February 18, 2021

Meeting Host:
CFAR Biostatistics & Bioinformatics Core

Count Me In! Click Here to Register

Bryan Shepherd Statistical Approaches to Improve HIV Studies with Error-prone Data

Bryan E. Shepherd, PhD

Professor (Biostatistics)
Vanderbilt University

From the speaker:
Observational data derived from patient medical charts and electronic health records are increasingly used for HIV/AIDS research. There are challenges (some recognized, some unrecognized, and some recognized but ignored) to using these data, in particular with regards to data quality.

I will describe some of our efforts to validate subsamples of data, in both international HIV cohort studies and in local studies using the electronic health record. I will then describe methods to incorporate validation data into analyses to obtain estimates closer to the truth while maintaining adequate statistical power/precision.

Methods to address measurement error and misclassification have been developed in the statistical literature, but they often address relatively simple settings, whereas the errors seen in our datasets span multiple variables (both predictors and outcomes), are correlated, and may even affect who is included in the study.

Finally, I will discuss validation sampling strategies, where researchers can strategically select which records to validate, often with multiple waves of sampling, to maximize information learned with a fixed number of records to validate. Examples and illustrations with HIV data will be used throughout.

4:50 — Program opens
5:00 — Totally Terrific Talk
5:45 — Q&A
6:00 — Adjourn

Meeting Location:
Via Zoom
Link provided after registration