Journal of Data Science

Abstract: Epidemiological cohort study that adopts a two-phase design raises serious issue on how to treat a fairly large amount of missing val ues that are either Missing At Random (MAR) due to the study design or potentially Missing Not At Random (MNAR) due to non-response and loss to follow-up. Cognitive impairment (CI) is an evolving concept that needs epidemiological characterization for its maturity. In this work, we attempt to estimate the incidence rate CI by accounting for the aforemen tioned missing-data process. We consider baseline and first follow-up data of 2191 African-Americans enrolled in a prospective epidemiological study of dementia that adopted a two-phase sampling design. We developed a multiple imputation procedure in the mixture model framework that can be easily implemented in SAS. Sensitivity analysis is carried out to assess the dependence of the estimates on specific model assumptions. It is shown that African-Americans in the age of 65-75 have much higher incidence rate of CI than younger or older elderly. In conclusion, multiple imputation pro vides a practical and general framework for the estimation of epidemiological characteristics in two-phase sampling studies.

Abstract: In longitudinal studies where the same individuals are followed over time, bias caused by unobserved data raises a serious concern, particularly when the data are missing in a non-ignorable manner. One approach to deal with non-ignorable missing data is a pattern mixture model. In this paper, we combine the pattern mixture model with latent trajectory analysis using the SAS TRAJ procedure, which offers a practical solution to many problems of the same nature. Our model assumes a stochastic process that categorizes a relative large number of missing-data patterns into several latent groups, each of which has unique outcome trajectory, which allows patterns with missing values to share information with patterns with more data points. We estimated the longitudinal trajectories of a memory test over 12 years of follow-up, using data from the prospective epidemiological study of dementia. Missing data patterns were created conditional on survival, and final marginal response was obtained by excluding those who had died at each time point. The approach presented here is appealing since it can be easily implemented using common software.

Abstract: Believe the Positive (BP) and Believe the Negative (BN) rules for combining two continuous diagnostic tests are compared with proce dures based on likelihood ratio and linear combination of the two tests. The sensitivity-specificity relationship for BP/BN is illustrated through a graph ical presentation of a ”ROC surface”, which leads to a natural approach of choosing between BP and BN. With a bivariate normal model, it is shown that the discriminating power of this approach is higher when the correla tion between the two tests has different signs for non-diseased and diseased population, given the location and variations of the two distributions are fixed. The idea is illustrated through an example.