How NNTs can be graphically illustrated in network meta-analyses

Selecting the best option out of a number of medical interventions can be difficult. This does not only apply for patients, but also for treating physicians and guideline authors. Network meta-analyses can provide support. With this approach, more than two treatment or diagnostic options can be compared, even if they have not (yet) been tested against each other in clinical studies. Experts commonly choose the number needed to treat ( NNT) as an effect measure, also because this measure is relatively easy to understand, even for non-experts. It specifies how many patients on average need to be treated to achieve treatment success in one patient.  However, the interpretation of the NNTs determined in network meta-analyses is by no means easy.

In the Journal of Clinical Epidemiology, a team of authors analysed how NNTs can best be graphically represented in network meta-analyses. The aim of the graphs is to facilitate interpretation of results. Using the example of drugs for Alzheimer’s dementia, the authors, including Ralf Bender (Head of the Medical Biometry Department at the Institute for Quality and Efficiency in Health Care, IQWiG), show which types of graphical representation are suitable for NNTs and under which conditions.  They also explain the strength and weaknesses of these graphs in the respective analytical context.

The article will be published in the July issue of the Journal of Clinical Epidemiology and is freely available on the Internet until the 3rd of June.