Agreement With Pediatric Suprapatellar Bursa Effusion Assessments by Point-of-Care Ultrasound After Remote Training

Background The ease of instruction for point-of-care ultrasound (POCUS) to detect suprapatellar bursa (SPB) effusions in pediatric patients is unknown. Considering in person limitations because of the coronavirus pandemic, strategies for POCUS education by remote learning are necessary. Methods We crafted a 90-minute didactic training that was presented via a remote learning format. The main outcome of interest was the interobserver reliability of SPB effusion assessment by novice sonologists compared with POCUS faculty. Novice sonologists were pediatric emergency medicine (PEM) fellows. Pediatric emergency medicine fellows interpreted longitudinal SPB examinations obtained in our pediatric emergency department from July 2013 to June 2020. Assessments were performed 2 months after the remote training. Pediatric emergency medicine fellows had a limited experience performing these musculoskeletal scans and were blinded to POCUS faculty and each other's assessments. Interobserver reliability was assessed with Cohen kappa coefficient. Second, we calculated test characteristics of knee radiography compared with PEM POCUS faculty determination of SPB effusion by ultrasound. We further explored how effusion size measured by POCUS impacted the diagnosis by knee radiography. A receiver operator characteristic curve of knee radiography diagnosis of SPB effusion was created using the maximal height of SPB effusion by POCUS as the predictor variable. Results A total of 116 SPB scans in 71 patients were assessed. From this group, 70 scans were of affected knees and 46 scans were of contralateral, asymptomatic knees. The mean age of patients was 10 years and 46% were girl. The prevalence of SPB effusions was 42%. The kappa coefficients between the 3 novice sonologists and POCUS faculty were 0.75 (0.62-0.87), 0.77 (0.65-0.89), and 0.83 (0.72-0.93) with 88%, 89%, and 91% agreement. Knee radiography exhibited an overall sensitivity of 65% (95% confidence interval [CI], 46-79%), specificity of 84% (95% CI, 60-97%), negative predictive value of 55% (95% CI, 43-66%), and positive predictive value of 88% (95% CI, 73-96%) to diagnose SPB effusions. The area under the receiver operator characteristic curve was 0.850. With an SPB height cutoff of 4 mm as true positives, radiography had a sensitivity of 81% and a specificity of 83%. Conclusions After a remote teleconference didactic session, PEM fellows were able to successfully diagnose SPB effusions using a longitudinal view with substantial interobserver reliability. Knee radiography exhibited limited sensitivity to rule out SPB effusions.