Interpreting a systematic review — branching viva

Opening (candidate)

I would treat this as a structured appraisal of the synthesis and the guideline, not a result handover. First I would confirm the question was focused and preregistered, then read the PRISMA flow for the breadth and reproducibility of the search, then score the review's risk of bias with AMSTAR-2 before reading the pooled effect, then quantify heterogeneity and check the funnel plot, and finally rate the certainty with GRADE and weigh applicability to the child. [1] [2] [10]

Branch A — Method before the result

Examiner: Why must you judge the method before the pooled effect? [2]

Candidate: A pooled estimate inherits the weaknesses of the search, the eligibility rules, and the risk of bias of the included studies. A biased or poorly conducted review produces a precisely wrong pooled estimate, however large or statistically significant it appears. I would score the review with AMSTAR-2, noting its seven critical domains, and read the PRISMA 2020 flow for the breadth of the search before I trusted a single number. [2] [1]

Branch B — Heterogeneity and publication bias

Examiner: The I-squared is 65 percent and the funnel plot is asymmetric. What do these tell you? [4]

Candidate: The I-squared of 65 percent indicates substantial heterogeneity, meaning the included studies disagree beyond chance, so the pooled estimate is fragile and I would look for a clinically sensible subgroup that explains the disagreement before I trusted it. The asymmetric funnel plot suggests publication bias — small negative studies failed to reach publication — so the pooled estimate overstates the true effect. Together they tell me the headline benefit is inflated, and I would rate the certainty down under GRADE. [4] [2]

Branch C — Conveying the benefit

Examiner: The abstract quotes a 50 percent relative pooled effect. How do you convey this to a family? [10]

Candidate: A relative figure quoted alone inflates perceived benefit, so I would give the absolute baseline risk and the absolute risk reduction, the number needed to treat, and the confidence interval. Where the certainty is low and the choice is preference-sensitive, I would present the residual uncertainty honestly and let the family share the decision. [10] [8]

Branch D — The guideline recommendation

Examiner: A guideline cites this review to issue a strong recommendation. How do you judge whether to follow it? [7]

Candidate: I would appraise the guideline with AGREE II across its six domains — scope and purpose, stakeholder involvement, rigour of development, clarity, applicability, and editorial independence — and check it rests on a systematic evidence base and is current. Then I would weigh whether the child sits inside the evidence, and decide whether to follow, adapt, or depart from the recommendation. A strong recommendation built on a biased review with low certainty is not automatically transferable. [7] [8]

Branch E — Applicability to the child

Examiner: Your patient is a preschool child, but the trials are all in adults. Your approach? [8]

Candidate: I would rate the evidence down for indirectness under GRADE, weigh age-specific pharmacology and safety, and treat the adult recommendation as not automatically transferable to a preschool child. I would adapt or depart from it with explicit reasons, share the uncertainty with the family through shared decision-making, and seek specialist input given the sparse direct evidence. [8] [10]

Close

Confirm understanding with teach-back, leave a written summary of the appraised synthesis and its certainty, name the next contact, and document the question, the evidence, its method and certainty rating, the shared decision, and the plan to reassess. [7] [10]