---
title: "Adverse nurse working hours and increased sickness absence: longitudinal e‑roster analysis"
id: "bmj-open-16-associations-between-adverse-working-hours-and-nurses-sickness-absence-a"
canonical_url: "https://medichelpline.com/clinical-feed/bmj-open-16-associations-between-adverse-working-hours-and-nurses-sickness-absence-a"
content_type: "clinical_feed_article"
specialty: "General"
source_name: "BMJ Open"
source_url: "http://bmjopen.bmj.com/cgi/content/short/16/7/e120066?rss=1"
published_at: "2026-07-21T12:12:52.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Adverse nurse working hours and increased sickness absence: longitudinal e‑roster analysis
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/bmj-open-16-associations-between-adverse-working-hours-and-nurses-sickness-absence-a
- **Specialty:** [General](https://medichelpline.com/clinical-feed/general.md)
- **Primary Source:** BMJ Open
- **Source URL:** [Original Journal Publication](http://bmjopen.bmj.com/cgi/content/short/16/7/e120066?rss=1)
- **Published At:** 2026-07-21T12:12:52.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- This retrospective longitudinal study analysed 1.4 million electronic roster records and 19,876 sickness episodes from 7,515 registered nurses on 95 adult acute wards across two NHS Trusts (April 2015–September 2020). - The primary aim was to examine associations between recent **shift work patterns** (7‑day rolling windows) and onset of sickness absence using logistic mixed‑effects models that adjusted for total hours, bank hours, previous sickness and part‑time status. - Main exposure variables included proportion of **long shifts** (≥12 hours), proportion of **night shifts**, spells of consecutive shifts (≥3 intense; ≥6 long), quick returns (≤11.5 hours), short returns after night‑to‑day rotations (≤48 hours) and shift rotations. - In full multivariable models, each additional quick return, intense spell and shift rotation in the previous 7 days increased odds of sickness absence (ORs: 1.23 for quick returns; 1.24 for intense spells; 1.09 for shift rotations). - Nonlinear modelling showed working about 70% of shifts as long (≥12 hours) was associated with greatest odds of sickness (OR 1.68, 95% CI 1.53–1.84) relative to none; working 100% night shifts modestly increased odds (OR 1.07, 95% CI 1.01–1.13). - Some rare configurations (long consecutive spells ≥6 shifts) were associated with reduced odds, but these were infrequent in the dataset and estimates were imprecise. - Sensitivity analyses with 28‑day lookback windows and pre‑March 2020 data showed the same directions of effect but attenuated magnitudes; the 7‑day window provided a more parsimonious fit (lower AIC/BIC). - Models accounted for nurse‑level clustering (ICC=0.710); ward‑level clustering was not significant. - Strengths: large electronic roster linkage, shift‑level exposure linked to sickness onset, modelling of nonlinear cumulative exposures. - Limitations: lack of demographic covariates and inconsistent overtime data due to governance restrictions; no source data available for sharing under data agreements. - Implication: organisational scheduling features — especially **short recovery times**, frequent rotations, consecutive intense shifts and high proportions of long/night shifts — are associated with higher odds of nurse sickness absence and can inform shift‑planning policies to support staff well‑being.
