Identification of Latent Variables Underlying Manual Dexterity In Middle-aged and Old Adults

Abstract

As a biomarker of neurologic health and function, manual dexterity quantifies the ability to coordinate and manipulate objects in a timely manner. The NIH Toolbox measure of manual dexterity is the Rolyan 9-hole pegboard test with the Lafayette 25-hole grooved pegboard test provided as a supplement. PURPOSE: To identify latent variables associated with pegboard times in middle-aged and old adults. We hypothesized that pegboard times would be slower for old adults and that latent variables would differ for the two groups. METHODS: Middle-aged (MA, 40-60 yrs; n=25) and old adults (OA, 65-89 yrs; n=28) performed the 9-hole pegboard test (9HPT) and the grooved pegboard test (GPT), as well as tests of maximal grip strength, tactile discrimination, force steadiness, and the NIH Toolbox cognition battery. Latent variables were identified using Independent Component Analysis from significant Spearman’s rank correlation coefficients between pegboard times and secondary measures. RESULTS: MA adults (51±7 yrs) performed significantly faster than OA (72±5 yrs) on both the 9HPT (18±3 s and 20±3 s, p<0.01) and GPT (60±9 s and 78±16 s, p<0.01). MA adults also had superior tactile discrimination (95±11 au and 72±20 au, p<0.01). The latent variables influencing manual dexterity differed for the two tests and age groups: (1) 9HPT - MA times were negatively correlated (r = -0.84) with the first Independent Component (IC), which explained 37.8% of the covariance and included the 10% double-action pinch force error as the second largest contributor (r = 0.45; scaled r = 0.54); OA times were positively correlated (r = 0.60) with the first IC, which explained 24.4% of the covariance and included wrist extension strength the second largest contributor (r = 0.52; scaled r = 0.88); (2) GPT - MA times were positively correlated (r = 0.71) with the first IC, which explained 44.7% of the covariance and included index finger abduction strength as the second largest contributor (r = -0.42; scaled r = -0.59); OA times were negatively correlated (r = -0.90) with the first IC, which explained 73.6% of the covariance and included age as the second largest contributor (r = -0.44; scaled r = -0.49). CONCLUSIONS: 9HPT times were explained by accuracy to reach a force target and strength for MA and OA respectively, whereas GPT times were explained by strength and age.