Abstract
ObjectivesNursing home (NH) Minimum Data Set (MDS) have frequently been used to measure medication use in epidemiologic studies, but there is little evidence on the accuracy of MDS-based medication records. We compared antipsychotic use estimated using 2 data sources—MDS and NH electronic medication administration records (eMAR). DesignCross-sectional comparison. Setting and ParticipantsThis analysis was based on MDS and linked eMAR data of 604 NH residents with dementia at 54 NHs in 10 states, participating in a cluster-randomized pragmatic trial (METRIcAL), from June 2019 to February 2020. MethodsOne admission, quarterly, or annual MDS assessment was chosen for each participant. The MDS assessment recorded the number of antipsychotic treatment days during a 7-day window. We then identified antipsychotic administrations during the corresponding window in the eMAR. We used Cohen kappa to assess agreement in the proportion of participants on antipsychotics during the week and used intraclass correlation coefficient (ICC) to assess the agreement of treatment days. We further used the eMAR data as a reference to calculate validity parameters. ResultsA total of 29.5% of study participants were identified as antipsychotic users based on the MDS vs 28.3% based on the eMAR data (kappa value: 0.96). MDS-based average treatment duration was estimated to be 2.0, consistent with eMAR-based estimate (1.8 days, ICC: 0.96). The sensitivity was 98.8% (95% CI 95.8%-99.9%), the specificity was 97.9% (95% CI 96.1%-99.1%), the positive predictive value was 94.9% (95% CI 90.8%-97.3%), and the negative predictive value was 99.5% (95% CI 98.2%-99.9%). Conclusions and ImplicationsAgreement between the MDS and eMAR in antipsychotic use is high, suggesting that the MDS is a valid tool to measure antipsychotic use in epidemiologic studies. Further studies with large and diverse populations are warranted to confirm our findings.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have