Controlling for provider type and specialty, significantly improved diagnostic accuracy was seen in immediate posttest scores compared with pretest scores for conditions shown in both lighter and darker skin types (lighter: 15.2 vs 13.6; darker: 13.3 vs 11.3; both P<.001)(Figure 2). The immediate posttest demonstrated higher mean diagnostic accuracy and confidence scores for skin conditions shown in lighter skin compared with darker skin (diagnostic accuracy: 15.2 vs 13.3; confidence: 3.0 vs 2.6; both P<.001), but the disparity between scores was less than in the pretest.
FIGURE 2. Mean scores for diagnostic accuracy overall and in lighter and darker skin following pretest, immediate posttest, and 3-month posttest. Single asterisk indicates P<.05; double asterisk, P<.01; triple asterisk, P<.001.
Following the 3-month posttest, improvement in diagnostic accuracy was noted among both lighter and darker skin types compared with the pretest, but the difference remained significant only for conditions shown in darker skin (mean scores, 11.3 vs 13.3; P<.01). Similarly, confidence in diagnosing conditions in both lighter and darker skin improved following the immediate posttest (mean scores, 2.7 vs 3.0 and 1.9 vs 2.6; both P<.001), and this improvement remained significant for only darker skin following the 3-month posttest (mean scores, 1.9 vs 2.3; P<.001). Despite these improvements, diagnostic accuracy and confidence remained higher for skin conditions shown in lighter skin compared with darker skin (diagnostic accuracy: 14.7 vs 13.3; P<.01; confidence: 2.8 vs 2.3; P<.001), though the disparity between scores was again less than in the pretest.
Comment
Our study showed that there are diagnostic disparities between lighter and darker skin types among interprofessional health care providers. Education on SOC should extend to interprofessional health care providers and other medical specialties involved in treating or triaging dermatologic diseases. A focused educational module may provide long-term improvements in diagnostic accuracy and confidence for conditions presenting in SOC. Differences in diagnostic accuracy between conditions shown in lighter and darker skin types were noted for the disease categories of infectious, cancerous, and inflammatory conditions, with the exception of conditions more frequently seen in patients with SOC. Learning resources for SOC-associated conditions are more likely to have greater representation of images depicting darker skin types.7 Future educational interventions may need to focus on dermatologic conditions that are not preferentially seen in patients with SOC. In our study, the pretest scores for conditions shown in darker skin were lowest among infectious and cancerous conditions. For infections, certain morphologic clues such as erythema are important for diagnosis but may be more subtle or difficult to discern in darker skin. It also is possible that providers may be less likely to suspect skin cancer in patients with SOC given that the morphologic presentation and/or anatomic site of involvement for skin cancers in SOC differs from those in lighter skin. Future educational interventions targeting disparities in diagnostic accuracy should focus on conditions that are not specifically associated with SOC.
Limitations of our study included the small number of participants, the study population came from a single institution, and a possible selection bias for providers interested in dermatology.
Conclusion
Disparities exist among interprofessional health care providers when treating conditions in patients with lighter skin compared to darker skin. An educational module for health care providers may provide long-term improvements in diagnostic accuracy and confidence for conditions presenting in patients with SOC.