Skip navigation

Attitudes to risk in OA

Clinical bottom line

Two thirds of patients with osteoarthritis were willing toaccept increased risk of heart attack, stroke, or stomach bleeding for a 2/10 point reduction in pain. half would accept risk at the level of 1 in 50.


People in this study were patients with osteoarthritis of the hip and knee aged between 45 and 74 years, and free of other conditions causing chronic pain. They were asked to consider two treatment options which were described in terms of:

The options differed only in the level of pain, namely a pain level of 5 for one option and 3 for the other. Patients chose the second, since it produced lower pain relief. They were then given a series of choices where the risk levels were changed, and where the minimum risk level to chose lower pain over increased risk could be measured.


In a detailed if somewhat complicated set of results, the most obviously interesting was that most of 196 patients were willing to accept some additional risk even for a 2/10 point reduction in pain, and that increasing the pain reduction to 5/10 points made little difference to this choice, as Figure 1 shows.

Figure 1: Percent of patients willing to accept additional risks of heart attack or stroke (MI) or stomach bleed (PUB) for 2 point and 5 point pain reduction

half of patients were willing to accept maximal acceptable risk increments of 2% or more. The skewed nature of the results suggest the possibility that not all of them fully understood the nature of all the questions, but a 2% absolute risk increase is equivalent to a risk of 1 in 50. Of course, half were not willing to accept this degree of risk, and there was probably greater willingness to accept a risk of a stomach bleed than a heart attack or stroke.


A couple of points are worth considering. First, patients are likely to respond to the information provided to them. The information provided on heart attack and stroke was fair in that it gave a sensible assessment of death of up to 1 in 5. Stroke risk is not increased with coxibs or NSAIDs, and most events are non-fatal heart attacks. The information on stomach bleeds was a little high at 2%, with complicated events occurring at more like 1 per 100 person years. Moreover, the information claimed that a small proportion may die, and there is a body of evidence suggesting that the risk of death may be as high as 1 in 10, though not everyone agrees with that.

In any event, the fact that two-thirds of patients find the risks presented acceptable is interesting, because the risk they would be prepared to accept is far above that measured in patients with arthritis even with the drugs most implicated. With most, the risk is far less than this.

Most interesting is that this is an example of how important patients in pain rate pain relief. After all, the were not presented with complete relief but a 2 point reduction on a scale of 10.


  1. CG Richardson et al. Pain relief in osteoarthritis: patients' willingness to risk medication-induced gastrointestinal, cardiovascular, and cerebrovascular complications. Journal of Rheumatology 2007 34:6.