24x7 Editing

‘Bits’ & ‘Pieces’ of Meta-Analysis: A statistical approach for Combining Data from Multiple Studies

Date : October 30, 2019

With new studies being constantly performed and published, the amount of information is available in a specific domain is overwhelming. Different studies published on a specific topic may provide contradictory theories, resulting in the misinterpretation of the results. But by combining data from individual studies, the accuracy of the results/estimates can be increased statistically.

Meta-analysis a statistical procedure includes a set of approaches used to combine qualitative and quantitative data from various studies and to arrive at a single conclusion of greater statistical power. This is because meta-analysis includes a number of subjects, accumulated results/effects, or greater subject diversity.

A meta-analysis, commonly used in medical research or health care, can be used for various purposes.

  • To provide a complex analysis of safety data and benefits
  • To develop a precise estimate of effect magnitude
  • To evaluate subgroups with statistical insignificant individual numbers
  • To develop the statistical significance of studies with conflicting results

How to perform a meta-analysis?

If the original review was flawed, unsymmetric or partial, then the meta-analysis may deliver unreliable quantitative information. The crucial requirement to perform the meta-analysis is to state the unbiased collection of original studies that examine a similar therapeutic question. The most recommended approach of assessment here is the QUOROM (Quality Of Reporting of Meta-analysis) statement method. 

On fulfilling the major requirement, the next step is to perform the meta-analysis. The stages involved in this procedure are: 

  • Identify a research question 

The initial step is to formulate a research question and propose the hypothesis. The significance of the study must be explained and analytical strategy should be justified. 

  • Locate studies

Meta-analysis demands a comprehensive search technique which interrogates various electronic databases. This involves looking for multiple databases of reliable indexes such as Scopus, the web of science, Embase, PubMed, and many more. Ensure to include only those sources with specific terms & follow certain research design and exclude papers that do not meet the study criteria.  

  • Decide selection criteria

Use inclusion and exclusion criteria to ensure that only high-quality and direct research question relevance evidence are included. To achieve this include randomized controlled trials and unpublished studies to eliminate publication bias. 

  • Data extraction 

Upon selecting the studies for inclusion in the meta-analysis, extract or summarize the data from each study. Also, based on the study and research question, include categorical or numerical measures. In addition, measure the data variability for control groups and intervention. 

  • Calculate Effect size

Having gathered the required data, the next step is to determine the summary measures from each study. Such measures are known as effect size. The effect size represents the difference in the average scores between the control groups and intervention. Ensure to standardize the units of measurements as they vary according to a study.  

  • Estimate the final effects 

In the last step, choose and apply an apt model (such as fixed effect or random effect model) to compare the effect size across various studies. Once the final estimate is obtained, represent then using a forest plot. 

 

Where does meta-analysis fit in the research process?

Meta-analysis popularly finds its application in the following aspects

  • Planning new studies – Meta-analysis plays a crucial role in planning new studies. It helps to identify the questions which have been answered and have been left unanswered, which populations can yield significant results and which variant intervention is powerful. 
  •  Grant application – Meta-analysis can be used in the grant application to justify the requirement of the new study. It puts the data in the context and displays the utility of new study. The graphical elements such as the forest plot capture the reader’s attention and clearly presents the data.
  • Publications – Often journals require scholars to submit meta-analysis that summarizes the body of evidence rather than a traditional narrative review. 

Meta-analysis provides a rationale to deal with practical complexities. Some of the added advantages of meta-analysis include:

  1. It helps to overcome bias due to unsystematic reviews
  2. It detects small yet significant effects
  3. It provides precise estimates of the size of effects uncovered 

Meta-analysis offers a systematic approach to synthesizing evidence to obtain answers to therapeutic questions. Nevertheless, pitfalls involved in the execution of meta-analysis can be limited by the quality of the underlying studies.


Leave a Reply

Your email address will not be published. Required fields are marked *