Impact of Internal Medicine Subspecialty Consultations on Length of Stay: A Pilot Retrospective Cohort Study.

Objective For many years, the average length of stay for inpatient hospitalization has been steadily increasing, with hospitalists citing many potential reasons for the increase. Some of the commonly known reasons are time spent waiting for consultant recommendations, procedure delays, and diagnostic testing delays. The objective of this study was to preliminarily examine the relationship between the number of internal medicine subspecialty consultations requested and inpatient length of stay (LOS) on a single hospital medicine service. Design, setting, and participants This pilot study was a retrospective cohort study at one large tertiary academic medical center. Included participants were adults who were admitted to one hospital medicine service line from January 1, 2022, to December 31, 2022. The main outcome was the impact of the total number of internal medicine subspecialty consultations on the length of stay (LOS) index (observed LOS divided by expected LOS). Secondary outcome measures were between-group baseline demographic differences, primary admission diagnoses, and discharge admission diagnoses.  Results Patients with internal medicine subspecialty consults had a significantly higher expected LOS (6.10 days vs 4.73; p<0.0001), observed LOS (7.67 days vs 5.04; p<0.0001), and LOS index (1.17 vs 1.03; p<0.0001). The most common admission and discharge diagnosis was COVID-19. In the univariate and multivariable analyses, demographic variables were not significantly associated with the LOS index. After controlling for covariates, the LOS index increased by 0.3178 for each additional consult. Non-diabetes endocrinology, palliative care, and rheumatology were the internal medicine subspecialty consultations associated with the greatest increase in the LOS index. Conclusion Patients in the consult group experienced a 1.5-day longer overall length of stay, with each additional consultation resulting in a 32% increase in the ratio of observed to expected LOS.