OBJECTIVES:

This study aims to assess the impact of depression on total healthcare services expenditures and total prescribed medicine expenditures among Medicare patients (age>=65 years) with ADRD in the US.

METHODS:

The analysis population (Medicare beneficiaries with ADRD; N=1,158) was derived from the Medical Expenditure Panel survey’s (MEPS) medical conditions and full year consolidated data files pooled from 2016-2020. Preliminary analysis involved the assessment of differences in outcome variables across target years to determine the suitability of data pooling. Univariate analyses of covariates examined the association between the outcome variables and independent variables using two sample t-test for binary categorical variables and analysis of variance for multi-level categorical variables. Multivariable analyses examined the effects of demographic (age, gender, race/ethnicity, marital status, region) and economic (education level, health insurance, and employment) covariates on the primary predictor (depression) using multiple regression models and goodness-of-fit which was evaluated using the Akaike Information Criterion (AIC). All analyses were conducted in SAS On Demand, academic version.

RESULTS:

For total healthcare services expenditures, univariate analysis showed that there was a statistically significant difference (P<0.05) by depression diagnosis, insurance status and education level. For total prescribed medicine expenditures, there was a statistically significant difference (P<0.05) by depression, race/ethnicity, and employment status. Multivariable analysis results did not indicate any covariate confounding the relationship between depression and outcomes. The best fitting models included depression, demographic and economic variables for total healthcare services expenditures, and depression alone for total prescribed medicine expenditures.

CONCLUSIONS:

This study shows that depression has a statistically significant impact on healthcare services and prescribed medicine expenditures for Medicare beneficiaries with ADRD in both univariate and multivariate analyses.