Critical appraisal top tips: Understanding confidence intervals

Have you ever wanted to feel more confident about interpreting the findings of clinical trials and reviews? We’re here to help with a new series of blog posts aiming to help you get to grips with understanding and interpreting biomedical research.

This month’s post focuses on 95% confidence intervals.

What is a confidence interval?

Confidence intervals can be calculated for most kinds of statistical test, e.g. p values, t tests, measures of risk and numbers needed to treat. They provide an indication of the strength of evidence e.g. for the effectiveness of a treatment. Confidence intervals define a range of values within which you can expect the statistical results of a study to fall.

If a study is carried out once it will obviously end with one set of results, and one set of statistics calculated from those results, e.g. a p value. If that study were to be repeated over and over again, you’d expect the results to be slightly different every time –if only because there will always be some variation.

So, you might find a different p value if a study were repeated. Statistics such as p values are often used to try and determine whether an intervention has a statistically significant effect, and it’s possible that one repeat of a study would find a non-statistically significant p value, even if the original study findings had been statistically significant. Confidence intervals can help you to understand how likely this is to happen.

What do confidence intervals mean?

Confidence intervals provide a range of values within which you would expect the results of a study to fall, were it to be repeated. Although the same trial repeated hundreds of times would not yield the same results every time, on average the results would be within a certain range.  95% confidence intervals are often calculated as standard. A 95% confidence interval means that there is a 95% chance that the true size of the intervention effect will lie within this range – so if a study were repeated there is a 95% chance that it would produce a result falling somewhere within the range defined by the confidence interval.

Find out more

In the next of this regular series of posts on interpreting medical statistics we’ll look at another common statistical technique, odds ratios. In the meantime the Library provides you with access to range of resources and papers on confidence intervals and other statistics. A short selection of further reading is given below.

Davies, Huw T.O., Crombie, Iain K. (2009) What are confidence intervals and p-values? Hayward Medical Communications

Sedgwick, Philip (2014). “Understanding confidence intervals” BMJ 2014; 349 :g6051

Don’t forget, we also run regular training sessions on critical appraisal for study types such as RCTs, systematic reviews and qualitative research.

Sources

Greenhalgh, Trisha (2014). How to read a paper : The basics of evidence-based medicine (5th ed.). Chichester: Wiley ; BMJ Books.