Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

NSAIDs and fracture risk


Laboratory exploration of the effect of NSAIDs on bone metabolism has demonstrated that bone resorption can be affected through prostaglandin inhibition. One implication is that NSAIDs potentially could reduce bone loss and hence fracture risk. A huge observational study using the UK general practice research database (GPRD) [1] tells us that this hope will not be realised.


A large study of aspirin and NSAID use on bone mineral density in 7,768 white women older than 65 years in the USA [2] concluded that bone mineral density was higher in users of these drugs. Risk of fracture was unaffected. The study had the benefit of being large, but aspirin and NSAID users were different from non-users. Osteoarthritis, rheumatoid arthritis, back pain and other conditions were much more common in NSAID users than nonusers. Adjustment of results for potential confounding can be difficult in circumstances where subjects and controls differ markedly.

NSAID use did not affect the rate of excretion of markers of bone resorption [3], but bone mineral density at some sites was again found to be affected by proprionic acid NSAIDs in 84 older women [4].

The background evidence that NSAIDs reduce fracture risk was thin, but the apparent effects on bone density meant that some reduction in fracture risk might be expected. A very large study would be needed to show this.


The large retrospective cohort study was conducted using the GPRD [1]. It looked at fracture risk in people using NSAIDs and compared that with people who did not use NSAIDs.

NSAID users fulfilling one or more prescriptions for an NSAID from 1987 up to end 1997 were included, and divided arbitrarily into those receiving three or more prescriptions and those receiving one or two prescriptions. A control group of people never having a prescription for NSAIDs was created by matching for sex, age, and practice (where possible). Systemic corticosteroid use was an exclusion criterion for users and controls. Information on about a dozen possible confounding conditions, and about a dozen possible confounding drug treatments was collected for each case and control.

After a prescription was filled follow up was until fracture or 91 days after the last prescription. Nonvertebral fractures were assessed by ICD codes and vertebral by radiography.


NSAIDs were prescribed for 501,000 patients, with 215,000 having three or more prescriptions for a median 3.4 years (regular users), and 287,000 having one or two prescriptions for a mean of 0.7 years (incidental users). There were 215,000 controls. Back pain and rheumatoid arthritis were more common in NSAID users than in controls. Incidental users were about 10 years younger than the mean age of 54 years for regular users and controls.

Nonvertebral fractures occurred more frequently in older women and the oldest men (Figure 1). For women fracture rates rose substantially after age 64.

Figure 1: Annual incidence of nonvertebral fractures in men and women by age in 215,000 UK GP patients not prescribed NSAIDs

Fracture rates with regular NSAID users were certainly not lower than controls. If anything, they were somewhat higher (Table 1) for vertebral and all nonvertebral fractures. Regular users had fracture rates no different from incidental users. No NSAID was associated with higher or lower rates of fracture. Restricting analysis to patients with a history of arthopathy reduced the difference between regular users and controls for nonvertebral fractures, with a relative risk of 1.2 (1.1 to 1.3).

Table 1: Occurrence and relative risk of regular NSAID users (215,000) and non-NSAID controls (215,000) in UK general practice

Regular NSAID user Control
Fracture type Number of fractures Rate per 100 years Number of fractures Rate per 100 years Relative risk
Vertebral 808 0.1 192 0.03 2.9 (2.5 to 3.4)
All nonvertebral 10505 1.5 5793 1 1.5 (1.4 to 1.5)
Forearm 2516 0.3 1556 0.3 1.3 (1.2 to 1.4)
Hip 973 0.1 686 0.1 1.1 (0.98 to 1.2)


This beautiful study tells us that there is no major effect of NSAIDs on risk of fracture. It also shows the problems with confounding. While many confounding factors could be taken into account, others, like diet, exercise or bone density could not. To properly take account of confounding factors you have to know what they are, and how much to adjust for them.

The impact of NSAIDs on bone metabolism, on fractures, and bone healing has become a hot topic. Bandolier has reviewed all the evidence available. This can be seen here on the Bandolier Internet site, either to be read on-screen, or downloaded as a printable document ( Bandolier Extra ).


  1. TP Van Staa et al. Use of nonsteroidal anti-inflammatory drugs and risk of fractures. Bone 2000 27: 563-568.
  2. DC Bauer et al. Aspirin and NSAID use in older women: effect on bone mineral density and fracture risk. Journal of Bone and Mineral Research 1996 11: 29-35.
  3. NE Lane et al. Aspirin and nonsteroidal antiinflammatory drug use in elderly women: effects on a marker of bone resorption. Journal of Rheumatology 1997 24:1132-1136.
  4. DJ Morton et al. Nonsteroidal anti-inflammatory drugs and bone mineral density in older women: the Rancho Bernado study. Journal of Bone and Mineral Research 1998 12: 1924-1931.
previous or next story in this issue