Esp 9.4.2 Summary of findings table

Systematic reviews of diagnostic studies should be accompanied by a summary of findings table, which should include the question being investigated, the index test, the reference test, the population, the estimates rate of true positives, false negatives, true negatives and false positives and the absolute difference between the index and reference tests for these values per 1000 patients,  the sample size as well as the number of studies which contributed to the sample, the GRADE (Grading of Recommendations Assessment, Development and Evaluation) quality of evidence for each finding, and any comments (including decisions as to why the reviewers assigned the final GRADE ranking) (Schünemann et al. 2013). These Summary of Findings tables can be created using the software program Guideline Development Tool (GDT, http://www.guidelinedevelopment.org/) and should appear in the Executive Summary section in JBI systematic reviews, following Implications for Research.

To determine a GRADE ranking of the evidence, the GRADE approach begins by assigning a starting level of quality to findings. For studies of diagnostic test accuracy, cross-sectional or cohort studies are considered to provide ‘high quality’ evidence, whereas for other quantitative studies they are ‘low’. There are two other levels in the GRADE systems, with four levels in total. These are high, moderate, low, and very low (Schünemann et al. 2013; Gopalakrishna et al. 2014; Atkins et al. 2004).

Different factors are then considered that lead to downgrading the GRADE ranking. These are: Risk of bias (as determined by critical appraisal; -1 if serious risk of bias, -2 if very serious risk of bias), Inconsistency or heterogeneity of evidence (-1 if serious inconsistency, -2 if very serious inconsistency), Indirectness of evidence (-1 if serious, -2 if very serious), Imprecision of results (-1 if wide confidence interval, -2 if very wide confidence interval) and Publication bias (-1 if likely, -2 if very likely) (Schünemann et al. 2013; Gopalakrishna et al. 2014; Atkins et al. 2004).

For other review types there are factors which can increase the GRADE quality of evidence (i.e. large magnitude of effect, dose response, all plausible confounding factors would reduce the demonstrated effect, or create a spurious effect where results suggest no effect). However, no such factors have been endorsed for studies of diagnostic test accuracy. For further guidance on the GRADE approach visit the GRADE working group website.

Table 9.6: Summary of Findings template

JBI endorses GDT for the development of Summary of Findings tables. All Summary of Findings tables created for JBI DTA reviews must use the GDT software.

When developing a Summary of Findings table within GDT, there are different format options for exporting the table. JBI reviews must use the Summary of Findings table (layer one) option (Table 9.6).