Skip to main content icon/video/no-internet
icon backReturn to Entries

Meta-Analysis

In epidemiology, the proliferation of multiple and sometimes contradictory studies can be a challenge for interpretation of health risk and health policy formulation. One approach to synthesizing the results of separate but related studies is meta-analysis—the systematic identification, evaluation, statistical synthesis, and interpretation of separate study results. For example, for many years, conflicting results were reported in observational studies of the effect of diet on breast cancer risk. The lower rate of breast cancer incidence for women in Asian countries suggested a protective effect for soy-based diets; yet migration patterns and changes in diet yielded conflicting results. A synthesis of epidemiologic studies showed a moderately protective effect for soy intake (odds ratio [OR] = 0.89, 95% confidence interval [CI] 0.75 − 0.99), with a stronger effect among premenopausal women (OR = 0.70, 95% CI = 0.58 − 0.85).

This entry reviews the elements of a well-conducted meta-analysis, summarizes recent research, and discusses two important examples of the use of metaanalysis in epidemiology: The Guide to Community Preventive Services and the Human Genome Epidemiology Network.

The technique of using a quantitative synthesis probably was used first by Karl Pearson in 1904 to increase statistical power in determining the efficacy of a vaccine for enteric fever; Gene Glass coined the term meta-analysis in 1976 to apply to systematic review and quantitative synthesis. From the social sciences, use of meta-analysis quickly spread to medicine in the 1980s. Later, meta-analysis was used increasingly to combine results from observational studies.

Over time, meta-analysis has become more prominent in epidemiology, extending to important policy decisions and determining the effectiveness of interventions. To address the quality of reporting of metaanalytic reviews, guidelines were developed for reporting randomized controlled trials (RCTs) to facilitate synthesis, meta-analysis of RCTs, and metaanalysis of observational studies.

Elements of a Well-Conducted Meta-Analysis

Stating the Problem and Conducting the Literature Search

A well-conducted meta-analysis should start with an explicit statement of the research problem, which can be framed by population, intervention (or exposure), comparison, or outcome. After specifying the study question, the next step is a systematic search for relevant studies. Computerized databases have aided this step, particularly in meta-analyses of RCTs. However, limiting a search to two or three electronic databases might produce incomplete evidence. A comprehensive search will include multiple databases, the reference lists of recent review articles and metaanalyses, and frequently contact with experts to find unpublished results.

With the proliferation of meta-analyses in the epidemiologic literature and the availability of electronic repositories of research, variably skilled researchers are conducting searches. Recognizing the importance of knowledge and skills in complex bibliographic retrieval and verification of information, the Medical Library Association has developed a policy that health science librarians should contribute to the search process for health and information research.

Collection of Data

Abstraction of data from the search should begin with explicit inclusion and exclusion criteria for studies. Commonly used criteria include period covered in the review, operational definitions of the variables, the quality of a study, and the language of publication. To the extent possible, inclusion or exclusion criteria should be based on valid scientific principles (e.g., treatment changes over time), not the convenience of the researcher.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution