Utility of digital photographs of the hand for assessing the presence of hand osteoarthritis.

OBJECTIVES Nodal osteoarthritis of the hand (hand OA) is a subset of OA with a strong heritable component. Multiple genetic analyses of this condition have been performed and are underway. Highest yield from any genetic study depends upon a clear clinical phenotype for case definition. Radiographs may provide the most detail about the nature of the lesion. Physical examination is an imperfect means of evaluating each patient, particularly when hundreds or thousands of patients are required for study. Our study evaluated the accuracy, relative to a radiograph, of a digital photograph of the hands for the presence of OA in a particular joint, as well as for the diagnosis of nodal hand OA. METHODS Consecutive patients were evaluated as part of the I-NODAL study (Investigation of Nodal Osteoarthritis to Detect an Association with Loci encoding Interleukin-1 [IL-1]). Evaluation included a physical examination by a trained rheumatologist, a postero-anterior radiograph of the hands, and a digital photograph of each hand. Radiographs were read by one trained observer using the Kellgren-Lawrence scale. Photographs were taken by one individual and were analyzed by an experienced rheumatologist. Kappa statistics were determined for each modality and accuracy was assessed using radiographic readings as a gold standard. RESULTS Intra-reader reliability for radiograph interpretation was good for the overall diagnosis of hand OA (kappa0.76 [0.45,1.07]), but varied widely for the presence or absence of K-L grades 2-4 in individual joints (median kappa0.70, range 0.49-0.87 for ACR index joints). Distal interphalangeal joint (DIP) nodes on physical examination were sensitive (median 96.27, 93.94-100), but not specific for radiographic hand OA in the corresponding joint (median 33.0, 17.24-42.86). Physical examination evidence of OA in the 1st carpo-metacarpal (CMC) and proximal interphalangeal (PIP) joints provided only moderate sensitivity and specificity. However, the negative predictive value of the examination of individual joints was good (median negative predictive value was 82.58 for IP joints with a range 68.29-100.00), particularly in the DIP joints. Specificity of a node visualized on hand photograph was variable (median for all IP joints and 1st CMC 83.77, range 53.37-96.97), with greatest specificity for radiographic OA in the corresponding joint found in the 1st CMC and the PIP joints. Clinical hand OA was sensitive, but not specific for the radiographic diagnosis of hand OA; while, photographic OA was moderately specific, but insensitive. CONCLUSION The visualization of a node on a digital photograph of the hand provides fair to moderate specificity for radiographic hand OA in the corresponding joint, with generally poor sensitivity. A photograph has limited value as a screening tool for the diagnosis of radiographic hand OA.