Research based on EHR data needs to account for, remove bias

Screening analyses using electronic health record data need to be refined to reduce the risk of bias and prevent "spurious" findings, according to a new report in eGEMs (Generating Evidence and Methods to Improve Patient Outcomes).  

The researchers, from the Group Health Research Institute in Washington state, noted that EHRs increasingly are being used to facilitate research and quality of care, such as in the review of cancer screening rates. However, the EHR data for cancer screening, such as colon cancer, is inaccurate because there is no separate code for screening; to that end, the data would include diagnostic colonoscopies, as well as screening colonoscopies. This bias can range in overestimation of colon cancer screening by 11.6 percent, according to the study's authors.

They recommended that potential bias be taken into account when studying screening utilization by adjusting the algorithms to reduce the misclassification of tests, and suggested different approaches to doing so. The researchers then adjusted the screening algorithm accordingly and tested it with data of 139,163 patients from Group Health, an integrated delivery system in Washington.

The unadjusted colon cancer screening estimate was 4 percent higher than the adjusted rate.

"As use of electronic medical records expands, EHR data will become increasingly valuable as a source of information on screening test utilization," the researchers said. "The considerations provided here allow for evaluation of the bias that will occur if algorithms for screening indication are used without correction for potential misclassification of screening tests as diagnostic tests."

EHRs continue to prove their potential worth in research. However, inaccurate data in the EHR and/or poorly designed methods to analyze the data can render the research and any results therefrom unreliable.

To learn more:
- read the study (.pdf)