Reliability and sensitivity of joint space measurements in hand radiographs using computerized image analysis

Objective. To establish the sensitivity and reliability of proximal interph alangeal (PIP) and metacarpophalangeal (MCP) mean joint space measurements using standard clinical radiographs of healthy subjects, in order to determ ine the limits at which a change in radiographic joint space could indicate a change in actual joint size. Methods. Repeat hand radiographs of healthy subjects were taken using stand ard techniques at 3-5 day intervals with the hands flat (5 posteroanterior radiographs in 8 subjects) or in 6 different flexed positions on a single o ccasion (8 subjects). The mean joint space was determined using custom soft ware and was validated manually. Measurement reproducibility within subject s, within films, and between hand positions was assessed by analysis of var iance. Results. In repeat radiographs taken in the standard clinical position, the precision of individual joint space measurements indicates that changes > 0.11 mm (similar to7%) would represent an actual physical change in joint s pace width (with 95% probability). Averaging measurements across fingers fo r a single subject decreases the detectable change to 0.05 mm. (similar to3 %). With increasing flexure, radiographic joint space tended to increase in MCP and decrease in PIP. Conclusion. Mean finger joint space measured from standard clinical radiogr aphs is a reliable and sensitive measurement in healthy subjects even with some change in hand position. Work is required to establish whether the joi nt space change measured from serial radiographs of patients with arthritis over a period of 6-12 mo exceeds the detectable limits of change derived i n this study.