Logistic Regression with Correction for HIV Self Reporting

Self report is one of the widely used methods of collecting information regarding individuals' health status. However it is surprising that this issue has not received, in relative terms more sustained attention. In these surveys patients might simply be mistaken, misremember or exaggerate the material covered. Thus reliability of self-reported is tenuous and the exposure assessments on which the associations between exposures and disease occurrence rely on are subject to either measurement error in a quantitative variable or misclassification in a categorical variable. Relatively few methods are available to handle misclassified categorical exposure variable (s) in the context of logistic regression models. The statistical model for characterizing misclassification is given by the transition matrix lambda from the true to the observed variable. In our research we aim at correcting the self reported data using the actual biomedical status of a sample of individuals from the given population. We exploit the relationship between the size of misclassification and bias in estimating the parameters of interest using logistic regression of self-report, the corrected logistic regression and the misclassified SIMEX (Simulation Extrapolation). We show that these methods are quite general and applicable to models with misclassified response and/or misclassified discrete regressors. We apply our methods to a study on the Kenya AIDS Indicator Survey data with a misclassified (Self-reported) longitudinal response.