Items | Effectiveness research | Accuracy research | Issues to consider |
The basics | |||
Question | Participants, Intervention, Comparator and Outcome(s) | Population, Index test and Reference standard | PICO structure does not readily apply in diagnostic research |
Study design | Randomised controlled trial (RCT) Non-randomised controlled studies | Cross-sectional Other designs for index test and reference standard comparison | RCT not required for accuracy evaluations |
Measure of performance | Effect estimate (OR or risk ratio) | Accuracy estimates (sensitivity and specificity) | Accuracy estimates are usually paired; global single accuracy measures are not very intuitive |
GRADE feature | |||
Risk of bias | Tools for the risk of bias typically assess: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting | Methodological quality part QUADAS-2 instrument includes: participant selection, implementation of the index test, reference standard, flow and timing | Various tools exist; there is no consensus on what is best for accuracy research |
Indirectness | Based on PICO question (see above) | QUADAS-2 instrument applicability part can be deployed | Accuracy research requires looking at participants, reference standard, flow and timing |
Inconsistency | I2 or χ² tests for heterogeneity | Visual assessment of overlap of CIs between studies | Assessed separately for sensitivity and specificity in accuracy research |
Imprecision | 95% CIs around an effect measure | 95% CIs around multiple accuracy measures | Assessed separately for sensitivity and specificity in accuracy research |
Publication bias | Test for funnel plot asymmetry | Test for funnel plot asymmetry | Test for funnel asymmetry requires particular caution in accuracy research |
PICO, Population Intervention Comparator Outcome; GRADE, Grading of Recommendations Assessment, Development and Evaluation; QUADAS-2, Quality Assessment of Diagnostic Accuracy Studies.