EHS
EHS

Author response: A predictive model to identify Parkinson disease from administrative claims data

We thank Dr. Kawada for his interest in our study, in which we used Medicare data to develop a predictive model designed to identify Parkinson disease (PD) in the prodromal period.1 As Medicare does not include self-reported information on smoking, we estimated the probability of ever smoking from the claims data. The objective method we applied was identical for cases and controls, and thus we expected that our PD–smoking odds ratio would be attenuated due to nondifferential exposure measurement error. While the PD model we presented was carefully developed for the prediction of PD, the dataset is also suitable for developing models for testing some hypotheses. We agree that a study with data on diet and human leukocyte antigen genotype—in addition to Medicare data on related conditions and incident PD—would allow for exploration of interesting hypotheses.

EHS
Back to top button