Background:

Readmissions after hospital discharge are a common and costly occurrence in our health care system. Published models describing risk factors for hospital readmission inadequately stratify low–risk and high–risk patients. Patients with psychiatric illness or a history of substance abuse may be at high risk for hospital readmission but have not been adequately studied. We aimed to evaluate the risk of preventable 30–day readmission in patients with psychiatric illness or active substance abuse after hospital discharge.

Methods:

We retrospectively reviewed the records of all patients discharged from the general medicine service of an urban teaching hospital from July 1, 2009 to June 30, 2010 (12,383 discharges). We collected data on all readmissions to surrounding hospitals in the hospital’s network. Preventable readmissions were differentiated from nonpreventable readmissions using a validated algorithm based on ICD–9 codes (SQLape). Admissions followed by a 30–day preventable readmission were compared to admissions not followed by a 30–day readmission, and the relative contribution of psychiatric disease or substance abuse was assessed. Patients were considered to have a psychiatric illness or active substance abuse, if they met any of three conditions: they had a diagnosis consistent with these diseases on their outpatient problem list and were prescribed a medication intended to treat the disease, they had an admission or discharge diagnosis of psychiatric disease or substance abuse on their index hospitalization, or they were referred to the addiction psychiatry team during their index admission.

Results:

There were 1,101 preventable readmissions over the time period (11% preventable readmission rate). Compared to patients who were not readmitted, patients with preventable readmissions were significantly more likely to have psychiatric disease or substance abuse. In multivariate analyses correcting for age, sex, length of stay, and Elixhauser comorbidity index, patients with psychiatric disease and/or substance abuse were at increased risk for preventable readmission (OR 1.25, CI 1.05–1.47), with each psychiatric diagnosis found on the outpatient problem list increasing readmission risk by ∼25% (OR 1.24, CI 0.99–1.57), while a decreased number of prescribed psychiatric medications lowered readmission risk (OR 0.73 (CI 0.57–0.94) comparing lowest to highest tertile).

Conclusions:

Our findings suggest that patients with psychiatric illness or active substance abuse are at increased risk for preventable hospital readmission. Inclusion of these patients into prediction models may improve their performance, while designing interventions that address these comorbidities may offer promise for decreasing preventable hospital readmissions.