night shifts.13 Several studies have documented an increased risk for auto accidents among medical interns returning home after working long hours.141516
Long-hour and non-standard shifts produce physical and mental fatigue that is thought to cause many of the errors, accidents and performance problems experienced by healthcare employees.1718 Affected workers may experience stress or various psychological and somatic complaints. Fatigue and stress from working long hours and non-standard shifts can interfere with concentration and memory and decrease ability to perform complex cognitive and monitoring tasks.1920 Night shifts and long-hour schedules can disrupt circadian rhythms and sleep patterns, thereby diminishing employees’ ability to work safely and productively.21
This study examines risks for job-related injuries among healthcare employees working long-hour or non-standard schedules. Injuries among healthcare personnel can indicate conditions (fatigue and stress) that could jeopardise the quality and safety of patient care.2223 Few previous studies have examined job-related injury risk among health workers, and none that we know of have attempted to assess risks across a range of healthcare occupations and clinical settings. Existing studies suggest that healthcare workers may be especially susceptible to the effects of fatiguing work. For example, working long-hour shifts (>12 h per day) or “off-hour” shifts (night or weekends) was found to increase the risk for nurses to sustain musculoskeletal injuries.2425 A study using Oregon workers’ compensation data found that evening and night shift hospital employees were at greater risk of sustaining occupational injuries than day workers.26 Gold et al27 found that nurses in rotating shifts were about twice as likely as those in day shifts to experience an “accident” (defined as automobile accidents, medication errors, on-the-job procedural errors or occupational injuries). Nurses in rotating shifts were about 39% more likely to suffer work-related injuries than those in day shifts, according to a 1979 study.28 However, existing evidence is somewhat equivocal. A 2004 study, for instance, found no statistically significant difference between healthcare workers in night shifts and day shifts with respect to the risk of automobile crashes or occupational injuries.29
Our study uses longitudinal data for 13 years (1987 to 2000), covering a range of healthcare professions and clinical settings to assess the relationship between various types of work schedules and the incidence of job-related injuries and illnesses among healthcare personnel. We calculate relative risks for 11 types of work schedule and four types of clinical setting. Having this detailed information available will provide insights about the hazards in specific work schedules and settings that might impede worker performance and thus potentially affect the care provided to patients.
This study is based on a retrospective review of data from the National Longitudinal Survey of Youth, 1979 (NLSY), which is sponsored and administered by the US Bureau of Labor Statistics. The NLSY contains self-reported employment data from a nationally representative sample of 12 686 men and women who were 14 to 22 years old in 1979 when the survey was first administered. The survey was performed annually until 1994 and biannually thereafter. Data for our study came from surveys administered between 1988 and 2000, when the cohort members were 22 to 43 years old. Information collected in the surveys includes demographic, employment and wage information. Detailed work histories concerning all jobs held during that period were obtained. Information was collected about the occurrence of work-related injuries on the job including the injury date and the type of injury.
For this analysis, we limited data to employees having jobs in the healthcare industry. Based on the 1970 US Census industry classification system adopted by NLSY, these jobs included codes 828 (physician offices), 829 (dental offices), 837 (chiropractic offices), 838 (hospitals), 839 (convalescent institutions), 847 (offices of health practitioners) and 848 (health services, not elsewhere classified). Occupational classification codes from the 1970 US Census system were also applied for workers in healthcare jobs. There were 24 occupational classification codes used covering physicians (codes 061–073), nurses (074, 925–926), health technologists and technicians (074, 076, 080–085), health administrators (212) and other health service workers (921–924).
Information about the incidence of workplace injuries was based on respondents’ answers to this question:
“I would like to ask you a few questions about any injuries or illnesses you might have received or gotten while you were working on a job. Since [date of last interview] have you had an incident at any job that resulted in an injury or illness to you?”
Individuals responding affirmatively were asked follow-up questions about the type of injury or illness, the date when it occurred and the effects of the injury on employment status, time missed from work, and return to work. For simplicity, we will use “injury” here as a general term applying to both injuries and illnesses.
Respondent-supplied information about job schedules, shifts, commuting time and the usual start and end time of work each day allowed for the employee's schedule to be classified into one of four types of “long-work-hour” schedules: extended hours per week (⩾60), extended hours per day (⩾12), overtime and extended commute time (⩾2 h). Schedules not containing any of those four “long-hour” demands were considered conventional schedules. Shifts were classified by the respondents into one of five types of non-standard shift (based on criteria provided by NLSY): night, evening, rotating, irregular, and split shifts or a conventional day shift.
