Situation: Staff nurses on an adult surgical ward want to know whether chewing gum after bowel surgery speeds the return of bowel function. They plan to search the literature and present the evidence to the ward's practice council. The search yields four sources on the question. Which source gives the STRONGEST level of evidence for changing practice?

In the hierarchy of evidence for interventions, systematic reviews and meta-analyses of randomized controlled trials rank highest, then randomized controlled trials, then cohort and case-control studies, with expert opinion lowest. A systematic review ranks at the top only when it pools randomized trials; a review of cohort studies remains observational evidence and ranks below a well-conducted randomized trial. A large sample improves precision but does not change the level of a design.

Evidence hierarchy for intervention questions
When a clinical question asks whether an intervention—such as chewing gum after bowel surgery—changes an outcome, the strength of evidence depends primarily on the study design, not on sample size or the reputation of the authors. For intervention questions, randomized controlled trials and systematic reviews of randomized trials sit above observational designs and expert opinion.

The four options differ in a way that is deliberately easy to misread. A systematic review sounds impressive, but its position in the hierarchy depends on what it pools. A systematic review of cohort studies is still observational evidence; it cannot produce a stronger causal inference than the primary studies it contains. Option 1 reviews 12 prospective cohort studies. Cohort studies follow groups over time but do not randomly assign the intervention, so confounding by indication, baseline differences, and unmeasured variables can distort the apparent effect of gum chewing.

Option 2 is a single prospective cohort study with a large sample of 5,000 clients. The large number improves precision—the confidence interval narrows—but a bigger sample does not upgrade an observational design to experimental evidence. Precision and validity are separate concepts in evidence appraisal.

Option 4, a consensus statement from a national panel of surgeons, represents expert opinion. Even when experts are highly respected, consensus statements sit near the bottom of the hierarchy for intervention questions because they are not generated through controlled comparison of groups.

Option 3 is a multicenter randomized controlled trial of 600 clients. Randomization balances known and unknown prognostic factors across groups, which is the key mechanism that supports causal inference. The multicenter feature also improves generalizability because the intervention is tested in more than one setting. Key point! For intervention questions, a well-conducted RCT outranks a systematic review of observational studies.

The evidence hierarchy can be summarized as follows:

Level	Design	Why it ranks there
Highest	Systematic review or meta-analysis of RCTs	Pools randomized evidence; reduces random error across trials
High	Single randomized controlled trial	Randomization supports causal inference
Moderate	Systematic review of cohort studies	Still observational; confounding remains possible
Low	Single cohort or case-control study	No random assignment; bias risk is higher
Lowest	Expert opinion or consensus statement	No controlled comparison of groups

The provided evidence base reinforces this reasoning. The strongest available sources on gum chewing and postoperative recovery are meta-analyses of randomized controlled trials ^[1][3][4], and an individual RCT ^[2]. These align with the principle that randomized evidence—whether pooled or in a single well-designed trial—provides the most trustworthy answer for an intervention question. A review of cohort studies, regardless of how many cohorts it includes, remains a lower level of evidence because the design cannot rule out confounding the way randomization can. Watch out! Do not equate “systematic review” with “highest evidence” automatically; always check whether the review pooled randomized trials or observational studies.