What Does Shift Work Do to Police Officers?

Rajaratnam et al. (2011) screened 4,957 North American police officers between July 2005 and December 2007. Two routes fed the sample: 3,693 officers answered an online survey of US and Canadian police, and 1,264 were screened on site at one municipal and one state police department. Sleep disorders were identified by validated screening questionnaires, including the Berlin Questionnaire for sleep apnea and the Athens Insomnia Scale, and 126 participants underwent polysomnography for comparison.

Among those screened, 40.4% screened positive for at least one sleep disorder. Sleep apnea accounted for most of that: 1,666 officers (33.6%) screened positive. A further 281 officers screened positive for moderate to severe insomnia, reported as 6.5%, a share taken from the officers who completed the Athens Insomnia Scale rather than from all 4,957, and 269 (5.4%) screened positive for shift work disorder, rising to 14.5% among officers who worked the night shift. Excessive sleepiness was reported by 1,312 of the 4,608 officers who completed the sleepiness scale (28.5%), a different denominator again, and 1,294 officers (26.1%) reported falling asleep while driving at least one time a month.

Two limits belong next to those percentages. A positive screen is not a diagnosis, and the authors note that the shift work disorder questionnaire remains to be validated, which makes the 5.4% and 14.5% figures the softest in the set. Selection is the larger problem: three quarters of the sample came from a self-selected online survey with an unknown denominator, and the authors concede they cannot confirm the sample represents all North American police officers. The 40.4% describes officers screened in this study, not the police workforce.

The prospective arm followed 3,545 officers with monthly surveys for up to two years, contributing 15,735 person-months. Officers who had screened positive for a sleep disorder were more likely to report an error or safety violation they attributed to fatigue (23.7% versus 15.5%, adjusted OR 1.63; 95% CI, 1.43-1.85), falling asleep during meetings (14.1% versus 7.0%, adjusted OR 1.95; 95% CI, 1.52-2.52), falling asleep while driving (14.4% versus 9.2%, adjusted OR 1.51; 95% CI, 1.20-1.90), and a serious administrative error (17.9% versus 12.7%, adjusted OR 1.43; 95% CI, 1.23-1.67).

Two further outcomes were weaker. Uncontrolled anger toward suspects was reported by 34.1% of screen-positive officers and 28.5% of screen-negative officers (adjusted OR 1.25; 95% CI, 1.09-1.43), and absenteeism by 26.0% versus 20.9% (adjusted OR 1.23; 95% CI, 1.08-1.40). An odds ratio of 1.63 means 63% higher odds of reporting the outcome, never 63% more incidents, and the authors themselves volunteer that the absolute difference in risk for some of these outcomes was relatively small. The gap between 23.7% and 15.5% is the honest version of the headline.

Nothing in this arm used administrative records. Every outcome was self-reported, from the fatigue-attributed errors to falling asleep at the wheel, and the authors flag recall bias, possible under-reporting because of stigma, and the possibility that officers preferentially completed a survey after a bad month. Physical activity and diet were not assessed and so remain uncontrolled. The comparison is also not a schedule comparison: it sets officers who screened positive against officers who screened negative, not night workers against day workers.

The other police-specific dataset in view is the Buffalo cohort. Violanti et al. (2025) followed officers across three examinations: 464 enrolled at baseline, 65.4% of the 710 invited, with 276 returning with baseline data at seven years and 191 at twelve. Metabolic syndrome prevalence was 10.7 percentage points higher at seven years (95% CI, 5.3-16.0; p=0.0001) and 7.4 points higher at twelve (95% CI, 0.1-14.0; p=0.028). Abdominal obesity was 15.1 points higher at seven years (95% CI, 9.6-20.7) and 16.1 at twelve (95% CI, 9.5-22.6), while glucose intolerance was 11.6 points higher (95% CI, 5.9-17.2) and then 11.7 (95% CI, 5.3-18.1).

Perception moved with the biology. The share of officers rating their health excellent or very good was 11.8 percentage points lower over seven years (95% CI, -17.8 to -5.9) and 17.3 points lower over twelve (95% CI, -24.2 to -10.4), and police-specific stress on the Spielberger scale was 3.4 points higher at twelve years (95% CI, 1.2-5.6; p=0.003). Not everything moved, though. Hypertension was 4.3 points higher at seven years (95% CI, -0.1 to 9.3) and 4.48 at twelve (95% CI, -1.65 to 10.6), and elevated triglycerides 1.8 points (95% CI, -3.2 to 6.8) and then 5.18 points (95% CI, -0.85 to 11.2). Those intervals cross zero, so a general claim that everything got worse is not supported.

The authors state that the paper is descriptive and that they are not examining exposure-outcome relationships, which is the sentence most likely to be dropped when the study gets quoted. A cohort measured at seven and twelve years cannot separate anything about the job from ordinary aging in the same people. Attrition compounds it: the twelve-year estimates rest on 191 officers, the interval on the twelve-year metabolic syndrome change nearly touches zero, and the twelve-year change (7.4 points) is smaller than the seven-year change (10.7 points), which most plausibly reflects who stayed in the study rather than a health improvement.

Neither study randomized anything, and neither compared one roster with another. The screening study associates a questionnaire result with self-reported outcomes over the following months, and the cohort study describes how one group of officers changed over a decade. A claim that shift work causes errors, crashes, anger, or metabolic disease in police officers cannot be sourced to either. The defensible reading is narrower: sleep disorders appear common and largely undiagnosed among the officers screened here, and screening positive was associated with reporting more fatigue-related problems.

The screening data were collected between 2005 and 2007 and published in 2011, so they are roughly two decades old, and policing schedules, fatigue policy, and awareness of sleep apnea have all moved since. JAMA issued a correction the following January (JAMA 2012;307(4):363), and reading it settles what it touched: it lists incorrect language in the text, in table footnotes, in row headings, in a column head, in the financial disclosures, and in the reference list, along with contributors omitted from the acknowledgments. No percentage and no odds ratio appears among the items it corrects, so the figures quoted above stand as published.

Independence is the one part still partly closed. The work was funded by the National Institute for Occupational Safety and Health, and the paper carries an extensive author disclosure statement whose visible opening records consulting work by the lead author. The rest of that statement sits behind the journal paywall, so the citation registry entry for this paper records it as read in part rather than in full, and a claim about the authors' commercial ties should not run past what the visible portion says.