Agreement of Medicare Part D and Minimum Data Set Reported Psychotropic Medication Use in Nursing Homes

Little evidence exists on the accuracy of the Minimum Data Set (MDS)-based medication items. We compared quarterly rates of antipsychotic, antidepressant, and hypnotic use between the MDS and Part D Prescription Drug Event file (PDE) in 2018.

Cross-sectional comparison.

All US Medicare- or Medicaid-certified nursing homes were included. Long-stay nursing home residents enrolled in Medicare Part D who were aged ≥65 years with psychiatric disorders or dementia identified in the MDS data (N = 580,340) were included. Two subcohorts included the following: residents with psychiatric disorders and no Alzheimer's disease and related dementias (ADRD), and residents with ADRD.

Psychotropic medication use was assessed using the share of study participants who received the medication in the quarter identified by the MDS. We used Cohen kappa to assess agreement in the share of residents using psychotropic medications during the quarter and used PDE data as a reference to calculate validity parameters.

The MDS-reported and PDE-reported antipsychotic (MDS vs PDE: 34.0% vs 34.3%) or antidepressant (MDS vs PDE: 73.4% vs 71.3%) users have high concordance as measured by the kappa value (antipsychotic: 0.9; antidepressant: 0.83). Sensitivity, specificity, positive predictive value, and negative predictive value of MDS data compared with PDE were 96.7%, 92.8%, 96.3%, and 93.6% for antipsychotic, and 84.5%, 96.7%, 91.1%, and 94.0% for antidepressant. Only 3.7% of study participants identified in the MDS data, vs 32.6% of PDE-based users (kappa value: 0.1), used hypnotics. By combining antianxiety medications with hypnotics, the rates of MDS hypnotic or antianxiety users increased to 35.0% (kappa value: 0.74). Sensitivity, specificity, positive predictive value, and negative predictive value were 89.5%, 85.6%, 92.8%, and 79.8%, respectively.

Agreement between the MDS and PDE in antipsychotic and antidepressant use was high, suggesting that the MDS is a valid tool to measure antipsychotic and antidepressant use. Additional work is needed to understand the disagreements between MDS and PDE in hypnotic use.