Objective: We aimed to describe progression to, identify independent prognostic factors for, and develop a valid prognostic model for, dependency and “death or dependency” in PD.

Background: Most previous prognostic studies in Parkinson’s disease (PD) have used unrepresentative cohorts with selection biases. Prognosis is best studied by long-term follow-up of community-based incident cohorts. Dependency, the need for help with activities of daily living is an important patient-orientated clinical outcome and “Death or dependency” is a useful measure of poor outcome. Little has been published previously on these outcomes in PD.

Methods: Data were derived from a community-based, incidence cohort of PD with lifelong prospective follow-up in North-East Scotland (cases identified over 4.5 years, during 2002-4 and 2006-9). Annual evaluations of activity limitation were performed with the Schwab & England scale (S&E) and death notifications were received from the NHS central register. Dependency was defined as S&E score<80. Weibull regression was used to identify independent prognostic factors measured at time of diagnosis, with careful selection to avoid over-fitting, and to create a prognostic model.

Results: Of 168 patients who were independent at baseline, 99 became dependent and 124 either died or became dependent, during follow-up up between 6 and 12 years. Independent baseline prognostic factors for both increased dependency and increased “death or dependency” were age, smoking, living alone, UPDRS bradykinesia score; ratio of axial to limb sign, and poorer cognition. Hazard ratios are given in the table

Prognostic model for dependency and "death or dependency"

	Model predicting dependency		Model predicting “death or dependency”
Baseline variable	Hazard ratio (95% confidence interval)	P-value	Hazard ratio (95% confidence interval)	P-value
Age at diagnosis (10-year increase)	2.47 (1.82–3.34)	<0.001	2.08 (1.62–2.68)	<0.001
Living alone (yes vs no)	0.59 (0.36–0.97)	0.04	0.58 (0.37–0.91)	0.02
Smoking history (10-pack-year increase)	1.14 (1.04–1.26)	0.007	1.16 (1.06–1.26)	0.001
Bradykinesia score (5-point increase)	1.54 (1.22–1.93)	<0.001	1.45 (1.18–1.78)	<0.001
Axial:limb ratio (1-unit increase)	1.52 (0.99–2.33)	0.05	1.72 (1.20–2.45)	0.003
MMSE score (1-point increase)	0.85 (0.76–0.96)	0.007	0.86 (0.77–0.95)	0.003
Shape parameter	1.86 (1.58–2.20)	<0.001	1.69 (1.46–1.96)	<0.001
Constant	2.45 (0.09–66.8)	0.49	4.21 (0.23–76.1)	0.33

. Prognostic models incorporating these factors had good discrimination (C-statistic 0.77 and 0.76 for models of dependency and “death or dependency” respectively) and very good calibration.

Conclusions: Identifying prognostic factors and their combination in a prognostic model can allow individualised risk prediction, facilitate stratified medicine and improve clinical trial design. Further work will include external validation in an international cohort.

« Back to 2016 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/predicting-poor-functional-outcome-in-parkinsons-disease/