Targeted partial validation to make EHR data au-dit they can be: Correcting for data quality issues in the learning health system

Music City Center 

The allostatic load index (ALI) is an informative summary of whole-person health that is predictive of downstream health outcomes. The ALI uses biomarker data to measure cumulative stress on five systems in the body for the general adult population. Borrowing data from electronic health records (EHR) is a promising opportunity to estimate the ALI and potentially identify at-risk patients on a large scale. However, routinely collected EHR data may contain missingness and errors, and ignoring these data quality issues could lead to biased statistical results and incorrect clinical decisions. Validation of EHR data (e.g., through chart reviews) can provide better-quality data, but realistically only a subset of patients' data can be validated. Thus, we devise a targeted study design ("targeted audit") to harness the error-prone surrogates from the EHR to identify the most informative patient records for validation. Specifically, the targeted audit design seeks the best statistical precision to quantify the association between ALI and healthcare utilization in logistic regression. In this talk, we detail the process of the targeted audit design and its application to EHR data from Atrium Health Wake Forest Baptist.