Q&A: Researchers assess role of linguistics in racial, ethnic bias in pregnancy outcomes

The Columbia University Data Science Institute recently awarded a two-year grant to researchers who will explore why racial and ethnic disparities in pregnancy outcomes exist in the context of linguistic biases.

“There is a growing body of qualitative literature describing obstetric racism,” Veronica Barcelona, PhD, MSN, RN, assistant professor at Columbia University School of Nursing, said in the school’s February Research Roundup. “Black birthing people have reported experiences of lack of autonomy, poor communication and feeling unheard. Unfortunately, generating solutions to improve this problem has proven difficult. Racism and discrimination are embedded in our society at large, and solutions must address racism in institutions and structures to improve the health of our most vulnerable.”

Supported by the grant, Barcelona and colleagues will develop natural language processing tools to examine linguistic biases in medical records and compare them with pregnancy outcomes.

Healio spoke with Barcelona to learn more about her research.

Healio: What prompted your research on this topic?

Barcelona: Black birthing people are twice as likely to experience severe pregnancy complications as those who are white, and [Latin Americans] also have an increased risk. Scientists have identified many individual-level risk factors that contribute to adverse pregnancy outcomes, often resulting in blaming the patient instead of the systems that perpetuate this harm. Black and [Latin American] birthing people face particularly toxic stressors, such as racism and discrimination, which contribute to poor pregnancy health.

Though there is a growing body of literature on patient experiences of racism in pregnancy health care, there is limited work on provider biases and pregnancy outcomes. I wondered how the relationship between individual and institutional racism is associated with pregnancy-related morbidity outcomes. I am fortunate to work with great colleagues at Columbia University School of Nursing who have complementary expertise to mine, in artificial intelligence, natural language processing (NLP) and data mining of the electronic health record. We designed a project to expand on a new area of inquiry: the use of NLP to identify stigmatizing language and its association with maternal morbidity outcomes. Previous work had focused on patterns of stigmatizing language use by race and ethnicity, and no studies had been applied to pregnant populations.

Healio: How will machine learning be implemented in your research?

Barcelona: We will refine and compare NLP approaches to identify stigmatizing language in clinical notes, then examine patterns of stigmatizing language use by patient race and ethnicity. Finally, we will examine associations with selected maternal morbidity outcomes.

Healio: How does linguistic bias play a role in pregnancy outcomes?

Barcelona: Previous work has identified racism and bias in stigmatizing language in outpatient internal medicine and home care settings by race and ethnicity. We will build upon these identified categories of stigmatizing language including stereotyping, questioning the veracity of the patient’s words and use of judgment words. We expect to find similar disparities in the use of stigmatizing language by race and ethnicity in pregnancy care settings as racism permeates health care. Language is important, and the way the bodies of pregnant people are described in terms of “failure speak” or “geriatric pregnancies” further compounds these harms.

Healio: Why is this topic important for clinicians?

Barcelona: The 21st Century Cures Act now allows patients to immediately view their medical records. Clinicians and institutions must ensure that their internal implicit biases are addressed so they are not reflected in the medical record and perpetuated from one provider to the next. Racism permeates American society, and every health care provider and institution has a responsibility to examine their own thoughts, behaviors and policies that contribute to unequal care and outcomes for patients.

Healio: What impact do you anticipate your findings will have?

Barcelona: Findings from this project will inform systems-level interventions to improve patient care and outcomes. For example, we hope to develop tailored interventions for clinicians and institutions to address racism and bias and improve the care of pregnant and birthing people.

Healio: Is there anything else you would like to highlight?

Barcelona: All women and birthing people deserve to be treated with respect and care in one of the most vulnerable moments of life. By improving the care for Black and [Latin American] patients, we ensure a healthier environment for all families to grow and thrive.

Reference:

• Columbia University School of Nursing February Research Roundup. https://www.nursing.columbia.edu/news/february-research-roundup. Published Feb. 17, 2022. Accessed Feb. 25, 2022.

For more information:

Veronica Barcelona, PhD, MSN, RN, can be reached at vb2534@cumc.columbia.edu.