The Tinetti test, or POMA, is among the most used in geriatrics, and the article that described it best is titled Babylon in geriatric assessment: across 37 publications, the name, the items, the scoring and the thresholds vary. This page therefore classifies no fall risk. It serves what the literature allows: tracking a balance impairment, with the right noise bound.

A gain of three points after a block of rehabilitation is measurement noise in a given patient, while the same gain averaged over a cohort makes a trial result. The detail is below.

Figures taken from the studies listed at the end of this page, every value carrying its source where it is written. The grid itself is not reproduced.

What the Tinetti measures, and why it does not exist in the singular

The Tinetti test, or POMA for Performance-Oriented Mobility Assessment, assesses mobility in older adults across two subscales, balance and gait, whose sum forms the total. It is scored by watching the patient perform position changes and walking manoeuvres of daily life, without equipment. A high score signals better mobility.

And that is about all that can be said in the singular. The article that best described this instrument is titled Babylon in geriatric assessment. Its authors systematically searched the literature on the Tinetti test and falls, found 37 publications, and observed wide variations bearing simultaneously on the instrument's name, its items, its scoring and its cut-off values. Their conclusion is unambiguous: this heterogeneity interferes with evaluating the test's validity, reliability and generalisability.

The shortest demonstration fits in one line of The Lancet. A Quebec team validates there, in 225 people aged 75 and over, a screening threshold of 36 or less on what it calls the Tinetti balance scale. Yet the only reference in this file whose abstract gives the bounds describes a scale with 28 points maximum. Thirty-six is arithmetically unreachable on twenty-eight. Two leading publications therefore call 'Tinetti scale' two different instruments.

A systematic review quantifies the scale of the phenomenon in today's literature: among the studies retained to assess fall prediction, 58.3% used a modified version of the POMA. In other words, the majority of the literature is not about the standard grid. Before reusing a figure, one must know which version it comes from, and the total out of 28 displayed by the tool on this page is that of the most widespread version, not a constant.

It does not predict falls, and that is not its job

This is the use the test is most often reached for, and the one the literature refuses.

At the classic threshold, below 20 points out of 28, sensitivity falls to 0.45 and specificity to 0.69: the test lets more than one faller in two through. In the same cohort, no difference in mobility or balance is found between subjects with and without a fall history, the area under the curve for the Tinetti being 0.598. The p of 0.762 often quoted alongside it does not bear on that absence of difference: it compares the Tinetti's area with that of the rival instrument tested in the same work, 0.570, and says only that the two are equivalent.

The most rigorous systematic review on this point retains only studies where falls were recorded prospectively, the only way to assess a prediction rather than a recollection. Its conclusion is written out in full: none of these balance tests individually are able to predict future falls. It recommends incorporating them in practice solely to identify and track balance impairment, never as a substitute for a comprehensive falls risk assessment.

A second review, devoted to the predictive accuracy of the POMA alone, concludes something else, not to be confused with the first: it is not possible to determine a specific cut-off point. Published values run from 15 to 26, with sensitivities of 24 to 91 and specificities of 37 to 97, in populations whose fall rate varies from 5 to 61%. A threshold without its population and its version is not a threshold, which is why the tool at the top of this page applies none.

Five points against zero point eight

What remains is the defensible use, tracking a balance impairment. That is where the literature becomes frankly counter-intuitive, and where the most useful figure on this page sits.

In older people in residential care, a change in the total score must reach at least 5 points to be judged reliable at the individual level. At group level, 0.8 point suffices. Same instrument, same data, and two thresholds more than six times apart.

The practical consequence is blunt: a gain of three points after a block of rehabilitation is measurement noise in your patient, while the same gain, averaged over a cohort, would make a publishable trial result. This is not a contradiction, it is the difference between a mean, whose error shrinks with sample size, and a single measurement, whose error does not.

And that individual threshold is not a constant either. After a recent stroke it rises to 6 points, in 55 patients of mean age 75. In bilateral knee osteoarthritis it falls to 0.97 point, in a convenience sample of 25 subjects. From one population to another, this instrument's noise varies by a factor of six. The table below puts each value in front of its own.

