Binary outcomes

Risk ratio, odds ratio and risk difference from event counts

Use this module when each study reports how many events occurred in each group and the size of each group. That covers yes or no outcomes: mortality, recurrence, response to treatment.

If your effect already comes calculated from the paper, the right module is pre-calculated effects, not this one.

This page covers what belongs to this module. The screens and the buttons are in the three-step flow.

Step 1, the data

Five required columns, with these exact names:

Column What it is
study the study label, for example “Silva 2019”
event.e events in the experimental group
n.e size of the experimental group
event.c events in the control group
n.c size of the control group

Any extra column is optional and becomes available later, for subgroup analysis and meta-regression. The example dataset carries Region, Design, RiskOfBias, MeanAge and FollowUpMonths.

Step 2, the parameters

Summary measure (sm)

Default RR. It decides what gets pooled:

  • RR, risk ratio. The ratio of the event risks in the two groups.
  • OR, odds ratio. Needed if you will compare against literature that uses OR, or if you intend to use the Peto or GLMM methods, which only work with OR. Remember that the OR drifts away from the RR when the event is common.
  • RD, risk difference, on the absolute scale: the difference between the two event risks, read as events per 100 patients.

Method

Default inverse. It is the pooling method, and it exists only in this module and in single proportions:

  • inverse, inverse variance. The app default. Each study is weighted by the inverse of the variance of its effect estimate.
  • MH, Mantel-Haenszel. Uses the Mantel-Haenszel weights, computed from the cell counts rather than from the variance of the effect estimate.
  • Peto. Uses the one-step Peto odds ratio. Only works with sm = OR.
  • GLMM, generalized linear mixed model. Avoids the continuity corrections the other methods apply to zero cells. Slower. Only works with sm = OR, and the forest plot comes out without the Weight column, because a GLMM fit produces no per-study weight.
Peto and GLMM require OR

If you leave sm = RR or sm = RD and pick Peto or GLMM, the analysis stops with an error message. Switch sm to OR first, or set the method back to inverse or MH.

The other fields

Outcome name, the outcome direction and the group names work as described in Outcome settings.

Analysis model, Tau method, Random CI method and Prediction method work as described in Statistical settings (advanced). In this module the Prediction interval is off by default.

Step 3, export and results

This module carries the full set of cards: forest plot, leave-one-out, funnel plot, small-study effects with Egger’s test, subgroup analysis, meta-regression and the two summary tables as XLSX. All of them are described in step 3 of the flow.