Do 12-Hour Nursing Shifts Hurt Patient Care?

The RN4CAST survey collected questionnaires from nurses on adult general medical and surgical wards in 488 hospitals across 12 European countries between June 2009 and June 2010. Of 54,140 questionnaires distributed, responses came back from 33,659 registered nurses (62%), and 31,627 of those worked the wards of interest and entered the analysis (Griffiths et al., 2014). Mean age was 38, 92% were female, 65% worked full time, and 76% reported on a day shift. Intensive care, high dependency, and long-term care units were deliberately excluded because their staffing and shift patterns differ, so none of these numbers transfer to ICU nursing or to any non-healthcare setting.

Half the sample had worked a shift of 8 hours or less (50%, n=15,930), 32% worked 8.1 to 10 hours (n=9,963), 4% worked 10.1 to 11.9 hours (n=1,160), 14% worked 12 to 13 hours (n=4,314), and 1% worked more than 13 hours (n=260). Those five counts account for all 31,627 nurses, with the percentages rounded. Shift length here means the hours worked on the most recent shift, not a contracted rotation. The survey did not capture usual pattern, fixed versus rotating work, weekly hours, break availability, or rest between shifts, all of which the authors name as unmeasured confounders. Total hours worked could not be measured directly either, so full-time versus part-time status stood in as a proxy.

Every outcome in both papers is a nurse's own rating: quality of care on the unit, a patient safety grade, and a count of care activities left undone on the last shift. There is no mortality, infection, fall, or medication error data anywhere in the study. The authors concede the outcomes are self-report, note that the clinical importance of the differences observed is unclear, and recommend that future work use objective measures. Nurse-reported quality and safety have been validated against mortality and failure to rescue elsewhere, but that is a separate body of work rather than a finding of this survey.

Against a reference group of shifts of 8 hours or less, nurses whose last shift ran 12 hours or more had 1.41 times the odds of grading patient safety poor or failing (95% CI 1.13 to 1.76) and 1.30 times the odds of rating quality of care poor or fair (95% CI 1.10 to 1.53). Those are odds, not case counts. An odds ratio of 1.41 means 41% higher odds of that rating, not 41% more safety incidents. The estimates come from multilevel models adjusted for shift type, ward type, patients per nurse, nurse age, full-time status, hospital size, high-technology status, and teaching status (Griffiths et al., 2014).

A dose-response gradient is largely absent for the patient-facing ratings. Odds of adverse quality and safety were raised at every shift length above 8 hours, but only the 12-hours-or-more category reached statistical significance, and the increase at 8.1 to 10 hours was described as marginal. A clean curve from 8 through 10 to 12 hours for quality and safety ratings is not supported by these estimates.

The one patient-facing outcome with a consistent gradient is care left undone. Nurses on shifts of 12 hours or more reported a 13% higher rate of care activities left undone on a 0 to 13 item count (RR 1.13, 95% CI 1.09 to 1.16), and that outcome reached significance across all shift lengths above 8 hours. This is a rate ratio on a self-reported count of omitted tasks, which makes it the closest thing in the study to a concrete operational measure and the most defensible one to replicate locally.

Working overtime on the shift, modeled independently of shift length, carried a stronger association with every outcome. Nurses who worked overtime had 1.67 times the odds of grading safety poor or failing (95% CI 1.51 to 1.86), 1.32 times the odds of rating quality poor or fair (95% CI 1.23 to 1.42), and a 29% higher rate of care left undone (RR 1.29, 95% CI 1.27 to 1.31). Every one of those estimates is larger than its shift-length counterpart (Griffiths et al., 2014).

Overtime was also more common than long shifts in this sample. A total of 27% of nurses (8,606) worked overtime on their last shift, ranging from 50% in England down to 12% in Poland, and from 0% to 80% across individual hospitals. The authors tested an interaction between shift length and overtime and it was not statistically significant, so in this data the two exposures appear to act independently rather than compounding one another.

The caveat that has to travel with the overtime numbers is that the survey could not distinguish mandatory from voluntary overtime, or paid from unpaid. That matters operationally, because a unit where nurses stay late by choice and a unit where they are held over are not the same problem even though they produce the same survey response. The very wide hospital range, from no overtime at all to 80% of nurses, shows how much the exposure varies between local settings.

The same cohort was analyzed for nurse-side outcomes. Compared with shifts of 8 hours or less, shifts of 12 hours or more were associated with higher odds of emotional exhaustion (aOR 1.26, 95% CI 1.09 to 1.46), low personal accomplishment (aOR 1.39, 95% CI 1.20 to 1.62), job dissatisfaction (aOR 1.40, 95% CI 1.20 to 1.62), and intention to leave the job due to dissatisfaction (aOR 1.29, 95% CI 1.12 to 1.48) (Dall'Ora et al., 2015). Two estimates sit on the edge of significance and should not be presented as settled: depersonalization (aOR 1.21, 95% CI 1.01 to 1.47) and dissatisfaction with work schedule flexibility (aOR 1.15, 95% CI 1.00 to 1.35). Job dissatisfaction did show a gradient below 12 hours, at aOR 1.15 (95% CI 1.05 to 1.25) for 8.1 to 10 hours and aOR 1.31 (95% CI 1.09 to 1.57) for 10.1 to 11.9 hours.

The wider literature is directionally consistent on long shifts but does not agree on where the worst point sits, and the authors say so themselves. They note prior US evidence that does not fit a simple linear hours effect: in one US study the odds of adverse reports of quality and safety were greater for nurses working 10 to 11 hours than for those working 12 hours or more, and a pediatric subsample showed elevated reports only above 13 hours. Both of those results reach this page through the RN4CAST discussion rather than through the primary US papers, so treat them as a reason the question stays open rather than as a settled US position. An earlier systematic review found insufficient evidence on shift length and nurse satisfaction, and nothing in the European survey rules out a mid-range peak.

Shift length in this survey is also heavily confounded with country, which is one reason the two literatures can point in different directions. Twelve-hour shifts were the norm in Poland (99% of nurses) and Ireland (79%), mixed in England (36% overall, 32% of day shifts and 37% of night shifts), and reported by fewer than 5% of nurses in Belgium, Germany, Greece, the Netherlands, Norway, and Sweden. The authors did not test interactions between country and shift work, so they can only estimate an average effect across all countries and cannot explore differences between them. Fieldwork also ran from June 2009 to June 2010, so this is evidence from a European survey now roughly 16 years old, predating post-2020 workforce conditions.