Treadmill walking underestimates real-world walking spatiotemporal parameters and overestimates physiological demand across inclined terrain: implications for mobility assessment in older adults.

Purpose: Treadmills (TM) are widely used for gait assessment in older adults (OA), yet their ecological validity across inclined terrain remains underexplored. This study compared spatiotemporal, physiological, and kinetic gait outcomes between TM and overground (OG) walking across flat, uphill, and downhill terrain in OA and younger adults (YA), and examined sensorimotor predictors of speed discrepancies. Methods: Twenty-six OA (70{+/-}6 years; 22 women) and 24 YA (26{+/-}5 years; 7 women), none with prior TM experience, completed matched TM and OG trials across three terrain conditions. Self-selected TM speed was determined using a bidirectional protocol. The modified Clinical Test of Sensory Interaction in Balance quantified sensorimotor profiles. Mixed-design ANCOVAs and multiple regression examined condition, inclination, and group effects with sex as a covariate. Results: TM speeds were consistently slower than OG across all conditions in both groups ({Delta} = -0.35 m/s, d = -1.67), with shorter stride length, lower cadence, and altered support phase timing; YA showed larger reductions and a greater shift toward double support than OA. Foot clearance at midswing was largely preserved across modalities. TM walking elicited higher heart rate and RPE despite slower speeds, most pronounced in OA uphill. Ground reaction forces and loading rates were substantially reduced on the TM. Sensorimotor profiles predicted the downhill speed discrepancy (R2 = 0.49), with vestibular and somatosensory contributions as independent predictors alongside age group. Conclusion: TM-derived speed, spatiotemporal, and physiological measures are not interchangeable with real-world ambulation data in OA across inclined terrain.