## Clinical Analysis & Structured Key Points
Skip to main content Intended for healthcare professionals Log In Basket Search for this keyword Advanced search Latest content Archive For authors About Browse by collection You are here Home Archive Volume 16, Issue 7 Email alerts Article Text Article info Citation Tools Share Rapid Responses Article metrics Alerts PDF Health services research Original research Associations between adverse working hours and nurses’ sickness absence: a longitudinal analysis of e-roster data from acute hospital wards http://orcid.org/0000-0001-5595-685XTalia Emmanuel1, http://orcid.org/0000-0003-2439-2857Peter Griffiths1,2, http://orcid.org/0000-0001-5329-7619Carlos Lamas-Fernandez3, http://orcid.org/0000-0002-6858-3535Chiara Dall’Ora1,2 Correspondence to Dr Talia Emmanuel; T.Emmanuel@soton.ac.uk Abstract Objective To examine the associations between registered nurses’ shift work patterns and sickness absence in hospital inpatient wards. Design Retrospective longitudinal study using 1.4 million e-roster and sickness absence records analysed with logistic mixed-effects models. Setting All adult acute inpatient hospital wards in two large healthcare organisations (Trusts) in England. Participants Registered nurses (N=7515) working across 95 adult acute wards in 2015–2020. Primary outcome measure Nurses’ sickness absence in relation to the shift pattern configurations worked in the previous 7 days. Logistic mixed-effects models were adjusted for total working hours, bank hours, previous sickness and part-time status. Results Higher odds of sickness absence were significantly associated with each additional rest period of fewer than 11.5 hours (OR 1.23, 95% CI 1.19 to 1.27, p<0.001), change between day and night shifts (OR 1.09, 95% CI 1.04 to 1.13, p<0.001) and spell of 3 or more consecutive shifts (OR 1.24, 95% CI 1.16 to 1.32, p<0.001) in the previous 7 days. By accounting for nonlinearity, analyses revealed that working 70% of shifts as ≥12-hour shifts and 100% of shifts as night shifts were associated with increased odds of sickness (OR 1.68, 95% CI 1.53 to 1.84 and OR 1.07, 95% CI 1.01 to 1.13, respectively) when compared with working none of these shift types. Sensitivity analysis using 28-day exposure windows did not change the direction of effects; however, the magnitude and significance of some associations attenuated. Conclusions This study highlights the organisational consequences of adverse working hours arrangements for registered nurses working in hospital wards. Short recovery periods, frequent shift rotations, consecutive working spells and high proportions of long and night shifts were associated with higher odds of sickness absence, with associations most apparent over 7-day exposure windows. Overall, these findings can inform shift planning policies that better support staff well-being through proactive management of sickness absence. Data availability statement No data are available. Due to the data sharing agreements with the providers, we are unable to share the source data. https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/. https://doi.org/10.1136/bmjopen-2026-120066 Request Permissions If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways. Request permissions Strengths and limitations of this study Analysis of registered nurses’ shift working patterns from comprehensive dataset of electronic roster records from two large health organisations. Linkage of roster and sickness data allowed working hours exposures to be examined in direct relation to the onset of sickness absence. Mixed-effects modelling that captured non-linear features of recent shift scheduling and repeated observations within individual nurses. To offset absence of demographic information and inconsistent records of overtime, other variables such as nurses’ prior sickness absence and total hours worked in exposure windows were included in models. Introduction In healthcare settings such as hospitals, nursing staff are routinely required to work non-standard hours (ie, outside of 07:00 and 19:00, Monday–Friday), extended work shifts (eg, working for 12 hours or longer) and rotating day–night schedules. These patterns of work are inherently disruptive to personal circadian rhythms, with implications for physiological functions such as hormone release, sleep–wake cycles and metabolism.1 Consistent with this, there is a wealth of international evidence that links shift work with negative outcomes for staff, including increased fatigue and burnout, poor work–life balance and development of chronic illness or cancer in the long term.2–5 Healthcare employers therefore have a duty to ensure that working hours are organised in ways that minimise these risks and protect the well-being of their nursing workforce. One organisational outcome useful for monitoring workforce well-being is staff sickness absence, as documented by historical rosters and/or payroll records. Administrative records of shifts cancelled due to sickness absence offer a readily accessible interpretation of staff wellness: when calling in sick, it can be inferred that most staff do not