How can I perform a meta‑analysis of oral antibiotics without mechanical bowel preparation in elective colorectal resection when no randomized controlled trials exist?

Conducting a Meta-Analysis on Oral Antibiotics Without MBP When No RCTs Exist

You can still perform a meta-analysis by including high-quality observational studies (cohort studies, case-control studies, or registry data) when RCTs are absent, but you must acknowledge the increased risk of bias and apply rigorous sensitivity analyses to assess the robustness of your findings. 1

Methodological Approach When RCTs Are Unavailable

Study Selection Strategy

The absence of RCTs does not preclude meta-analysis, but requires careful consideration of study design hierarchy:

• Include longitudinal cohort studies with exposure (oral antibiotics without MBP) measured before outcomes, controlling for baseline confounders 1
• Prioritize studies with propensity score matching or adjustment to minimize confounding, as demonstrated in the ACS NSQIP database studies referenced in the guidelines 2
• Exclude cross-sectional studies where temporal ordering cannot be established, as these are highly prone to confounding bias 1

The ERAS guidelines explicitly acknowledge this situation: "However, to date no RCTs have been performed to support this observation [oral antibiotics without MBP], and as such, further high-quality evidence is necessary to inform the debate" 2. Yet the guidelines cite meta-analyses that successfully combined RCTs with observational studies—23 RCTs and 8 cohort studies totaling 63,432 patients 2.

Addressing Bias in Non-Randomized Studies

Critical considerations for your meta-analysis:

• Confounding control: Ensure included studies adjust for key confounders (age, smoking, operative time, surgical approach, comorbidities) as demonstrated in multivariate analyses 3
• Selection bias: Use registry data or population-based cohorts rather than single-center convenience samples
• Measurement bias: Verify consistent definitions of outcomes (SSI, anastomotic leak) across studies

The guidelines note that "oral antibiotic decontamination alone in patients with no bowel preparation has not been studied and any potential effect remains unknown" 2, making this an area where observational data may be the only available evidence.

Practical Implementation Steps

1. Define Eligibility Criteria Explicitly

Preregister your protocol with clear inclusion/exclusion criteria 1:

• Study design: Prospective cohorts, retrospective cohorts with adequate confounder adjustment, large registry studies
• Population: Adults undergoing elective colorectal resection
• Intervention: Oral antibiotics WITHOUT mechanical bowel preparation (plus systemic IV antibiotics)
• Comparator: No bowel preparation (with systemic IV antibiotics alone)
• Outcomes: SSI (primary), anastomotic leak, mortality, ileus, length of stay

2. Extract and Synthesize Data

• Use adjusted effect estimates (odds ratios, risk ratios) rather than crude rates when available
• Document the confounders adjusted for in each study
• Extract data on both benefits (reduced SSI) and harms (C. difficile, antibiotic resistance)

3. Conduct Sensitivity Analyses

Essential analyses to assess robustness 1:

• Restrict to studies with propensity score matching: Compare results from matched vs. unmatched cohorts
• Stratify by study quality: Use tools like ROBINS-I to assess risk of bias in non-randomized studies
• Perform leave-one-out analyses: Remove each study sequentially to identify influential outliers
• Assess publication bias: Use funnel plots and Egger's test, though acknowledge limited power with few studies

The guidelines demonstrate this approach: when comparing all studies vs. RCTs only for MBP alone, results remained consistent 2.

4. Grade Evidence Certainty

Apply GRADE methodology 4:

• Start at "low certainty" for observational studies
• Downgrade further for serious risk of bias, inconsistency, or imprecision
• Consider upgrading if large effect sizes exist or dose-response relationships are evident

The ERAS guidelines rated combined MBP+oral antibiotics as "low quality evidence, weak recommendation" precisely because RCT data were limited 2.

Reporting Your Findings

Transparency Requirements

Explicitly state in your manuscript 4:

• "No RCTs comparing oral antibiotics without MBP versus no preparation were identified"
• "Meta-analysis includes X observational studies with inherent risk of confounding"
• Present both pooled estimates AND prediction intervals to show heterogeneity
• Discuss limitations prominently: "Findings should be interpreted cautiously given reliance on non-randomized evidence"

Contextualize Within Existing Evidence

The guidelines provide a model: they acknowledge that "much of these new data have been derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP®) targeted colectomy database, with a likely degree of cross-reporting of patient populations" 2, yet still synthesize these data while noting limitations.

Alternative Approaches If Observational Data Are Insufficient

If even high-quality cohort studies are lacking:

• Conduct a systematic review without meta-analysis: Narratively synthesize available case series or institutional reports
• Perform indirect comparisons: Use network meta-analysis to compare oral antibiotics without MBP to other strategies through common comparators 5
• Advocate for future RCTs: The guidelines explicitly state "further high-quality evidence is necessary to inform the debate" 2—your review can justify the need for such trials

Key Pitfalls to Avoid

• Do not combine RCTs and observational studies in the same pooled estimate without sensitivity analysis separating them
• Do not ignore heterogeneity: If I² >50%, explore sources through subgroup analysis (colon vs. rectal surgery, laparoscopic vs. open)
• Do not overstate conclusions: Use language like "suggests" or "may reduce" rather than "reduces" when evidence certainty is low
• Do not omit harms: The guidelines note that "no recommendation for the use of oral antibiotic decontamination can be given for patients having no bowel preparation" 2 partly due to unknown effects on antibiotic resistance