作者: P Zaninotto
DOI:
关键词: Gerontology 、 Binary outcome 、 Imputation (statistics) 、 Longitudinal study 、 Healthy population 、 Missing data 、 Coronary heart disease 、 External validation 、 Medicine 、 Older people
摘要: This study explored gender differences in quality of life and depressive symptoms over time among people with coronary heart disease (CHD) compared them a healthy population. Using three waves the English Longitudinal Study Ageing (2002-03, 2004-05 2006-07) methodological problems such as missing data sources error uncertainty which may arise from reliance on self-reported measure CHD were addressed. A simulation was set up to compare techniques for dealing data: full information maximum likelihood, multivariate normal imputation two-fold fully conditional specification. Results supported use latter technique outperformed other two recovering targeted parameters, especially binary outcome. Results based imputed showed that population, men women had average lower levels life. Men also at higher risk having than Women likely population have symptoms. Trajectories different shape trajectories after onset CHD. Men’s declined no changes found. Women’s only between baseline four year follow-up, while same period their reduced. A sensitivity analysis an external validation deterministic helped understand impact misclassification could results this thesis. It found reliability presented be affected by false positive negative cases