Letters

Changing perceptions in osteoporosis

Several risk factors are important

Replacing bone mineral density with bone turnover is not a solution

Markers should be used as adjunct to bone densitometry

Author's reply

Several risk factors are important

EDITORWilkin argues for broadening the indication for treatment of osteoporosis to "infirm older people."¹ There are several problems with the specific case he makes but evidence nevertheless to support a move in this direction. He deduces that frequency of impact is the main risk factor for fracture. Falling is indeed a recognised risk factor, but the evidence suggests that it is just one among several predictors.²

Surprisingly, few researchers have attempted to combine their results to produce a risk score. We have identified one study that produced a score with a sensitivity of 70% and specificity of 98% on the basis of three factorsbone mineral density, body sway, and muscle strength.³ This prediction is for fractures occurring within a fairly short follow up time and probably overestimates the potential for effective intervention. Our preliminary results from a similar cohort provide support for a risk score approach over a longer period.⁴

We suggest that there is a case for focusing on elderly people but that the approach should be based on firm evidence and involve a specific risk score rather than a nebulous concept of infirmity. Hormone replacement therapy and bisphosphonates are costly and have important side effects and should not be used indiscriminately in any age group.

C W McGrother, senior lecturer in epidemiology. 
sk29{at}leicester.ac.uk

M M K Donaldson, research associate. 
Department of Epidemiology and Public Health, University of Leicester, Leicester LE1 6TP

^a Competing interests: None declared.

Replacing bone mineral density with bone turnover is not a solution

EDITORLow bone mineral density is a disease surrogate; as Eastell points out in his commentary on Wilkin's article on osteoporosis, it is unfortunate that this particular surrogate has been given the status of a true diagnosis.¹ But Wilkin's suggestion of replacing bone mineral density with another surrogate, bone turnover, is not a solution to the deficiencies of bone mineral density.

An important use for a surrogate is to identify groups of high risk patients so that treatment is directed most economically at those who can benefit most. Evidence suggests that measurements of bone mineral density are rather better at risk stratification than bone markers.² For instance, 36% of hip fractures occur among women with high bone turnover,² who are about a quarter of the older population, while for low bone mineral density about half of hip fractures occur in a similar population fraction.³ Thus Wilkin's suggestion that treatment should be targeted at infirm people with high bone turnover would result in a greater number needed to treat for every fracture prevented than would targeting treatment in the same people on the basis of bone mineral density.

Wilkin's argument turns on the premise that bone marker levels can normalised whereas abnormally low bone mineral density can be increased only slightly. This, however, is to fall into the trap of treating a surrogate rather than the disease. Thus an "infirm" 75 year old woman who has already had Colles and vertebral fractures and whose mother and older sister have both had hip fractures but who has "normal" levels of bone markers should not be treated, because her risk factors cannot be normalised. In contrast, a woman of the same age but whose only risk factor for fracture is high bone turnover, which can be normalised, should be offered treatment even though she is almost certainly at lower risk of hip fracture than the other patient. Such a situation is as bad as treating a woman whose only risk factor for fracture is low bone mineral density but not treating a patient who has multiple other risk factors but who has bone mineral density slightly greater than 1 SD below the premenopausal mean.

The aim of treatment should be not to normalise bone mineral density or markers of bone turnover but to reduce risk of fracture. One way forward might be to use both markers and measurements of bone mass to improve the specificity of treatment and thereby its cost effectiveness.²

David J Torgerson, senior research fellow. 
Centre for Health Economics and Department of Health Studies, University of York, York YO10 5DD

^b Competing interests: Dr Torgerson has been reimbursed by Johnson & Johnson Clinical Diagnostics, which manufactures tests for markers of bone turnover, for attending internal symposiums.

Markers should be used as adjunct to bone densitometry

EDITORWilkin builds a case against using bone densitometry to diagnose osteoporosis by taking out of context a point from Marshall et al's meta-analysis of how well measures of bone mineral density predict the occurrence of osteoporotic fractures. ^{1 2}

Marshall et al indeed concluded that "bone mineral density measurements ... cannot identify individuals who will have a fracture." But of course bone density measurements cannot identify individuals who will have a fracture. Neither do cholesterol concentrations tell who will have a heart attack or blood pressure readings tell who will have a stroke. Although these measures do not predict the future for an individual, they clearly permit identification of individuals at high (and low) risk of complications and allow interventions to be targeted at high risk groups and high risk individuals.

