Year: 2010
Type: dissertation
Abstract: (pdf)Quantification of the impact of exposure to different risk factors on mortality or morbidity at the population level is a fundamental issue in epidemiologic research.Population Attributable Fraction (PAF) is a statistical concept that can be used to quantify this impact.PAF assesses the proportion of outcome that could be avoided if the current exposure distribution was replaced by a hypothetical, presumably preferable exposure distribution.So far, the methods for the estimation of PAF have been developed for and applied in case-control and cross-sectional studies.The development of methods for the estimation of PAF from cohort studies, which properly take into account the time perspective, has started only recently.In the estimation of PAF for a certain followup time interval, the type of outcome of interest (mortality vs. morbidity) has not, however, been taken into account.In this study, the statistical methodology for the estimation of PAF in cohort studies will be extended to cover both the estimation of PAF for total mortality and disease incidence.Research 34/2010 National Institute for Health and Welfare Population Attributable Fraction (PAF) in Epidemiologic Follow-up Studiesdeath increased the overestimation of PAF, especially when the impact of risk factors on mortality was strong and the follow-up time long.A new program for the estimation of PAF both for total mortality and disease incidence, implementing the new methods, was developed using SAS/IML language.This program was shown to be flexible and fast.An application of PAF to evaluate the relative importance of the risk factors of type 2 diabetes and the potential effectmodifying role of metabolic syndrome or its components in a meta-analysis of two representative Finnish cohorts was carried out using this program.As a result, the use of PAF provided further evidence of weight control being the primary diabetes prevention method.The pooling of the PAF estimates increased the power to detect associations in smaller subpopulations defined by the metabolic syndrome or its components, establishing new evidence on the importance of early lifestyle changes in the prevention of type 2 diabetes.In conclusion, it is essential to take time perspective into account in the estimation of PAF.Different estimators of PAF for a certain time interval, taking into account different sources of censoring, are needed, depending on the outcome of interest.PAF is a useful measure in cohort studies for providing population-level information on the effects of predictor modifications on the outcome in time and has wide applications in many different fields of research.
Author: Maarit A. Laaksonen
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
Funder: Terveyden ja hyvinvoinnin laitos
OpenAlex ID: W1153462348