Quantitative ultrasound and risk factor enquiry as predictors of postmenopausal osteoporosis: comparative study in primary care
BMJ 2003; 326 doi: https://doi.org/10.1136/bmj.326.7401.1250 (Published 05 June 2003) Cite this as: BMJ 2003;326:1250All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Editor – Hodson and March [ref 1] conclude that peripheral ultrasound
enhances the prediction of osteoporosis when added to risk factor
profiles, but closer inspection of the data shows that this approach will
require two out of three women to be referred to DXA while still missing
one in ten cases of osteoporosis. The reasons are the following:
1)
Combining two screening tools will always increase sensitivity unless
their results are completely overlapping. It is unsurprising, therefore,
that combining two referral instruments enabled identification of 90% of
the women with osteoporosis. For the individual patient, the pre-test
probability of a normal DXA was 84%. This could be increased moderately to
95% by the combined use of QUS and risk factor assessment.
2) Using the
combined criteria of low QUS score OR risk factors, two-thirds of women
attending the primary care clinic will need DXA, but almost 80% of those
referred will not have osteoporosis (as the NPV for the combination is
only 22%). This study confirms that risk factors have poor sensitivity and
specificity for osteoporosis, but does not suggest that QUS – despite its
role in assessing overall fracture risk - will be effective in reducing
referrals to DXA. Whilst the World Health Organisation criterion (bone
mineral density T score of -2.5 or less) remains the standard method of
diagnosing osteoporosis, the role of peripheral ultrasound will be limited
[ref 2]. If however, as predicted by some [ref 3], the osteoporosis field
moves towards the expression of an absolute risk of fracture over a
defined time period, such as ten years, then peripheral ultrasound in
conjunction with other risk factors such as previous fracture history,
could develop an important role. Until then, there is no convincing
evidence for the widespread use of peripheral ultrasound technology.
Bo Abrahamsen
senior researcher
Osteoporosis Clinic, Department of Endocrinology,
Odense University Hospital,
DK-5000 Odense C, Denmark
Kim Brixen
consultant
Osteoporosis Clinic, Department of Endocrinology,
Odense University Hospital,
DK-5000 Odense C, Denmark
Tahir Masud
consultant
Department of General Medicine,
Nottingham City Hospital NHS Trust,
Hucknall Road, Nottingham NG5 1PB
1. Hodson J, Marsh J. Quantitative ultrasound and risk factor enquiry
as predictors of postmenopausal osteoporosis: comparative study in primary
care. BMJ 2003; 326: 1250-1
2. Masud T, Francis RM. The increasing use of peripheral bone
densitometry. BMJ, 2000; 32: 396-8.
3. Kanis,JA, Johnell O, Oden A, et al. Ten year probabilities of
osteoporotic fracture according to BMD and diagnostic thresholds.
Osteoporosis Int 2001; 12: 989-95.
Competing interests:
None declared
Competing interests: No competing interests
The current recommendation to primary care physicians is that they
should refer for DXA if any of the RCP guideline criteria are present.
However, there is little evidence from primary care as to whether this
method of selection is effective. The purpose of our study was to
determine how well this method predicts for osteoporosis on DXA BMD (not
osteoporotic fracture) and to compare with ultrasound as an alternative.
Our results showed that risk factor enqiry, as it stands, was a poor
predictor. If , therefore. risk factor enquiry is to be used as a
selection method for DXA the information needs to be used more effectively
and we agree that weighting individual risk factors is sensible. Indeed,
the ideal would be to assess 10 yr fracture probability when determining
overall risk rather than relying on BMD alone. As previously stated, we
wanted to examine what GPs are currently doing and clearly it is not
ideal.
We acknowledge that the frequency of osteoporosis may be higher than
average in our population as many women were referred because of a
perceived risk factor. However,16.3% is similar to the findings of a
previous primary care survey (16%)Versluis RG, Br J Gen Pract 2001;51:806-
810 and of course higher than the incidence of fracture which we did not
eaxamine.
The query raised re number of risk factors is explained as follows:
the number of women with at least one risk factor was 113 but some women
had several RF so that when the frequency of each RF was reported the
total was 169.
