New BMJ policy on economic evaluations.

Editor—We, the NHS Economic Evaluation Database Research Team, agree with Smith that economic evaluations should contain comprehensive reporting of both clinical effectiveness and economic analysis and that the BMJ is right to implement this new policy. How the clinical trial results (which inform the economic evaluation) are obtained is often paramount to the understanding and quality of the economic analysis conducted. Research reports are included and abstracted in full on the NHS Economic Evaluation Database (www.york.ac.uk/inst/ crd)—if they explicitly report costs and clinical outcomes for an intervention and at least one comparator. However, to critique the method adopted in the effectiveness study underpinning the economic evaluation appropriately, our template requires information that is often omitted in the report of the economic evaluation. When the parent clinical study has been previously published elsewhere, we obtain the study and use that alongside the economic research when writing the abstract. The abstract on the database then provides information on sample selection, study design, method of analysis, and so on, with the fact that the relevant information is cited from the parent study. Adhering to published guidelines, such as those provided by the BMJ, should produce publications of the highest quality, but authors are still likely to feel the need to be selective in their reporting, given word limits. If authors are required to report more effectiveness data other crucial aspects of the economic evaluation might receive less attention. The focus for BMJ editors should be to ensure that reporting of both important components of economic evaluations receives appropriate attention from the authors. If the policy results in full reporting of both clinical and economic results in one place—for example, two papers in one issue of the journal—this will constitute an improvement. If, however, the new policy results in the combination of clinical and economic results in one short paper, this may be a step backwards.