PubMed Central, : RMMJ Rambam Maimonides Medical Journal Rambam Health Care Campus 2026; 17(4): e0024. ISSN: 2076-9172Published online 2026 October 26. doi: 10.5041/RMMJ.10585

Table 2.

Risk-of-Bias, Certainty, and Decision-Relevance Appraisal by Evidence Group.

Evidence Group	Dominant Bias Concerns	Risk-of-bias Judgment	Certainty of Evidence (GRADE)^*
RCT post hoc / pooled analyses	Psychiatric exclusion criteria; post hoc harms; rare-event underpowering	Some concerns (RoB 2)	Moderate
RCT meta-analyses	Rare-event ascertainment; heterogeneous harms definitions	Not re-appraised with RoB 2; risk of bias as judged within each source synthesis	Moderate
Active-comparator cohort / target-trial emulations	Residual confounding; outcome misclassification; treatment channeling	Moderate (ROBINS-I)	Moderate
Broad EHR / passive-comparator studies	Comparator imbalance; surveillance bias; obesity-severity confounding	Serious to critical (ROBINS-I)	Low
Pharmacovigilance / disproportionality	Notoriety bias; no denominator data; cannot estimate incidence	Not formally scored (qualitative appraisal only)	Very low
Regulatory reviews	Dependence on quality of underlying data; integration of multiple sources	Not applicable (regulatory evidence synthesis)	Not applicable ^†

* Certainty ratings in this column are GRADE assessments made at the level of each evidence group; outcome-specific GRADE ratings are given in Table 3.

† Regulatory reviews are not assigned a GRADE certainty rating; they are instead of high decision relevance, because they integrate multiple evidence streams (randomized controlled trials, large electronic health record/Sentinel cohorts, and spontaneous reports) under a formal regulatory risk–benefit framework and directly inform prescribing and labeling decisions. Per-study domain-level RoB 2/ROBINS-I judgments are reported in Supplement B.

EHR, electronic health record; GRADE, Grading of Recommendations Assessment, Development and Evaluation; RoB 2, Cochrane Risk of Bias 2; ROBINS-I, Risk of Bias in Non-randomised Studies of Interventions.