Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Screening for falls in the ED - which test is best?

Three Part Question

in [ambulatory elderly patients presenting to the ED]which [falls screening tool] the [most useful to predict Falls Risk]?

Clinical Scenario

You have just reviewed an 87 year old lady who has presented to the emergency department with a minor laceration after striking her head on a shelf at home. She lives independently in the community, and thinks she may have tripped on a loose carpet. You note that she appears to be quite frail and you wonder if there is an appropriate tool to use, to assess her risk of falling before her discharge.

Search Strategy

MEDLINE [1950- May week 3 2010], EMBASE [1947-2010 May 28], Psycinfo [1806-May week 4 2010] all three simultaneously via OVID interface. CINAHL [1950-2010 May 28 2010 using, and the Cochrane Library May 2010
[[elderly OR aged] AND falls] AND [screening test OR risk assessment OR assessment tool] AND [emergency department OR community dwelling OR ambulatory]LIMIT to humans and English language

Search Outcome

392 citations were identified using the strategy. 19 were relevant to the question. 10 of these were of sufficient quality to include in this review.

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Bergland et al
307 community dwelling females aged 75 to 93 years Patients were asked to rise from a supine position on the floor unaided. The outcome measured was whether they managed to rise from the floor unaided without stumbling or falling. Patients were followed for 12 months to measure incidence of falls prospectively, using daily falls calendars and with phone contact. incidence of all falls57%Patients who could not reach the research office alone or who could not stand for 60 seconds, or who had major cognitive impairment were excluded.
Prediction of all fallsThe sensitivity for all falls was 20%, specificity 77%. The Likelihood Ratio (LR) was 0.87.
Bogle Thorbahn L et al
66 adults aged 69 to 94 years, 2 independent life-care communitiesThis prospective cohort study used a convenience sample of volunteers. Subjects were excluded if they were unable to follow the directions for the test after three repetitions. Each subject had a berg balance test performed; the authors used a cut-off score of 45. 6 month follow-up was via a postal questionnaire (85% return rate). incidence of all falls28%A convenience sample was used, wich may not be representative of general population
The ability of Berg Balance score of 45/56 to predict falls at 6 months sensitivity was 53%, specificity 92. The LR was 7.25. The authors concluded that the Berg balance test is not sensitive in identifying fallers but may identify those that are unlikely to fall.
Cwikel J et al
361 members of a community primary care clinic aged 60 years or older This prospective cohort study recruited members of a community primary care clinic. Each subject completed an Elderly fall screening test(EFAST). Follow-up was to “approximately” 12 months (modal follow-up was 12 months), with an exit interview (78.4% response rate). incidence of all falls35%Subjects were excluded if they were bed or home bound and if they were unable to take part in the interview. It is likely that a more active group were thus included in the study.
The ability of EFAST to predict falls at 12 monthssensitivity was 49%, specificity 83%. The LR was 2.88.
Kario K et al,
266 community dwelling adults aged 65 years and older This was a prospective cohort study. Each subject had computerized dynamic posturography measured. 12-month follow-up was via monthly-preaddressed postcards. Information collection was completed by telephone. incidence of all falls22%Exclusion criteria were extensive excluding large numbers of patients that would typically present to the ED.
ability of 'loss of balance' on testing to predict falls at 12 monthsThe sensitivity was 37%, specificity 75%. The LR was 1.46.
Lin M et al
1,200 community dwelling adults aged 65 years and olderThis prospective cohort study recruited 1,200 community dwelling adults living in one of six rural villages in Taiwan. Each subject performed four balance tests; Timed up and go, Functional reach, One leg stand, Tinetti Balance. 12 month follow-up was via preaddressed postcards, filled after a fall. Subjects were contacted via telephone every three months and asked if they had fallen. TUG was best performing of the tools studiedOR 1.02 (1.01-1.03) AUC 0.61 Results were expressed as Odds ratio and Area under the receiver operating characteristics curve (AUC), but no confidence interval for AUC was presented, nor was the best performing cut-off score given. The incidence of falls at 1 year follow-up was not clearly displayed in the paper.
