The 2x2 table is indispensable for interpreting clinical trials but it is difficult to produce from meta-analysis: most analyses are reported by odds ratio (OR) and cumulative totals are not weighted. We present a simple method to construct a 'typical' 2x2 table from OR, which allows easy derivation of ERR (event rate ratio) and ERD (event rate difference) using the conventional table:

Outcome Event No Outcome Event
Intervention

a



b

No Intervention

c



d


1. Let n1=n2=n and assign a convenient value (e.g. n = 100).
2. Assign a control rate Rc = c/n. Control odds (Oc:d) are Rc/(1- Rc).
3. Calculate the odds Oa:b, for the intervention: Oa:b = OR.Rc/(1- Rc).

Calculate a = Oa.n/(1+ Oa). As c=Rc.n, the remaining cells b and d can be produced and ERR and ERD and NNT calculated.

The effects of different values of Ac can examined. Note that the original control rate Rc for the meta-analysis can only be calculated if ERR is known, in which case Rc = (ERR-OR)/(ERR(1-OR)). Example: The typical OR for cesarean section with simple fetal monitoring vs. intermittent auscultation in labour is given as 2.7.1 Assume a 'control' risk of 15%. The 'typical event table' is:

Cesarian Section No Cesarean Section
Fetal Monitoring 32 68
No Monitoring 15 85

The treatment risk of cesarean section of 32% is easily visible: ERR is 2.1. For 'control' rates Rc of 5% to 40% the risk with monitoring ranges from 12% to 64 % for a fixed odds ratio. 'Typical event tables' can be easily generated for different prevailing 'control' rates, are useful for explaining options in health care, and should be included in reports of meta-analysis.

1. Nielson JP. EFM alone vs. intermittent auscultation in labour. CCPC, lssue2 1995.