Dissociation Between Instrumented Mobility and Self-Reported Function in Adults Aged 70 Years or Older After Unilateral Total Knee Arthroplasty: A Preliminary Longitudinal Study Using an Inertial Sensor and the WOMAC

Abstract

Background and Objectives: Total knee arthroplasty (TKA) reliably reduces pain and improves self-reported function, yet objective mobility may remain impaired in older adults. We examined the relationship between inertial-sensor-based mobility and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) in adults aged 70 years or older after unilateral TKA. Methods: In this observational longitudinal study, 25 participants were assessed at up to three postoperative time points (M1, first available assessment; M2, 2024 follow-up; M3, 2025 follow-up). Outcomes were the instrumented Timed Up and Go (TUG) test with segmented turning variables, the 2-Minute Walk Test (2MWT), spatiotemporal gait variables recorded with a trunk-mounted inertial measurement unit, and the WOMAC. Paired changes in the instrumented outcomes were tested with Wilcoxon signed-rank tests under Benjamini–Hochberg false-discovery-rate (FDR) control and are reported with bootstrap confidence intervals (CIs); longitudinal change in the WOMAC was not tested, because the response format used at the first assessment could not be established from the available records, so WOMAC scores are used only within time point; associations used Spearman correlations with bootstrap CIs and leave-one-out influence diagnostics. Individual-level agreement was examined by standardising each of three externally validated objective outcomes against the inverted WOMAC total score within time point and classifying discordance across a range of cut-offs. Observed changes in the instrumented outcomes were benchmarked against published minimal detectable change (MDC) values. Results: Participants were 75.56 ± 4.00 years old and 64.0% were women; 14 were reassessed at M2 and 2 at M3. No instrumented global outcome changed after FDR adjustment, and the median changes in total TUG time (+0.07 s) and 2MWT distance (+12.80 m) remained below published MDC95 values (2.27 s and 14.96 m). Mid-turn duration shortened by a median of 0.80 s (95% CI −0.93 to −0.28) and mid-turn peak and mean angular velocities increased by 16.55°/s (8.00 to 30.10) and 7.40°/s (1.30 to 14.15), all with FDR-adjusted p = 0.036. At M2, the WOMAC total score was inversely associated with walking speed (rho = −0.762, 95% CI −0.973 to −0.307), cadence (rho = −0.761) and 2MWT distance (rho = −0.644); these three associations retained their sign and significance in every leave-one-out refit and changed little on adjustment for age, body mass index or postoperative duration, whereas all M1 associations were weak. Between 42.9% and 57.1% of participants fell outside a ±0.5 standardised concordance band at M2 depending on which validated outcome was used, with no consistent asymmetry in either direction, and the estimated prevalence ranged from 7.1% to 85.7% across plausible cut-offs. Conclusions: Group-level association between patient-reported and instrumented outcomes coexists with frequent individual-level discordance, but the magnitude of that discordance is not stably estimable in a cohort of this size. WOMAC scores should not be assumed to substitute for objective mobility testing after TKA, and confirmation in larger prospective cohorts is required.