Date of Award


Document Type

Open Access Dissertation


Health Services and Policy Management


Norman J. Arnold School of Public Health

First Advisor

Janice C. Probst



The likelihood of a depression diagnosis may differ based on whether the patient was seen in an inpatient or outpatient setting. Depression has been associated with an increased risk of a 30-day readmission. The purpose of this study was to determine the level of agreement (concordance) between depression diagnosis as identified by inpatient and outpatient records and examine the relationship between depression diagnostic concordance and 30-day readmission.


Using universal administrative claims data from South Carolina, we examined inpatient and outpatient records. The analysis was restricted to Medicaid recipients aged 55 years and older with a primary admitting diagnosis of acute myocardial infraction (AMI), chronic obstructive pulmonary disease (COPD), heart failure (HF), and pneumonia (PN) from 2013 – 2015 (n = 8,621 patients). Depression diagnostic concordance was determined by comparing secondary diagnosis codes in inpatient records and primary or secondary diagnosis codes in prior outpatient records. Diagnostic concordance was analyzed as concordant/not concordant and as a categorical variable (concordant-no depression, concordant-depression, not concordant-inpatient only, and not concordant-outpatient only). Using log-binomial regression, we modeled diagnostic concordance and 30-day readmission, while adjusting for covariates of interest in both models.


The agreement between inpatient and outpatient data for a recorded depression diagnosis was poor. Diagnostic concordance was significantly associated with patient age, sex, race/ethnicity, and health condition. The risk of a 30-day readmission was significantly associated with an outpatient history of depression, but not for patients with depression recorded in inpatient data. Patients with not concordant-outpatient only had a 30-day readmission rate of 8.1%, adjusted relative risk, 1.42 (p = 0.001). Patients with concordant-depression had lower 30-day readmission rates, but not statistically significant.


Patients may be at a greater risk of a 30-day readmission when a history of depression was not detected during a hospitalization. Depression, particularly when it is not detected during a hospitalization, leads to a higher risk of a readmission. Diagnostic concordance between depression recorded in inpatient and outpatient settings needs improvement to ensure patients with depression receive appropriate care during and after a hospitalization. Improving diagnostic concordance for depression may reduce untimely hospital readmissions.