Skip navigation

Calcium and vitamin D for preventing hip fractures

One of the topics covered in the Bandolier conference on osteoporosis was that of supplementing the diets of elderly at-risk populations with calcium and vitamin D. Some health authorities have large populations of elderly nursing home residents who are mobile but fragile, and formulating a policy on this may be important to such an authority.

There is accumulated evidence that positive calcium balance - promoted by adequate calcium and vitamin D intake - is beneficial in the prevention of bone loss in older people. Stronger bones should result, and stronger bones should be more resistant to fracture if older people fall.

The question is whether positive action to ensure adequate dietary calcium results in fewer broken bones. Chapuy and colleagues [1] have undertaken a randomised controlled trial to test that. The Chapuy study has been criticised because it is a selected group, of women only, which might not reflect the whole elderly population (the average age, for instance, was in the mid-80s), but it probably does reflect any similar UK population of elderly but mobile people living in homes.

Randomised trial

The trial took place in France in a selected group of elderly women ranging in age from 69 to 106 years at entry. Randomisation was between 1.2 grams of elemental calcium (as an aqueous suspension of tricalcium phosphate) plus 20 ug vitamin D3 daily at lunch time and matched placebos. Vitamin D3 was included in the trial because dairy produce in France is not fortified with vitamin D.

The inclusion criteria were:-

  • Living in nursing home or apartment house for elderly people.
  • Ambulatory - walking outdoors easily to walking indoors with stick.
  • No serious medical condition.
  • Life expectancy of at least 18 months.

Exclusions were:

  • Taking corticosteroids, thyroxine or anticonvulsants within last year.
  • Treated with fluoride salts for more than three months.
  • Treated with vitamin D or calcium in last six months.


Clinical status and dietary calcium intake was assessed at baseline. They were then re-assessed at 6, 12 and 18 months when they clinical status was recorded, together any information about fractures or adverse events. A further assessment was made 36 months after the trial started [2].

The main outcome measures were hip fractures and all non-vertebral fractures, though a small group of women also had biochemical variables and bone density measured.


Evaluation was made on three bases:-

  • For women treated and followed up for the full 18 (or 36) months.
  • For women who had received treatment for varying lengths of time when they had a fracture, dropped out, or died (active-treatment analysis).
  • For all women (intention to treat), but censoring data by excluding those who died from causes other than fracture.


There was a significantly reduced number of fractures (all fractures and hip fractures) for all methods of analysis at 18 months and at 36 months. Odds ratios are not given here as they are not particularly informative.

1728 women were treated and followed for the full 18 months. Dropout rates were similar in the two groups. Adverse events causing drop-out were mainly gastrointestinal (nausea, diarrhoea or epigastric pain); they occurred in 68 women with a similar distribution between the two groups.

The number-needed-to-treat to prevent one fracture at 18 months and at 36 months is given in the table on page 6 with 95% confidence intervals. For hip fractures, the NNT of 20 means that 20 women have to be treated with calcium and vitamin D for 3 years to prevent a hip fracture in one who would have had a fractured hip had they all not been treated. The NNT to prevent fractures at any non-vertebral site was 14.

NNT values at 36 months were about twice as good as those at 18 months, indicating that continued treatment increases the benefit. Over longer periods, it is likely that the NNT would continue to fall (see [3]).

The data on fractures is buttressed by information from bone density measurements - an average 3% increase in the treated group compared with a 5% decrease in the untreated group. Biochemical measures (reduced PTH and increased vitamin D) also supported improved bone metabolism.

What are the holes in these studies? Few, actually. The data are there for the most conservative of determinants of efficacy. It might be possible to quibble about the inclusion criteria, because a small minority of elderly women will be treated with corticosteroids, thyroxine or anticonvulsants, but not many. No men were studied, probably because their lifetime risk (5%) is about half that of women (12%).


A systematic review of vitamin D supplementation in the Cochrane Library [3] comes to a different conclusion. It calculates the NNT at 36 months as 40 (95% CI 22 - 210). The reason for the difference seems to be that these reviewers have taken an "all patients randomised" view of the denominator for the intention to treat analysis, whereas the in original paper information from women who died from causes other than fracture were censored. Increasing the denominator without increasing the numerator will give larger NNTs and less good effects.

This is an interesting issue. There is a purist line which would support not censoring, but in a population where all cause death rates are high, reducing hip fractures in those still living is clearly the clinical outcome desired. Perhaps the key lesson is that if different interventions were to be compared to prevent hip fracture, then they should be compared on the same basis, whether that was censored or not censored data. This may be particularly important when it comes to cost-effectiveness analysis.

Cost effective?

Depends, especially with the NNT consideration above.

The cost of a broken hip has been estimated at [sterling]5,000 (strictly hospital medical costs [4]), but others have suggested much higher figures. At the Bandolier conference, figures as high as [sterling]17,500 were suggested. Prophylactic treatment of high-risk patients would be worthwhile if the cost of prophylaxis was about the same or less than the cost of treating a fractured hip. The combination used in the Chapuy study was Ostram, costing about 20p a day or [sterling]73 per person for a year.

With a NNT of 20, twenty patients would have to be treated for three years to prevent a hip fracture in one of them, so the cost of prevention would be [sterling]4,400.

With a NNT of 40, but accepting that some people would not use three year's worth of supplement, the cost would be likely to be less than [sterling]8,800 - perhaps [sterling]6,600 in an elderly population.

So it looks as if there is a balance between spending money on supplements or spending money on treating broken hips. But that is just the cost basis. Most elderly people with broken hips have a much reduced quality of life, over and above the trauma and hurt. They rarely get back to the same degree of independent living as they enjoyed before they fell and broke their hip. Where there is an identifiable population at risk, purchasers may want to do their own computations for benefit in their own area.

Who is most at risk?

Of patients admitted with hip fracture, 87% are over the age of 65, and 82% of these are women.

Risk factors for osteoporosis are many, and include :


  1. MC Chapuy, ME Arlot, F Debouef etc. Vitamin D3 and calcium to prevent hip fractures in elderly women. New England Journal of Medicine 1992 327: 1637-42.
  2. MC Chapuy, ME Arlot, PD Delmas, PJ Meunier. Effect of calcium and cholecalciferol treatment for three years on hip fractures in elderly women.
  3. WJ Gillespie, DA Henry, DL O'Connell, J Robertson. Vitamin D and Vitamin D analogues in the prevention of fractures in involutional and post-menopausal osteoporosis. The Cochrane Library 1997, issue 1.
  4. Bandolier 25 , March 1996.

previous or next story in this issue