Letter to the editor: association between delays to patient admission from the emergency department and all-cause 30-day mortality

We read with interest the recently published article by Jones et al. This paper provides important data to inform the debate around the current 4hour target and the currently considered changes laid out in a recent government white paper. The use of national admissions data linked with robust mortality information is a considerable undertaking and the authors should be congratulated on their methods. This study potentially represents the best information available to make the judgements required in planning for the future of emergency care. Because of the significant implications of this work, we would like to highlight some important considerations for its interpretation. In particular, while the authors have presented convincing data to demonstrate an association between length of ED stay and 28day mortality, it is imperative to understand that this does not necessarily imply that long ED stays cause the increased mortality rate. The authors calculated a standardised mortality rate adjusting for available confounders. They acknowledge that, due to lack of available data, they were unable to account for acuity or severity of illness. However, in our experience, this is the most important potential confounder. Patients with high acuity will often remain in the ED beyond 4 hours either because (a) remaining in the ED adds value to their care (for example, allowing more intensive monitoring and/or resuscitation in the early phase after initial presentation); or (b) they must wait for the availability of scarce high dependency inpatient beds. Therefore, it is possible that the effect demonstrated is largely explained by critically ill patients being more likely to stay in the ED beyond 4 hours, rather than those patients dying because they received suboptimal care in the ED. We recognise that there is a paucity of available data to account for disease severity and acuity. This highlights the inherent challenges with investigating the effectiveness of target times. We do, however, note that the Hospital Episode Statistics Admitted Patient Care database includes data on admission to critical care beds. We wonder if the analysis reported could be extended to evaluate any association between length of stay in the ED and admissions to critical care areas. Such an analysis may help to partly understand the potential confounding effect of acuity. The 4hour target is representative of a system built around the target itself. ED processes, specialty admission teams, bed managers and porters are all geared to preventing 4hour breaches. In a sense, after the 4hour target was introduced, an intentional Hawthorne effect was triggered where stakeholders knew they were being observed so they changed their behaviour. As such, when examining the effect of the 4hour target on mortality, we are observing not just an association between length of ED stay and outcome but in fact the effect of the entire healthcare system on outcomes. This is pertinent since the data used were from April 2016 to March 2018—where the 4hour target has been implemented throughout this period. Therefore, it is hard—if not impossible—to make (counterfactual) generalisations about what might happen to outcomes (mortality) should future ED target times change. Any change in target times would mean the system would again adapt around that time point. Comparing between historical data where the target time did not exist, and matching these (eg, using propensity scores and clustering by ED) to contemporary data, might be informative for future work. In summary, the paper by Jones et al demonstrates a fascinating and valuable insight into the association between time and mortality, but as acuity has not been accounted for, we must not infer a causal relationship. The study also raises several interesting questions about how we measure the effectiveness and harm of processes within our systems, about the impacts of time targets and how we can address their limitations in future.