Long-term outcome of hand pain and functional difficulty in middle and old age: identifying distinct trajectories using latent class growth modelling

Problem

Musculoskeletal disorders affecting the hands are common in middle and old age but their severity, impact, and prognosis is highly variable and not well-explained by diagnosis. We sought to identify and predict different long-term trajectories in hand pain and functional difficulty among community-dwelling symptomatic adults aged over 50 years.

Approach

Latent class growth models were fitted to data from the Clinical Assessment Study of the Hand (CAS-HA) – a prospective population-based cohort of adults aged 50 years and over reporting hand pain in the last 12 months at baseline and followed up at 18, 36, 63, and 89-months respectively – to identify subgroups of participants with similar trajectories of (1) hand pain and (2) hand function over time. At each time point, hand pain and functional difficulty were measured using the AUSCAN questionnaire (0-10; increasing score worse pain or more functional difficulty). Model selection was based on goodness-of-fit statistics and clinical interpretability and was used to identify the model with the optimal number of subgroups. Multinomial logistic regression was used to predict trajectory group membership from baseline predictors including demographic, lifestyle, health, and psychological measures along with measures of participants’ hand condition including onset, severity, chronicity, side affected and number of joints with radiographic osteoarthritis.

Findings

A total of 623 participants were recruited to the CAS-HA study (mean age = 64 (standard deviation = 8.2) years, 62% female) and follow-up rates were 96, 91, 71, and 66% at each follow-up time point respectively. Five sub-groups were found to be optimal in the hand pain data: “Severe” (N=81), “Mild deterioration” (N=159), “Moderate” (N=198), “Mild” (N=144), and “Episodic” (N=39). Correspondingly, “Mild”, “Moderate” and “Severe” trajectory groups were also found in the hand function data, however, an additional three groups were also defined: “Mild/Moderate” (N=138), “Progressively deteriorating” (N=64) and “Improving” (N=70). For both outcomes, the strongest predictor of subgroup membership was the baseline measure for the outcome of interest. Hand function, pinch strength, and sudden symptom onset were also significant predictors of trajectory group membership for hand pain; and physical function, grip strength and frustration, for hand function. However the addition of such extra predictors did not greatly improve overall model fit (e.g. Nagelkerke’s pseudo R-square: Hand pain, baseline only 0.64; with additional predictors 0.70; Hand function, 0.80 and 0.83 respectively).

Consequences

Accurate identification of individuals with a differing future course of hand pain and hand function has important implications for clinical management; to both reassure those with mild problems of the likely unchanging future course of their symptoms and to identify those with severe pain, unlikely to improve, where early onward referral/treatment is most needed. Simple measurement of pain and functional limitation severity at a given point in time may be the most useful single prognostic factor.