Fall Risk: Balance Confidence Predicted It, Tests Did Not

Adult children usually arrive at this question after something specific: a stumble on the stairs, a bruise nobody explains, a parent who has stopped coming to things.

The instinct is to look for proof. The evidence says the proof you are looking for is the weakest signal available.

What predicted falling, and what did not

The Physical Therapy analysis entered fall history, pathology and a battery of measures as candidate predictors, then followed participants to see who actually fell. Three variables survived into the final model:

And the sentence that inverts the usual advice:

“Fall history, presence of pathology, and physical tests did not predict falling.”

Not fall history. Not a diagnosis. Not the physical tests — with the partial exception of Timed Up and Go, which contributed under 5%.

What carried nearly four-fifths of the model’s predictive power was a questionnaire asking people how confident they felt doing ordinary things: walking around the house, reaching for something on a shelf, stepping onto an escalator.

The authors’ own reading is worth quoting, because it is the practical takeaway: participants “may have had a better sense of their fall risk than with a test that provides a snapshot of their balance.”

They knew. They were just never asked.

The quiet symptom: what they have stopped doing

The second-strongest predictor was not confidence but behaviour — the avoidance that follows from low confidence.

A separate 12-month prospective cohort in European Geriatric Medicine followed older adults to see who newly began restricting activity because of concerns about falling. Over the year, 10.1% did. Three things predicted it:

• Greater frailty — Fried Frailty Index, OR 1.58 (95% CI 1.09 to 2.30)
• Experiencing a fall during the study — OR 2.22 (95% CI 1.13 to 4.38)
• Poorer functional mobility — Timed Up and Go, OR 1.08 (95% CI 1.01 to 1.15)

A fall roughly doubles the odds that someone starts pulling back from their own life.

This is the part families miss, because avoidance does not look like a symptom. It looks like preference. She doesn’t really like going out any more. He’s stopped doing the garden. They’ve started using the downstairs loo.

Those are not preferences. On this evidence they are the observable form of the strongest predictor there is.

What to actually ask

Since confidence outperformed the clinical measures, the useful conversation is about confidence — and it works better as specifics than as “are you worried about falling?”, which invites a reflexive no.

Ask how confident they feel, right now, doing each of these:

• Walking around the house
• Getting up from a low chair without using their arms
• Walking on an uneven pavement
• Reaching for something above head height
• Stepping into and out of the bath or shower
• Walking outside when it is icy or wet
• Going up and down stairs without holding on

Then ask the avoidance question, which is often more honest: what have you stopped doing in the last year, and why?

A confident answer across the list is reassuring in a way a clean fall history is not. A hesitant one is a signal that outperformed every physical test in this analysis.

What this does not mean

It does not mean physical assessment is useless. Timed Up and Go still contributed, and it predicted the onset of activity restriction in the second study. A falls clinic does things this page cannot: medication review, blood pressure, vision, neuropathy, strength.

It does not mean a fall history is irrelevant. It means that in this analysis it did not predict who fell next, which is a narrower claim. A previous fall still doubled the odds of activity restriction, and it remains the reason most families start looking.

And this is one study with real limits. The authors note falls were recorded by participant recall rather than a diary, which is a genuine weakness — memory for falls is imperfect. One prospective analysis in one sample is a strong signal, not a settled fact.

Where a medical alert system fits, honestly

These findings change the case for a call button, and not in the direction the advertising usually goes.

The usual pitch is rescue: something goes wrong, you press it, help comes. That is real, and we went through what the evidence says about how often people actually press it in medical alert systems: most who needed one did not press it.

But if the strongest predictor of falling is low balance confidence, and the behaviour that follows is withdrawal from ordinary activity, then the more defensible argument for a device is that it lets someone keep doing things. The walk, the garden, the shower with the door closed. Not because the button prevents a fall — it does not — but because the alternative to confidence is avoidance, and avoidance is itself in the model.

That framing also happens to be the one older people accept. The objection recorded in the falls literature is not price. It is that accepting the device feels like conceding something. An argument built on staying active meets that objection; an argument built on rescue confirms it.

The short version

Balance confidence explained 38.7% of the variance in who fell. Avoidance behaviour added 5.6%. Timed Up and Go added 4.9%. Fall history, diagnoses and the physical tests predicted nothing in this analysis.

Meanwhile one in ten older adults newly restricts their activity over a year because of falling concerns, and a fall roughly doubles those odds.

So do not start with the records. Ask how confident they feel, and notice what they have stopped doing. Then take that to a doctor, because the causes of low confidence — weakness, dizziness, medication, poor vision — are the treatable part.

Frequently Asked Questions

How can I tell if my elderly parent is at risk of falling?

Ask them how confident they feel, and take the answer seriously. In a prospective analysis published in Physical Therapy, the Activities-specific Balance Confidence Scale alone explained 38.7% of the variance in who went on to fall, with fear-of-falling avoidance behaviour adding 5.6% and the Timed Up and Go test 4.9%. The same analysis found that “fall history, presence of pathology, and physical tests did not predict falling.” The authors suggest participants “may have had a better sense of their fall risk than with a test that provides a snapshot of their balance.” Practically: ask about confidence walking around the house, getting out of a low chair, stepping into the shower and managing stairs, and ask what they have stopped doing this year.

My parent has never fallen. Does that mean they are fine?

Not on this evidence, and this is the most counterintuitive finding here. Fall history was entered as a candidate predictor in the Physical Therapy analysis and did not predict who fell. That is a narrow claim about one prospective sample rather than a general rule, and a previous fall still matters for other reasons — in a 12-month cohort it roughly doubled the odds of newly restricting activity (OR 2.22, 95% CI 1.13 to 4.38). But a clean record is weaker reassurance than most families assume. Someone who has never fallen and is quietly avoiding the stairs is showing you the stronger signal.

