Clinical Grading of Trachoma From Photographs

Abstract

Purpose: : The WHO program to eliminate trachoma uses a simplified clinical grading system to identify communities with high baseline prevalence and to monitor the efficacy of interventions. Clinical activity does not correlate well with ocular chlamydia infection, partly because it is difficult to standardize and validate the grading system. Grading from photographs allows the comparison of multiple observers outside the field and over time. Here we validate grading from photographs.

Methods: : Subjects were children age 1–5 years living in 44 villages in the Gurage Zone of Ethiopia, where trachoma is endemic. Half of the subjects lived in villages that had undergone periodic community–wide azithromycin treatment. Photographs of the superior tarsal conjunctiva were taken using a macro lens at 1:1 magnification. Six of the authors, with varying levels of experience grading trachoma in the field, then evaluated each photograph for 1) total number of follicles >0.25mm diameter, 2) total number of follicles >0.5mm diameter, 3) papillary hypertrophy and 4) quality of photograph. Each photograph included 0.25mm and 0.5mm reference circles. Ten percent of the photographs were duplicated and inverted in the set in order to evaluate intra–rater reliability.

Results: : The kappa statistics (with 95% confidence intervals) for inter–rater reliability were as follows: 0.50 (0.40 –0.60) for 0.25mm follicular trachoma, 0.26 (0.08–0.52) for 0.5mm follicular trachoma, and 0.57 (0.46–0.67) for severe papillary hypertrophy. Mean intra–rater reliability was 0.68 (0.46–0.90) for 0.25mm follicular trachoma and 0.90 (0.78–1.02) for severe papillary hypertrophy. Photograph quality was rated ungradeable by more than one grader for 2% of the photographs.

Conclusions: : Photographic grading is a reliable method of assessing clinical activity and may be useful to improve standardization of clinical grading. The large majority of photographs taken by amateur photographers with a macro lens are of sufficient quality to allow reliable grading. It may be more appropriate to count all follicles greater than 0.25mm, as surprisingly few follicles are greater than 0.5mm.

Keywords: trachoma • bacterial disease • clinical (human) or epidemiologic studies: systems/equipment/techniques