Risk of death or readmission among people discharged from hospital on Fridays.

The timing of patient interventions can significantly affect outcomes. A study in the United Kingdom showed that patients discharged from intensive care units at night had a higher hospital mortality than those discharged during the day. In a study of acute care admissions from emergency departments in Ontario, patients with some serious medical conditions were more likely to die in hospital if they were admitted on a weekend than if they were admitted on a weekday. Do patients discharged on Fridays have worse outcomes than those discharged on other days? Friday is the most common hospital discharge day. More discharges could result in patients receiving fewer discharge instructions from hospital staff. Perhaps because of decreased staffing on weekends and physician cross-coverage, patients may be preferentially discharged on Fridays rather than on subsequent weekend days. Some patients discharged on Fridays could therefore leave hospital before they are fully stable. Also, new home health and social support services for weekend discharges often are not initiated until the following Monday. Such a delay may result in poor outcomes for patients discharged on Fridays who need these services initiated immediately. For this study we used anonymous data from population-based administrative databases for Ontario. Data for all adults discharged from hospital to the community between Mar. 1, 1990, and Mar. 1, 2000, were extracted from the Discharge Abstract Database (DAD), which records all discharges from Ontario hospitals. For patients with 2 or more admissions, we randomly chose 1 admission for each patient using a randomnumber generator. Only nonelective admissions were included in the study. We used proportional hazards modelling to determine the association between discharge day and nonelective readmission to hospital (measured using the DAD) or death (measured using the Registered Patient Database) within 30 days after discharge while controlling for potential confounders. These confounders were determined from the DAD and included age, sex, comorbidities (measured using the Charlson–Deyo score6), nonelective hospital admission during the 6 months before the index admission, length of stay, whether a procedure was performed and whether a complication occurred. In the proportional hazards model, patients were observed for 30 days after discharge or until the occurrence of an event (nonelective readmission or death). Databases were linked using common patient identifiers. The study was approved by the Sunnybrook & Women’s College Health Sciences Centre Research Ethics Board. A total of 2 403 181 patients met our inclusion criteria. Friday was the most common discharge day (Fig. 1). Overall, 7.1% of the paRisk of death or readmission among people discharged from hospital on Fridays Research letter