• accreditation
• attitudes
• audit and feedback

‘Time is tissue’ has become a mantra in emergency care: delays in treatment limit the opportunity to minimise damage caused by trauma and ischaemia, particularly for conditions like acute stroke or myocardial infarction where drugs and techniques can restore blood flow. The therapeutic time window may have widened for some forms of acute ischaemic stroke,1 2 but that does not mean that time is no longer critical. For ST-segment elevation myocardial infarction (STEMI), survival rates are higher if balloon angioplasty is performed earlier than 90 min from hospital presentation.3 4 Efforts to improve patient outcomes further by reducing time to treatment have taken a ‘whole pathway’ approach, from initial symptoms in the community to definitive intervention in the hospital. These have included raising public awareness,5 evaluating different routes of urgent referral to hospital6 and investigating the timing of admission to the hospital. There is now an extensive, but contradictory, literature on the impact of ‘off-hours’ admission on care processes in patients with a STEMI. A systematic review in 20147 reported a higher mortality and longer door-to-balloon times for weekend admissions, while a 2017 review8 found neither. A recent analysis of 4590 patients between 2005 and 2016 in the multicentre Melbourne registry9 has reported longer door-to-balloon times for weekend admissions (median 83 min vs 60 min weekday), but a similar case mix and no difference in outcomes. What are we to make of these results?

In this issue of BMJ Quality & Safety, Jayawardana and colleagues10 have examined the relationship between admission timing (nights and weekends vs weekday days), hospital and 30-day postadmission mortality rates, and door-to-balloon times for 42 677 patients with ST-elevation myocardial infarction whose details were available in a national clinical registry of patients with acute coronary syndromes presenting to hospitals in England and Wales between 2007 and 2012. They adjusted for severity of illness using the validated adjusted mini-Global Registry of Acute Coronary Events (GRACE) risk score,11 which includes acute physiology, troponin, creatinine and diuretic use. The authors find no difference in unadjusted outcomes between in-hours and off-hours admissions (30 day mortality rates 3.91% vs 3.86%). However, they identify a higher GRACE score for the in-hours admissions, indicating that these patients were sicker, and when they adjust outcomes for severity, this reveals a higher than expected mortality rate in the off-hours population. Median door-to-balloon times were also 7 min longer in the off-hours population (45 min vs 38 min), and longer door-to-balloon times were associated with an increased mortality risk. Incorporating time to intervention in the model attenuated the mortality difference by 38%.

This is an interesting inversion of the finding by several other groups that the higher mortality observed on weekends (the so-called weekend effect) reflects increased severity of illness among patients seeking care on weekends.12–15 In other words, these studies have suggested that increased hospital mortality on weekends results from increases in risk of bad outcomes (ie, having sicker patients), not delays in care due to inadequate staffing or other quality problems potentially present on weekends. By contrast, Dr Jayawardana and colleagues report their out-of-hours population as having lower risk of death, slightly younger and with lower incidence of stroke or cerebrovascular disease and chronic renal failure. Yet, the authors find no difference in crude mortality when, according to their baseline risk factors, the mortality rate on weekends should have been lower because the out-of-hours admissions were less severely ill.

The authors conclude that the higher-than-expected mortality in the off-hours admissions may in part be attributed to the delay in coronary intervention. There are perhaps two additional comments to make. The first is that their UK centres’ door-to-balloon times were commendably short, whether in-hours or out-of-hours. The second is that they had to exclude from analysis 20 234 patients from whom complete GRACE data were unavailable, and although the door-to-balloon time and the mortality in this excluded population were in the same direction as those with complete GRACE scores, the magnitude of the differences was larger in the non-GRACE off-hours population, suggesting that the absence of these data might not be a random event.

The weekend effect was first described by Bell and Redelmeier in 2001,16 who reported a significantly higher weekend admission mortality rate for 23 of 100 leading causes of death in Canada. It is a complex metric, a ratio of two ratios or rates: the mortality rate for weekend admissions divided by the mortality rate for weekday admissions. It therefore has two numerators and two denominators at work, each of which can be influenced by different factors in the causal pathway. The majority of publications on this topic have quantified the magnitude of the weekend effect. A recent systematic review identified more than 600 studies, of which 68 analysed outcomes for 640 million unselected emergency and elective weekend admissions to the hospital: the pooled excess relative risk of death for weekend admissions was 16%.17 However, very few examine causation, and the study by Jayawardana et al therefore contributes usefully to this scanty literature. In the perceptive accompanying editorial to Bell and Redelmeier’s publication, Halm and Chassin18 stated that ‘Disentangling the potential causal pathways would require painstaking detective work… of first accounting for the biologic and social determinants of risk and then identifying the precise differences in processes of clinical care that explain the differences in risk-adjusted outcomes’.

This ‘painstaking work’ has started. A detailed case-record review study comparing quality of care and outcomes for patients in two epochs, before (2012–2013) and after (2016–2017) health policy changes to promote 7-day services,19 will appear shortly. However, we already know that there does not appear to be a relationship between weekend consultant staffing and outcomes20 or 7-day service standards,21 that weekend admissions tend to be more acutely ill,12–15 that fewer patients are admitted at weekends despite the same numbers of patients attending hospital emergency departments12 22 and that there is a marked reduction in the proportion of patients referred to hospital by their general practitioners.12 22 The search for the cause of the weekend effect clearly needs to include community services, but Jayawardana and colleagues provide a timely reminder that it is the whole patient pathway which is of interest, not just isolated segments.