Lindeman U, et al,
56 community dwelling, non-disabled persons aged 65 years and overThis prospective cohort study recruited 56 community dwelling adults aged 65 years and over at a seniors meeting club. Each subject had Modified Maximum Step Length measured. 12 month follow-up was via fall diaries. The mean age of participants was 67.7 years. In the follow-up year 54% of participants experienced falls. Two participants who had incomplete falls diaries were excluded from the final analysis. incidence of all falls54%Small study size No calculation of different cut-off points subjects were recruited at a seniors meeting club, amoung the exclusion criteria was inability to come to the research centre. This may have selected a more active group of people, than us present at baseline.
Ability of Modified Maximum Step Length score to predict any falls. The sensitivity for using a score of 0.64 to predict falls at 12 months was 63%, specificity 77%. The LR was 2.03. The sensitivity for using a score of 0.66 to predict falls at 12 months was 70%, specificity 69%. The LR was 2.26.
Morris R, et al,
104 community dwelling females aged 60, with history of vertebral fractures This prospective cohort study recruited women with a history of vertebral fractures aged 60 years and over at a seniors meeting club. Each subject had Five meter timed up and go measured. 12 month follow-up was via fall diaries. The mean age of participants was 67.7 years. In the follow-up year 46% of participants experienced falls. incidence of all falls46%Several other tests were also studies but data was not presented for these.
ability of 5m Timed-up-and-go to predict any falls at 12 monthsThe sensitivity was 33.3%, specificity 85.1%. The Positive LR was 2.26.
Muir S et al ,
210 community dwelling ”older adults” from RCT on falls intervention randomized, to receive geriatric assessment.(mean age 79.47 years) This prospective cohort study recruited 210 community dwelling older adults. The data was collected as part of the Project to Prevent Falls in Veterans a community based trial on prevention of falls. All participants who received comprehensive geriatric assessment during the course of the trial, which included measurement of a Berg Scale, were thus included in this study. 12 month follow-up was via monthly fall calendars. incidence of all falls38%Volunteers for study on prevention of falls Exposed to data on falls prevention as part of this Majority of participants were male military veterans 23 patients lost to follow-up – majority of these(19) had high BBS scores
ability of berg balance scale to predict all fallsThe sensitivity to predict falls at 12 months was 25%, specificity 87%. The Positive Likelihood Ratio (LR) was 1.92.
Trueblood PR et al.
198 adults aged 65-90 yearsThis was a prospective cohort study. All participants had Tinetti gait and balance scales, 3m Timed up and go (TUG), Modified Clinical test for sensory interaction on Balance, and 75% Limits of stability(75% LOS) measured. 6 month follow-up was via telephone survey at 4 and 6 months. The investigators were blinded to subjects previous falls history and results of balance assessment scores.incidence of all falls16.5%
Ability of the tests studied; TUG, 75%LOS, Tinetti balance and gait scores and Modified Clinical test for sensory interaction on Balance, to predicit falls at 6 monthsThe 75% LOS was the most sensitive ranging from 0.33 to 0.60 depending on the cut-off used. The TUG and Tinetti Gait tests were more sensitive, all above 0.92.
Verghese et al,
60 community dwelling adults aged 65 years and over.This prospective cohort study comprised of consecutive participants in a longitudinal aging study in New York. All participants performed the following assessments; Tinetti balance and mobility screen, Timed Gait ,Walking while talking (WWT) tasks; simple and complex 12 month follow-up was via structured phone interviews at 6 and 12 months. Researchers were blinded to participants’ history of falls and scores on assessment tools at the follow-up.incidence of all falls22% The mean age for all participants was not quoted.
ability of the tests studied to predict any falls at 12 months.Baseline walking speed using a cut-off of 18 seconds, had good specificity (85.7%) but modest sensitivity (38.4%) (LR2.06).


