Industrial action and emergency department attendances in England, 2022–2024: a retrospective multicentre causal mediation analysis
Steven Wyatt, Mohammed A. Mohammed, Jonathan Clarke, Marta Garcia-Fiñana, Iain BuchanObjectives:
Between December 2022 and July 2024, the English National Health Service (NHS) experienced 27 periods of industrial action spanning 78 days, involving hospital and ambulance staff. We examined the direct and indirect impacts of these strikes on emergency department (ED) performance
Design:
A retrospective causal mediation analysis. Effects were estimated using mixed effects accelerated failure time, linear and generalised linear regression models with patient- and hospital-level covariates within a causal mediation framework.
Setting:
20 major (Type 1) EDs in England with high-quality linkage across ED, inpatient, and imaging data.
Participants:
Patients attending these EDs between August 2022 and July 2024.
Main outcome measures:
A patient’s time in ED analysed in relation to strike activity at the mid-point of attendance by one or more staff groups: (1) resident/junior doctors, (2) consultants, (3) both resident and consultant doctors, (4) paramedics and (5) nurses.
Results:
Average duration of ED attendances reduced on days affected by residents’ strikes (-12.6%, 95% CI -10.4% to -14.8%), consultant strikes (-5.6%, 95% CI -2.7% to -8.4%), paramedic strikes (-4.5%, 95% CI -2.0% to -6.8%), but increased when residents and consultants were on strike simultaneously (+7.2%, 95% CI 3.8%–10.7%). There were no significant changes to ED durations during nurse strikes (-1.7%, 95% CI -1.5% to 4.7%). Changes in emergency inpatient occupancy explained a proportion of the change in average ED durations (resident doctor 10.4%, consultants 53.3%, combined doctors 5.4% and paramedics 14.3%). The impact of strikes on ED attendance volumes was mixed. The net effect of these changes explained 1.5%, 16.8% and 28.8% of the change in average ED durations observed during residents, combined doctor and paramedic strikes, respectively. During the residents’ strikes, patients were more likely to receive treatments in ED (adjusted incident risk ratio 1.013, 95% CI 1.008–1.018), but less likely to be admitted (adjusted odds ratio 0.963, 95% CI 0.946–0.980). Efforts to meet the 4-h target intensified during the resident, consultant and combined doctors strike, but reduced during the nurse and paramedic strikes. This explained 2.7% and 17.1% of the changes in average ED durations observed during the resident and consultant strikes, respectively. There were substantial reductions in many forms of planned hospital care during the resident, consultant and combined doctor strikes, but these did not contribute to the observed reductions in ED durations.
Conclusion:
Industrial action – particularly by resident and consultant doctors – was associated with shorter ED stays, partly mediated by greater emergency bed availability, lower demand and intensified throughput efforts. However, much of the reduction during resident strikes remained unexplained, suggesting unmeasured operational or behavioural adaptations. Substantial reductions in planned care did not significantly affect ED performance.