Gene V. Glass coined the term “meta analysis” for, “The statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings”
Systematic assessment of earlier evidence-based studies regarding a similar topic of research to derive a pooled conclusion using a quantitative and formal design can be defined as meta analysis. Conclusion and results drawn from any meta analysis are much more robust and have a higher statistical significance than the individual studies included in the meta analysis. It minimizes bias, since; the validity of any hypothesis cannot be based on a single study. It also helps identify similar patterns, disagreements, errors and interesting relationships among the various constituent studies. Reporting a relevant meta analysis in the introduction section of a new primary study corroborates the significance of the new study. It also helps in solving some unresolved aspects of the individual studies, settling conflicting evidence of contrasting studies. A meta analysis might lead to establishing and performing some new hypothesis.
Factors affecting the success of a meta analysis depend upon the objective of the meta analysis, selection of the studies that are to be included, literature search for the studies, dealing with the incomplete data, statistical methods used, analyses of the data and minimizing publication bias of the meta analysis.
For any meta-analysis, the protocol development is the initial and the most important step. It provides the researchers a systematic and methodological approach to conduct the meta-analysis. It details methodology to screen and evaluate the studies based on the predefined inclusion and exclusion criteria. For literature search, the protocol defines the keywords; the filters to be used (like language of the papers to be included; year range, if applicable; regional or global studies, etc.). It also outlines the steps for data extraction and analysis, reporting, and interpretations. The primary purpose of a protocol is to prepare a blueprint for the meta analysis. This also helps in minimizing the publication bias and favors the most conducive collaborating conclusion of all.
PRISMA-P guidelines specifies a checklist (http://www.prisma-statement.org/documents/PRISMA-P-checklist.pdf) for writing a protocol for meta analysis:
- Title: This includes identification of the protocol and/or its update.
- Registration number: The registration number of the meta analysis should be given under this section.
- Authors: The contact details of the authors and their contribution in the review should be mentioned.
- Amendments: Any update or amendment in the previous version of the protocol is given under this section.
- Support: Information regarding the financial sources, role and contact of the sponsor or the institution should be given here.
Introduction: This section includes rationale (6) and objectives (7) of the review along with the references to participants, interventions, comparators and outcomes (PICO).
- Eligibility criteria: The inclusion and exclusion criteria for the screening of the studies and reports to be included in the meta analysis should be specified.
- Information sources: Describes all the information sources (electronic databases) to be searched along with the planned dates of coverage.
- Search strategy: Atleast for one electronic search database, presenting the draft of search strategy including the search limits used.
- Study records: The methods to manage data (11a), basis of study selection (11b) and strategies to extract data (11c) from the selected studies is explained in the study records section.
- Data items: Provides the list and definitions of variables, assumptions and simplifications to be used to sort data.
- Outcomes and prioritization: Lists the primary and secondary outcomes of the study.
- Risk of bias in individual studies: Describes methods for biasness risk assessment.
- Data synthesis: this section contains description of additional analysis (such as Sub group analysis), planned summary measures, quantitative synthesis, etc.
- Meta-bias(es): Specifies any planned assessment of meta-bias (such as publication bias across studies, selective reporting within studies)
- Confidence in cumulative evidence: Describes the method to assess the strength of the body of evidence (such as, GRADE).
The rigorous approaches and stringent criteria used while designing a protocol for the concerned meta analysis determines its authenticity and success. So, the protocol must be carefully delineated before initiating any meta analysis on a research study or hypothesis.
Anoop Inder Kaur
Dr. Shivali Arora