Is the weekend effect really ubiquitous? A retrospective clinical cohort analysis of 30-day mortality by day of week and time of day using linked population data from New South Wales, Australia

Objective To examine the associations between day of week and time of admission and 30-day mortality for six clinical conditions: ischaemic and haemorrhagic stroke, acute myocardial infarction, pneumonia, chronic obstructive pulmonary disease and congestive heart failure. Design Retrospective population-based cohort analyses. Hospitalisation records were linked to emergency department and deaths data. Random-effect logistic regression models were used, adjusting for casemix and taking into account clustering within hospitals. Setting All hospitals in New South Wales, Australia, from July 2009 to June 2012. Participants Patients admitted to hospital with a primary diagnosis for one of the six clinical conditions examined. Outcome measures Adjusted ORs for all-cause mortality within 30 days of admission, by day of week and time of day. Results A total of 148 722 patients were included in the study, with 17 721 deaths within 30 days of admission. Day of week of admission was not associated with significantly higher likelihood of death for five of the six conditions after adjusting for casemix. There was significant variation in mortality for chronic obstructive pulmonary disease by day of week; however, this was not consistent with a strict weekend effect (Thursday: OR 1.29, 95% CI 1.12 to 1.48; Friday: OR 1.25, 95% CI 1.08 to 1.44; Saturday: OR 1.18, 95% CI 1.02 to 1.37; Sunday OR 1.05, 95% CI 0.90 to 1.22; compared with Monday). There was evidence for a night effect for patients admitted for stroke (ischaemic: OR 1.30, 95% CI 1.17 to 1.45; haemorrhagic: OR 1.58, 95% CI 1.40 to 1.78). Conclusions Mortality outcomes for these conditions, adjusted for casemix, do not vary in accordance with the weekend effect hypothesis. Our findings support a growing body of evidence that questions the ubiquity of the weekend effect.