CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1381
Title: Measurement error case series model of infection-cardiovascular risk: Application to database of patients on dialysis Authors:  Danh Nguyen - University of California, Irvine (United States) [presenting]
Abstract: The accumulation of patient data across practices over time provides a rich source of patient-level data to design and conduct population-based studies. Examples include national registries, electronic medical data and medical claims linked across health systems, and adverse events reporting systems of adverse drug reactions. We introduce the case series method and discuss its use for big data analytics in biomedicine. As an example, an application of the measurement error case series method (MECS) using the national registry, United States Dates Renal Data System (USRDS), which collects data on >99\% of all patients in the U.S. is presented. Infection and cardiovascular (CV) disease are leading causes of hospitalization and death in patients on dialysis. A challenge in modeling the infection-CV risk is that the exact time of infection onsets cannot be ascertained based on hospitalization data. Only imprecise markers of the timing of infection onsets are available. The new MECS model is used to account for measurement error in time-varying exposure onsets. We describe the general nature of bias resulting from estimation that ignores measurement error and proposed a bias-correction method. Hospitalization data from the USRDS is used to illustrate the new method. The results suggest that the estimate of the CV incidence following the 30 days after infections is substantially attenuated in the presence of infection onset measurement error.