We agree that ROC curve for ultrasound at different T cut-offs is
valuable and had in fact prepared this analysis. Unfortunately the
restrictions of a short report did not allow inclusion of this data. We
also agree that studies examining cost effectiveness of ultrsound are
essential and hope these will be forthcomong. Thank you for your interest.
Competing interests:
None declared
Competing interests: No competing interests
Editor,
Hodson and Marsh suggest that ultrasound may be a promising screening or diagnostic test for detection of osteoporosis in primary care.1 Clarification is necessary before their findings can be accepted. The population included in their study, selected on the basis of perceived risk or interest, is likely to be higher risk than would normally be expected in primary care setting. This is confirmed by the high prior probability of 16%.2 Interpretation of diagnostic data is better presented as positive likelihood ratios (LR) with their confidence intervals.3 Given a prior probability of 16%, this form of data presentation allows estimates of post test probability. In this study, ultrasound alone gives a post test probability of less than 50% (Table). The post test probability of osteoporosis is less impressive if a prior of 4% is used (estimated to be the true prior of osteoporosis fracture in primary care).2
They have reported the combined diagnostic value of nine risk factors overall. This is likely to be misleading. When osteoporosis is the target disorder, individual risk factors taken from the history and physical examination have very different diagnostic test properties.4 For example, a past history of steroid use has a positive LR of 12, whilst age greater than 70 has a positive LR of 5.5.2;4 It would have been more informative to calculate the diagnostic value of each separate risk factor and the diagnostic value of combined symptoms and signs in individual patients. Furthermore, the number of risk factors cited in the text (n=169) and in the table (n=113) don't add up. This discrepancy should be clarified.
A receiver operating characteristic curve for different cut-points for ultrasound with their corresponding likelihood ratios could be constructed. A more informative interpretation of ultrasound could then be made as a screening or diagnostic test.3 Finally, the introduction of ultrasound as either a screening or diagnostic test should not be considered until randomised trials have demonstrated effectiveness and cost effectiveness in prevention of fractures and improved quality of life.
Yours sincerely,
Joseph Wong, Medical Student, University of Dundee
Vivien Wong, Medical Student, University of Dundee
Peter Brindle, Wellcome Fellow in Health Services Research, University of Bristol
Alastair Hay, Clinical Lecturer in Primary Health Care, University of Bristol
Tom Fahey, Professor of Primary Care Medicine, University of Dundee
Table
Test accuracy of risk factors, ultrasound examination and both risk factors
and ultrasound. Adapted from Hodson and March.1
Assessment |
Positive likelihood ratio (95% CI) |
Post-test probability of osteoporosis (95% CI) if prior probability 16%1 |
Post-test probability of osteoporosis (95% CI) if prior probability 4%2 |
Risk factors present
|
1.2 (0.9 to 1.5) |
18.6 (14.8 to 23.1) |
4.7 (3.6 to 6.0) |
Ultrasound T score < -1.7 |
4.8 (2.8 to 6.3) |
44.9 (35.1 to 55.1) |
14.8 (10.4 to 20.8) |
Risk factors and Ultrasound T score < -1.7 combined |
1.5 (1.2 to 1.7) |
22.0 (19.3 to 25.1) |
5.7 (4.9 to 6.7) |
References
1. Hodson J,.Marsh J. Quantitative ultrasound and risk factor enquiry as predictors of postmenopausal osteoporosis: comparative study in primary care. BMJ 2003;326:1250.
2. Jarvik JG,.Deyo RA. Diagnostic evaluation of low back pain with empahsis on imaging. Annals of Internal Medicine 2002;137:586-97.
3. Black ER, Bordley D, Tape TG, Panzer RJ (Eds). Diagnostic Strategies for Common Medical Problems. Philadelphia: American College of Physicians, 1999.
4. Deyo RA, Rainville J, Kent DL. What can the history and physical examination tell us about low back pain? JAMA 1992;268:760-5.
Competing interests: �
None declared
Competing interests: No competing interests
We thank Drs Knol and Grove for their agreement with our comclusions.
