Grip strength, walking pace, and hospital-record-defined incident degenerative cervical myelopathy: a UK Biobank cohort study

Previous UK Biobank work associated grip strength and usual walking pace with degenerative cervical myelopathy (DCM). We evaluated whether these associations were robust to more specific hospital-record definitions, competing mortality, alternative exposure modeling, and internally validated incremental discrimination.

We analyzed 501,936 UK Biobank participants. Grip strength was instrumentally measured, and usual walking pace was self-reported. The primary outcome was any-position hospital-record-defined incident DCM; primary-position DCM and paired hospital diagnosis and cervical procedure records were confirmatory outcomes. Cause-specific Cox models adjusted for age, sex, ethnicity, smoking, body mass index, alcohol intake, diabetes, and hypertension. Grip strength was examined continuously and with restricted cubic splines; walking pace was modeled categorically. Fine-Gray competing-risk models, lagged analyses, interaction tests, and 200-resample bootstrap internal validation were performed.

During a median 16.43 years (interquartile range 15.58-17.16; 7,945,865 person-years), 978 participants developed hospital-record-defined DCM (12.31 per 100,000 person-years). The main complete-case analysis included 489,344 participants and 923 events. Each 5-kg higher grip strength was associated with a lower hazard (hazard ratio [HR] 0.859, 95% confidence interval [CI] 0.823-0.896). Relative to brisk pace, steady/average pace (HR 1.274, 95% CI 1.087-1.495) and slow pace (HR 2.543, 95% CI 2.049-3.156) were associated with higher hazards. The grip-strength association was approximately linear (P for nonlinearity = 0.724), and no interaction was evident by sex or age. Primary-position, strict 0-90-day surgery, lagged, and Fine-Gray analyses were directionally consistent. In exploratory internal validation, adding both markers increased the optimism-corrected C-index from 0.650 to 0.680 and the censoring-adjusted 10-year AUC from 0.642 to 0.679; the 10-year Brier-score improvement was numerically very small.

Instrumentally measured grip strength and self-reported walking pace were associated with subsequent hospital-record-defined DCM. The results extend previous UK Biobank findings by applying more specific hospital-record definitions, competing-risk methods, evaluation of functional form, and bootstrap internal validation. These nonspecific functional measures may inform future prospective diagnostic research but should not be used as stand-alone screening or triage tests.