Format

Send to

Choose Destination
Neurology. 2016 Mar 15;86(11):986-93. doi: 10.1212/WNL.0000000000002437. Epub 2016 Feb 17.

Development and external validation of a prognostic model in newly diagnosed Parkinson disease.

Author information

1
From the Department of Neurology (D.C.V., R.M.A.d.B., L.W.), Department of Medical Psychology (B.S.), and Clinical Research Unit (R.J.d.H.), Academic Medical Center, Amsterdam, the Netherlands; John Van Geest Centre for Brain Repair (J.R.E., S.L.M., R.A.B., C.H.W.-G.), Department of Clinical Neurosciences, University of Cambridge; Institute of Neurology (T.F.), University College London, UK; Department of Brain and Cognition (B.S.), University of Amsterdam; and Department of Neurology (B.P.), Radboud University Medical Center, Nijmegen, the Netherlands. d.c.velseboer@amc.uva.nl.
2
From the Department of Neurology (D.C.V., R.M.A.d.B., L.W.), Department of Medical Psychology (B.S.), and Clinical Research Unit (R.J.d.H.), Academic Medical Center, Amsterdam, the Netherlands; John Van Geest Centre for Brain Repair (J.R.E., S.L.M., R.A.B., C.H.W.-G.), Department of Clinical Neurosciences, University of Cambridge; Institute of Neurology (T.F.), University College London, UK; Department of Brain and Cognition (B.S.), University of Amsterdam; and Department of Neurology (B.P.), Radboud University Medical Center, Nijmegen, the Netherlands.

Abstract

OBJECTIVE:

To develop a prognostic model to predict disease outcomes in individual patients with Parkinson disease (PD) and perform an external validation study in an independent cohort.

METHODS:

Model development was done in the Comorbidity and Aging in Rehabilitation Patients: The Influence on Activities (CARPA) cohort (Netherlands). External validation was performed using the Cambridgeshire Parkinson's Incidence from GP to Neurologist (CamPaIGN) cohort (UK). Both are longitudinal incident cohort studies that prospectively followed up patients with PD from the time of diagnosis. A composite outcome measure was made in which patients were classified as having an unfavorable prognosis when they had postural instability or dementia at the 5-year assessment (or at the last assessment before loss to follow-up), or had died before this time. The final model was derived with a backward selection strategy from candidate predictor variables that were measured at baseline.

RESULTS:

In the resulting model, higher patient age, higher Unified Parkinson's Disease Rating Scale motor examination axial score, and a lower animal fluency score were all associated with a higher probability of an unfavorable outcome. External validation confirmed good discriminative ability between favorable and unfavorable outcomes with an area under the receiver operating characteristic curve of 0.85 (95% confidence interval 0.77-0.93) and a well-calibrated model with a calibration slope of 1.13 and no significant lack of fit (Hosmer-Lemeshow test: p = 0.39).

CONCLUSION:

We constructed a model that allows individual patient prognostication at 5 years from diagnosis, using a small set of predictor variables that can easily be obtained by clinicians or research nurses.

PMID:
26888991
PMCID:
PMC4799715
DOI:
10.1212/WNL.0000000000002437
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for HighWire Icon for PubMed Central
Loading ...
Support Center