They rightly point out the absurdity of diluting an efficient test by
diluting it with a poor one.The information provided by ultrasound
assessment is diminished by combining with risk factors and our
presentation of the combination, alongside the individual tests, was
intended to illustrate this point. Ultrasound has roughly the same
sensitivity as risk factor enquiry but halves the number of false positive
diagnoses; this has obvious economic implications. The higher predictive
values of ultrasound are more useful to the clinician in a practical
sense. However, the guidelines for general practitioners currently advise
risk factor enquiry for selection for DXA and the overall assessment of a
patient must include attention to clinical history. Further studies
examining the cost effectiveness of these methods of assessment of risk
for osteoporosis would be useful to help solve this dilemma.
Competing interests:
None declared
Competing interests: No competing interests
I read Hodson and Marsh's paper (1) with interest.
In common with Dr. Knol (2), I'm not sure their conclusion "The
combination of quantitative ultrasound scanning and enquiry about risk
factors ... had slightly better predictive values than risk factors alone"
is supported by their data.
Combining the two measures certainly improves the sensitivity (as one
would expect) but at the cost of considerable specificity. Sensitivity and
specificity are traditionally used to assess tests because they are
unaffected by disease prevelance (casemix) ... but in the clinical
context, positive predictive and negative predictive values are more
important (although often not available).
Imagine yourself as Dr. Hodson's nurse, about to assess patient
number 201. Look at the positive and negative predictive values of their
tests. USS alone (PPV 45% NPV 94%) is clearly much more useful than the
combination of measures (PPV 22% NPV 95%). Using both measures results in
a 23% loss of positive predictive value for a 1% gain in negative
predictive value.
The bottom line is that Dr Hodson will have to decide whether to
refer for DEXA based on these tests. If she relies on USS alone she will
miss 9 cases of osteoporosis and incorrectly refer 27 cases. If she uses
both measures, she will only miss 3 cases ... but at the cost of
incorrectly referring on a further 99 cases. I'm not sure this constitutes
the "effective selection process" referred to in the introduction to the
paper.
(1) Quantitative ultrasound and risk factor enquiry as predictors of
postmenopausal osteoporosis: comparative study in primary care. Hodson J
and Marsh J, BMJ 2003,1250-1
(2) BMJ Rapid responses, http://bmj.com/cgi/eletters/326/7401/1250
Competing interests:
None declared
Competing interests: No competing interests
Hodson and Marsh (BMJ 2003,1250-1) investigated the combination of
risk factor enquiry and ultrasound T-score <_-17 as="as" a="a" better="better" predictor="predictor" for="for" osteoporosis="osteoporosis" than="than" the="the" individual="individual" tests..="tests.." their="their" argument="argument" is="is" that="that" they="they" have="have" correctly="correctly" diagnosed="diagnosed" more="more" patients="patients" with="with" osteoporosis.="osteoporosis." in="in" table="table" fraction="fraction" of="of" diagnosis="diagnosis" under="under" heading="heading" risk="risk" factors="factors" _0.7="_0.7" indexgroup="indexgroup" and="and" _0.6="_0.6" control="control" group.="group." this="this" an="an" insignificant="insignificant" difference.="difference." confirmed="confirmed" by="by" odds="odds" ratio="1.5" ci="ci" _0.7-3.5.="_0.7-3.5." it="it" surprising="surprising" combination="combination" poor="poor" good="good" test="test" give="give" result="result" tests.="tests." authors="authors" use="use" mc="mc" nemars="nemars" comparison.="comparison." comparison="comparison" gold="gold" standard="standard" presented="presented" gives="gives" z="9.4" _="_" ultrasound="ultrasound" t-score="t-score" _-1.7="_-1.7" respectively="respectively" statistic="statistic" test.="test." p="p"/>The absolute difference in P is not impressive, but the best test (with
the lowest Z) is the ultrasound one. The reason is the huge fraction of
approximately 50% of the patients with a negative DXA and positive
"Risk factors" and "Risk factors, ultrasound T-score <_-1.7 _.in="_.in" the="the" ultrasound="ultrasound" test="test" this="this" is="is" considerably="considerably" lower="lower" i.e="i.e" _17.="_17." diagnostic="diagnostic" yield="yield" not="not" only="only" measure="measure" both="both" diagnoses="diagnoses" osteoporosis="osteoporosis" and="and" have="have" to="to" be="be" taken="taken" into="into" account.="account." mc="mc" nemars="nemars" statistic="statistic" evaluates="evaluates" balance="balance" of="of" disagreement="disagreement" between="between" two="two" diagnoses.="diagnoses." risk="risk" factors="factors" t-score="t-score" _-1.7="_-1.7" acceptable.="acceptable." quantitative="quantitative" measurement="measurement" with="with" seems="seems" more="more" appropriate.="appropriate." p="p"/>Competing interests:
None declared
Competing interests: No competing interests
My dear James Walmsley,
At first, I am really surprised about your question on what does it mean
"semeiotic", rather than "biophysical-semeiotic"...
