Article Text
Abstract
Objectives To develop a prognostic model to predict mortality in Parkinson's disease (PD).
Background Prognostic models can be used for individual prediction, case-mix correction, and clinical trial design. No prognostic model has been published for use in PD.
Methods The PINE study (198 patients) and ParkWest study (192 patients) are community-based, incidence cohorts of PD with lifelong prospective follow-up in North-East Scotland and Western Norway, respectively. In each study area, all incident cases were identified with community-based ascertainment strategies (2004–6 and 2006–9 in PINE; 2004–7 in ParkWest). Clinical predictors measured at diagnosis in the PINE cohort were combined to create a prognostic model of mortality using Weibull regression. It was externally validated by measuring calibration (observed versus predicted risk in quantiles of risk) and discrimination in the ParkWest cohort.
Results Independent baseline prognostic factors in the PINE study were age, male sex, higher co-morbidity, more bradykinesia, and more axial relative to limb features. The model combining these factors had good discrimination in the ParkWest cohort (C-statistic=0.77) but tended to underestimate survival.
Conclusions The model discriminated well and can therefore be used for case-mix correction or patient selection/stratification in trials. Recalibration may be necessary before making individualised predictions in other settings.