Every few days, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.
The article featured today is from Statistics in Medicine and the full article, published in issue 38.14, is available to read online here.
Leahy, J, Thom, H, Jansen, JP, et al. Incorporating single‐arm evidence into a network meta‐analysis using aggregate level matching: Assessing the impact. Statistics in Medicine. 2019; 38: 2505– 2523. doi: 10.1002/sim.8139
Two important questions are regularly asked when a new medical treatment is developed:
1. Is it beneficial?
2. Is it more beneficial than the other treatments available?
To answer the first question, careful testing is required. The best way to do this is through randomised controlled trials. In this type of trial one group of patients is given the new treatment, while another group is given some other treatment, and they are compared. This other treatment may be a placebo (a pill designed to have no effect), or another commonly available treatment. The advantage of the new treatment is assessed relative to the comparator.
To answer the second question, it is necessary to find trials that have treatments in common. For example, if drug A is compared to placebo in trial 1, and drug B is compared to placebo in trial 2, then drug A can be compared to drug B by measuring how beneficial or harmful both these drugs are compared to a placebo. To do this a method called network meta-analysis is used. A network is connected if a link can be made between any two treatments through a number of trials.
However, in some studies all patients are given the new treatment without a comparison. These are referred to as single armed trials. These may be done when a company is trying to see how beneficial the treatment might be in different types of patients, or if many similar treatments are being developed at the same time, there may not have been a chance to compare the new treatment to these other treatments. Trials where patients are not assigned to one treatment on a completely random basis, are thought to be of lower quality, or not as valuable, as trials where patients are selected for different treatments at random. Therefore, it is difficult to say whether the new treatment is truly beneficial in this case.
The second question is even more difficult to answer using single armed trials, as the network will be disconnected if a treatment has only been studied in a single arm trial. Therefore, a traditional network meta-analysis cannot be done. This paper puts forward a method of connecting a disconnected network by creating a match between trials that are most alike. A “fake” match is made based on extra information about patient types in the various trials. The trials with the most similar patient types are matched together, almost as if they were in the same trial. The suitability of this method is assessed and advice on situations when it may or may not be appropriate is provided. Even in the ideal set up of the simulation study, this method can add additional bias to the results. Therefore, this method should be used with care, and should not be thought of as a substitute for doing randomised controlled trials. The method is shown using a mix of randomised controlled trials and single arm trials for treatments of infection with the hepatitis C virus.