Risk of Hospital Readmission for Older Adults Discharged on Friday

Hospital readmission is an important indicator of quality of care, a significant driver of healthcare costs, and a target for improving delivery of complicated healthcare services for the growing older adult population. Hospital readmission (hereinafter readmission) is common in older adults; a recent study reported that nearly 20% of all Medicare recipients were readmitted within 30 days of hospital discharge.1 Unplanned readmissions alone are estimated to cost Medicare between $12 billion and $17 billion annually.1, 2 Anecdotally, inpatient discharges occurring on Fridays are thought to be associated with greater risk of readmission because of higher discharge volume, less time for discharge planning, and longer wait time to receive postdischarge services. Although previous research has demonstrated that system-level admission factors, including time and season, have effects on quality of care and mortality,3-6 there is a dearth of evidence about the relationship between timing of discharge and patient outcomes.7 One Canadian study found that the largest volume of discharges occurred on Fridays and that patients discharged on Fridays had higher rates of readmission and mortality.8 Identification of day of discharge as a modifiable readmission risk factor would provide an opportunity to improve care. The current study aimed to describe patient characteristics according to day of discharge and to determine whether Friday discharges are associated with hospital readmission. It was hypothesized that, for hospitalized older adults, Friday discharge may result in increased risk of readmission. This was a retrospective cohort study of hospitalized older adults aged 65 and older admitted to an academic teaching hospital from January to December 2007 (N = 6,580). The primary outcome was risk of hospital readmission within 30 days after discharge. A Cox proportional hazards model was used to compare hospitalized older adults discharged on Fridays with all others, adjusting for participant characteristics, healthcare utilization, primary diagnosis, and discharge disposition. Variables were included in a multivariate model if they were deemed conceptually necessary (day of discharge, age, sex, race, and Medicaid status). Backward stepwise selection (P = .20) was used to select the remaining variables for inclusion. There was 80% power (α = .05) to detect a 35% or greater change in the hazard ratio between those discharged on Fridays and all other discharge days. Hospitalized older adults in the study had a mean age ± standard deviation of 74.2 ± 7.0 and were 52.2% male and 68.9% white. Hospitalized older adults discharged on Fridays were significantly more likely to have been treated on a surgical service (P = .04) and have had a shorter length of stay (P < .001) than those discharged on all other days. Hospitalized older adults were similar in other demographic, hospital stay, and discharge locations. Friday was the most frequent day of discharge, with 20% of all discharges occurring on Fridays (P < .001). The fewest discharges occurred on Mondays (11.2%) and Sundays (7.5%). Discharge location was similar between groups, with discharges to rehabilitation and skilled nursing facilities being slightly higher for those discharged on Fridays (10.1%) than on all other days (8.4%) (P = .05). In adjusted time-to-event analysis, hospitalized older adults discharged on Fridays did not have a significantly greater risk of 30-day readmission (hazard ratio (HR) = 1.02, 95% confidence interval (CI) = 0.84–1.24; P = .83) (Table 1). The effect of alternative day of discharge definitions on the multivariate model was investigated. Models comparing weekends (Friday–Sunday, Saturday–Sunday) with all other days did not produce significant differences in risk of readmission. When examining all days of the week separately, Monday discharge was found to have the greatest risk of readmission within 30 days (HR = 1.28, 95% CI = 0.92–1.76). In this large, retrospective cohort of more than 6,500 older adults, it was found that Friday was the most common day for hospital discharge. Discharge on Friday was not associated with greater risk of readmission than discharge on other days of the week after adjusting for individual and hospital stay factors. The results indicate that the volume of discharges varies according to day of the week, possibly reflecting patients' characteristics and hospital practice patterns, but there were insufficient data to determine that the greater volume of discharges resulted in greater readmission rates for hospitalized older adults. Although it is reassuring that this study did not find that greater volume of discharges compromised the quality of care, it may be hard to assess a small difference in risk with this study. To further investigate the relationship between Friday discharge and readmission, future studies using large, comprehensive databases are needed. We would like to thank David Marion for help with the administrative database and Qilu Yu for help with statistical concepts. Conflict of Interest: Dr. Graham was supported by Johns Hopkins Pre-doctoral Clinical Research Training Program Grant 1TL1RR-025007 from the National Center for Research Resources, a component of the National Institutes of Health. Dr. Arbaje is a former Robert Wood Johnson Clinical Scholar at Johns Hopkins University, supported by Robert Wood Johnson Foundation Grant 047945, Princeton, New Jersey. She is also a former Clinical and Research Fellow in the Johns Hopkins Division of Geriatric Medicine and Gerontology, supported by National Institute of Aging Grant T32-AG-000120. At the time this research was conducted, Dr. Arbaje was a Robert Wood Johnson Foundation Harold Amos Medical Faculty Development Program Scholar supported by Grant 63518. She also received support from the John A. Hartford Foundation. Author Contributions: Dr. Graham: Study concept and design; acquisition, analysis, and interpretation of data; preparation of manuscript. Dr. Leff: Study concept and design, interpretation of data, preparation of manuscript. Dr. Arbaje: Study concept and design, acquisition and interpretation of data, preparation of manuscript. Sponor's Role: None.