The information elements were organized into several groups, namely intervention parameters, population description, outcomes, and funding sources.
Information on study design, demographics, recruitment methodology, intervention description, outcomes and key results were extracted from the studies by one author and checked by a second.
(2) A narrative description of the study design, aims, participant characteristics, sample size, intervention type, control group description, outcomes measures, and statistical methods.
We also designed several questions to assess methodological rigor of the studies in each of five domains: representativeness, bias/confounding, intervention description, outcome assessment, and analysis.
With regard to studies that did not contain a randomized controlled design, the following aspects were evaluated: definition of study population, selection bias, intervention description, outcome definition, blinded outcome assessments, completeness of the dataset/follow-up, loss to follow, confounders, and a judgment of the validity and applicability of the study.
The data extracted included information on the quality of the study (based on the guidelines from the Centre for Reviews and Dissemination [ 10]) and information on the participants, intervention and control descriptions, outcomes and outcome measures as well as the results (Additional file 1: Appendix 2).
High risk of bias was considered present when no descriptions of patient selection or setting, or no description of outcomes, predictors, or effect modifiers were found in the report.
In addition the longitudinal design of this study permitted an assessment of outcome effect, rather than being limited to a simple description of outcomes.
Limitations of the study included the lack of description of outcomes and the lack of a priori designation of outcomes as primary versus secondary.
For the NK argument to work, then, Malebranche has to restrict the description of outcomes in KP to the highest level of detail.
The setting up of clinical trials registers and the advance publication of detailed protocols with an explicit description of outcomes and analysis plans should help combat these problems.
