Low dose aspirin for primary prevention-how change in disease risk impacts an intervention's net benefit

This week a huge randomized trial in Japan (14,464 subjects) of low dose aspirin (100 mg enteric coated tablet once a day) in people aged 60-85 years with hypertension, dyslipedemia, or diabetes published findings of no difference in the composite outcome of cardiovascular death, nonfatal stroke, and nonfatal MI.   study link

The study was initially limited by lower mortality from cardiovascular events than predicted based on data from prior studies. This is good news for residents of Japan. But the smaller incidence of cardiovascular deaths, which was discovered in a planned review after the first year of study, meant the study was underpowered to detect important differences between those taking and not taking the daily aspirin. So the researchers increased trial enrollment from the planned 10,000 to 14,000 subjects.

Even so, the trial which was planned for 6.5 years of follow-up was stopped early with a median 5 years of follow-up. Less than half of the predicted number of events had occurred by that time. The independent committee assigned to evaluate study discontinuation determined the study was not likely to be able to detect a difference in the additional short time left. And, they reasoned, continuing the trial could put subjects at unnecessary risk of aspirin related adverse events.

And there were adverse events reported in the study. An increase in serious extra-cranial hemorrhage that required transfusion or hospitalization was reported in the aspirin group - Hazard Ratio of 1.85 (95% CI 1.22-2.81). The aspirin group also had more intracranial hemorrhage (23/7220 compared to 10/7244) and subarachnoid hemorrhage (8/7220 compared to 4/7244).

The authors acknowledge that their study was underpowered and may have found a small benefit if it had continued to the estimated number of events.

This study provides us with an opportunity to examine important factors to consider when continuing an intervention believed to prevent disease in people who do not yet have the disease i.e. primary prevention. Let's look at a couple.

Risk to benefit can change over time

1) When we consider the risk to benefit we must recognize that changes in clinical practice or behavioral habits over time can impact the underlying risk of disease. In this case, the authors report that the prevalence of smoking, a very important risk factor for cardiovascular disease, has decreased in Japan. In addition, they report blood pressure control has improved significantly. It is not surprising, then, that the incidence of cardiovascular deaths has decreased.

2) When the disease we seek to prevent decreases, so does the benefit from the preventive intervention. If the incidence of side effects resulting from the intervention does not change, the risk from the intervention relative to the benefit increases.

Risk to benefit can be different in different populations

1) In this example, the authors report that hemorrhagic stroke is more common in Japanese than in Western populations. If populations are at greater risk of side effect from the intervention, their risk from the intervention relative to the benefit also increases.

This study provides a reminder that we must not only assess but also re-assess, net benefits of all interventions to balance the impact from side effects against the expected benefits and especially for interventions given to people without disease.