No validated French version, but a French grid in circulation

No validated French version of the POMA is indexed in PubMed. The absence was searched for seriously: six search formulations, then nine more built specifically to defeat the claim. None brought up a French validation.

But the contrary signal must be recorded, because it is real and a reader will find it. A 2021 article validating the French translation of another instrument uses as comparator a French Tinetti Assessment Tool, with a correlation of at least 0.83. A Tinetti grid in French therefore does circulate, and serves in published research. Simply, no study of its validation exists.

Three languages do have one: German, Chinese and Korean. The German version is worth pausing on, because it also settles the rights question. Its authors write that no officially authorised German version existed before their 2017 work, and present theirs as the first time authorized German translation. There is therefore a rights holder, and their permission must be sought.

What this means for a French department: the grid in use is a working translation, with no published validation and no known authorisation, and nothing guarantees that it scores the same items as the English version the applied thresholds come from.

Three administration pitfalls

Comparing two scores without checking the version

The name, items, scoring and thresholds vary from one publication to another, and 58.3% of the studies retained in a systematic review used a modified version. A Tinetti score without its version is not comparable to another.

Reading a two or three point improvement as progress

In older people in residential care, at least 5 points are needed to conclude at the individual level, against 0.8 at group level. The threshold rises to 6 points after a recent stroke.

Using it as a fall prediction test

Below 20 points out of 28, sensitivity falls to 0.45. The review limited to prospectively recorded falls concludes that none of these tests, taken individually, predicts future falls, and recommends using them solely to identify and track balance impairment.

Frequently asked questions

What does the Tinetti test measure?

It assesses mobility in older adults across two subscales, balance and gait, scored by observation and added into a total. In the most widespread version, that total runs from 0 to 28, where 28 is the lowest fall risk.

Does the Tinetti test predict falls?

No. The systematic review limited to prospectively recorded falls concludes that none of these tests individually predicts future falls, and at the classic threshold of below 20 points sensitivity is only 0.45.

Which threshold should be used for the Tinetti score?

None transfers. Published cut-off points run from 15 to 26, and the review that gathers them concludes that a specific cut-off point cannot be determined.

How many points make a real improvement?

It depends on the population. In older people in residential care, 5 points are needed at the individual level against 0.8 at group level; 6 points after a recent stroke; 0.97 point in bilateral knee osteoarthritis.

Is there a French version of the Tinetti test?

No validated French version is indexed. A French grid circulates and even serves in published research, but with no validation study and no known authorisation.

