The study
A cohort study published in BMC Geriatrics on 6 November 2026 by researchers from LMU University Hospital Munich and Helmholtz Munich, Germany, using data from KORA-Age.
In 2008 and 2009, 1,079 community-dwelling adults aged 65 and over from the Augsburg region had three tests: the Timed Up and Go (TUG, 3 m, timed with a stopwatch), gait speed on a GAITRite walkway, and grip strength with a JAMAR dynamometer. About three years later, a phone interview asked whether they had fallen in the previous 12 months. Fall status was available for 819 people, with a median age of 75. Of these, 148 had fallen: 90 once and 58 more than once.
Falls were asked about in two steps. People were first asked about dizziness, balance problems or falling, and anyone who said no was counted as a non-faller without being asked about falls directly. Some interviews were done with a proxy.
Reading the numbers
Odds ratio (OR): how much the odds of a fall change per unit of the test. An OR of 1.10 means 10% higher odds.
Area under the curve (AUC): how well a model separates fallers from non-fallers. 0.5 is a coin toss and 1.0 is perfect. Around 0.7 is modest.
Sensitivity: the share of fallers a cut-off correctly flags. Specificity: the share of non-fallers it correctly clears.
What they found
Each extra second on the TUG raised the odds of a fall by about 12% after accounting for age and sex (OR 1.12, 95% CI 1.06 to 1.18). After also adjusting for previous falls, health conditions, cognition, opioid use, physical activity and frailty, it was about 10% (OR 1.095, 1.02 to 1.18).
Gait speed was linked to falls until frailty was added. Each 0.1 m/s slower raised the odds by 14% after age and sex adjustment (1.05 to 1.25). In the fully adjusted model, the result was 1.09 (0.98 to 1.21), no longer significant.
Grip strength wasn't linked to falls in any model.
Adding the TUG to age and sex moved the AUC from 0.680 to 0.700 (p = 0.067, so not a statistically significant improvement). Adding gait speed gave 0.692 and grip strength 0.675. All three tests together gave 0.700, the same as the TUG alone.
At a TUG cut-off of more than 12 seconds, sensitivity was 39% and specificity 83%. At the more-than-15-second cut-off used in the 2022 World Falls Guidelines, sensitivity was 21% and specificity 94%.
For recurrent falls, AUCs were higher: 0.79 for gait speed and 0.78 for the TUG. Grip strength reached 0.74 and added nothing beyond age and sex. This analysis was exploratory, based on the 58 people who fell more than once.
Funding and conflicts
Funded by Helmholtz Munich (through the German federal research ministry and the State of Bavaria) and German federal research grants for KORA-Age. The authors declared no conflicts of interest.
What it means for your practice
At the 15-second cut-off, the TUG missed about four in five people who reported a fall. At 12 seconds, it still missed about three in five.
A slower TUG stayed linked to falls after allowing for frailty and previous falls, which gait speed and grip strength didn't.
Adding gait speed and grip strength to the TUG didn't improve the model.
Falls were self-reported for one 12-month window, about two to three years after testing. Falls earlier in the follow-up weren't captured, and the two-step questions may have counted some fallers as non-fallers. The 260 people without fall data were older and frailer, though a statistical correction for this gave similar results.
The tests were done in 2008 and 2009 in one German region.
Bottom line
In 819 older adults in Germany, a slower TUG was linked to a higher chance of reporting a fall, even after allowing for frailty and previous falls. On its own it separated fallers from non-fallers only modestly, and at the 15-second cut-off it missed about four in five people who fell. It seems like a normal TUG result tells us less about falls risk than we might assume.