Due to the fact, however, that I cannot explain perfectly my ideas in a
fluent English, may I write my response in Italian language? Many thanks:
1) Lei con la sua, tanto provocatoria quanto desolante, domanda, mi fà
sentire fiero di avere avuto, come antenati, i romani, che discutevano di
logica e di diritto quando altre popolazioni erano interessate
semplicmente alla loro, personale,animalesca sopravvivenza.
2) Il latino "signum" o il greco "semeion", ed il "logos" intorno ad essi
sono noti persino allo "scolaretto" di Gregory Bateson (G.B. Nature and
Mind"): non offro di essi, pertanto, alcuna mia spiegazione, anche perchè
Lei, dotato esclusivamente di humour becero, non ne capirebbe nulla
comunque.
3) La conoscenza della "semeiotica fisica e di quella biofisica" ha
permesso ad un medico generico, con 46 anni di esperienza clinica, di
riconoscere al letto del malato pericolose condizioni patologiche senza
evidente fenomenologia clinica, come il "mio" impending infarction, motivo
per cui ora Le posso rispondere, anche se le speranze di farmi capire da
Lei sono veramente nulle.
Mi permetta, infine,di porLe - con l'occasione - una domanda: "Perchè non
mi ha scritto privatamente?" Forse, io penso, per ragione di humour! Mi
avrebbe, ad ogni modo, evitato di perdere il mio tempo per risponderLe,
senza ottenere da parte sua,alcun risultato.
Cordialmente La saluto
Sergio Stagnaro MD, Member New York Acàdemy of Sciences.
Competing interests:
None declared
Competing interests: No competing interests
Sir,
Sergio Stagnaro's rapid response has given me an opportunity to ask about the word "semeiotic" (sometimes spelt "semiotic").
I would love to be able to use this word to baffle colleagues, but I don't know what it means. At first I thought it was might be a neologism formed out of the words "seminal" and "idiotic" - an adjective referring to a ground-breaking piece of foolishness perhaps. When I asked a colleague what "semiotic" meant, he suggested it referred to half an ear. When I finally realised he was joking, both he and I looked the word up in separate dictionaries... And having read the definitions (which were different) I still don't know what it means.
Can anybody help? I'm afraid it might come up as the Countdown conundrum.
Competing interests:
None declared
Competing interests: No competing interests
Sir,
In order to identify women (and men, of course) at high risk of
osteoporosis we need first of all a bed-side method, appliable rapidly on
very large scale, i.e. a method which relies on the assessment of
osteoporotic constitution as a clinical selection method for referral,
e.g., for dual energy x ray absorptiometry.
A 46-year-long well-established clinical experience allowed me to suggest
such as clinical method as well as an original theory onosteoporosis
pathogenesis, I illustrated in previous papers (1,2). A congenital
functional mitochondrial cytopathology, I called Congenital Acidosic
Enzyme-Metabolic Histangiopathy (CAEMH)(2) represents the "conditio sine
qua non" also of osteoporosis.