The overall trend in all the falls risk screening tools looked at was for them to be more specific than sensitive; meaning that the tools are more likely to correctly identify those who are unlikely to fall than to identify those who are likely to fall. 6 of the tests had a sensitivity of greater than 0.80; the Berg Balance Scale, Elderly Falls screening test (EFAST), two components of the modified clinical test for sensory interaction on balance, TUG, Tinetti Balance and gait scales and the walking while talking (WWT) tasks. None of the tests studied has sensitivities of greater than 0.80, only the Modified mean valid step length approached this level (0.77), however it showed only moderate specificity (0.62). The highest likelihood ratios were calculated for the walking while talking complex (9.75) and the berg balance scale (7.25). From the emergency department point-of view, any test considered needs be practically easy to use and quick to administer, ideally with minimal or no training and importantly not require specialist treatment. Only WWT and TUG do not require specialist training to complete. TUG takes less than a minute to perform whereas WWT complex takes up to 5 minutes. There are no inter-rater reliability data available for WWT complex, whereas TUG had good inter-rater reliability. It is not clear which tool is best for assessing falls risk in an emergency department population, however taking both tool performance and practicability into account it seems that the TUG offers the best option at the moment. It has moderate sensitivity and specificity, minimal training is required to be able to administer it.

Clinical Bottom Line

Further research is required to look at the performance of falls screening tools in ED populations as data for this patient population is lacking.


  1. Bergland A, Laake K. Agi Concurrent and predictive validity of Aging Clin Exp Res 2005;17(3):181-85
  2. Bogle Thorbahn LD, Newton RA. Use of the Berg Balance Test to predict falls in elderly persons. Phys Ther. 1996;76(6):576-83.
  3. Cwikel JG , Fried A, Biderman A, Galinsky D. Validation of a fall-risk screening test, the Elderly Fall Screening Test (EFST), for community-dwelling elderly Disabil Rehabil 1998;20(5):161-67
  4. Kario K, Tobin JN, Wolfson LI, Whipple R, Derby CA, Singh D, Marantz PR, Wassertheil-Smoller S. Lower standing systolic blood pressure as a predictor of falls in the elderly: a community-based prospective study J Am Coll Cardiol 2001;38(1):246-52
  5. Lin MR, Hwang HF, Hu MH, Wu HD, Wang YW, Huang FC. Psychometric comparisons of the timed up and go, one-leg stand, functional reach, and Tinetti balance measures in community-dwelling older people. J Am Geriatr Soc. 2004;52(8):1343-48.
  6. Lindemann U, Lundin-Olsson L, Hauer K, Wengert M, Becker C, Pfeiffer K Maximum step length as a potential screening tool for falls in non-disabled older adults living in the community. Aging Clin Exp Res 2008; 20(5):394-399
  7. Morris R., Harwood R.H., Baker R., Sahota O., Armstrong S., Masud T. A comparison of different balance tests in the prediction of falls in older women with vertebral fractures: A cohort study. Age and Ageing. 2007. 36(1)(pp 78-83),
  8. Muir S.W., Berg K., Chesworth B., Speechley M. Use of the Berg Balance Scale for predicting multiple falls in community-dwelling elderly people: A prospective study. Physical Therapy 2008. 88(4)(pp 449-459)
  9. Trueblood PR, Hodson-Chennault N, McCubbin A, Youngclarke D. Performance and impairment-based assessments among community dwelling elderly: Sensitivity and specificity Issues Aging. 2001;24(1):2-6.
  10. Verghese J, Buschke H, Viola L, Katz M, Hall C, Kuslansky G , Lipton R. Validity of divided attention tasks in predicting falls in older individuals: A preliminary study. J Am Geriatr Soc. 2002;50(9):1572-76