Physician Gender and Patient Perceptions of Interpersonal and Technical Skills in Online Reviews.
Summary
In 345,053 reviews of 167,150 US physicians, female physicians were more likely to receive negative comments on interpersonal manner and technical competence, and these negatives disproportionately reduced their odds of high star ratings—especially for female primary care physicians and surgeons. The findings highlight pervasive gender bias in online assessments.
Key Findings
- Female physicians had higher odds of receiving negative comments on interpersonal manner (overall OR 1.22; 95% CI, 1.18–1.26).
- Negative comments on technical competence disproportionately reduced the odds of high star ratings for female PCPs (OR 0.60; 95% CI, 0.50–0.73) and female surgeons (OR 0.67; 95% CI, 0.50–0.89).
- Female PCPs were less likely than male PCPs to receive high star ratings even when receiving positive technical competence comments (OR 0.82; 95% CI, 0.70–0.96).
Clinical Implications
Clinicians and institutions should contextualize online ratings with bias-aware frameworks, adjust quality metrics to mitigate gender bias, and consider educational messaging to patients about responsible reviewing.
Why It Matters
Provides large-scale, quantitative evidence of gender bias in patient reviews, informing how rating platforms and institutions should interpret and mitigate biased feedback—relevant for cosmetic/plastic surgery where online reputation heavily influences patient choice.
Limitations
- Cross-sectional design; causality cannot be inferred and unobserved confounding (e.g., case mix) may persist.
- Selection bias of online reviewers and platform-specific effects may limit generalizability.
Future Directions
Test interventions that de-bias ratings (e.g., calibrated displays, reviewer prompts), link review data to clinical outcomes, and replicate across platforms and countries.
Study Information
- Study Type
- Cohort
- Research Domain
- Diagnosis
- Evidence Level
- IV - Large cross-sectional observational analysis using NLP and multilevel models.
- Study Design
- OTHER