From Prosit
Jump to: navigation, search

Study Question

In order to meet the standard outlined in this document the study should concern itself only with relevant and comparable topics. At the beginning of the study it should be explained why the research question was chosen and how it is important to the key potential decision makers. When mentioning the key potential decision makers it should also be made clear which point of view the study takes, e.g. health care system or health insurance, and why this point of view was chosen for the study in order to make it apparent to the reader if the study can be of importance to him or her. Lastly, it should be explained how the costs used for calculation are made up.



Selection of alternatives

When comparing two treatments it should clearly be stated why the alternative was chosen, furthermore the alternative should either be the most widely used treatment or the treatment that provides best cost effectiveness. If current common practice is "doing nothing" then a comparison with placebo should be conducted as it may open up the possibility of comparing the study with other studies that used placebo as comparator.

The chosen alternatives should be described in sufficient detail to provide the reader with enough information to relate the information on costs and outcomes to the alternative courses of action. In addition, the chosen alternative must be designed to help to get as close a measure as possible of the opportunity cost of the new treatment.[1] Decision trees or other decision analytic techniques can help the reader to follow the paths the treatments might take and provide them with information about costs and outcomes associated with each path. Information about probabilities, cost, and outcomes linked to paths of each treatment will help the reader to make their own decisions about each treatment.




Data of one economic study is closely bound to a specific setting. Decision makers need to generalize results or transform them so that it suits their point of view. Transparency of data helps alleviate this process.

Economic evaluations should be executed under conservative assumptions. Report results of sensitivity analysis which help the readers assess the robustness of study conclusions.

Furthermore every conclusion should be exclusively drawn from published data, which enables the reader to understand and verify it.

Report on disaggregated or aggregated data?

At first report components of costs or benefit (direct costs, indirect costs, life years gained, improvement in quality of life) in a disaggregated form. Afterwards combine them in appropriate indices or ratios.

Readers should be able to calculate other ratios which might fit better regarding their respective purpose or point of view.



Reporting about cost data

Estimates of resources used and their prices have to be reported separately to help the reader judge their relevance to his/her setting. Beside of this, the costs of all procedures done to a patient during a clinical trial should be included because these costs could possibly influence outcome.

The dates of both the estimates of resource quantities and prices should be recorded, along with details of any adjustments to a more recent price level. Adjustments to the inflation of preceding years, modifications to show the relevant perspective as well as discounting is to be reported.[2]



Rating Data

If possible only rely on clinical trials which are randomized, double-blinded and controlled. This type of study design is called gold standard and ensures highest internal validity.

In order to judge the quality for publications it might be useful to allocate the Jadad score.

The study protocol of a clinical trial should report the design and result of the study, method of allocation of subjects, effect size with confidence intervals, whether it is an intention-to-treat analysis or an evaluable cohort analysis. If it is a trial with placebo control take into account to mention suitable active comparators.

Overviews should report details about the method of synthesis or meta-analysis, search strategy, criteria for inclusion or exclusion of studies.

At the end of the publication / report provide the reader with information how you assessed the sources used in this publication. For example give the full table of Jadad scores you obained from your sources.



Missing Data

Missing data can be acquired by modelling.

If there is no available source (study, overview) it is justifiable to use other sources like expert opinion. To collect opinions of several experts a delphi panel could be the method of choice. Naturually information gained under such conditions is not as reliable as outcomes from a clinical trial for example.

Maybe not every information was reported in the publication. Contact the author to gain addtional knowledge and information you might need.

There is also the possibility that the clinical trial was too insignificant or maybe the result was not positive or neutral (p-Value > alpha value). Those trials are likely to be never published in journals although their data might be essential for an economic evaluation. This phenomenon is called publication bias. Funnel plots help to assess whether there is publication bias or not.

Regardless which kind of source is used in an economic evaluation it should be considered that any limitations which weaken the assessment of effectiveness (e.g. clinical trials, overview) weaken any economic evaluation.



Information about tests used

The results depends on the Statistical tests used. That means a sample can be interpreted in different ways in wihich all of them look evident. So it is important to jusify the Selection of Statistical Tests. Especially if a approximated test is used. Strengths and Weak Points of the Test should be alluded, also resulting consequences for the sample and the Research Question. Reporting results does not mean reporting the final conclusion. For better understanding it is helpful the present all intermediate results. During this the reader can follow the chain of evidence by himself. confidence intervals around the main variables should be specified[3].

The methods used for sensitivity analysis should be explained. It is not recommended to approximate one way simple sensitivity analysis by multi way simple sensitivity analysis. The influence of every parameter stays clear.[4]



Reporting about Uncertainty

If a sample is used to gain model parameters uncertainty will be a side effect. This kind of uncertainty can be uncovered by confidence intervals or statistical ditributions. [5]To figure out how uncertainty effects the model and the results a sensitivity analysis should be made. Therefore it is nescessary to explain why the sensitivity analysis is used. It is also nescessary to justify the chosen method. Sensitivity analysis can be used for different types of target sizes. Various sensitivity analysis are more useful for a better understanding of the influence of the model parameters than one sensitivity analysis which shows the total influence. Several model parameters can boost others or negate them. So a multi-way sensitivity analysis should be used carefully. A picture is always fine.SenAn.gif[6]

It should be clear whether the precision of the quantitative results or the robustness of the conclusion drawn out of them is more important, as the importance of uncertainty depents on that.




Authors ofter compare their cost effectiveness ratios with other studies using league tables.

Those rankings should be acted cautiously.

  • Firstly differences might arise from different methodological approaches rather than real differences in intervention.
  • Secondly take into consideration how the cost effectiveness or cost-utility ratio of each figure in the league was generated; maybe one intervention was compared to placebo control. Another intervention was compared to current goldstandard in healthcare.

Encapsulating only comparisons with health care interventions should be made which are quite similar in setting and study design.



Report Contents

Health-economic publications and reports follow a different line of argumentation as for instance when reporting a prospective clinical trial. We suggest following structure of a sound HE publication:

Introduction / Background
Methods: Patient characteristics, Cost data set, clinical data etc.
Health-economic study type and study design
Economic perspective
Model used for the analyses
Scenarios investigated
Data sources and data handling (incl. synthesis)
Underlying assumptions (if any)
Results: Medical outcomes, Economic outcomes, combined health-economic outcomes (e.g. ICERs, CLYG), other outcomes
Validation(Sensitivity analyses, extreme value analysis)
Influence of discounting

return short