References

11 sources, PMIDs included

1. Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 1986;34(2):119-26. PMID 3944402. The original publication. A reading caveat applies straight away: its abstract is not indexed in PubMed. No figure on this page is therefore attributed to it, and everything that follows comes from later work, including work that departs from the original grid.
2. Köpke S, Meyer G. The Tinetti test: Babylon in geriatric assessment. Z Gerontol Geriatr 2006;39(4):288-91. PMID 16900448. The title says it all, and it is its authors'. A systematic search on the Tinetti test and falls finds 37 publications and wide variations bearing simultaneously on the instrument's name, its items, its scoring and its cut-off values. Their conclusion: this heterogeneity interferes with evaluating the test's validity, reliability and generalisability.
3. Faber MJ, Bosscher RJ, van Wieringen PC. Clinimetric properties of the performance-oriented mobility assessment. Phys Ther 2006;86(7):944-54. PMID 16813475. The most complete clinimetric reference in older people in residential care, 245 participants, 78% women, mean age 84.9. It carries the most useful figure on this page: a change in the total score must reach at least 5 points to be reliable at the individual level, against 0.8 point at group level. It also concludes, in 72 participants, that the accuracy of the total in predicting falls is poor, with sensitivities and specificities of 62.5 to 66.1% even at optimal cut-off points.
4. Canbek J, Fulk G, Nof L, Echternach J. Test-retest reliability and construct validity of the Tinetti performance-oriented mobility assessment in people with stroke. J Neurol Phys Ther 2013;37(1):14-9. PMID 23389388. The same question asked after a recent stroke, in 55 patients of mean age 75, starting physiotherapy at eight days on average. Test-retest reliability there is 0.84 and the minimal detectable change rises to 6 points. A reading caveat, against the use one would be tempted to make of it: its authors conclude that reliability and minimal detectable change are similar to those published in long-term care residents. One point between 5 and 6 does not show that noise varies from one population to another; the gaping gap is between the individual and the group level.
5. Parveen H, Noohu MM. Evaluation of psychometric properties of Tinetti performance-oriented mobility assessment scale in subjects with knee osteoarthritis. Hong Kong Physiother J 2017;36:25-32. PMID 30931036. The value that shatters the idea of a constant measurement noise. In 25 subjects with bilateral knee osteoarthritis, a convenience sample, the minimal detectable change of the total falls to 0.97 point, for a standard error of measurement of 0.35 and a reliability of 0.96. Five times less than in older people in residential care, on the same instrument.
6. Knobe M, Giesen M, Plate S, Gradl-Dietsch G, Buecking B, Eschbach D, van Laack W, Pape HC. The Aachen Mobility and Balance Index to measure physiological falls risk: a comparison with the Tinetti POMA Scale. Eur J Trauma Emerg Surg 2016;42(5):537-45. PMID 27287271. The only reference in this file whose abstract explicitly carries the scale's bounds: 28 points max., lowest fall risk. That is where the total out of 28 used by the tool on this page comes from. It also measures the classic threshold: below 20 points, sensitivity falls to 0.45 and specificity to 0.69, and no difference in mobility is found between subjects with and without a fall history.
7. Raîche M, Hébert R, Prince F, Corriveau H. Screening older adults at risk of falling with the Tinetti balance scale. Lancet 2000;356(9234):1001-2. PMID 11041405. The clearest proof that 'the Tinetti scale' names several instruments, and it fits in one line of The Lancet. This Quebec team, in 225 people aged 75 and over followed for a year, retains a threshold of 36 or less, which identifies seven fallers in ten, sensitivity 70%, specificity 52%. But 36 is unreachable on a scale that tops out at 28. The abstract gives no bounds: this is therefore a deduction, but an arithmetic one.
8. Jahantabi-Nejad S, Azad A. Predictive accuracy of performance oriented mobility assessment for falls in older adults: A systematic review. Med J Islam Repub Iran 2019;33:38. PMID 31456962. The quantified translation of the disorder. Of 121 studies identified, 12 retained, with fall rates from 5 to 61%: the cut-off point varies from 15 to 26, sensitivity from 24 to 91 and specificity from 37 to 97. The authors conclude that it is not possible to determine a specific cut-off point, notably because different versions are used.
9. Omaña H, Bezaire K, Brady K, Davies J, Louwagie N, Power S, Santin S, Hunter SW. Functional Reach Test, Single-Leg Stance Test, and Tinetti Performance-Oriented Mobility Assessment for the Prediction of Falls in Older Adults: A Systematic Review. Phys Ther 2021;101(10):pzab173. PMID 34244801. The review that retains only studies with prospectively recorded falls, the only way to assess a prediction. Two results. First, 58.3% of studies used a modified version: the majority of the literature is not about the standard grid. Second, its conclusion, in full: none of these tests individually are able to predict future falls. It recommends using them solely to identify and track balance impairment.
10. Schülein S, Pflugrad L, Petersen H, Lutz M, Volland-Schüssel K, Gaßmann KG. [German translation of the performance-oriented mobility assessment according to Tinetti]. Z Gerontol Geriatr 2017;50(6):498-505. PMID 27312196. The reference that settles the licence question, by an unexpected route. Its authors write that no officially authorised German version existed before their work, and present theirs as the first time authorized German translation of the POMA. There is therefore a rights holder whose permission must be sought to publish a translated version. Their method confirms it: three forward translations, two back-translations by native speakers.
11. Jérôme V, Esfandiar C, Morten Tange K, Amandine L, Harold J, Matteo L, Lindsey H, Jacques H, Thibaut T. Psychometric properties of the Cumulated Ambulation Score French translation. Clin Rehabil 2021;35(6):904-10. PMID 33371731. The only published clue that a French Tinetti grid exists. This article, validating the French translation of another instrument, uses as convergent validity comparator a French Tinetti Assessment Tool, with a correlation of at least 0.83. A French grid therefore circulates and serves in published research, without any validation study of it being indexed. Author names are reproduced as PubMed carries them.