Conducting PRISMA-Compliant Systematic Reviews and Meta-Analyses
A methodological framework for clinical researchers and medical students conducting evidence synthesis.
Jaber Husam Jaradat, M.D.
Conducting PRISMA-Compliant Systematic Reviews and Meta-Analyses
In evidence-based medicine, systematic reviews and meta-analyses occupy the apex of the evidentiary hierarchy. Having led and contributed to multiple meta-analyses published in peer-reviewed journals (such as our work on ibrutinib in CNS lymphoma in Critical Reviews in Oncology/Hematology), I frequently receive questions from medical students and residents on how to execute rigorous reviews.
Here is a step-by-step roadmap for clinical researchers.
1. Registering the Protocol (PROSPERO)
Never initiate a systematic review without a registered protocol:
- Define your PICO question (Population, Intervention, Comparator, Outcomes).
- Search PROSPERO and the Cochrane Library to confirm an identical review does not already exist.
- Transparently register your search strings, inclusion/exclusion criteria, and primary endpoints to prevent selective outcome reporting.
2. Comprehensive Multi-Database Searching
A rigorous search should span at least 3 major bibliographic databases:
- PubMed / MEDLINE
- Embase
- Cochrane Central Register of Controlled Trials (CENTRAL)
- Supplementary registries (ClinicalTrials.gov, WHO ICTRP)
Always combine controlled vocabulary (MeSH terms in PubMed, Emtree terms in Embase) with free-text keywords using Boolean operators OR and AND.
3. Dual-Independent Screening & Risk of Bias
Following PRISMA 2020:
- Two independent reviewers screen titles/abstracts, followed by full texts.
- Discrepancies are resolved by consensus or a third senior reviewer.
- Risk of Bias Assessment:
- Randomized Controlled Trials: Cochrane RoB 2 tool.
- Non-randomized Studies: ROBINS-I or Newcastle-Ottawa Scale (NOS).
4. Quantitative Synthesis: Fixed vs. Random Effects
When pooling odds ratios (OR), relative risks (RR), or hazard ratios (HR):
- DerSimonian and Laird Random-Effects Model: Accounts for both within-study error and between-study heterogeneity ($\tau^2$). Given clinical diversity across real-world cohorts, random-effects models are almost always preferred over fixed-effect models in oncology.
- Heterogeneity Assessment: Evaluate $I^2$ statistics and Cochran’s $Q$ test. If $I^2 > 50%$, explore subgroup analyses and meta-regression.
Conclusion
Evidence synthesis is not merely summarizing literature—it is extracting quantitative clarity from conflicting data. Adhering to rigorous standards ensures our findings can safely guide clinical guidelines and patient counseling.
About Dr. Jaber Husam Jaradat
Physician-scientist and medical intern pursuing U.S. Internal Medicine training with a focus in cancer bioinformatics, single-cell/spatial omics, and translational immuno-oncology. Recipient of the 2026 AACR Global Scholar-in-Training Award (GSITA) and founder of BrighterUs.