Videum
 
 
Blausen Medical
Blausen Medical

Blausen Medical

Medically accurate and informative 3D animations

0 Followers
Share
73

Matching-Adjusted Indirect Comparison

In clinical research, randomized controlled trials provide the best evidence for comparing different treatments. In many trials, new treatments are compared only with placebo. In the absence of direct drug comparisons, how can clinicians compare the effects of these two drugs? They could rely on their clinical experience to interpret the data or consider other real-world sources of evidence. Another approach is to perform an indirect comparison. This represents a more rigorous approach that is increasingly accepted by healthcare decision makers. One method of indirect comparison is network meta-analysis, or NMA. With this method Drug A and Drug B could be compared indirectly, as both trials used a common comparator of placebo. However, NMA assumes that effect modifiers in the trial populations are similar (homogenous). If the populations are not similar, due to differences in eligibility criteria, such as prior treatment use, the comparison may be biased. Another method is matching-adjusted indirect comparison, or MAIC. MAIC adjusts for differences in study populations by taking individual patient data from one trial and weighting it to match the aggregate data from another trial. Therefore, MAIC can be used to compare outcomes in trials with heterogenous study populations. If a common comparator, such as placebo, is available, the MAIC is said to be anchored. If no common comparator is available, the MAIC is unanchored and weighted outcomes in single arms are compared. Suppose we wanted to compare Drug A, evaluated in the Trial A, and Drug B, evaluated in Trial B. Unlike Trial B, Trial A included patients with previous biologic failure and therefore represents a more difficult-to-treat patient population. The baseline characteristics which we can select for adjustment may include demographic factors, disease severity and use of previous treatment. MAIC assigns more weight to patients in Trial A who are similar to those in Trial B, and less weight to patients who are not similar. This requires individual patient data from Trial A and aggregate data from Trial B. Now the populations are matched, creating a smaller but more homogenous population for comparison. Note that MAIC requires some overlap between study populations. The clinical context, or target population, of MAIC is that of the matched trial, and outcomes should be interpreted in this population only. Different comparisons can be made depending on the availability of the individual patient data from different trials and swapping the roles of the studies. For example using individual patient data from Trial B and aggregate data from Trial A. This comparison will have a different clinical context and may come to different conclusions. To summarize, MAIC allows for indirect comparisons between trials with heterogeneous study populations. The clinical context of the treatment effects is well-defined, which helps when interpreting results. This method has limitations. The weighting interferes with randomization in anchored MAIC while unanchored MAIC is not randomized. Weighting also reduces the effective sample size. Lastly, MAIC can only adjust for reported characteristics, so heterogeneity may persist.Where does MAIC fit in the overall picture of clinical evidence? While randomized controlled trials remain the best source of evidence, MAIC provides complementary evidence and is an increasingly accepted approach. The number of publications citing MAIC is increasing and it has been applied in many therapeutic areas. Recognition of MAIC among regulatory bodies is also increasing. In the UK, the National Institute for Health and Care Excellence Decision Support Unit has recently published guidelines for the use and reporting of MAIC.

Duration: 04:59
Licence: CC - Attribution
Original Language: English
Published: 8/9/2019
Format: 3D Animation

Other videos in this channel

 

Related Videos

Matching-Adjusted Indirect Comparison

Comments (0)

Login to add your comment
Be the first to comment!