At the 2019 ACR Annual Meeting in Atlanta, almost 500 presentations discussed new randomized controlled trials (RCTs). We are lucky to have so much research activity in our field, but applying an RCT to your daily practice can be challenging. Here are three tips to supercharge your critical appraisal!
1. Go Beyond the Abstract: Busy practicing clinicians often stop after reading the abstract. If you want to apply an RCT to patient care, though, you have to answer three important questions:
Does this apply to my patient? (inclusion and exclusion criteria)
What is the risk of bias? (allocation concealment, unblinding, data shenanigans, etc.)
How large is the benefit and how great are the risks? (primary endpoint, adverse events)
The abstract almost never provides an answer to #1 or #2 and frequently omits important aspects of #3. Avoid superficially skimming three manuscripts. Instead, pick out one paper that matters, skip right to the Methods and the Results sections, and answer all three questions before you move on to the next.
2. Randomization and Blinding – Trust but Verify: Well conducted double blind RCTs mitigate selection bias, information bias, and confounding, all of which plague observational data. Unfortunately, threats to validity creep in as trials progress. (1,2)
For starters, I always check to make sure that the randomization process and subsequent concealment of allocation were performed carefully. If that seems appropriate, I look for things that could unblind patients or physicians, such as an injection that burns or a revealing side effect. In particular, I always evaluate loss to follow up – this may not be equal across groups and could falsely amplify (or reduce!) the treatment effect. Finally, be wary of open label extensions. These can provide important information about harms but are essentially well funded cohort studies. (3)
3. Use the NNT and NNH: Evaluating the magnitude of benefit from therapy can be challenging. I recommend simplifying the process by calculating the number needed to treat (NNT) for the primary outcome measure and the number needed to harm (NNH) for important adverse events.
Calculating these numbers is easy! Find the percentage of response in both the treatment group and the control group; the difference between them is the “absolute risk reduction.” If you take the inverse of the absolute risk reduction, you have the NNT. The NNT translates easily into clinical practice, as it answers the question, “How many patients will I need to treat before one experiences the purported benefit?” Using this approach also helps you understand the risk/benefit ratio of a therapy, as you can directly compare the NNT to the NNH.
It is an exciting time to be in rheumatology. At times it can be overwhelming, but the trials we saw this weekend will be published gradually over the next year. If you take a few minutes each week to critically analyze them when they arrive, you will come to the next annual meeting in Washington DC with a deep understanding of the information presented this year. For more information, I recommend the citations below and spending time with all the great content at RheumNow.com!
1. Hill CL, LaValley MP, Felson DT. Secular changes in the quality of published randomized clinical trials in rheumatology. Arthritis & Rheumatism 2002;46:779–784.
2. Cohen N, Lavie RG, Manor Y, Mimouni M, Furst DE, Amarilyo G. Questioning a publication bias between industry-funded and non-industry-funded randomized controlled trials on biological and small molecule therapy for rheumatoid arthritis. Seminars in Arthritis and Rheumatism 2019:S0049017219300733.
3. Taylor GJ, Wainwright P. Open label extension studies: research or marketing? BMJ 2005;331:572–574.