OBJECTIVES : To develop a statistical model describing the longitudinal changes after diagnosis in the Hoehn and Yahr (HY) staging scale, a measure of Parkinson’s disease stage, as a function of the Movement Disorder Society sponsored Unified Parkinson´s Disease Rating Scale Part III (MDS-UPDRS III). The aim was to develop a statistical model able to capture the associations between these two scales in a simulation of Parkinson's disease progression.

METHODS : Longitudinal data were obtained from the Parkinson's Progression Markers Initiative study (www.ppmi-info.org/data, accessed January 2018, N=423, mean age 61.7 years, disease duration 6.6 years at baseline followed for up to 6.8 years). The data were analyzed using ordinal logistic regression (OLR) with repeated measures to predict changes in patients’ HY stage based upon previous values of MDS-UPDRS III and HY stage. Higher HY stages are indicative of greater impairment and disability. Due to insufficient sample sizes in each individual HY stage, grouping of HY stages (0-1, 2, and 3-5) was required. The OLR model with repeated measures computed the log-odds of cumulative predicted probabilities for belonging to the modeled HY stages.

RESULTS : The significant predictors for HY stage included baseline HY stage (2+ vs 1), prior HY stage (0-1 vs 2 and vs 3-5), baseline MDS-UPDRS III, and prior MDS-UPDRS III. At baseline, most of the analyzed subjects were HY stage 2 (56.03%), with an average MDS-UPDRS III score of 25.32; a 37.61% relative risk increase of progressing to HY stages 3+ was predicted for those with a 5-point MDS UPDRS III increase post-baseline. The OLR model performed well in validation analyses—average probabilities of belonging in HY stages were compared for each post-baseline visit.

CONCLUSIONS : The OLR model captures the longitudinal relationship between HY stage and MDS-UPDRS III. Future research should focus on further evaluating the correlation between these measures.