#meta-analysis

Conduct a meta-analysis following PRISMA guidelines. Steps: 1. Formulate PICO question (Population, Intervention, Comparison, Outcome). 2. Systematic literature search. 3. Study selection and quality assessment. 4. Data extraction (effect sizes, sample sizes, moderators). 5. Calculate pooled effect size (fixed or random effects). 6. Assess heterogeneity (I² statistic, Q-test). 7. Test for publication bias (funnel plot, Egger's test). 8. Conduct sensitivity and subgroup analyses. Use software: R (metafor), Comprehensive Meta-Analysis. Report with forest plots.

Conduct quantitative meta-analysis following best practices. Data preparation: 1. Extract effect sizes and standard errors from each study. 2. Code study characteristics (sample size, population, methodology quality). 3. Handle multiple effect sizes from same study (average, select one, or use robust variance estimation). Statistical analysis in R metafor package: 1. Fixed effects model: assumes one true effect size. 2. Random effects model: assumes distribution of true effects. 3. Test for heterogeneity using Q-statistic and I² (>75% = high heterogeneity). 4. Moderator analysis: meta-regression or subgroup analysis to explain heterogeneity. 5. Publication bias assessment: funnel plots, Egger's test, trim-and-fill method. Report: Forest plot showing individual study effects and pooled estimate with 95% CI. Address limitations and clinical significance of findings.

Design a systematic literature review following PRISMA guidelines. Protocol steps: 1. Define research question using PICO framework (Population, Intervention, Comparison, Outcome). 2. Develop search strategy: identify 3-5 databases (PubMed, Scopus, Web of Science), create Boolean search terms, set inclusion/exclusion criteria. 3. Screen titles/abstracts independently by 2 reviewers, resolve conflicts with third reviewer. 4. Full-text review using predefined criteria. 5. Data extraction using standardized forms (study design, sample size, outcomes, bias assessment). 6. Quality assessment using appropriate tools (Cochrane Risk of Bias, Newcastle-Ottawa Scale). 7. Synthesize findings narratively or through meta-analysis if appropriate. Document decisions transparently. Register protocol in PROSPERO before starting.