The Old Cheap Drug Keeps Winning

A network meta-analysis of 41 trials in children and adolescents found semaglutide producing the largest reduction in BMI. Its authors nonetheless conclude that phentermine–topiramate — an older, cheaper combination — may be the optimal addition to lifestyle treatment.^[1] The numbers underneath explain why, and they contain a discordance nobody accounts for.

The two ways of measuring disagree

The review pooled 41 randomized trials covering 3,923 children and adolescents, comparing drug treatments added to lifestyle modification against lifestyle modification alone. It reported both the average reduction and how many additional people crossed meaningful thresholds.

The larger average effect produced fewer people crossing the higher threshold.

At the 10% threshold, phentermine–topiramate produced nearly twice as many additional responders as semaglutide — 734 against 399 — while its mean reduction was two percentage points smaller.

What that discordance implies, and what nobody explains

That is the most plausible reading and it is not established. The review does not explain the pattern, and network meta-analyses can also generate discordant continuous and binary estimates for technical reasons — different studies contributing to different outcomes, different follow-up lengths, different definitions of response. We are reporting the discordance rather than resolving it.

It matters practically because the two measures answer different questions. “How much will the average child’s BMI fall” and “what are the chances this child gets a meaningful result” are not the same, and a family is usually asking the second.

The same drug came out ahead in adults, for a different reason

This is the second independent analysis to place phentermine–topiramate above semaglutide on a measure that is not raw efficacy. The American College of Physicians’ economic review found semaglutide to be low value compared with phentermine–topiramate in adults — covered in we know what they do, not what they’re worth.

Different populations, different methods, different endpoints, converging on the same underrated option. Neither analysis says the older drug produces more weight loss. Both say that once tolerability or cost enters the comparison, the ranking changes.

The pediatric reviewers make the tolerability point explicitly, citing it as the reason for their conclusion. Gastrointestinal side effects and discontinuation are the recurring cost of the GLP-1 class throughout this register, and in adolescents — where treatment may run for years and adherence is harder — that cost weighs more.

Why this is not a recommendation

• The outcome is BMI, not health. These are surrogate measures in a group whose outcomes lie decades away — what changed in the teenagers’ bloodwork.
• Long-term safety in growing bodies is unstudied for every drug here.
• Only semaglutide is FDA-approved for adolescent weight management, which constrains the real choice regardless of what a meta-analysis ranks.
• Trials in this population are small — 3,923 participants across 41 trials averages fewer than a hundred each.

For what these drugs do to weight in adolescents specifically, see GLP-1 drugs in teenagers; for the diabetes question in the same age group, tirzepatide in ten- to seventeen-year-olds.