麻豆传媒

There's no sign of a gender disparity in multiple-choice test scores at the end of ophthalmology education, but women do slightly lag behind men in a more subjective mid-year evaluation, according to a retrospective, cross-sectional study.

Among 452 residents graduating in 2019, there were no statistically significant differences by gender in average mid-year or year-end patient care ratings, year-end medical knowledge ratings, or year-end Written Qualifying Examination (WQE) scores.

However, the average mid-year medical knowledge ratings were slightly lower for women compared with men (3.76 vs 3.88), a significant -0.12 point difference (95% CI -0.18 to -0.03, P=0.006), Divya Srikumaran, MD, of Wilmer Eye Institute at Johns Hopkins University School of Medicine in Baltimore, and colleagues detailed in .

The findings are heartening overall, Srikumaran said in an interview. But the mid-year data do raise questions about whether bias by evaluators may be affecting the scores of women, she said. "That's the data that may impact where you match for fellowship or your job placement -- your future postgraduate plan. You can imagine it creates a domino effect later."

As the study authors noted, women are underrepresented in ophthalmology, and a "suggested that the number of women applicants in ophthalmology may be trending downward." In addition, research has shown a gender gap in areas such as salary, cataract surgery volume, authorship and publication rates, and grant funding.

The authors launched the new study to evaluate how men and women performed on "Milestone" evaluations during their last year of postgraduate medical education (PGY-4) plus the year-end multiple-choice WQE. Overall, men constituted 61% of the residents included in the study and 39% were women.

Researchers presumed the written test to be objective, while the Milestone evaluations are more subjective and potentially subject to personal bias because they're conducted by faculty, Srikumaran said.

She questioned whether the lower numbers for women at that point are truly a reflection of their knowledge. By the end of the year, just 6 months later, "that difference was gone," she said. "Did the female residents really not know as much as their male counterparts [at mid-year], or was there some bias in play at that point? I don't know that you can catch up that much in a few months."

Srikumaran also noted that while the gender difference in mid-year medical knowledge evaluations was statistically significant, it was quite small. "It's not necessarily a meaningful difference in performance," she said.

Either way, the findings "are not confirming a major bias on a national level," she said.

While the Milestone evaluations are more subjective than written tests, the fact that multiple faculty are involved may actually protect against gender bias, because the ratings aren't just up to a single person, she said. "We have multiple committees looking at the data points," she said. "Maybe that's our strategy for mitigating some of this bias -- have a diverse group of faculty members who are doing these evaluations."

In an , Steven Gedde, MD, of Bascom Palmer Eye Institute at the University of Miami Miller School of Medicine, praised the "well-designed and clearly reported" study and wrote that its findings "raise some concern about gender bias in ophthalmology Milestone assessment."

He added: "Initiatives to reduce gender bias should involve sensitizing faculty to potential biases in their perception of the competence of female residents and emphasizing the need for consistent performance feedback."

The researchers started with 2019 data from 481 PGY-4 residents from 120 ophthalmology programs and whittled the number down to 452 (median age 30) who had matching WQE scores. The medical knowledge scores were most closely matched to year-end written test scores.

Srikumaran and colleagues noted several limitations to their study, such as the possibility that the year-end written exam is not "gender-neutral." And the researchers noted that they couldn't look for race/ethnic bias because those data weren't available.

Comment