Marshall et al also came to an important conclusion that Wilkin chose to ignore: "[the] ability of decreases in bone mass [to predict fracture] was roughly similar to (or, for hip or spine measurements, better than) that of a 1 SD increase in blood pressure for stroke and better than a 1 SD increase in serum cholesterol concentration for cardiovascular disease."² Because the risk of fracture rises dramatically after the first fracture,³ early identification and treatment are important. We believe that bone density testing is the best way to identify patients at high risk of fracture before the first fracture occurs.

Clearly, there are risk factors for fracture besides bone density. We agree that biochemical markers of bone turnover provide clinically useful information, but we use markers as an adjunct to bone densitometry and fail to see convincing evidence or logic to abandon densitometry in favour of markers.

Wilkin states that "the small increases in bone density gained during the early years of treatment with hormone replacement therapy or bisphosphonate drugs soon level off." One study showed a 10% increase in spinal bone density with treatment with alendronate⁴this is not small by our standards. Our long term study of intermittent cyclical etidronate showed continued gains in spinal bone density in the sixth and seventh years of treatment.⁵ We believe that seeing bone density stable or increasing with treatment is a good indication that treatment is effective.

Wilkin's arguments have several flaws. There are certainly things to be learnt to advance the field, but Wilkin is pointing us in the wrong direction.

Nelson B Watts, director, osteoporosis and bone health programme. 
Emory Clinic, Atlanta, GA 30322, USA nwatts{at}emory.edu

Paul D Miller, director. 
Colorado Center for Bone Research, Lakewood, Colorado USA

^c Competing interests: Dr Watts is the president elect and Dr Miller is the immediate past president of the International Society for Clinical Densitometry. Dr Watts and Dr Miller have both received research support, honoraria for lectures, and consulting fees from several pharmaceutical companies.

Author's reply

EDITORWatts identifies a key issue relating to my article questioning the value of bone densitometry in predicting bone fracture. I quoted from Marshall et al's meta-analysis that "bone densitometry measurements cannot identify individuals who will have a fracture,"¹ to which Watts points out that it nevertheless identifies risk. There are two questions of crucial interest to clinicians: how much of the risk does bone densitometry really identify, and how far can bone density predict risk of fracture 10, 20, and 30 years into the future? These questions are important because bone densitometry is expensive and response in its surrogate measure (bone density) may not reflect the improvement sought (reduction in risk of fracture).

The risk of fracture is low (though not absent) between the ages of 50 and 70, so that the case for undertaking bone densitometry in this age group in order to prescribe an antiresorptive drug with the promise of no more than halving an already minimal risk is inherently weak. Fractures first become common after the age of 65-70, but in one of the few longer term studies with sufficient numbers Gandsell et al found only poor prediction of fracture from bone densitometry, which was poorer still in the older age group.² De Laet et al's observations in 5814 subjects indicate that, whereas the risk of hip fracture increases 13-fold from ages 60 to 80, the changes in bone density associated with age contribute at most 1.9.³ They conclude that "the contribution of decline in bone density to the exponential increase in risk of fracture with age is relatively small."

According to De Laet et al's study, some 85% of the age related risk of fracture in women is related to factors other than bone density, but they are not accounted for in epidemiological studies such as that of Cummings et al.⁴ In that study, bone densities were corrected for age by Z score but other age related factors that contribute to risk of fracture were not. Here lies the flaw: the age related (and possibly remediable) factors that mostly account for risk of fracture were ignored and their contribution ascribed erroneously to bone density. The fundamentally important observation of Hui et al, that the same bone density predicts an eight-fold difference in risk of fracture for people of 45 compared with those of 80 makes bone densitometry an uncertain tool.⁵ Osteoporotic fractures occur mostly in elderly people, and most particularly in those who are infirm. Bone density adds little.

Terence J Wilkin, professor of medicine. 
Plymouth Postgraduate Medical School, University Medicine, Derriford Hospital, Plymouth PL6 8DH terry.wilkin{at}phnt.swest.nhs.uk

^d Competing interests: None declared.