feel able to work because they are not well. Furthermore, significant upticks in rates of sickness absence among different working environment exposures can provide clues as to where targeted strategies for improvement are warranted. Elevated sickness absence rates place considerable strain on healthcare organisations6 and can force reliance on costly temporary staffing or worsen patient well-being and other measures when staffing levels fall below planned thresholds.7 8 Recent national data on sickness absence rates among registered nurses working in England’s National Health Service (NHS) show some concerning trends in this regard.9 These data reveal higher levels of sickness when compared with those prior to the start of the COVID-19 pandemic, as well as when compared with other health professions and the public sector overall. The most common recorded reasons for sickness absence (when measured by percentage of full-time equivalent (FTE) days lost due to sickness) were related to anxiety, stress, depression and other psychiatric illness (25%).9 Furthermore, 46% of registered nurses responding to the annual staff survey conducted by the NHS10 reported feeling unwell as a result of work-related stress specifically over the last 12 months. Although preventing all sickness absence is not possible, any harmful contribution of nurses’ working environments should be minimised. Previous research analysing administrative records of shifts and sickness absence has shown increased rates when nursing staff are working certain configurations of shifts, including long shifts (ie, ≥12 hours), night shifts, long weeks (ie, ≥48 working hours per week) and quick returns (ie, <11 hours of inter-shift recovery time).11–15 However, there is a gap in understanding the effects of cumulative patterns of work and recovery, particularly those that occur across multiple days. The present study builds on this previous research by examining how sequences of shift intensity across consecutive days, total weekly working hours and intershift rest periods are associated with nurses’ sickness absence. Methods Study design, setting, participants We conducted a retrospective longitudinal analysis of historical shift and sickness absence data recorded in electronic staff rostering systems from all adult acute inpatient wards in two NHS hospital Trusts in England. Original data were collected and anonymised as part of a larger project exploring the staff-related, patient-related and cost-related consequences of different healthcare staffing configurations.16 Administrative pay bands were used to identify registered nurses (band 5 or above). Unique study identifiers were used to link shifts and sickness episodes to the same nurse across the study period, and therefore, all variables and analyses were calculated at the shift-per-nurse level. Shifts that were not worked due to sickness were aggregated into episodes, starting on the first day that a nurse was absent from work and finishing on the day they returned for at least one shift. Demographic information for staff (eg, age, year join/left hospital, number of years in current role) were not available due to data governance restrictions from participating hospitals. For the present study, we analysed the roster and sickness absence records of all registered nurses scheduled to work on wards between April 2015 and September 2020. Variables We created a series of shift pattern variables that accounted for the following configurations: long working hours, night work, spells of consecutive working days, inadequate recovery time and shift rotations. Each variable was defined by a 7-day exposure period, that is, the measurements taken in rolling windows of 7 days prior to each worked shift and sickness absence episode. This exposure length reflects the weekly structure through which legal and contractual working-time limits are typically defined. Variables were also created in accordance with an established framework for characterising working time exposure in relation to staff well-being,17 as well as with nurses’ more recent reports of the difficulties of working several consecutive shifts or transitioning from night to day duties with inadequate recovery time.18 19 Specifically, our main exposure variables included: Proportion of shifts worked as long shifts (shifts lasting 12 hours or more). Proportion of shifts worked during the night (shifts that finish at 08:00 or earlier). Number of spells of consecutive shifts, that is, ‘long’ spells (≥6 consecutive shifts) and ‘intense’ spells (≥3 consecutive long or night shifts). Number of inadequate rest periods, that is, ‘quick’ (≤11.5 hours rest between consecutive shifts) and ‘short’ returns (≤48 hours rest between a night-to-day shift rotation). Number of shift rotations (including night-to-day and day-to-night rotations). Proportions were used to analyse exposure to long working hours and night work to enable consistent comparison of exposure effects across lookback windows with varied total working hours. Initial exploratory (categorical) analysis of these variables indicated that relationships may not be