*Geek Box: Reviews
Reviews are common in research, and have the potential to be valuable to both peers in a given field as well as practitioners, by providing a synthesis of the current evidence and relevant citations. There are generally two types of review which are most common: narrative reviews and systematic reviews.
A narrative review is, in effect, a review which discusses the scientific literature, but in narrative form these reviews may be highly prone to investigator bias, cherry-picking of studies, misreporting of findings [i.e., downplaying certain results and overstating other results], and often lack a direction as the specific research question being asked. Narrative reviews may be done well, but they are certainly prone to the aforementioned issues.
A systematic review attempts to use objective and transparent methods to qualitatively synthesise evidence in relation to a specific research question. Systematic reviews should use predefined criteria to select, analyse, and synthesise available research. The analysis should include an assessment of the validity of the included studies, i.e., by using a study assessment tool, a popular example of which is the Grading of Recommendations, Assessment, Development, and Evaluation [GRADE] framework, although this has limitations for nutrition.
Many of these assessment tools were designed to assess the effectiveness of drug treatments for disease, and do not always translate neatly to food/diet/nutrients as the exposure of interest. A good systematic review should also rate the overall quality and confidence of the body of evidence, to provide conclusions in relation to the potential health effect of the exposure of interest.