Show simple item record

dc.contributor.authorShebl, Fatma Mohamed
dc.date.accessioned2012-03-05T21:35:42Z
dc.date.available2012-03-05T21:35:42Z
dc.date.issued2008
dc.identifier.urihttp://hdl.handle.net/10713/976
dc.descriptionUniversity of Maryland, Baltimore. Epidemiology and Preventive Medicine. Ph.D. 2008en_US
dc.description.abstractBackground. Use of an imperfect measurement would possibly result in misclassification of health status, and biased incidence estimates. Most statistical methods handling misclassified incident data assume that conditional on the true disease status, test results obtained from the same person at different time points are independent. Objective. (I) To examine the effect of conditional dependence and misclassification on cumulative incidence estimates in the simple situation of two equally spaced observations per person. (II) To develop a likelihood based method to obtain incidence rate estimates adjusted for misclassification in the presence of unequally spaced observations, differential misclassification and interval censoring. Methods. (I) We obtained expressions defining estimates of the adjusted incidence in terms of naive incidence, misclassification and conditional dependence probabilities. Mean squared error and variance were estimated to measure bias and variation of naive and adjusted rates. (II) We developed a parametric likelihood-based method to adjust for incidence rate, allowing for handling differential misclassification, and variance estimation. R and SAS codes were developed to employ the proposed methods to a hepatitis C disease dataset. Results. (I) In the presence of low incidence rate, specificity had the greatest impact on the magnitude of bias, and the higher the conditional dependences, the smaller the magnitude of the bias. Remission was the most sensitive estimate to the presence and degree of misclassification probabilities. The MSE of adjusted rates were generally much smaller than MSE of naive rates. (II) The results of the application in HCV data indicated that, in the presence of misclassification, incidence estimates are always biased. We detected a higher impact of specificity on the magnitude of bias. False positive rates and thus imperfect specificity induced greater bias than imperfect sensitivity due to low incidence rate in this population. Conclusion. To obtain the most accurate incidence estimates, it is of great importance to choose the appropriate diagnostic test at the early stages of study design, and to properly adjust for any possible misclassification and conditional dependence in the analysis stage.en_US
dc.language.isoen_USen_US
dc.subjectHealth Sciences, Epidemiologyen_US
dc.subjectconditional dependenceen_US
dc.subjectmisclassificationen_US
dc.subject.meshBiasen_US
dc.subject.meshHepatitis Cen_US
dc.titleIncidence estimation when the disease assessment is subject to misclassification error, with application to hepatitis Cen_US
dc.typedissertationen_US
dc.contributor.advisorMagder, Laurence S.
dc.identifier.ispublishedYes
 Find Full text

This item appears in the following Collection(s)

Show simple item record