A smartphone-based consultation system for acute burns – methodological challenges related to follow-up of the system
AIM This study assessed whether photographs of burns on patients with dark-skin types could be used for accurate diagnosing and if the accuracy was affected by physicians' clinical background or case characteristics. METHOD 21 South-African cases (Fitzpatrick grades 4-6) of varying complexity were photographed using a camera phone and uploaded on a web-survey. Respondents were asked to assess wound depth (3 categories) and size (in percentage). A sample of 24 burn surgeons and emergency physicians was recruited in South-Africa, USA and Sweden. Measurements of accuracy (using percentage agreement with bedside diagnosis), inter- (n=24), and intra-rater (n=6) reliability (using percentage agreement and kappa) were computed for all cases aggregated and by case characteristic. RESULTS Overall diagnostic accuracy was 67.5% and 66.0% for burn size and depth, respectively. It was comparable between burn surgeons and emergency physicians and between countries of practice. However, the standard deviations were smaller, showing higher similarities in diagnoses for burn surgeons and South-African clinicians compared to emergency physicians and clinicians from other countries. Case characteristics (child/adult, simple/complex wound, partial/full thickness) affected the results for burn size but not for depth. Inter- and intra-rater reliability for burn depth was 55% and 77%. CONCLUSION Size and depth of burns on patients with dark-skin types could be assessed at least as well using photographs as at bedside with 67.5% and 66.0% average accuracy rates. Case characteristics significantly affected the accuracy for burn size, but medical specialty and country of practice seldom did in a statistically significant manner.