Accounting for Differential Error in Time-to-Event Analyses Using Imperfect Electronic Health Record-Derived Endpoints

Hubbard, Rebecca A.; Harton, Joanna; Zhu, Weiwei; Wang, Le; Chubak, Jessica

doi:10.1007/978-3-319-69416-0_14

Cited by 2 publications

(2 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because classification accuracy is dependent on the intensity and type of interaction a patient has with the health care system, exposures that tend to result in more frequent contact with the health care system or that are correlated with health care seeking behavior are likely to exhibit differential misclassification. Previously, we found that differential misclassification induced notable bias in effect estimates in the setting of second breast cancers . On the basis of our results, type I error rates for naïve analyses of such exposures will also be substantially inflated.…”

Section: Discussionsupporting

confidence: 50%

See 1 more Smart Citation

Inflation of type I error rates due to differential misclassification in EHR‐derived outcomes: Empirical illustration using breast cancer recurrence

Chen

Wang

Chubak

et al. 2018

Pharmacoepidemiology and Drug

Self Cite

View full text Add to dashboard Cite

Purpose: Many outcomes derived from electronic health records (EHR) are not only imperfect but may suffer from exposure-dependent differential misclassification due to variability in the quality and availability of EHR data across exposure groups. The objective of this study was to quantify the inflation of type I error rates that can result from differential outcome misclassification. Methods: We used data on gold-standard and EHR-derived second breast cancers in a cohort of women with a prior breast cancer diagnosis from 1993–2006 enrolled in Kaiser Permanente Washington. We simulated an exposure that was independent of the true outcome status. A surrogate outcome was then simulated with varying sensitivity and specificity according to exposure status. We estimated the type I error rate for a test of association relating this exposure to the surrogate outcome, while varying outcome sensitivity and specificity in exposed individuals. Results: Type I error rates were substantially inflated above the nominal level (5%) for even modest departures from non-differential misclassification. Holding sensitivity in exposed and unexposed groups at 85%, a difference in specificity of 10% between the exposed and unexposed (80% vs 90%) resulted in a 36% type I error rate. Type I error was inflated more by differential specificity than sensitivity. Conclusions: Differential outcome misclassification may induce spurious findings. Researchers using EHR-derived outcomes should use misclassification-adjusted methods whenever possible or conduct sensitivity analyses to investigate the possibility of false-positive findings, especially for exposures that may be related to the accuracy of outcome ascertainment.

show abstract

Section: Discussionsupporting

confidence: 50%

“…We found that, when misclassification of second breast cancers was nondifferential, parameter estimates were only minimally biased. However, under differential misclassification bias became relatively severe …”

Section: Introductionmentioning

confidence: 99%