Cox proportional hazards analyses were performed to compare the incidence of reported injuries per 100 worker-years of time on the job among each of the four “long-hour” schedules to a conventional-hour schedule and among each of the five types of non-standard shifts to a conventional day shift. Covariates in the regression models included age, sex and US geographical region. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated. Additional subgroup analyses were performed to estimate HRs and 95% CIs for workers in schedules involving overtime or 60 or more hours per week (the schedules found to have statistically significant elevated risk) for each of four healthcare settings (medical provider offices, hospitals, convalescent facilities and other) and four occupational groupings (physicians and nurses, support personnel, administrators and other).
The Cox proportional hazard analyses used weighted data, with sample weights that were calculated and applied by NLSY separately for each survey period. Calculations were performed using SAS statistical software V.9.0). To account for sampling effects, 95% CIs were estimated around the HRs by applying Taylor expansion approximation techniques using SAS’s SURVEYLOGISTIC procedure.30 Additional details on the methods we used in this study have been published elsewhere.31323334 NLSY study information, item response rates, sample weights, standard errors and survey documentation can be obtained at the NLSY website: http://www.bls.gov/nls/nlsy79.htm.
During the observation period for this study, 545 work-related injuries were reported by healthcare personnel. Demographic characteristics of the injured workers are summarised in table 1.
The injured healthcare workers were predominantly (81.4%) female, were more likely than other injured workers to be black and less likely (15.5% vs. 22.2%) to be unionised. Among injured healthcare workers, about one-third were nurses (registered or licensed practical), one-third administrators and the remainder mostly nursing aids, orderlies, attendants and technicians. Only 2.6% were physicians (see table 2). More than half (50.6%) of the injured healthcare employees worked in a hospital setting, one-quarter worked in rehabilitation or long-term care facilities and the rest worked in provider offices and other settings.
Table 3 summarises the kind of injuries suffered by healthcare workers and the types of schedules in which the injuries occurred. Musculoskeletal disorders were the most common type of injury, representing 37.7% of the cases. Cuts, bruises and other traumatic injuries accounted for 33.4% of the total. Occupational diseases (principally cases of carpal tunnel syndrome) were reported by 17% of the injured healthcare workers. More than 40% of the injuries occurred in schedules that involved overtime.
Relative risks for suffering a work-related injury among the various types of work schedule are summarised in table 4. Among healthcare workers, none of the non-standard shift schedules were found to have a statistically significant elevated risk for injury compared with a conventional day shift schedule. Evening shifts had the highest HR (1.30) of any shift type, but that HR was not statistically significant (p = 0.20). The greatest risks for healthcare workers were observed in jobs requiring overtime (HR 2.11) or 60 or more hours per week (HR 2.02). Interestingly, similar risks were not observed among employees working 12 or more hours per day (HR 1.22, 95% CI 0.88 to 2.92).
Table 5 presents the results of the subgroup analysis in which data for people working the schedules identified as especially hazardous (overtime or at least 60 h per week) were stratified to see how the HRs vary by setting and occupation. Among care settings, medical provider offices were most hazardous with a HR of 2.86 (significant at a level of 94.7%, 95% CI 0.99 to 8.32). By comparison, hospitals had a HR of 1.64 (CI 1.16 to 2.33). HRs for workers in other settings were not statistically significant.
For occupational categories, the greatest risk of injury among employees working overtime or at least 60 h per week was observed in the miscellaneous grouping that included dieticians, therapists, dental assistants, health aids (other than nursing), health trainees and lay midwives (HR 2.23). Statistically significant risks were also observed among support personnel (nursing aids, orderlies, attendants and technicians) with a HR of 1.88 and among physicians and nurses (HR 1.72). The HR among health administrators (1.43) was not statistically significant.
This study of occupational injury risks among healthcare personnel has revealed new information about the magnitude of the risk in each of nine types of shift-work and extended-hour schedules. The data corroborate other studies showing that healthcare workers in schedules involving overtime and long hours (⩾60) per week face elevated injury risks compared with workers in conventional schedules.
However, there were several surprises from this study. For example, schedules involving 12 or more hours per day had a HR of only 1.22, which was not statistically significant. This suggests that the greatest potential source of fatigue and stress for nursing and other healthcare workers may arise because of accumulated hours over the course of a week or longer periods, rather than from occasional 12-hour shifts worked on a few (3–4) days during a week.
Another surprise was that none of the non-standard shift-work schedules (night, evening and rotating shifts) posed a statistically significant elevated risk compared with a conventional day schedule. In fact, rotating shift schedules, which have been the subject of considerable concern among workers and safety officials, actually were found to have a (statistically non-significant) negative association with injuries, with an HR of 0.73 (CI 0.39 to 1.39). Here again, it may be that the accumulated stress of working many long-hour shifts over the course of a week is a stronger determinant of injury risk than the particular shift pattern (day, evening, night or rotating).
