Explicitly state the viewpoint and consistently derive all cost information from that viewpoint. Previously 75-82% of studies failed in that task. Two treatment options are always considered: the intervention and the comparator. One needs to define sites of intervention, the target population (demographics, socioeconomic, behavioral and clinical factors), and frequency of the intervention.

Analysis boundaries need to be defined - that is, how far did you go in assessing health effects and costs? For example, in a health program directed at pregnant women, was the effect on the fetus or siblings or father of the fetus tallied as well? The time frame of the analysis should be indicated. (Over what time period were costs and health effects tabulated?) The time period should be long enough to evaluate all significant effects/costs.

Data and Methods Section:

An overview of the disease process being studied and interventions applied should be described. If, as is likely, a model is used to estimate costs and effects of the health intervention, that model should be well described. If clinical trial data are used to generate CE ratios, the specifics of the trial should be well described, so that the adequacy of the clinical studies can be assessed.

For health states defining quality of life, either the entire scale, or a specific reference to a published scale is required. The population (or populations) examined to weight different quality of life and the type of weighting (standard gamble, time trade off techniques) should be defined in the paper as well. For cost data, state the inflation rates and how they were determined. For health care costs, state how utilization was determined and how time costs were valued. Any controversial sources of data should be identified and discussed. The actual discount rates should be noted, and the extent of sensitivity analyses defined.

The Results Section:

With a simulation model, validation should be reported here. The results of the CEA are next. If a viewpoint other than the societal viewpoint is taken, an additional section should be constructed to follow the societal viewpoint. Doing an analysis from the hospital's viewpoint, the most common viewpoint in anesthesia literature, would also require a second, much more complex analysis of the cost to society in order for that CEA to be a useful addition to the CE literature.

A basic set of reference case results includes total costs and effectiveness, incremental costs and effectiveness, and incremental cost- effectiveness ratios with both 3 and 5% discount rates. The QALY benefit should be broken down into longevity versus enhanced quality of life, allowing society to distinguish between life saving effects versus quality of life effects.

Health interventions that affect quality of life primarily can then be balanced against non-health care interventions that similarly affect quality of life. Key sensitivity analyses should follow the results. As noted above, those sensitivity analyses should entail disparate effects on socioeconomic, gender, or age based groups. They should also report how values of important variables were varied.

Discussion Section:

The quantitative results should be expressed in the context of the disease process being studied, and how the results address the concerns that motivated the study. A full discussion of study limitations is considered essential. Comparison to other studies that are relevant are encouraged (e.g. CE of CABG vs. angioplasty vs. smoking cessation vs. diet modification).

How the health effects are distributed - primarily benefiting one population, or providing great benefit to one population but harm to another - must be addressed. Ethical problems relating to the intervention - access to care issues, age/disability issues, and invasion of privacy issues (e.g. in mandatory urine sampling for drugs) - should be discussed.

If a paper is too long for publication based on the requirements set forth here, the suggestion is to provide the journal with a paper of appropriate length and a technical addendum for review. The journal may then decide whether or not to publish the addendum.

Conclusion:

Standard reporting is essential if our society is ever to make good use of CEA's. These guidelines provide a difficult standard to meet, but it is a standard that will give our society valid data on which to base resource allocation.

If you desire a more in-depth discussion of the uses and conduct of cost- effectiveness analyses as decision-making aids in the health and medical fields, see a copy of the book written by the panel: Cost-Effectiveness in Health and Medicine, edited by Marthe R. Gold, Joanna E. Siegel, Louise B. Russell, and Milton C. Weinstein, published by Oxford University Press in 1996.

Read Part 1 of this series.
Read Part 2 of this series.