Return to search

Predicting mortality and dependency in Parkinson's disease

This thesis aimed to improve our understanding of prognosis in Parkinson's disease (PD) in terms of two imporatant outcomes: death and dependency. 88 studies were included in a systematic review of mortality in PD. Inception studies (recruiting patients at diagnosis) provided more consistent results than non-inception studies, with 50% higher mortality that in people without PD. Survival declined, on average, by 5% per year, but hospital-based studies consistently reported higher survival than community-based studies. 23 studies were included in a systematic review of activity limitation (difficulty with activities of daily living) and progression to dependency (the need for help with basic activities of daily living). Heterogeneity prevented quantitative analysis. Recommendations for future studies were developed. The rest of the thesis consists of analyses of the PINE study, a community-based, incident cohort of PD in North-East Scotland with 198 patients followed for up to 12 years. The mortality rate in PD was 8 per 100-person-years and was increased 1.5-fold compared to poulation mortality. Survival probabilities were lower than most previously reported. The rates of development of sustained dependency and "death or sustained dependency" were 14 and 16 per 100-person-years, respectively, in those independent at diagnosis. Older age, male gender, worse bradykinesia, more severe axial signs relative to limb signs, and higher co-morbidity (witih effect only early in the disease course) were independent baseline predictors of mortality in multivariable Cox regression. Increasing age, more smoking, worse bradykinesia, more axial signs relative to limb signs, and poorer cognitive function were independent baseline prognostic factors for both increased dependency and "death or dependency". These prognostic factors were combined into prognostic models for the three outcomes using Weibull parametric survival modelling and were internally vali There are several potential important uses for these models, in clinical practice and in research, subject to external validation.

Identiferoai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:690573
Date January 2015
CreatorsMacleod, Angus Donald
PublisherUniversity of Aberdeen
Source SetsEthos UK
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Sourcehttp://digitool.abdn.ac.uk:80/webclient/DeliveryManager?pid=230050

Page generated in 0.0018 seconds