The purpose of conducting a randomised controlled trial, or a systematic review of randomised controlled trials, is to determine the size of the treatment effect (or the difference in outcome for the treatments being compared). The mean (or average) difference in outcome between the treatment groups is called the point estimate. The point estimate is our best guess of the true value of the treatment effect.
But the point estimate comes with some uncertainty, and this uncertainty can be quantified using confidence intervals. At the centre of the confidence interval is the point estimate, but now there is some room on either side of the point estimate for uncertainty. Confidence intervals always have a lower and an upper limit, which indicates that the true effect may be somewhere within this interval, and the width of the confidence interval represents the precision of the treatment effect estimate. If the confidence interval is narrow, the size of the treatment effect is known more precisely. Different levels of confidence intervals can be calculated (eg, 95%, 99%), but the type most commonly reported in trials and reviews is the 95% confidence interval.
Interpretation of confidence intervals will be explained using a hypothetical trial that found a mean difference of 2 points in a 0-10 pain scale between treatment A and treatment B (control) with a 95% confidence interval ranging from 1 to 3 points. That is, at the end of the trial patients receiving treatment A had 2 (out of 10) points lower pain, on average, compared to patients receiving treatment B. A simple interpretation of the confidence interval is that if the same trial was repeated 100 times, in 95 of the repeats the point estimate would fall between 1 and 3 points. Alternatively, we can say that we are 95% confident that the true effect of the intervention lies somewhere between 1 and 3 points.
Reporting of the point estimate and its confidence interval provides physiotherapists reading trials (and reviews) with richer information than just reporting the probability value (or p-value) produced by statistical testing. P-values only indicate if the observed difference is statistically significant (p<0.05) or not (p>0.05). Point estimates and confidence intervals indicate the magnitude and precision of the effect. The confidence interval also indicates the results of statistical testing – if the 95% confidence interval includes “0” (no effect) there is no statistical difference between the groups (p>0.05).
The good news for physiotherapists is that the use of confidence intervals in trial reports is increasing steadily over time. In 2016, 42% of physiotherapy trials reported confidence intervals.
Your ability to read scientific articles will improve with practice. Make the commitment to read at least one article per month and share your reading with the global physiotherapy community in #MyPTArticleOfTheMonth.