Evidence-based Medicine Toolkit

Appraising therapy articles 55Confidence intervals (CIs)Any study can only examine a sample of a population. Hence, wewould expect the sample to be different from the population. Thisis known as sampling error. Confidence intervals (CIs) are used torepresent sampling error. A 95% CI specifies that there is a 95%chance that the population’s ‘true’ value lies between the twolimits.Look to see if the confidence interval crosses the ‘line of no difference’between the interventions. If so, then the result is notstatistically significant.The confidence interval is better than the p value because itshows you how much uncertainty there is around the stated result.Quantifying the risk of benefit and harmOnce chance and bias have been ruled out, we must examine thedifference in event rates between the control and experimentalgroups to see if there is a significant difference. These event ratescan be calculated as shown below.ControlExperimentalTotalEvent a b a + bNo Event c d c + dTotal a+c b+dEvent rateRelative riskControl event rateCER = a/(a + c)EER/CERExperimental eventrate EER = b/(b + d)Absolute riskreductionRelative riskreductionCER – EER(CER – EER)CER