# How To Use Comprehensive Meta-Analysis Software For Meta-regression: A Tutorial And Review

## Meta-regression in Comprehensive Meta-Analysis: a review and tutorial

Meta-regression is a technique that allows meta-analysts to explore the sources of heterogeneity among the effect sizes estimated from different studies. Meta-regression can be used to test whether the effect size varies according to some study-level characteristics, such as sample size, study design, intervention type, or quality score. Meta-regression can also be used to adjust the pooled effect size for potential confounding or moderating factors.

## How to use Comprehensive Meta-Analysis software for meta-regression: a tutorial and review

Comprehensive Meta-Analysis (CMA) is a popular software program for conducting meta-analyses. CMA has a user-friendly interface and offers a wide range of options for data entry, analysis, and presentation. CMA also has a meta-regression module that allows users to perform univariate and multivariate meta-regression analyses with fixed-effect or random-effects models.

In this article, we will review the basics of meta-regression and provide a step-by-step tutorial on how to use CMA for meta-regression. We will use an example dataset from a meta-analysis of the effect of Bacillus Calmette-GuÃrin (BCG) vaccination on tuberculosis prevention [^1^]. We will show how to perform a meta-regression analysis to test whether the effect of BCG vaccination varies according to latitude, year of publication, or study quality.

## Data entry

The first step is to enter the data into CMA. CMA can accept different types of effect size data, such as odds ratios, risk ratios, mean differences, standardized mean differences, correlations, or Fisher's z values. For this example, we will use the odds ratio as the effect size measure. We will also enter the corresponding standard errors or confidence intervals for each study. In addition, we will enter the study-level covariates that we want to use as predictors in the meta-regression analysis: latitude, year of publication, and quality score.

CMA has a spreadsheet-like interface for data entry. We can enter the data manually or copy and paste from another source. Alternatively, we can import the data from an Excel file or a text file. CMA also allows us to edit or delete the data easily. Here is how the data entry screen looks like for our example:

We can also add labels or notes for each study or each column. CMA will use these labels or notes in the output tables and graphs.

## Basic analysis

Before performing a meta-regression analysis, it is advisable to perform a basic meta-analysis to obtain the pooled effect size and test for heterogeneity among the studies. CMA can perform both fixed-effect and random-effects meta-analyses. The fixed-effect model assumes that all studies share a common effect size and that any variation among them is due to sampling error. The random-effects model assumes that the effect size varies across studies and that there is a distribution of true effects in the population. The random-effects model is more appropriate when there is substantial heterogeneity among the studies or when there are reasons to expect variation in the effect size due to differences in study characteristics.

To perform a basic meta-analysis in CMA, we need to select the effect size type and model from the drop-down menus at the top of the screen. For this example, we will select "Odds ratio" as the effect size type and "Random effects" as the model. Then we click on "Calculate" to obtain the results.

The results are displayed in three tabs: "Forest plot", "Table", and "Statistics". The forest plot shows the effect size and confidence interval for each study and for the pooled estimate. The table shows the same information in numerical form. The statistics tab shows some summary statistics and tests related to the meta-analysis, such as Q statistic, I-squared statistic, tau-squared statistic, z-test, p-value, and fail-safe N.

For our example, we can see that the pooled odds ratio is 0.49 with a 95% confidence interval of 0.39 to 0.61. This means that BCG vaccination reduces the risk of tuberculosis by 51 04f6b60f66

- +