Hosp Pract (1995). 2021 Jan 27. doi: 10.1080/21548331.2021.1882238. Online ahead of print.
BACKGROUND: Given high costs for inpatient stays, hospital systems are investigating ways to decrease lengths of stay while ensuring high quality care. The goal of this study was to determine if patients in teaching teams (hospitalist teams with residents and interns) had a higher length of stay after adjusting for relevant confounders compared to hospitalist-only teams (staffed only by attending physicians).
METHODS: Using a retrospective design, we investigated difference in length of stay for 17,577 inpatient encounters over a 2-year period. Length of stay was calculated based on the time between hospital admission and hospital discharge with no removal of outliers. Encounters were assigned to teams based on the discharge provider. Teams were grouped based on whether they were teaching teams or non-teaching teams. Since length of stay was not normally distributed, it was modeled first using generalized linear models with gamma distribution and log link, and secondly by quantile regression. Models were adjusted for age, gender, race, medicine vs. non-medicine unit, MS-DRGs, and comorbidities.
RESULTS: Using gamma models to account for the skewed nature of the data, length of stay for encounters assigned to teaching teams were 0.56 days longer (β =0.10 95% CI 0.06 0.14) than non-teaching teams after adjustment. Using quantile regression, teaching teams had encounters on average 0.63 days longer (95%CI 0.44 0.81) than non-teaching teams at the 75th percentile, and 1.19 days longer (95%CI 0.77 1.61) compared to non-teaching teams at the 90th percentile after adjustment.
CONCLUSIONS: After adjusting for demographics and clinical factors, teaching teams on average had lengths of stay that were over ½ day longer than non-teaching teams. In addition, for the longest encounters, differences between teaching and non-teaching teams were over 1-day difference. Given these results, process improvement opportunities exist within teaching teams regarding length of stay, particularly for longer encounters.
PMID:33499682 | DOI:10.1080/21548331.2021.1882238