Comprehensive Meta-Analysis is a powerful computer program for meta-analysis. The program combines ease of use with a wide array of computational options and sophisticated graphics.
Work with a spreadsheet interface
You can type data directly into the spreadsheet, much as you would with any spreadsheet-based program. Or, if you are currently using another program for meta-analysis, you can either copy data directly from that program or import it using a Wizard.
Compute the treatment effect (or effect size) automatically
In every meta-analysis you start with the published summary data for each study and compute the treatment effect (or effect size). For example, if a study reports the number of events in each group you might compute the odds ratio. Or, if a study reports means and standard deviations you might compute the standardized mean difference. This process of computing effect sizes is typically tedious and time consuming. In some cases, especially when studies present data in different formats, the process is also difficult and prone to error.
With CMA you enter whatever summary data was reported in the published study, and the program computes the effect size from that summary data. For example, you could enter events and sample size, and the program would compute the odds ratio. Or, you could enter means and standard deviations, and the program would compute the standardized mean difference.
Perform the meta-analysis quickly and accurately
This display is an interactive forest plot that yields a clear sense of the data - How many studies are included in the analysis, how precise is each of the studies, whether the effect is consistent from study to study or varies substantially across studies, and so on. You can then customize this display as needed. Add or remove columns, set computational options, open tables with additional statistics.
Create high-resolution forest plots with a single click
A key element in any meta-analysis is the forest plot – a plot that shows the effect size and precision for each study as well as the combined effect. This plot puts a face on the analysis – it shows whether the combined effect is based on a few studies or many, whether the effect size is consistent or varies, and so on. As such, the forest plot plays a central role in helping the researcher to understand the data, and also to convey the findings to others.
Most other meta-analysis programs use graphics engines that were developed for other purposes and push them into service for creating forest plots. By contrast, the plotting engine in CMA was developed specifically for the purpose of meta-analysis. It is very easy to use and provides a wide range of important options.
Create a high-resolution plot in one click and then customize any element on the plot. Select a symbol for studies, for subgroups, and for overall effect. Optionally, specify that symbols should be proportional in size to study weights, so the studies that contribute the most to the combined effect are easy to spot. Set colors and fonts for each element on the graph, and then export to Word™ or PowerPoint™ in a single click!
Use cumulative meta-analysis to see how the evidence has shifted over time
A cumulative meta-analysis is actually a series of meta-analyses, where each analysis in the sequence incorporates one additional study. For example, the first row in the analysis might include a study published in 1990, the next row would include studies published in 1990 and 1991, and so on. A cumulative meta-analysis may be done retrospectively, to show how the body of evidence has shifted over time, or prospectively, with new studies being added to the body of evidence as they are completed.
While cumulative meta-analysis is most often used to track evidence over time, it can also be used to show how the evidence shifts as a function of other factors. For example, we could sort the data by study size and run a cumulative analysis. In this case the program would show the combined effect with only the largest studies included (toward the top) and how this effect shifted as smaller studies were added to the analysis. Similarly, we could start with the higher quality studies and see how the effect shifts as other studies are added.
Use a “Remove-One” analysis to gauge each study’s impact
As part of a sensitivity analysis we might want to assess the impact of each study on the combined effect. For example, what was the impact on the combined effect of an outlier or of an especially large study? Or, did a small study have any impact at all?
To address these kinds of questions the program will automatically run the analysis with all studies except the first, then all studies except the second, and so on. The resulting plot shows the impact of each study at a glance.
Additionally, you have the option of running the analysis with any study or set of studies removed – these can be selected by name, or by the value of a moderator variable.
Work with subsets of the data
When running the analysis you can select by (or filter by) any variable or combinations of variables. You could include or exclude studies by study name. You could include studies that had been rated “Yes” for “Double-blind”. You could include studies where the age had been coded as “Elderly” and the patient type as “Chronic.
Work with multiple subgroups or outcomes within studies
The program allows you to enter data for more than one subgroup, outcome, time-point, or comparison within studies, and offers various options for dealing with these in the analysis.
Assess the impact of moderator variables
When the effect size varies substantially from study to study an important goal of the meta-analysis could be to understand the reason for this variation.
Use analysis of variance to assess the impact of categorical moderators. For example, “Is the treatment more effective for acute patients than for chronic patients?” or “Is homework a more effective intervention than tutoring?”
Use meta-regression to assess the impact of continuous moderator variables. For example, “Does the treatment effect increase as a function of dosage?”, or “Is the magnitude of the effect size related to the age of the students?”
Assess the potential impact of publication bias
Meta-analysis provides a mathematically accurate synthesis of available data, but there may be concern that significant studies were more likely to be published than non-significant studies, and therefore the pool of available data may be biased. The program includes a set of functions that can be used to assess the potential impact of this bias, as a kind of sensitivity analysis.
Video tutorials
We have developed videos of case studies that show how to run an analysis from start to finish. This includes how to enter data, how to run the analysis, how to create plots, how to compare the effect size in different subgroups, and so on.
Critically, each section of the video explains now only how to perform specific functions, but what purpose these functions serve in the context of the analysis, and how to understand the meaning of the statistics.
Each case study runs about ninety minutes. You can watch one from start to finish to learn how to perform a meta-analysis and report it properly. Or, from any screen in the program you can jump to the part of the video that explains all functions on that screen.
Common Mistakes and How to Avoid Them
We recently published a book called Common Mistakes in Meta-Analysis and How to Avoid Them.
This book includes mistakes in such areas as choosing a statistical model, statistics related to heterogeneity, comparing subgroups of studies, publication bias, and more.
From any screen in the program, you can click a link that will open a PDF with the relevant sections of the book.
