BACKGROUND

Cutaneous leishmaniasis (CL) is clinically classified into localized (LCL), mucocutaneous (MCL) and diffuse (DCL) types. While conducting a treatment study on CL at two sites in Ethiopia (Boru Meda and Gondar), differences in opinion in the classification of CL became apparent. The lack of uniformly understandable classifications has made comparison and generalizability of research findings challenging. We wanted to investigate the extent of agreement in CL classification between CL experts and describe on which points there was disagreement.

METHODOLOGY/PRINCIPAL FINDINGS: Thirteen CL experts in Ethiopia were provided with 26 high-quality photographs of CL lesions from patients enrolled in a clinical study on miltefosine treatment. Patients were selected for this sub study due to potential classification difficulties. The most common (majority vote) classification category was noted, and the proportion of experts which choose this (% agreement). Majority vote classification was used to reclassify patients compared to the original study classification. Among included patients, sixteen were originally classified as DCL, eight as MCL, one as LCL, and one as Leishmania recidivans. There was no unanimous consensus for any of the selected cases. The maximum agreement was 80%, which was seen for 38% (10/26) cases. Disagreements existed on whether patients had mucosal involvement, how to classify patients with singular large diffusely swollen lesions and when to classify patients with multiple lesions as DCL. According to the majority vote, 62% of selected patients would be reclassified to a different CL classification compared to the original study from which photographs were selected.

CONCLUSIONS/SIGNIFICANCE: There is no clear understanding and harmony in the classification of CL. Reproducible classification guidelines and training of healthcare providers on CL are needed to ensure consistency in further classification to allow generalizability and comparison of clinical trial findings. Simpler classifications with direct links to treatment decision-making would be valuable.