Case-control analysis of bone resorption markers, disability, and hip fracture risk: the Rotterdam study

BMJ 1996; 312 doi: (Published 24 February 1996) Cite this as: BMJ 1996;312:482
  1. Paulus L A van Daele, resident in internal medicinea,
  2. Markus J Seibel, senior investigatorb,
  3. Huibert Burger, resident in epidemiologya,
  4. Albert Hofman, professor of epidemiologya,
  5. Diederick E Grobbee, professor of clinical epidemiologya,
  6. Johannes P T M van Leeuwen, senior investigatorc,
  7. Jan C Birkenhager, professor of internal medicinec,
  8. Huibert A P Pols, senior endocrinologistc
  1. a Department of Epidemiology and Biostatistics, Erasmus University Medical School, Rotterdam, Netherlands
  2. b Department of Endocrinology and Metabolism, Ruprecht-Karls-Universitat Heidelberg, Germany
  3. c Department of Internal Medicine III, Erasmus University Medical School, PO Box 1738, 3000 DR Rotterdam
  1. Correspondence to: Dr Pols.
  • Accepted 23 November 1995

Several factors besides bone mineral mass have been related to the risk of hip fracture. Bone quality, the rate of bone loss, and non-skeletal factors have been identified as important.1 2 High rates of bone resorption may be associated with disruption of the trabecular network as well as with an increased rate of bone loss. Furthermore, immobility associated with disability induces bone resorption not followed by increased bone formation.3

Urinary pyridinium crosslinks are markers of bone resorption. We investigated whether these were associated with the risk of hip fracture and also whether such an association was attributable to disability.

Subjects, methods, and results

This nested case-control analysis was conducted as part of the Rotterdam study, a prospective cohort study of the incidence of and risk factors for chronic disabling diseases.4 Briefly, all 10275 residents of a district of Rotterdam aged 55 or over were invited to participate. The study consisted of an initial home interview followed by a series of medical examinations carried out during two visits to our research centre. The study was approved by the medical ethics committee of Erasmus University, and written informed consent was obtained from all participants. The overall response rate was 78%.

Between January 1990 and February 1994 all 17 women residents with an incident hip fracture after baseline assessments were identified by the computerised, general practice based follow up register. Each case was individually matched to three controls selected at random from women residents in the same one year age group. All fractures were verified radiologically.

At baseline participants brought one overnight urine sample to the research centre. Samples were stored at -20°C until analysis. Analyses were performed without knowledge of the subjects' fracture status. Urine samples were unavailable for five controls. Urinary creatinine excretion was measured by standard laboratory methods. Both total and free urinary pyridinoline and deoxypyridinoline crosslinks were measured by high performance liquid chromatography and corrected for creatinine excretion. Free deoxypyridinoline concentration was also measured by a recent enzyme linked immunosorbent assay (ELISA).

We measured bone mineral density at the femoral neck by dual energy x ray absorptiometry. Disability was assessed with the disability index of the Stanford health assessment questionnaire.5 The lower limb disability index was the mean score (0=no impairment, 3=unable to perform) of six component questions on rising, walking, bending, and getting in and out of a car.

The associations of the levels of urinary (deoxy)-pyridinium crosslinks with incident hip fractures were expressed as relative risks by conditional logistic regression. In an additional analysis we adjusted for disability.

The mean age of the women with hip fractures was 80.5 (SD 8.6) years. Table 1 shows the relative risks for the biochemical markers. One standard deviation increase in total and free pyridinoline concentrations was associated with a substantially higher risk of hip fracture. Total and free deoxypyridinoline concentrations as measured by high performance liquid chromatography were not significantly associated with hip fracture risk. However, a significant relation was observed for free deoxypyridinoline concentration measured by ELISA. A higher disability score was similarly associated with a considerably higher risk of hip fracture. Adjustment for disability resulted in substantially lower odds ratios for all urinary crosslinks. Though cases had lower bone mineral density than controls (0.72 v 0.76 g/cm2 (SD 0.14 g/cm2), relative risk per standard deviation decrease 1.3 (0.6-2.7)), this difference was not statistically significant. Adjustment for bone mineral density did not affect the risk estimates.

Table 1

Relative risks for hip fractures

View this table:


A considerable body of evidence has accumulated over the past few years confirming the validity of pyridinium crosslinks as parameters primarily of bone resorption. Most of the pyridinium crosslinks measured in this study showed a significant association with the risk of hip fracture, though there was a considerable difference in the gradient of risk associated with each marker. This was probably due to the comparatively small sample size and the fact that the various markers measure slightly different aspects of collagen degradation. The association between urinary pyridinium crosslinks and hip fracture risk appeared to be related to disability at baseline. Other investigators have shown that immobility increases bone resorption.3 This study provides evidence that disability and consequent immobility are followed by increased bone resorption, which then leads to increased bone fragility.

We conclude that urinary pyridinium crosslinks can be used in the prediction of hip fractures. Nevertheless, it remains to be seen whether simple measurements of risk factors for hip fracture1 may be more cost effective than the comparatively expensive measurement of resorption markers.

We thank the participants of the Rotterdam study, all the fieldworkers in the Ommoord Research Centre, and the general practitioners who collaborated.


  • Funding The Municipality of Rotterdam, the Netherlands Organisation for Scientific Research (NWO), and Proeventifonds (No 28-2435).

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract