Grip adds little to disability forecasts in studied arthritis cohorts

Researchers developed prediction models in 3,658 participants from the Survey of Health, Ageing and Retirement in Europe. They tested them unchanged in the Health and Retirement Study (696 participants), English Longitudinal Study of Ageing (639) and China Health and Retirement Longitudinal Study (328). Adding grip strength raised the development cohort’s prediction score from 0.681 to 0.692. This measured how well models distinguished people who later developed disability from those who did not. Grip strength was linked to disability but added little reproducible predictive value elsewhere. Other ways of expressing grip strength did not consistently improve performance. Predicted risks also differed from observed risks across cohorts. The authors called for local testing and adjustment before using predicted absolute risks elsewhere.