What does it mean if they have stopped doing things?

It is worth treating as a finding rather than a preference. Avoidance driven by concern about falling was the second-strongest predictor in the model, and a separate 12-month prospective cohort found 10.1% of older adults newly began restricting activity because of those concerns, predicted by frailty (OR 1.58), having a fall (OR 2.22) and poorer functional mobility (OR 1.08). The trap is that withdrawal is easy to misread — giving up the garden or moving to the downstairs bathroom looks like slowing down, not like a symptom. Ask directly what they have given up in the last year and why.

Is the Timed Up and Go test worth doing?

Yes, but as a supporting measure rather than the answer. In the falls-prediction analysis it entered the final model and contributed 4.9% of explained variance — real, and roughly an eighth of what balance confidence contributed. In the activity-restriction cohort it independently predicted new restriction (OR 1.08, 95% CI 1.01 to 1.15). So it carries information. What the evidence argues against is treating a physical test as the decisive check while never asking the person how they feel about their own balance, because in this analysis the questionnaire outperformed the test substantially.

Should we get a medical alert system?

That depends on what you expect it to do. It does not prevent falls, and the evidence on whether people press the button when it matters is sobering — a BMJ cohort of people over 90 found call alarms were widely available but went unused in most of the cases that ended in a long wait on the floor. The stronger argument from the findings on this page is different: if low confidence predicts falling and avoidance follows from it, a device that lets someone keep walking, gardening and showering with the door shut is addressing the predictor rather than only the aftermath. Buy it as something that supports activity, and choose one the person will actually wear.

What should we do first?

Take low confidence to a doctor rather than to a shop. The reasons an older person loses confidence in their balance are frequently treatable: muscle weakness, medications that cause dizziness or drops in blood pressure, poor vision, inner-ear problems, neuropathy, undertreated pain. A falls assessment looks for those. Nothing in this article substitutes for that, and a device installed while an unnoticed medication side effect goes unaddressed is solving the wrong problem. The order that follows from the evidence is: ask, then have it assessed, then decide about equipment.

How reliable are these findings?

Reasonably, with stated limits. Both are prospective — participants were measured first and followed forward, which is stronger than asking people afterwards why they fell. The falls-prediction analysis carries an explicit limitation from its own authors: “Falls were based on participant recall rather than a diary,” and recall for falls is known to be imperfect. It is also a single sample, so the specific percentages should be read as one good study rather than a settled figure. The 12-month activity-restriction cohort reports confidence intervals that are wide but exclude no effect for all three predictors. Treat the direction as well supported and the exact numbers as provisional.

Does this apply to someone in their sixties?

The mechanism plausibly does; the numbers are from older samples and should not be transplanted wholesale. What travels well is the practical point, which is not age-specific: how confident a person feels about their own balance carried more predictive weight than their fall history or a snapshot physical test, and what they have stopped doing is observable evidence of that confidence. A 65-year-old who has quietly stopped using a ladder or walking on uneven ground is telling you something worth following up, whatever the studies’ mean age was.

Sources

• Landers MR, Oscar S, Sasaoka J, Vaughn K. Balance confidence and fear of falling avoidance behavior are most predictive of falling in older adults: prospective analysis. Physical Therapy 2016 Apr;96(4):433-42. PMID 26294679. DOI 10.2522/ptj.20150184. Verbatim: “Using multiple regression, fall history, pathology, and all measures were entered as predictor candidates. Three variables were included in the final model, explaining 49.2% of the variance: Activities-specific Balance Confidence Scale (38.7% of the variance), Fear of Falling Avoidance Behavior Questionnaire (5.6% additional variance), and Timed ‘Up & Go’ Test (4.9% additional variance).” Conclusions verbatim: “Balance confidence was the best predictor of falling, followed by fear of falling avoidance behavior, and the Timed ‘Up & Go’ Test. Fall history, presence of pathology, and physical tests did not predict falling. These findings suggest that participants may have had a better sense of their fall risk than with a test that provides a snapshot of their balance.” Stated limitation, quoted because it belongs next to the figures: “Falls were based on participant recall rather than a diary.” Checked 25 August 2026. https://pubmed.ncbi.nlm.nih.gov/26294679/
• Frailty, falls and poor functional mobility predict new onset of activity restriction due to concerns about falling in older adults: a prospective 12-month cohort study. European Geriatric Medicine 2023 Apr;14(2):345-351. PMID 36739560. PMCID PMC10113287. DOI 10.1007/s41999-023-00749-2. Verbatim: “10.1% of the sample started to restrict activity due to concerns about falling at Timepoint 2. Three key predictors significantly predicted activity restriction group status at 12-month follow-up: greater frailty at Timepoint-1 (Fried Frailty Index; OR = 1.58, 95% CI 1.09-2.30), experiencing a fall between Timepoint-1 and 2 (OR = 2.22, 95% CI 1.13-4.38) and poorer functional mobility at Timepoint-1 (Timed up and Go; OR = 1.08, 95% CI 1.01-1.15).” Authors list no conflicts of interest. Checked 25 August 2026. https://pubmed.ncbi.nlm.nih.gov/36739560/
• Scope note, stated as a limitation rather than a finding. Both studies are prospective, which is the appropriate design for a prediction question, but they describe specific samples of older adults and the percentages above should not be read as population rates. The falls-prediction figures come from a single analysis whose own authors flag recall-based fall recording as a weakness. No product was tested, compared or endorsed in either paper, and neither study evaluated medical alert systems or any other device. Nothing on this page is a substitute for a clinical falls assessment. Checked 25 August 2026. https://pubmed.ncbi.nlm.nih.gov/26294679/