Digital photography as an operational tool for assessing corticosteroid-induced lipodystrophy

Abstract 

Background

Corticosteroid-induced lipodystrophy (CIL) is exclusively diagnosed in a subjective manner.

Objective

To evaluate the reliability of digital photographs in the diagnosis of CIL.

Methods

All consecutive patients starting long-term, high dosage corticosteroid therapy were photographed at baseline and after 3 months of therapy. At the end of the study, 3 physicians with expertise in corticosteroids classified patients as lipodystrophic yes/no/unclassifiable. Photographs analyses performed by 9 medical readers and evaluation of CIL using visual analog scale (VAS) performed during the M3 visit were compared to this classification.

Results

Eighty-eight patients were monitored. Fifty of them were classified by the 3 experts as lipodystrophic and 30 as not lipodystrophic (8 were unclassifiable). Their intra- and inter-observer agreements were moderate or fair (κ coefficient≤0.57) when month 3 photographs were analysed alone and substantial or near perfect (κ coefficient≥0.75) when M3 photographs were analysed beside baseline ones. By comparison with expert consensus, only 3 out of 4 patients were correctly classified using VAS. The AUROC curve and inter-observer agreement significantly improved with experience for the 9 non-experts.

Conclusion

The use of digital photographs do better than VAS to evaluate CIL. The accuracy of diagnosis improves with experience. Morphological changes are more important than morphological phenotype.

Keywords: Corticosteroids, Adverse events, Digital imaging