The 3-m backwards walk and retrospective falls: diagnostic accuracy of a novel clinical measure

Key points

1. 01People who fell walked the 3MBW in a mean 4.8 seconds versus 3.5 seconds for non-fallers (p=0.029)
2. 02A cutoff of 3.5 seconds gave the best overall accuracy (AUC 0.707, sensitivity 74%, specificity 61%)
3. 03Faster than 3.0 seconds: more than 75% had no reported falls; slower than 4.5 seconds: 81% had reported falling
4. 043MBW AUC was significantly higher than the TUG at 8 s and 13.5 s, and the 4-square step test
5. 05The 3MBW showed the largest age-related slowing from the 70s to 80s decade (61% slower on average, versus 36% for TUG)

How it was conducted

Design

Retrospective cohort study

Participants

59 community-dwelling older adults (37 female, 22 male) recruited from 3 retirement communities; mean age 71.5 years (SD 7.6); no neurological deficits; no assistive devices

Primary test

3-metre backwards walk (3MBW): average time of 3 trials walking backwards as fast and safely as possible over a taped 3 m course

Comparator tests

Timed Up and Go (TUG), 5 times sit-to-stand, 4-square step test

Primary outcome

Self-reported fall history in the past 12 months (yes or no)

Analysis

ROC analysis with DeLong method to compare AUCs; sensitivity and specificity at multiple cutoffs; Mann-Whitney U for group comparisons

What they found

• People who fell had a significantly slower mean 3MBW time (4.8 s vs 3.5 s, p=0.029); the TUG (9.3 vs 8.0 s, p=0.077), 4-square step test (9.5 vs 8.1 s, p=0.056), and 5 times sit-to-stand (12.5 vs 10.3 s, p=0.121) did not differ significantly between groups
• 3MBW at 3.5 s: AUC 0.707 (95% CI 0.570-0.821), sensitivity 74%, specificity 61%
• 3MBW at 3.0 s: AUC 0.619 (95% CI 0.479-0.745), sensitivity 78%, specificity 45%
• 3MBW at 4.5 s: AUC 0.665 (95% CI 0.527-0.786), sensitivity 39%, specificity 94%
• 3MBW at 3.5 s had a significantly higher AUC than TUG at 8 s (AUC 0.560, p=0.023), TUG at 13.5 s (AUC 0.528, p=0.011), and 4-square step test at 15 s (AUC 0.522, p=0.004)
• 3MBW AUC was not significantly higher than TUG at 10 s (AUC 0.586, p=0.098), 5 times sit-to-stand at 12 s (AUC 0.592, p=0.092), or 5 times sit-to-stand at 15 s (AUC 0.637, p=0.276)
• More than 75% of people faster than 3.0 s on 3MBW did not report falling; 94% of non-fallers completed 3MBW in under 4.5 s; 81% of those slower than 4.5 s reported falling
• 3MBW correlated strongly with TUG (r=0.823), moderately with 4-square step test (r=0.651) and 5 times sit-to-stand (r=0.608); all p<0.001
• 3MBW slowed 61% from age 70-79 (mean 3.7 s) to 80-89 (mean 5.9 s), compared with 36% for TUG, 8% for 5 times sit-to-stand, and 45% for 4-square step test

Limitations

• Small sample of 59 participants from active retirement communities, limiting generalisability to older adults with more complex conditions
• Study was underpowered to detect AUC differences smaller than 0.15, and 3 of the 6 AUC comparisons were non-significant
• Fall history was self-reported and retrospective, introducing recall bias
• Participants who used assistive devices or had neurological conditions were excluded, so results may not apply to those groups

Why it matters

For patients

If you can walk 3 metres backwards in under 3 seconds you are relatively unlikely to have fallen recently, while taking more than 4.5 seconds is a strong signal that you may be at high fall risk.

For clinicians

The 3MBW is a quick, equipment-free test that performs as well as or better than several standard fall-risk screens; a cutoff of 3.5 seconds offers reasonable sensitivity and specificity for identifying retrospective fallers in healthy community-dwelling older adults.

For readers

This small preliminary study suggests backwards walking speed captures fall risk information beyond what forward-walking tests detect, but prospective trials in larger and more diverse older-adult samples are needed before the 3MBW can be adopted in routine practice.