Abstract
Objectives: The aim of this study was to determine the relationship between the time to referral (TTR) to psychiatric consultation and the length of stay (LOS) after adjusting for medical comorbity. Methods: Using a naturalistic study design, we collected and analyzed inpatient consultation-liaison psychiatry service data over a 12-month period from 2 Canadian hospital sites. Data collected included demographic characteristics, referral characteristics, Charlson Comorbidity Index to measure medical comorbidity severity, psychiatric diagnoses, type of psychiatric intervention, and time variables, namely TTR and LOS. We modeled the relationship LOS and TTR after adjusting for Charlson Index using a 3-component finite mixture of exponential regression models. Results: A total of 814 patients were included. The median LOS was 12 days (interquartile range : 4-28 days). Median TTR was 3 days (inter quartile range: 1,9), and median Charlson Index was 5 (interquartile range 3,6). Bivariate analysis indicated a strong positive correlation among LOS and TTR (Spearman correlation: 0.77, p < 0.0001) and Charlson Index(Spearman correlation: 0.34, p < 0.0001), respectively. After controlling for Charlson Index, we observe that TTR was signcantly associated with LOS in each of the 3 components of the mixture of exponential regression models. Persons with longer TTR have longer expected LOS. Graphical summaries suggest that the mixture of exponential regression model provides a good fit to these LOS response data. Conclusions: Patients with longer TTR had significantly longer LOS. The association between TTR and LOS holds after controlling for severity of medical comorbidity. Our results support the role of integrated and proactive consultation-liaison psychiatry programs aimed at reducing TTR to improve LOS outcomes.