Objective: To explore the ability for electronic health records to identify persons with prodromal Parkinson’s disease in a cohort of those with RBD in the US.

Background: Neurodegenerative processes leading to Parkinson’s Disease (PD) may begin decades prior to diagnosis, and may become extensive before PD manifests.¹  Early identification of prodromal individuals who may benefit most from earlier disease modification therapy remains a critical need, and future disease modification clinical trials will likely need to utilize multiple key factors to recruit people with prodromal PD at the highest possible accuracy.

Method: We adapted the Movement Disorders Society (MDS) Prodromal PD calculator² to electronic health record (EHR) data to estimate the probability of prodromal PD among participants with REM sleep behavior disorder (RBD; i.e., a group with high expected rates of conversion to PD that is not diagnostically specific to PD) from the All of Us cohort.³ We identified persons with RBD using validated algorithms⁴ and calculated the rate of conversion from prodromal PD to diagnosed PD (or ‘phenoconversion’) associated with a range of Prodromal PD probabilities using Cox proportional hazards models.  We also determined the optimal cut-point for prodromal probability in this population for rates of conversion using 10-fold cross-validation (i.e., optimizing C-statistic and Youden’s index).

Results: We identified 544 individuals with RBD (average age: 51.6±17.1 years; 55% male), of which 83 converted to diagnosed PD over an average follow-up of 3.7 (SD:1.9) years. Relative to individuals in this cohort with estimated prodromal PD probability less than 20%, those with probabilities exceeding 80% had an over 23-fold increase in rates of phenoconversion (23.79; 95% CI: 18.45, 39.69). We found a cut-point of 70% probability of prodromal PD as the optimal threshold with the best discrimination between those that phenoconverted and those that did not, with a c-statistic value of 0.72.

Conclusion: These results demonstrate that the MDS prodromal calculator can provide important discriminatory information in people with RBD when applied to available EHR data. This may have implications for improved discovery of higher risk individuals both in timeliness and accuracy.

References: 1. Roos DS, Klein M, Deeg DJH, Doty RL, Berendse HW. Prevalence of Prodromal Symptoms of Parkinson’s Disease in the Late Middle-Aged Population. J Parkinsons Dis. 2022;12(3):967-974. doi:10.3233/JPD-213007

2. Heinzel S, Berg D, Gasser T, Chen H, Yao C, Postuma RB. Update of the MDS research criteria for prodromal Parkinson’s disease. Movement Disorders. 2019;34(10):1464-1470. doi:10.1002/mds.27802

3. Ramirez AH, Sulieman L, Schlueter DJ, et al. The All of Us Research Program: Data quality, utility, and diversity. Patterns. 2022;3(8). doi:10.1016/j.patter.2022.100570

4. Chahine LM, Ratner D, Palmquist A, et al. REM Sleep Behavior Disorder Diagnostic Code Accuracy and Implications in the Real-World Setting. Neurol Clin Pract. 2025;15(1). doi:10.1212/CPJ.0000000000200387

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/finding-prodromal-parkinsons-disease-with-the-movement-disorder-societys-prodromal-calculator-and-electronic-health-records/