linear; we therefore added quadratic and cubic terms to our models. We also included other working time-related covariates that may confound nurses’ sickness absence: total number of working hours and total number of hours worked as bank (ie, when voluntarily working shifts that cover temporary shortfalls in settings that are usually different to one’s ‘home’ role or ward). We additionally controlled for the number of previous sickness absence episodes, as well as part-time status, or working fewer than a median of 0.75 FTE hours per week in the previous quarter (ie, median of ≤26 hours per week in the previous 13 weeks).20 Sickness episodes that were preceded by zero working hours in the previous month were removed. Following this predefined exclusion, all records and exposure windows were retained in models. Statistical methods, sensitivity analyses We used random intercept logistic mixed models to estimate the association between shift pattern variables and the outcome of sickness absence. Repeated shift-level observations were clustered within nurses and accounted for by using nurse-level random intercepts. Intraclass correlation coefficients (ICC) were calculated to quantify clustering within nurses and wards. Significant clustering was observed at the nurse-level (adjusted ICC=0.710) but not at the ward-level; therefore, additional ward-level effects were not included in models. We tested the associations between shift pattern variables on nurses’ sickness absence via: univariable models, which examined each main shift pattern variable independently, and full multivariable models, which included all shift pattern and control variables. Akaike information criterion (AIC) and Bayesian information criterion (BIC) values were used to compare alternative model specifications, including exposure window lengths and quadratic/cubic term forms, with lower values indicating improved model fit. To test for multicollinearity between predictors, we calculated variance inflation factors (VIF), where values <5 indicated low multicollinearity.21 Two sensitivity analyses were conducted: (1) testing variables with 28-day lookback windows to assess whether any observed effects persisted over a longer exposure period and (2) re-analysing with pre-March 2020 data exclusively to compare and examine if associations were influenced by the onset of the COVID-19 pandemic in England. Shift variables were created with the pandas22 and datetime Python packages; modelling was undertaken using the lme4 package in R (V.4.4.0).23 Results Descriptive statistics The final dataset contained 1 367 497 worked shifts and 19 876 sickness absence episodes from 7515 registered nurses across 95 wards. The majority of worked shifts were from nurses working full-time (3789 nurses working 821 681 shifts (60%)) and sickness episodes lasted a median of 4 days (IQR 2–8 days). The average number/proportion of each shift pattern configuration for full-time and part-time nurses for the entire study period is shown in table 1. Median shift length and the proportion of long shifts worked were similar between the groups; however, full-time nurses worked more night shifts, quick returns and shift rotations compared with part-time nurses. Part-time nurses had greater variability in their schedules, suggesting more heterogeneous work patterns. VIEW INLINE VIEW POPUP Table 1 Shift patterns worked by full-time (FT) and part-time (PT) nurses in the previous 7 days Online supplemental table S1 presents a yearly snapshot of the shift pattern variables worked by full-time nurses specifically. Statistics demonstrate a stable pattern over the 5-year study period. Long (≥6 shifts) and intense (≥3 long or night shifts) consecutive spells were rare across the dataset, however, a slight increase in counts for the latter is noted from 2019 onwards. Similar increases are seen for the mean number of long shifts (2.4 in 2015 vs 2.5 in 2020), the mean proportion of night shifts (0.3 in 2015 vs 0.4 in 2020), and the number of quick returns (mean of 0.9 in 2015 vs 1.0 in 2020), indicating that these shift configurations became more frequent over the study period. Supplemental material [bmjopen-2026-120066supp001.pdf] Multivariable models In the full multivariable model (table 2), multicollinearity was low across all predictors (VIF <5, excluding polynomial terms where elevated values were expected). Reduced odds of sickness absence were observed among nurses who had experienced a sickness episode within the preceding 7 days. In contrast, other covariates either had negligible or nonsignificant associations with sickness absence. VIEW INLINE VIEW POPUP Table 2 Shift pattern configurations worked in the previous 7 days and odds of sickness For every intense spell of work, quick return and shift rotation, there was a 24%, 23% and 9% respective increase in the odds of sickness. Working long consecutive spells significantly and considerably decreased odds of sickness (OR 0.09, 95% CI 0.01 to 0.66), however, this configuration was rare with only 1937 cases across the dataset. Download figure Open in new tab Download powerpoint Figure 1 Cubic and quadratic