Changes in body weight and incidence of hip fracture among middle aged Norwegians

BMJ 1995; 311 doi: (Published 08 July 1995) Cite this as: BMJ 1995;311:91
  1. Haakon E Meyer, research fellowa,
  2. Aage Tverdal, research directora,
  3. Jan A Falch, consultantb
  1. aNational Health Screening Service, PO Box 8155, Dep N-0033 Oslo, Norway
  2. bDepartment of Internal Medicine, Aker Hospital, N-0514 Oslo, Norway
  1. Correspondence to: Dr Meyer.

    Lean body stature is an important risk factor for hip fracture.1 We assessed prospectively the relation between intrapersonal change in body weight and the incidence of hip fracture.

    Subjects, methods, and results

    We followed up 21 510 women and 21 157 men born between 1925 and 1940 attending both the first (1974-8) and the similar second (1977-83) cardiovascular screening in three Norwegian counties (85.2% of all invited)2 on average 11.3 (range 0.01-13.8) years after the second screening to study the incidence of hip fracture. We identified hip fractures (cervical or trochanteric) as described elsewhere1 at all hospitals in the three counties. We calculated the observation time for each person from the second screening to hip fracture, emigration, death, or end of follow up (in that order). We matched the file to the cancer registry of Norway, which has information on all diagnosed cancers in Norway, and to the register of death and emigration form “Statistics Norway.” Adjustment was made for potential confounders as described in a previous study of this cohort.1

    During follow up we identified 227 hip fractures, excluding fractures associated with high energy traumas and metastatic bone disease. The mean age at fracture was 57.2 (range 46.7-65.9) years in women and 55.5 (42.9-65.0) years in men.

    The mean weight in the total cohort increased by 1.3 (SD 4.3) kg between the first and second screening. The women losing more than 3 kg (1 SD from mean change) or gaining >/=5.6 kg (1 SD from mean change) had a distinctly higher risk of hip fracture (table). The same pattern, although not significant for those gaining >/=5.6 kg, was found in the men. Excluding all the subjects in whom cancer had ever been diagnosed and all those who died during follow up gave only minor changes in the relative risks. The same applied to additional adjustment for changes in physical activity and smoking habits between the first and second screening. If the whole study population had been exposed to the age adjusted rates of those gaining only 1.3-5.5 kg in weight then the incidence of hip fracture would have been reduced by 35% in the women and 26% in the men.


    We found that both weight loss and excess weight gain, calculated from standardised weight measurements at two screenings of the same population, were strong predictors of hip fracture. This was in addition to body mass index, which is also a strong predictor of fracture.1

    A relation between weight loss and hip fracture has previously been shown, and weight and bone loss is also associated with bone loss.4 The raised risk of fracture in the subjects gaining >/=5.6 kg was unexpected. The mean weight in this group, however, was lower at the first screening and higher at the second compared with the subjects losing more than 3 kg. These people may be more prone to repeated weight changes (including weight loss) than those whose weight is stable, and a resulting fluctuation in bone mass may produce permanent micro-architectural damage. A relation--independent of obesity--between coronary heart disease, total mortality, and fluctuations in body weight has also been reported,5 suggesting that among those with great fluctuations in weight different adverse health outcomes are overrepresented.

    Rates and relative risks of hip fracture by change in body weight between first and second screening* in 42 600 middle aged women and men in three Norwegian counties

    View this table:

    Adjustment for self reported physical activity did not have any substantial impact on the estimates of weight changes. Increase in weight was related, however, to a decrease in physical activity from first to second screening,2 and thus may indicate low levels of physical activity during follow up.

    People with chronic diseases might be more exposed to dramatic weight changes than more healthy people. Additional analyses that took account of this as far as our data permitted, however, did not substantially influence the strong association between weight change and hip fracture found in this cohort.


    We thank Randi Selmer for constructive comments about the analysis of data.


    • Funding None.

    • Conflict of interest None.


    View Abstract