An indirect comparison can estimate how medicines might differ when no head-to-head trial is available. It cannot make people from separate trials become randomized to the competing treatments. A 2024 comparison involving lisdexamfetamine illustrates that boundary.
Where did the data come from?
Schein and colleagues used patient-level information from a long-term centanafadine study and published aggregate information from separate lisdexamfetamine, methylphenidate and atomoxetine studies. They applied matching-adjusted indirect comparisons using propensity-score weighting.
For the lisdexamfetamine comparison, the centanafadine safety population contained 492 people and its efficacy population 490; the published lisdexamfetamine population contained 349. These are source-population counts, not a single randomized head-to-head sample.
What was matched?
Weights sought to align reported characteristics such as age, sex, race, ethnicity, baseline symptom severity and selected physical measures. The available variables differed among comparisons.
Matching aggregate means cannot establish balance on unmeasured characteristics, differences in study conduct, adherence, dropout, outcome collection or treatment expectations. Apparent balance in reported variables is therefore a conditional achievement, not proof of exchangeability.
What did the estimates suggest?
The adjusted symptom-score comparison at up to 52 weeks favored lisdexamfetamine over centanafadine by 6.15 points, with a 95% confidence interval of 4.31–7.99. Several reported adverse-event rates favored centanafadine instead.
For example, the estimated insomnia difference was 12.47 percentage points, not a 12.47% relative reduction. Some adverse-event comparisons, including anxiety and nasopharyngitis against lisdexamfetamine, were not significant. The analysis did not show a uniform advantage on every outcome.
Why safety rankings remain uncertain
Adverse-event collection, exposure duration and study populations may differ between trials. A reported infection or symptom is not automatically treatment-caused. Weighting cannot retrospectively standardize how every investigator elicited and recorded events.
The analysis also treated symptom scales from different trials as comparable. That assumption needs evaluation rather than being treated as an observed head-to-head measurement.
How should this be used?
This was an indirect evidence synthesis, sponsored in the context of centanafadine development, not a randomized demonstration that one medicine is categorically safer or more effective. It generates comparative estimates under explicit assumptions.
The useful questions are which variables were balanced, which were unavailable, how outcomes were defined and whether direct evidence corroborates the result. It does not establish a current prescribing hierarchy, individual tolerability or driving safety.
Scientific source
Schein J, Cloutier M, Gauthier-Loiselle M, et al. A matching-adjusted indirect comparison of centanafadine versus lisdexamfetamine, methylphenidate and atomoxetine in adults with attention-deficit/hyperactivity disorder: long-term safety and efficacy. Journal of Comparative Effectiveness Research. 2024:e240089. doi:10.57264/cer-2024-0089. Original commentary on the supplied paper.
Leave a Reply