Most existing literature in this field focuses on risks among nurses working in hospital settings. However, we observed that the injury risk for hospital employees working overtime or long-hours per week was comparatively lower than for staff working in non-hospital medical offices. Possibly, hospitals have more staffing and infrastructure to devote towards injury prevention compared with smaller work settings with limited resources. This finding suggests that it is important to emphasise safety efforts in ambulatory settings as well as in hospital environments.
Similarly, although nurses and nursing associations have been at the forefront of campaigns to limit hours and enhance working conditions in the healthcare industry, it is noteworthy that in this study, the injury risks faced by physicians and nurses working long hours were relatively lower than for support personnel, including aids, attendants, technicians, therapists, dieticians and other healthcare workers besides nurses and physicians. This is perhaps the first study that has examined long-hour injury risks among those adjunct worker classifications. Our findings indicate that it is important to focus on appropriate shift scheduling, work hours, fatigue and stress management among healthcare support personnel as well as among nurses and doctors.
As in any study based on a national survey, we were limited to a secondary analysis of the existing data. For example, NLSY did not contain any information about the effects of staff injuries on patient care or clinical outcomes. Numerous studies have linked worker injury with diminished performance.2223353637 We adopted the general approach of using worker injury as an indicator of potential fatigue and stress that might jeopardise patient care. However, we could not directly assess the impact of work injuries on patient care given the data available from NLSY.
The NLSY contained a large amount of information covering 10 793 workers who sustained 5313 work-related injuries over a 13-year period. However, only 545 injuries during that period involved healthcare personnel. Thus, after stratifying the data by type of work schedule, care setting and occupational classification, some analyses we performed had limited statistical power to detect small effects. Nevertheless, all of our major findings (HR⩾1.7) were statistically significant or, in the case of people working overtime or long-hour schedules in medical offices, very nearly significant (HR 2.86, with p = 0.053).
The small number of injuries and limited exposure time in some work schedules also precluded us from examining some dimensions of risk—for example, combinations of an unconventional shift and a long-hour schedule (working overtime on the night shift). Similarly, due to small numbers, we calculated risks for occupational and care-setting groupings rather than for specific job titles or particular types of medical offices (eg physician office, dental office, chiropractic office).
The NLSY survey questionnaire asked respondents to recall the occurrence of work-related injuries or illness during the preceding 1-year (in surveys conducted before 1994) or 2-year (in survey conducted since 1994) period. Some respondents may not have been able to recall the nature or timing of their injury precisely. In the USA, the ability to recall this information accurately is enhanced by mandatory reporting of the incident by the injured employees to the employer that is required by federal Occupational Safety and Health Administration regulations as well as the filing of a separate claim form with the employer’s workers’ compensation insurance carrier. In this study, any potential difficulties in recalling injury information would only bias the study results if respondents’ recall ability varied systematically between shift schedules. There does not appear to be any a priori reason to think that recall would vary according to the type of schedule worked.
Also, in the USA, many minor accidents and injuries that do not result in lost work time may not require the filing of injury reports or insurance claims. However, this study was not aimed at determining an accurate estimate of healthcare workers’ annual injury and illness injury rates. Rather, the purpose of this study was to determine the relative or comparative risks of injury among healthcare workers in specific kinds of demanding work schedules compared with workers in conventional-length or conventional day schedules. Here again, there is no evidence suggesting that the ability of workers’ to remember minor or major injury incidents systemically varies between kinds of work schedules.
Considerable attention has been given to dangers faced by healthcare personnel working long hours and demanding shift schedules and to possible implications for patient safety and quality of care. This study helps to refine the issue by identifying particular schedules, care settings and occupational classifications that appear to constitute the greatest risk. Healthcare managers responsible for quality improvement and patient safety programmes must pay special attention to those factors when considering interventions for optimising workforce performance and safety at their facilities.
The authors are deeply grateful to our colleague, Steven M. Banks, PhD, who served as the statistician on this project before his untimely death in August 2007. We also appreciate the helpful comments provided by Pamela Salsberry, PhD, RN, and Esther Chipps, PhD, RN.
Funding This research was supported by the WE Upjohn Institute for Employment Research.
Competing interests None.
Ethics approval This study was based entirely on secondary analysis of de-identified national survey data in the public domain and thus was not considered to be human subjects research requiring review by the Ohio State University Institutional Review Board.
Provenance and Peer review Not commissioned; externally peer reviewed.