Screening for hand osteoarthritis (OA) using a postal survey

The aim of this study was to determine the feasibility of screening for hand osteoarthritis (OA) using a postal survey.

Men and women aged 40-79, selected at random from a general practice in Nottingham, were sent a postal survey concerning arm pain. Subjects were asked to identify presence of any nodes, based on a diagrammatic example, and to document specific distal (DIP) and proximal (PIP) nodes. One hundred and thirty-nine subjects were subsequently assessed by a nurse metrologist, blinded to the subjects assessment. Subject-observer agreement was measured by calculating unweighted kappa (kappa). In addition, sensitivity and specificity were calculated, taking observer-identified nodes as the 'gold standard'. Intraobserver reproducibility was assessed on 10 subjects with nodes.

Intraobserver reproducibility was reasonable at all sites (kappa>/=0.59). Of the 56 subjects with observer-identified 'any nodes', 40 subjects had correctly identified them (sensitivity 71.4%); with three subjects incorrectly identifying nodes (specificity 96.4%). Corresponding kappa was 0.70. Subject-observer agreement was higher for presence/absence of any DIP nodes than PIP nodes (kappa=0.67 and 0.36 respectively). As with nodes overall, specificity was high (100% and 96.6% respectively). Agreement was generally poor for specific joint involvement.

A postal survey is a useful screening tool for identifying subjects with nodes. Although sensitivity may not be sufficient to permit assessment of prevalence of hand OA in the community, it has potential applications in terms of identifying and defining cases for further study.