A new study using artificial intelligence has uncovered troubling disparities in how hospital staff document labor and delivery experiences—disparities that disproportionately affect Black, Hispanic, and Asian/Pacific Islander (API) patients.
Researchers at Columbia University School of Nursing analyzed over 18,000 clinical notes from two major hospitals using natural language processing. They found that Black patients were 22% more likely than white patients to have stigmatizing language in their records. This included being labeled “difficult,” having their credibility questioned, or being described in ways that reflected bias toward their identity. In fact, 33% of Black patients were described as “difficult,” compared to 28.6% overall.
Lead researcher Veronica Barcelona, Ph.D., emphasized that clinical documentation can both reflect and perpetuate bias, potentially influencing future care. “These findings underscore the importance of implementing targeted interventions to mitigate biases in perinatal care,” the authors wrote.
While Black patients were also 19% more likely to have positive language in their notes, such as being portrayed as active participants in their care, the presence of stigmatizing language raises concerns about how bias may shape medical decision-making.
Hispanic patients were 9% less likely than white patients to be labeled “difficult,” but also 15% less likely to receive positive documentation. API patients were 28% less likely to be described using marginalized identity language and 31% less likely to have notes reflecting power or privilege.
The study highlights how subtle language choices in medical records can reinforce systemic inequities in maternal care.
See: “AI analysis of hospital labor and delivery notes finds racial disparities in biased language” (May 13, 2025)