Internal validity of non-randomised studies

EUNetHTA has opened a public consultation of the draft methodological guideline “Internal validity of non-randomised studies (NRS) on interventions”.  Interestingly, NICE was involved in the review and validation of the guidance.  You can access the guidance and consultation here, or skip to the recommendations below:

1. As the inclusion of non-randomised studies (NRS) in an HTA report requires large efforts (but often fails to increase the validity of the report’s conclusion), the decision to do so should be made only after careful consideration of all advantages and disadvantages.
2. Internal validity (or risk of bias) should be assessed separately from quality of reporting and external validity (or applicability).
3. Assessment of risk of bias (RoB) ideally covers at least 6 different types of bias: bias due to confounding, selection bias, performance bias, detection bias, attrition bias, and reporting bias.
4. At present, ACROBAT-NRSI (A Cochrane Risk of Bias Assessment Tool) should be used for the RoB assessment of NRS.
5. Adequate methodological and clinical knowledge is required for valid and reliable RoB assessment in NRS, because a full understanding of both, bias mechanisms and possible confounders, is necessary. In addition, clear and consistent decision rules should be agreed on to achieve acceptable reproducibility.
6. RoB assessment requires that NRS evidence is first subdivided into cohort and case-control studies. Registry studies usually fall into the category of cohort studies.
7. In HTA reports, RoB assessment of NRS should be described in meticulous detail in order to enable readers to understand the process and the results.

The consultation is open until the 15th May.