Benefits of pharmacotherapy for preventing hip fracture

Although falls and fractures are common in nursing homes, the guideline on preventing fracture in long-term care in CMAJ emphasizes therapies of unproven benefit and small effect size.1 The authors recognize that the evidence for their recommendations for pharmacotherapy in preventing fractures in long term care is weak; however, even with stronger evidence, the effect size remains very small and possibly not significant, at 2.5% for hip fracture prevention over two years.

The studies used to develop this guideline typically involved community-dwelling patients nearly 20 years younger than the average patient in a nursing home. In our homes in Calgary (Intercare Corporate Group Inc.), the average annual incidence of hip fracture over the past 9 years has been 7.2/100 000 resident days, or approximately 18 hip fractures a year across 700 beds.

For a medium-sized home (200 residents), we could therefore expect 5 hip fractures a year, or 10 hip fractures over 2 years. If, as the authors suggest, we could reduce the fractures by 25 for every 1000 residents treated (number needed to treat = 40) over 2 years, we would need to treat every patient (assuming 100% residents are at high risk and that all survive 2 years after admission) for 2 years to prevent 5 hip fractures in a 200-bed home. However, the median length of stay is 1.9 years in Calgary homes, so 50% of our patients would not survive to see a benefit at 2 years. This, in crude terms, would therefore reduce the number of “prevented hip fractures” by only 2.5.

Such a recommendation is not an attractive or realistic proposition to most patients or providers, especially without any reference to numbers needed to harm. To further complicate matters, we are poor at estimating prognosis in the nursing home setting — only really understanding that a third of our patients will die in any given year. Selecting patients who are likely to survive one year is challenging in those with end-stage chronic disease such as dementia. The authors arbitrarily place a one-year decision point in Figure 1 without any rationale to support it. Also, the broad application of this guideline, as illustrated in Figure 1, makes no reference to the wishes of patients and therefore the appropriateness of the interventions.

More attention perhaps should have been given to targeted high-risk populations in nursing homes who might be expected to experience realistic benefits. We need a better understanding of which patients to treat and what the actual effect size is in the nursing home.

I look forward to the results of future research using complex, frail nursing home patients with multiple comorbidities. It goes without saying that we would dearly love to reduce our rates of hip fracture morbidity and mortality.