Impact of In-Person and Electronic Training by Breast Radiologists on Rwandan General Practitioners' and Nurses' Skills in Diagnostic Breast Ultrasound

Background: Ultrasound (US) is a key tool in evaluation of palpable breast masses and can help refine the likelihood of malignancy and the need for further diagnostic studies. US technology is available in many low-resource settings, but there are few specialized radiologists. We launched a diagnostic breast ultrasound training program for general practitioner doctors (GPs) and nurses at a rural Rwandan district hospital that serves as a cancer referral facility. Aim: Assess GPs' and nurses' skill in diagnostic breast ultrasound over 23 months of intensive in-person and online supervision and mentorship. Methods: 4 rotating breast radiologists from Brigham and Women's Hospital trained 5 nurses and 4 doctors in Rwanda over 9 weeks of in-person training and 21 months of weekly remote case consultations and mentorship using electronic review of images with emailed feedback. During in-person trainings, trainees and radiologists evaluated patients separately, while radiologists' electronic assessments were based on emailed images and assessments from trainees. Among breast lesions with documented radiologist and trainee assessments, we compared written trainee and radiologist assessments to calculate the sensitivity of trainee assessments, with radiologist assessments as the gold standard. We used paired t-tests to examine whether the sensitivity varied between the first 14 months (stage I) and the last 9 months (stage 2), after the final in-person training. Results: Of 323 breast and axillary lesions assessed by trainees and radiologists, 279 were breast lesions. Of these, 114 (41%) were evaluated by radiologists in-person, and 165 (59%) through electronic evaluation. 237 (85%) were determined to be breast masses by the radiologists, with 164 of these solid masses, 25 complex solid/cystic lesions, 15 definite or probable simple cysts, 31 normal intramammary lymph nodes, and 2 other masses. Sensitivity of trainees' assessments for identifying a solid mass was 90.2% (95% CI 85.9-94.9) overall. Among trainees who scanned ≥ 10 lesions each, mean sensitivity was 90.6% in stage I, and 94.0% in stage 2 ( P = 0.3). In cases where both radiologists and trainees perceived solid masses (n=148), trainees' assessments had a sensitivity of 81.4% (95% CI 72.3-90.5) overall for detecting masses suspicious for malignancy, or probably benign but needing further evaluation (versus benign with no further evaluation needed). Among trainees who scanned ≥ 10 lesions each, sensitivity was 79.1% during stage I and 96.2% during the stage 2 ( P = 0.03). Conclusion: Nurses and GPs in a rural sub-Saharan African facility built strong skills in diagnostic ultrasound over 23 months of combined in-person training and distance learning via electronic case reviews. The sensitivity of their assessments for identifying masses concerning for malignancy showed significant improvement after sustained mentorship. Assessment of impact on patient care and outcomes is ongoing.