Moreover, all patients, affected by derangements in collagen metabolism of
whatever nature as well as location,including perivascular GAG, I visited
in the past 45 years, were or are CAEMH-
positive.Consequently, mitochondrially altered cells, even from the
functional view-point, due to low endocellular free energy, become a-
social elements causing possibly the most common human disorders (2,3).
Interestingly, with the aid of a new physical semeiotics, i.e.Biophysical
Semeiotics, I described
earlier in 150 papers (See: BMJ.com 30 March 2001, Rapid Responses and
Bibliography in my site:
http://digilander.iol.it/semeioticabiofisica), it is easy to recognize at
the bed-side and "quantify" the
osteoporotic risk in CAEMH-positive individuals as well as in osteoporotic
patients. In fact, this
original semeiotics allows doctors to diagnose also osteoporosis since
its early stage, i.e. osteoportotic constitution
(http://digilander.libero.it/semeioticabiofisica/Documenti/Eng/Costituzio...
).
The original semeiotics, ie. Biophysical Semeiotics, useful to the doctor
at the bed-side, is, unfortunately, overlooked until now.
In a few words: in healthy, digital pressure on lumbar or caudal vertebrae
brings about the Gastric Aspecific Reflex after a latency time of 8 sec.
(in the stomach both fundus and body are dilated, while antral-pyloric
region contracts). In addition, the preconditioning
(after 5 sec. exactly doctors applies a second examination) brings about a
favorable latency time: 10
or more sec. On the contary in both osteoporotic constitution and
osteoporetic patient, basal latency time is <8 sec. in relation to
underlying disorders severity. Moreover, the bone preconditioning appears
to be pathological, i.e. shorter than the basal one
(http://digilander.libero.it/semeioticabiofisica/Documenti/Eng/Precondizi...).
1) Stagnaro-Neri M., Stagnaro S., Diagnosi Clinica Precoce
dell’Osteoporosi con la Percussione Ascoltata. (Pub-Med) Clin.Ter. 137, 21
-27 1991.
2) Stagnaro S., Stagnaro-Neri M., Una patologia mitocondriale ignorata: la
Istangiopatia Congenita
Acidosica Enzimo-Metabolica. Gazz. Med. It. Arch. Sci. Med. 149, 67,
1990.
3) Stagnaro-Neri M., Stagnaro S., Auscultatory Percussion Coenzyme Q
deficiency Syndrome. VI Int.Symp., Biomedical and clinical aspects of
Coenzyme Q. Rome, January 22-24, Chairmen K. Folkers, G.L. Littaru, T.
Yamagani, Abs., pg. 105, 1990.
Competing interests:
None declared
Competing interests: No competing interests
Re: QUS:combining two screening tools does not reduce DXA referrals.
The main purpose of our study was to examine the current method of
DXA selection used by general practitioners ie the presence of a risk
factor according to RCP guidelines, and to compare with one type of
peripheral densitometry in view of the increasing popularity of these
investigations in the community. This was a primary care study, basically
to see whether either of these screening tests are of any value when
predicting for osteoporosis. Risk factors proved to be to be a very poor
screening test with many women without osteoporosis being referred for
DXA. As you point out, combining the two screening tests increased DXA
referrals which is not unexpected. It is also unlikely that such
combination, although increasing sensitivity, would detect all cases of
osteoporosis. Unless the definitive test ie DXA is performed for all women
(clearly not possible or recommended)then there will always be some cases
missed. In combination the tests increased negative prediction for
osteoporosis and this could be useful in areas with little or no DXA
service.
Ultrasound alone would reduce the number of DXA referrals for
approximately the same sensitivity but we accept that clinical risk
factors must be included in the overall assessment of risk. This presents
a dilemma as the poor specificity of risk factors will undoubtedly result
in higher cost for DXA scanning.
We entirely agree that assessment for osteoprosis must move forward to a
more useful and comprehensive expression of risk such as ten year fracture
prbability.In the meantime primary care physicians are faced with a list
of risk factors, inadequate DXA service and no clear guidance as how to
improve upon a clearly inefficient selection process.
Competing interests:
None declared
Competing interests: No competing interests