Systematic Reviews and Meta-Analysis in the Medical Sciences: Best Practice Methods for Research Synthesis

Systematic reviews (SRs), which compile data from several research on a topic, are becoming a more vital type of scientific communication: Since 2010, the number of reports has increased by nearly 200 %. The Preferred Reporting Items for Systematic Reviews and Meta-Analysis services reporting standards and guidelines have just been approved by Social Science & Medicine for writers to utilize when creating review papers for publication. To enable the highest calibre research synthesis to be published, to allow readers to judge if a specific SR “embodies mega illumination” or “mega error,” to promote improved scientific understanding and significant changes in practice. Although the methodologies are taken in total from all of science, the examples are taken from literature on health.

Meta-analysis is the ‘original big data.’

In essence, SRs combine the findings of two or more separate studies that were conducted on the same topic. SRs might not have a quantitative component to highlight the papers’ conclusions under review. The phrase “meta-analysis” is frequently employed in customary practice to imply that the writing of A Meta-analysis evidence has already been thoroughly obtained and analyzed. Writing Meta-analysis provide a different type that combines qualitative data acquired from several research on the same issue. Meta-reviews, in turn, are reviews about reviews. All SRs are a kind of evidence or research syntheses, whether they be SRs, meta-analysis, met synthesizes, or meta-reviews.

Assumptions involved in systematic reviews

The history that just ended omitted a concrete explanation of how a lack of rigour may jeopardize SRs. We discuss the fundamentals of systematic reviewing, divided into seven primary processes, which Fig. 1 succinctly summarizes to put these presumptions into context. The assumptions used in systematic reviews are organized per the SR process phase in this section. SR teams frequently improve their approaches as the process goes on, which requires going back and repeating previous parts of the process until the SR is finished with enough quality. This is the first sign that the methods are highly synergistic. The advice we offer in this article’s following paragraphs is summarized in Table 1.

Fig:01 The meta-analysis process is depicted in seven steps that build on each other and sometimes must be repeated as feedback learned during the process emerges.

Table 1 Methodological steps necessary to conduct systematic reviews (SRs), along with best-practice recommendations (the text expands on these points)

Description	Do	Don’t
1. Formulate the research problem	Use meta-reviews of prior SRs to identify the most exciting questions   Build an SR research team with broad expertise          Rely on the SR team members’ understanding of the literature Create answerable questions by utilizing TOPICS + M (see text) elements or alternative formulations	Conduct an SR entirely on your own          Assume rationale for decisions is evident to readers                          Decide on research questions without first scoping the literature
2. Find and select studies	From an early point, seek librarian advice to improve searches Pilot search terms                   Search multiple databases and electronic hosts	Ignore unpublished literature such as thesis, dissertations, conference abstracts, or organizational reports         Rely solely on the previous SR team’s search strategies to update a review
3. Code studies for crucial features	Assess methodological quality using a standardized inventory                             Code descriptive and methodological features of studies for descriptive purposes and potential post hoc observations	Assume coders will understand the data extraction process without training      Code without first creating a coding form
4. Calculating effect sizes	Check assumptions (e.g., triangulate calculations)        Meta-analyses with between-group comparisons should put study outcomes on a single metric, if necessary	Use unstandardized mean difference effect sizes unless all studies have the same metric and you are sure all assumptions are justified
5. Analyze the systematic review database	Know the assumptions in play Consider appropriate adjustments such as for clustering, small sample sizes, and outlier analysis/adjustment	Base your modelling decisions off results from analysis                 Ignore heterogeneity if present
6. Re-analysis, development, or criticism	Update SRs more often with rapidly developing literature Use quantitative means to re-examine an SR’s conclusions	Reinvent the wheel – instead, return to existing SRs to see what may be improved or advanced

Formulating the research problem

The SR team formulates the research challenge in Step 1, which depends on the members’ comprehension of the literature from both a substantive and techniques viewpoint (including statistical assumptions). Importantly, the SR will not be worthwhile to complete if the team has this clear notion. From a practical aspect, it is important to note that Step 1 is essential: The more resources required to review within a reasonable timeline depends on how extensive the research problem is. Therefore, a poorly constructed SR might result in losing essential resources.

Finding and selecting studies

Step 2 of the SR process involves comprehensive literature searches to identify as much relevant research as possible. As previously said, a well-developed research statement will speed up the finding of studies that meet the inclusion criteria and are eligible for evaluation. There are several exceptions to this rule. Of course, writing a Meta-Analysis manuscript: Since studies may not include mental health features in their titles and abstracts, it was essential to acquire many more full-text reports for analysis in HIV preventive SR.

Coding studies for substantive and methodological features

The most intriguing parts of the investigations, which the SR team anticipates would attenuate impacts, are captured through coding methods resulting from a well-formulated research challenge for the Clinical Meta-Analysis Experts. For instance, SRs of treatments frequently look at the behaviour modification strategies used to enhance participants’ health; another frequent factor is the treatment dose.

Calculating effect sizes

SRs pool the findings, either qualitatively or quantitatively. Effect sizes may look at connections between variables, mean levels of phenomena, or both in a meta-analysis, which pools findings from several studies. Authors should also include individual effect estimates for each research or the available quantitative data from the reports in an SR without meta-analysis that focuses on outcomes rather than just qualitative explanations of results.

Analyzing the Systematic review database

Non-independence across studies in a review is another issue that may need to be addressed at various stages in a traditional meta-analysis. The first instance occurs when effect estimates are determined because, if non-independence is neglected, improper study weighting may follow.

Re-analysis, development, or criticism

Some literature evolves quickly, superseding existing SRs and increasing the value of updated SRs. If the original SR’s methodologies were of good quality, the prior SR’s database, if accessible, may be reanalyzed to assess these hypotheses. Alternatively, an SR team may hypothesize that dimensions not taken into account in a published SR could assist explain observed heterogeneity.