Age, risk-benefit trade-offs, and the projected effects of evidence-based therapies

Background Physicians underutilize evidence-based therapies in the elderly, perhaps because of concerns about the generalizability of clinical trial results in elderly patients given that the relative efficacy of therapies may vary with age. We compared the estimated effects of age and efficacy of treatment on survival among patients with acute coronary syndromes. Methods Baseline risk, defined as mortality in the year after hospitalization for acute coronary syndromes, was determined for different age strata among 81,584 patients who had been discharged between April 1, 1997, and March 31, 2000, in Ontario, Canada. We calculated the relative efficacy (relative risk reduction) needed to achieve a clinically meaningful absolute survival benefit, using a number needed to treat of 50 patients for the different age strata. We also evaluated risk-benefit trade-offs in the elderly versus the young by modeling different levels of the relative efficacy and rates of fatal complication by age. Results Baseline risk (1-year all-cause mortality) was 12-fold lower in the youngest patients (age <50 years) than in oldest patients (age ≥75 years). Given this gradient, a therapy would have to have a relative efficacy of 88% (i.e., a relative risk of 0.12) in the youngest age group, and 7% (a relative risk of 0.93) in the oldest age group, to generate a number needed to treat 50 patients. For a therapy whose relative efficacy was 25%, the fatal complication rate would have to be sevenfold greater in the oldest compared with the youngest age group to outweigh the survival benefits associated with treatment. Conclusion For acute coronary syndromes, baseline mortality is so much higher for elderly patients that neither sharp reductions in the relative efficacy of therapies nor increases in the rates of serious complications are likely to negate the benefits of therapy. More attention should be paid to overall trial results and less to age-specific subgroup data, unless the latter provide very clear evidence for substantial reductions in absolute efficacy or net harm.