Disparities in Incident Parkinson's Disease: A Comparison of Health Outcomes Using Inpatient versus Outpatient Claims...

The availability of large population-based medical claims datasets, such as Medicare, has driven an explosion in studies of neurodegenerative diseases including Parkinson's disease (PD). However, the high cost of these data forces some investigators to purchase limited datasets, most commonly inpatient samples, resulting in potential biases. We sought to examine differences in demographics, outcomes, and use of treatment between PD cases first diagnosed in inpatient versus ambulatory settings.

Using Medicare data from 2014 to 2019, we identified all incident PD cases (ICD-10 G20) aged 67 years and older from 2016 to 2018. We classified cases as inpatient if their first PD diagnosis was from an inpatient claims file or ambulatory if the first PD diagnosis was from ambulatory files. We calculated the odds ratio (OR) and 95% confidence interval (CI) for demographic characteristics between inpatient and ambulatory PD cases using binary logistic regression. Lastly, we used competing risks regression to calculate survival subhazard ratios (SHR) and 95% CI comparing inpatient and ambulatory PD cases from diagnosis through the end of 2019.

We identified 286,589 total incident PD cases from 2016 to 2018. There were 40,425 (14.1%) inpatient PD cases and 246,164 (85.9%) ambulatory cases. Inpatient PD cases were more likely to be older (OR = 1.028, 95% CI: 1.026-1.029, per year), male (OR = 1.15, 95% CI: 1.13-1.18), black (OR = 1.38, 95% CI: 1.32-1.44), Hispanic (OR = 1.14, 95% CI: 1.09-1.20), Native American (OR = 1.34, 95% CI: 1.14-1.58), smokers (OR = 1.13, 95% CI: 1.10-1.17), and reside in more deprived areas (OR = 1.30, 95% CI: 1.27-1.34). Inpatient PD cases were also at a higher risk of death compared to ambulatory PD cases (SHR = 1.85, 95% CI: 1.81-1.90) following the initial diagnosis date.

Administrative data studies using inpatient files exclusively to identify people with incident PD have significant demographic biases that could affect the outcome of the study.