UCSF

Discrimination drives racial and ethnic disparities in maternal mortality and morbidity, imposing both a human and economic burden. Yet, discrimination is challenging to measure accurately. Little is published in perinatal literature about nonresponse bias to experiences of discrimination and how to mitigate this effect to obtain more complete information. In this secondary analysis of the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be (nuMoM2b), we used the Experiences of Discrimination (EOD) scale to examine missing responses to a question about racial or ethnic discrimination during medical care. We categorized responses as "yes," "no," or "nonresponse." Independent variables included racial and ethnic identification, age, body mass index (BMI), education, and poverty level. Logistic regression estimated the odds of nonresponse. Among 9289 participants, 121 (1.3%) reported discrimination, 8598 (92.6%) reported no discrimination, 570 (6.1%) did not respond. In the final adjusted model, racial and ethnic identification was the only statistically significant predictor of nonresponse (p < 0.001). Compared to the most numerous racial group, non-Hispanic white participants, Hispanic participants had 1.8 times the odds of nonresponse [95% CI: 1.3-2.3], and non-Hispanic Black participants had 1.6 times the odds [95% CI: 1.1-2.2]. When individuals most likely to experience discrimination are also the least likely to disclose it, nonresponse bias may result, leading to underestimation of prevalence and obscuring lived experiences of those most affected. Understanding the predictors of nonresponse in patient experience surveys can inform improvements in measurement of discrimination through participant recruitment, survey design, execution, and interpretation.