curves for proportion of long shifts and night shifts in the previous 7 days. Furthermore, all variable terms for proportion of long shifts (linear, quadratic, cubic) and proportion of night shifts (linear, quadratic) returned as statistically significant (p≤0.001) (table 3). Table 4 and figure 1 show how odds of sickness fluctuated across proportions of long shifts and night shifts. Working ≥40% of shifts as long increased odds of sickness (relative to working no long shifts), with the highest odds observed around the 70% mark (OR 1.68, 95% CI 1.53 to 1.84). Working 100% of shifts as night also slightly increased odds of sickness (relative to working no night shifts) (OR 1.07, 95% CI 1.01 to 1.13). VIEW INLINE VIEW POPUP Table 3 Non-linear terms for proportion of long shifts and night shifts in the previous 7 days VIEW INLINE VIEW POPUP Table 4 Plotting values for non-linear curves—proportion of long shifts and night shifts in the previous 7 days As a sensitivity analysis, we re-estimated associations using a 28-day exposure window, the results of which are reported in online supplemental tables S2, S3 and figure S1. The 28-day multivariable model showed higher AIC and BIC values, indicating that the 7-day exposure window model provided a more parsimonious fit. For most shift pattern variables, the direction of effects was the same across the two exposure windows, although effect sizes were attenuated in the 28-day model. Both measures of consecutive working spells were not statistically significant in the 28-day model, indicating that the influence of this element of shift intensity was more pronounced over weekly timeframes. Further sensitivity analyses restricting data to pre-March 2020 records showed minimal differences in effect sizes (ie, consistent direction and magnitude of effects with similar CIs), confirming that our primary findings were not influenced by any unrealised, pandemic-related changes in working hours configurations. Discussion The purpose of this observational study was to gain a comprehensive understanding of how adverse working time configurations are associated with sickness absence among registered nurses. This was achieved by analysing 1.4 million historical shift and sickness absence records collected from acute inpatient wards in two NHS hospital Trusts between 2015 and 2020. Compared with previous research on the relationships between shift type configurations and indicators of workforce well-being, this analysis explored the cumulative influence of patterns of work and rest, showing that higher proportions of long hours and night work, quick returns (having fewer than 11.5 hours of inter-shift recovery), intense spells of consecutive shifts (≥3 long or night shifts), and shift rotations were linked with increased odds of sickness absence. In our study, nurses working high proportions of long shifts and night shifts were associated with increased odds of sickness absence, with the highest odds observed when 70% of shifts were long (≥12 hours) or when working all night shifts in lookback windows. These findings mirror those of previous studies conducted within England. For example, in an analysis of 601 282 shift records from a large acute hospital, when 75% or more of shifts were worked as long shifts or night shifts in the past 7 days, the odds of sickness absence were increased when compared with working no long shifts (24% increase in odds) or day shifts only (12% increase in odds).11 12 Similarly, in a study that examined a pre-versus-post change in sickness absence rates in a large mental health hospital, an increase in the percentage of sickness hours per week (ranging from 0.73% to 0.98%, or, 1 shift per ward per week) was found following the organisational implementation of long shifts.15 However, the non-linear relationships observed in this study provide new and
## Related Clinical Research

- [Fear of COVID-19 and Coping Strategies in Female University Students: Role of Physical Activity an](https://medichelpline.com/clinical-feed/plos-one-20-factors-associated-with-fear-and-coping-strategies-during-pandemic-in-female.md)
- [Nurse Practitioner Trainees Face Clinical Mentor Shortage and Preceptor Challenges](https://medichelpline.com/clinical-feed/stat-news-3-nurse-practitioner-trainees-find-themselves-stymied-by-shortage-of-clinical.md)
- [Limits of Postpartum Psychiatry Revealed by the Lindsay Clancy Case](https://medichelpline.com/clinical-feed/stat-news-1-opinion-the-lindsay-clancy-case-shows-the-limits-of-postpartum-psychiatry.md)
- [Impact of Drug Decriminalization on Mental Health: Insights from Oregon and Washington](https://medichelpline.com/clinical-feed/medrxiv-1-drug-decriminalization-and-population-mental-distress-evidence-from-oregon-and.md)
- [Impact of population-level alcohol reduction to low-risk drinking on anxiety and depression preval](https://medichelpline.com/clinical-feed/medrxiv-21-prevalence-of-anxiety-and-depression-symptoms-under-population-level-alcohol.md)

## Navigation
- [← Back to General Feed](https://medichelpline.com/clinical-feed/general.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.