The idea behind this study is genuinely useful. If different drugs act on different aspects of eating — the pleasure of food, eating in response to emotion, sheer quantity, compulsive patterns — then matching the drug to the pattern would be a real advance, and a more meaningful way to choose than the brand comparisons in the super-responder brand comparison. A clinic in São Paulo tried to map it in adults treated for six months (n = 66)[1].
The reported effects are striking. Emotional eating fell with a standardized effect size of 2.04 on naltrexone-bupropion, 1.77 on tirzepatide, 1.54 on topiramate and 1.52 on semaglutide. Hedonic eating fell at 2.06 with tirzepatide. In most fields, effect sizes above 0.8 are called large; several of these are more than double that.
The selection compounds it. Analyses focused on early responders — people who had already lost at least 5% of their body weight by three months. So every behavioral change recorded belongs to someone for whom the drug was already working. That cannot describe what a drug does in general, and it certainly cannot identify who will respond, which is what the title's promise of precision pharmacotherapy implies.
There is also no control group and no randomization. These are paired before-and-after scores on a self-report scale, collected from people who had lost noticeable weight and knew it. Recalling your previous eating as worse than it was is the same problem that governs food noise measured by memory.
None of that makes the hypothesis wrong. Matching a drug to an eating pattern is a sensible thing to test, and if it holds up it would change how these products are chosen and sold. What it needs is a randomized comparison with hundreds of participants per arm, which does not exist — so for now the useful question to ask a prescriber is which pattern describes you, not which number in a heatmap is biggest. How much real-world response actually varies is documented in responding is not one thing and slow responders at week eight.