Slow Walking and Fuzzy Thinking Together Reveal Who Will Fall After 90

For people over the age of 90, a fall is rarely just a bruise. It can mean a broken hip, a head injury, a hospital stay, and the beginning of a downward spiral toward immobility, nursing home admission, or death. Yet the fastest-growing segment of the United States population—the oldest old—has been largely left out of the research that shapes how doctors decide who is most likely to fall. A new study published in GeroScience by researchers at the University of California, Irvine, drawing on The 90+ Study, one of the largest longitudinal investigations of aging and dementia ever conducted, now offers a sharper way to separate those at genuine risk from those who can keep moving safely. The answer, it turns out, lies not in how fast someone walks or how well they think, but in the interplay between the two.

The problem with existing fall risk assessments is stark. Most screening tools rely on physical performance measures such as gait and balance testing, with cutoff points derived from studies of adults aged 65 and older that included few or no participants over 90. Applied to a 93-year-old, nearly every one of these tools would flag the patient as high risk. But fewer than half of people aged 90 and older actually fall each year. Labeling everyone as high risk is not merely imprecise; it is actively harmful. Hospital patients tagged as fall risks are often kept in bed without assistance, which accelerates deconditioning and immobility. A high-risk label can also make it harder for clinicians to justify discharge home or to intensive rehabilitation, limiting access to restorative care for an already vulnerable group and feeding the already high rates of institutionalization among the oldest old.

To cut through this diagnostic fog, the research team led by Katherine A. Colcord and María M. Corrada analyzed 1,099 participants from The 90+ Study—749 women and 350 men with a mean age of 93.2 years. The cohort, largely drawn from surviving members of the Leisure World Cohort Study, a landmark epidemiological project begun in 1981 in a California retirement community, was evaluated every six months. At each visit, examiners recorded falls reported by participants or their informants, administered the Four-Meter Walk Test, and assigned a cognitive diagnosis ranging from normal to cognitive impairment no dementia (CIND) to dementia. The researchers then classified each participant into four groups: neither slow gait nor impaired cognition, slow gait alone, impaired cognition alone, or both.

The gait measure was deliberately tailored to this age group. Rather than using the conventional 1.0 meter-per-second threshold developed in younger populations—a cutoff that would have classified 93 percent of the study participants as slow walkers—the team defined slow gait as a speed below the sex-specific median: 0.56 meters per second for women and 0.69 meters per second for men. These values closely match published reference percentiles for adults over 90, allowing the investigators to distinguish relatively slower walkers within the cohort rather than simply sorting nearly everyone into the same bucket. The Four-Meter Walk Test itself is elegantly simple: participants walk four marked meters at their usual pace, timed from the moment the first foot crosses the starting tape to when it crosses the finish line, with buffer zones for acceleration and deceleration.

Using generalized linear mixed regression models with a negative binomial distribution and log person-years as an offset, the team calculated incidence rate ratios comparing each gait-cognition group against the reference group with neither deficit. The results, published in September 2026, revealed a striking sex difference. Among women, slow gait alone raised fall risk by 33 percent and the combination of slow gait and impaired cognition by 35 percent, but impaired cognition by itself showed no significant association. Among men, all three categories mattered: slow gait alone raised risk by 70 percent, impaired cognition alone by 75 percent, and the combination by a full 134 percent compared with men who had neither deficit. Adjusting for demographics and for a comorbidity index spanning seventeen chronic conditions did not meaningfully change these patterns, suggesting the associations were not driven by overall disease burden.

The sex difference persisted in sensitivity analyses designed to probe its robustness. Among the 604 participants who walked without any assistive device—people who might not otherwise be flagged as vulnerable—the combination of slow gait and impaired cognition was a powerful predictor, raising fall risk more than twofold overall, with significant effects in both women and men. In the subgroup of 596 participants who reported no falls at or before their baseline visit, impaired cognition and the combined deficit remained significant predictors in men but not women, indicating that the male associations were not simply an artifact of a prior history of falling. When the researchers re-included 110 participants who were unable to complete the walk test—counting them as slow walkers, since inability to finish likely reflects poor physical function—the combined group predicted falls in both sexes, and the magnitude of the risk estimates grew slightly, hinting that the true associations may be even stronger than observed.

The longitudinal trajectories added another layer of insight. In men, fall rates climbed steadily over time in the groups with neither deficit, slow gait alone, or impaired cognition alone, but remained consistently high from the start in the combined group. In women, fall rates rose significantly only in the group with neither deficit and stayed relatively flat elsewhere. This pattern carries a double message: the combination of slow gait and impaired cognition identifies people whose risk is persistently elevated, while even those who appear entirely healthy at 93 face rising risk as time passes. Continued fall screening, the authors argue, must not stop at age 90.

Why should cognition and gait be more predictive together than either alone? Several mechanisms have been proposed. Cognitive impairment erodes judgment and self-awareness, leading people to place themselves in precarious situations they can no longer safely navigate. Neuropathological changes in the brain that drive cognitive decline may simultaneously drive physical decline, so cognition testing can catch subtle deterioration in balance, postural control, and reaction time that gait speed alone misses. Medications matter too: cholinesterase inhibitors prescribed for cognitive impairment can cause dizziness and lightheadedness, indirectly raising fall risk. The findings echo earlier work in younger seniors—a meta-analysis of 6,204 participants with a mean age of 74.9 found a 44 percent higher fall risk in people with both slow gait and subjective cognitive complaints—but the new study extends this evidence into the ninth decade of life, where the stakes are highest.

The sex difference itself may have a surprisingly practical explanation. In the cohort, women used walkers more often than canes, while men favored canes. Canes demand greater cognitive capacity for sequencing and timing during gait and require more energy expenditure, and previous work from the same group found that walker use was associated with a declining fall rate over time while cane use was associated with an increasing one. A cane may simply be the wrong prescription for a cognitively impaired 93-year-old, and the authors suggest that referral to a clinician trained in gait analysis before prescribing an assistive device could itself become a fall prevention strategy. The study is not without limitations—falls were self-reported at six-month intervals, which may have led cognitively impaired participants to underreport and thus underestimate the cognitive effect, and the cohort was 98.6 percent White—but its inclusion of homebound participants, with examiners traveling even out of state, makes it unusually representative of the real 90-plus population.

The implications reach well beyond the clinic. As the oldest old become a larger share of society, distinguishing who truly needs fall precautions from who does not could preserve mobility, independence, and dignity for millions. This study shows that a stopwatch, a four-meter hallway, and a careful cognitive assessment—together, not separately—can do what decades of age-based screening could not: find the people who are genuinely about to fall, before they do.

Subject of Research: The combined role of gait speed and cognitive impairment in predicting fall risk among adults aged 90 and older

Article Title: Enhancing fall risk assessment in the oldest old: The interplay of cognition and gait speed

Article References: Colcord, K. A., Kristinsson, H. B., Jiang, L., Melikyan, Z. A., Al-darsani, Z., Falvey, J. R., Kawas, C. H., & Corrada, M. M. (2026). Enhancing fall risk assessment in the oldest old: The interplay of cognition and gait speed. GeroScience. https://doi.org/10.1007/s11357-026-02516-0

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02516-0

Keywords: fall risk, oldest old, gait speed, cognitive impairment, dementia, The 90+ Study, geroscience, fall prevention, aging, sex differences, Four-Meter Walk Test, assistive devices