Table 1 Some of the options in metapreg

model()

Specifies the type of the model.

There are three types of the model; model(hexact), model(random) which is also design(mixed), and model(fixed).

design()

Specifies the structure of data and the covariance structure of the random effects.

There are five different structures of data anticipated; design(general), design(comparative), design(pcbnetwork), design(mcbnetwork) and design(abnetwork). When there are two random effects in a comparative analysis, the possible covariance structures are design(comparative, cov(independent)), design(comparative, cov(unstructured))

smooth()

Requests for the model-based study estimates.

Provides a visual assessment of whether the model is a good data summary.

cimethod()

Specifies the type of confidence intervals for the study-specific estimates as displayed in the forest plot.

For proportions, the possibilities are cimethod(exact), cimethod(wald), cimethod(wilson), cimethod(agresti), and cimethod(jeffreys). For comparative and paired data, only cimethod(cml) is allowed when computing the CI for the probability ratios. For matched data, the only possibility for probability ratios is cimethod(koopman). For the odds ratios, the possibilities are cimethod(exact), cimethod(cornfield) and cimethod(woolf).

nomc

Instructs the program not to conduct model comparison in meta-regression.

Not fitting simpler models than what is requested saves time.

by()

Requests the summary estimates at the unique values of the by variable.

Can be useful when the model contains multiple covariates and grouped summary estimates are required.

stratify

Together with the option by(), the stratify option makes it possible to fit separate models in each group of data in the by variable, but present the results in one table and forest plot.

In contrast; the prefix command by: would print out separate tables and graphs for each sub-analysis.

sumtable()

Indicates the type of summary estimates to display.

The possibilities are sumtable(logit), sumtable(abs). When there are categorical data in the model, sumtable(rr) or sumtable(or) are also allowed. All tables are displayed with sumtable(all). By default, none of the tables are displayed.

outplot()

specifies which statistics to display on the forest plot.

outplot(abs) displays the study-specific and summary proportions. outplot(rr) and outplot(or) displays the study-specific probability and odds ratios when data are from comparative, paired, or matched studies. In the case of network meta-analysis, outplot(rr) and outplot(or) displays the summary probability and odds ratio(s) of the multiple treatments relative to the specified reference.