How valid is it to conclude that long-term heparin use is not associated with osteoporosis?


David M. Reid, Department of Medicine and Therapeutics, University of Aberdeen, Polwarth Building, Foresterhill, Aberdeen AB25 2ZD, UK.
Tel.: +44 1224 554761; fax: +44 1224 551154; e-mail:

For many years there has been controversy as to the effect of long-term unfractionated heparin therapy used in pregnancy on bone mineral density (BMD) and hence its potential as a risk factor for future osteoporotic fracture. Indeed, previous studies have suggested an excess fracture risk during pregnancy and a reduction in BMD. The paper by Rodger et al., published in this issue of the Journal [1], addresses the additional question of whether low-molecular-weight heparins (LMWHs), in this case dalteparin, are associated with loss of BMD. The authors conclude that there is no such loss; a conclusion that on the face of it is reassuring for those women who are required to take this treatment during pregnancy. While the results of the study are apparently robust in that they are reported as part of a randomized controlled trial, they do require to be considered with some caution due to some of the vagaries of assessing osteoporosis risk.

Osteoporosis is essentially a risk factor for fracture and some would argue not a disease entity in its own right [2]. However, fractures related to osteoporosis occur in one in every two women over the age of 50 and in one in every five men [3], and cost the UK Exchequer around £1.7 billion [4]. A consensus conference organized by the National Institutes of Health in the USA defined osteoporosis in 2001 as a ‘skeletal disorder characterized by compromised bone strength predisposing a person to an increased risk of fracture’ [5]. The statement further explained that bone strength reflects the integration of bone density and bone quality. BMD can be measured in vivo, most commonly by dual-energy X-ray absorptiometry (DXA) [6], but there is no clear agreement on how to assess bone quality, which cannot be assessed in vivo.

It is recognized that BMD assessment measured by DXA is the best method of predicting future risk of fractures, especially at the site of any future fracture [7]. It is now recognized that risk of future fracture is best assessed at the BMD site where the fracture will occur. Hence, hip fractures are best predicted by a measurement of BMD at the hip, most often now the total hip site, which includes the neck of the femur and the greater trochanter. This site is recognized as the most reproducible area at the proximal femur. It is pleasing therefore that Rodger et al. [1] chose to use this site when they reported the measurements at the proximal femur. The study by Rodger et al. meets two other important criteria that are often ignored by investigators unfamiliar with this methodology. The two major manufacturers of DXA equipment, GELunar and Hologic, although expressing BMD in g cm−2, have a systematic difference in their methods of comparison with a standard and hence have an absolute difference in the values between 10% and 15%, with GELunar devices returning the higher values. However, methods have been devised to cross-standardize between devices [8], and although over 90% of the participants in the study by Rodger et al. were measured on Lunar devices, the use of standardization equations excluded an erroneous result due to inappropriate cross-calibration. On the face of it, the difference in scanner characteristics could also have been averted by expressing the results of the scans as T-scores that related to the number of standard deviations different from the young normal mean, with osteoporosis being diagnosed according to the World Health Organization definition of a T-score ≤2.5 [9]. Rogers et al. demonstrated no difference in T-scores at the end of pregnancy between those women who received dalteparin and those who did not. However, the T-score results are only as good as the normative range used by the scanner manufacturers. Rodger et al. [1] make no comment on this although, as the study was presumably completed relatively recently, the omission may not be too serious. In 1996, Faulkner et al. [10] reported that Hologic normative ranges significantly over-reported osteoporosis at the hip sites in comparison with Lunar scanners. However, the manufacturers subsequently agreed to use a single normative range derived from the NHANES III database [11]. Regretfully, no such standard normative range exists for spine BMD and so the manufacturers use their own. This potential problem is unlikely to be a factor in the study by Rodger et al., not only because of the timing of the study but also because there were no differences in absolute standardized BMD. In addition, the vast majority of subjects were measured on one type of scanner.

The second pleasing feature of the study is the allowance made for apparent differences in body mass index (BMI) between the two otherwise fairly well-matched groups of patients. BMD is affected by a large number of factors including age, weight and body size, albeit the last factor is part-corrected by expressing the results as areal BMD, i.e. bone mineral content corrected for the area of bone assessed. However, such a correction does not always negate a relationship between BMI and BMD, as Rodger et al. showed in their study [1]. When comparing absolute BMD between populations it is therefore always important to allow for differences in weight and BMI between subject groups, as was apparently done in this study.

How might we summarize the reassuring results of this study? The strengths are that it was carried out within the confines of a randomized controlled trial, that there were few apparent differences between the populations studied and those that were seen were considered in the analysis, and that any differences in the measurement tools used were standardized. The weakness of course is that there was no longitudinal element to the study. Women were simply assessed at 6-weeks post-partum, hence the need to carefully allow in the analysis for differences in the populations studied. That having been done, it is probably safe to conclude that LMWHs, or at least dalteparin, produce no major deficiency in BMD in those requiring such treatment. There may be no need to include long-term heparin therapy in the indications for BMD scanning, which has been advocated by some authorities [12] but not in the more recent review by Poole and Compston [13].

Disclosure of Conflict of Interests

The author states that he has no conflict of interest.