Clinical sciences and statistics for MRCPI Part 1

A randomised controlled trial of a new drug reports a hazard ratio for death of 0.82 with a 95% confidence interval of 0.64 to 1.05, p = 0.11. What is the most accurate interpretation?

1. The confidence interval shows the range within which 95% of individual patient outcomes will fall Wrong. A confidence interval describes the precision of the estimate of the population effect, not the spread of individual outcomes.
2. The drug reduces mortality by 18% and should be adopted Overclaims. That is the point estimate, but the confidence interval means the true effect could plausibly be a small harm.
3. The result is compatible with a clinically important benefit and also with no benefit; the trial has not excluded either correct Correct. The point estimate suggests an 18% relative reduction, but the interval crosses 1, so a meaningful benefit and a null effect are both compatible with the data. The honest reading is uncertainty, most often from inadequate power.
4. There is an 11% probability that the null hypothesis is true A misinterpretation of the p value. The p value is the probability of data this extreme or more so IF the null were true, not the probability that the null is true.
5. The drug has been proven to have no effect on mortality Wrong, and the commonest misreading in medicine. Absence of statistical significance is not evidence of absence of effect.

The point: Read the confidence interval before the p value. A wide interval crossing the null in a small trial means underpowered, not disproven. A narrow interval tightly around the null is genuine evidence of no important effect.

Source: RCPI curriculum — research methodology and critical appraisal Royal College of Physicians of Ireland · tier 0, exam blueprint / regulator