How Reliable is Cost-Effectiveness Analysis?
Though the basic cost-effectiveness calculation appears to be simple, choices about units of measurement, definitions of interventions, scope of costs, and prices to be included not only will alter the numerical results but also will affect the interpretation of the cost-effectiveness ratio. In many cases the differences are so large that refining the underlying analyses is unnecessary. For instance, no amount of refinement will make coronary artery bypass grafting (>US$25,000 per DALY averted) more cost-effective than using new antimalarial drugs where resistance to older ones has developed (US$8 to US$20 per DALY averted) or taxing tobacco products (US$3 to US$50 per DALY averted) (table 3.1). For this reason, readers of DCP2 are encouraged to pay attention to different orders of magnitude, distinguishing extremely or moderately cost-effective interventions from those interventions that are not cost-effective.
[Table .]
When cost-effectiveness ratios are within a similar range, policy decisions become more difficult. In such situations, closer scrutiny of the cost-effectiveness ratios may be warranted to improve confidence that the measures are close. This would entail verifying whether the units of measurement, the definition of interventions, and the scope of costs that are included were similar.
Note also that the quality of the evidence available to assess cost-effectiveness varies, especially given the wide range of interventions being looked at. DCP2 notes that the best evidence comes from studies with randomized controls or systematic overviews and that the next best available evidence comes from nonrandomized studies that were nevertheless able to use rigorous statistical methods. The weakest evidence comes from limited case studies or surveys of expert opinion. However, a lack of evidence does not mean that an intervention is not cost-effective. It simply means that researchers do not know how cost-effective the intervention is. Nor does it mean that readers should ignore the cost-effectiveness numbers. Rather, readers should be cautious, should not rely heavily on point estimates, and should pay attention to orders of magnitude and quality of evidence.
