Why does reta look weaker for weight loss in this study?

keangkong

keangkong

Explorer
Joined
Sep 2, 2024
Messages
23
Reaction score
0
Location
Ryongsong Residence, Pyongyang, North Korea
In this research paper, reta appears less effective for weight loss than a number of other drugs, tirzepatide included — does anyone know why? Nong, et al. (2026). Comparative effects of drugs for adults with overweight or obesity - systematic review and network meta-analysis. BMJ, 394, e372161. doi.org/10.1136/bmj-2026-372161. Most of what I've come across regarding reta — research results as well as anecdotes — points the other way. An accompanying editorial, Gasoyan & Rothberg (2026). Comparing benefits and harms of obesity drugs. BMJ 394, e725349, doi:10.1136/bmj-2026-725349, points out that the Nong systematic review and network analysis found no weight loss superiority for subcutaneous semaglutide over oral semaglutide, and that subcutaneous semaglutide has been linked to better overall health outcomes than the other weight loss drugs.
 
Not something I like admitting, but these days I'd put the typical research paper on roughly the same reliability footing as the typical internet blog. Fancy graphs, tables, and a lengthy reference list may make them look nicer, yet competence is frequently missing. That goes double for analysis-type papers, where nothing was actually studied and the authors are just compiling and tabulating findings from other papers into a summary.

I've lost count of the papers I've come across where what the results claim clashes with the data shown, or where the researchers committed truly basic calculation mistakes that nobody ever caught.
 
Is what that paper reports consistent with what we're seeing on here? Retatrutide has me down 70 pounds, and I've managed to keep it off so far. 7 months.
 
When it comes to the Reta, the available data isn't sufficient to reach a conclusion or to set it against the Tirz.

Take a look at this reply under the opening article of this thread:




Comparative effects of drugs for adults with overweight or obesity: systematic review and network meta-analysis − BMJ



https://www.bmj.com/content/394/bmj-2026-372161 Published 08 July 2026

051f0e2f5b7f6ffbb4a2f10b404546ba44b6197464dfbacec1b043fb189c69a3.png



GLP1Chat.com

Additional data will come in time.

Getting input from the people on this forum would be a valuable thing.

Personally, 14% came off with Reta over 3 months.
 
Each person responds differently. For 12 months, Sema gave me nothing except a lighter bank account. Whatever small appetite reduction it produced, I could still eat and drink alcohol straight through it. With Reta, the experience was completely unlike that, and I see it as a modern miracle. My weight is down 78 lbs, equal to 29% BW, and it has stayed there. I reached that in 36 weeks, which is quicker than what the trials report. Any day, I would choose real-life experience over a paper.
 
Chili777 said:

Each person responds differently. For 12 months, Sema gave me nothing except a lighter bank account. Whatever small appetite reduction it produced, I could still eat and drink alcohol straight through it. With Reta, the experience was completely unlike that, and I see it as a modern miracle. My weight is down 78 lbs, equal to 29% BW, and it has stayed there. I reached that in 36 weeks, which is quicker than what the trials report. Any day, I would choose real-life experience over a paper.
Click to expand...
In 28 weeks I dropped 55 lbs — that works out to 2 lbs per week on average. Your average came to 2.1 lbs per week. That's great.
 
I spent 3 months on semaglutide and reached the full dose, but my weight didn't budge at all. My wife dropped roughly 12 pounds on it before it quit helping. A couple of years afterward, I gave reta a try and I'm down 35 lbs. My wife also used reta and lost 20, then moved to tirz because reta gave her joint pain; on tirz she has shed another 20lbs. There must be something off in that trial for reta to perform this poorly.
 
I haven't gone back to re-read it carefully, so I can't be 100% certain, but my recollection is that the data they used didn't include the latest reta trial findings — the ones reporting 29% weight loss at 12mg over 72 weeks. Also, if they're pooling every study together without properly adjusting for how long each one ran, then older, shorter trials would naturally show smaller weight loss when they only lasted 3-6 months, while there's plenty of tirz data spanning 1 year or more to pull from.
 
Reta actually works….. I’ve got roughly 30lbs to drop altogether, yet during the initial 4 weeks on a low dose I’ve already shed 9 of those pounds. This is pretty incredible.
 
tubby said:

Not something I like admitting, but these days I'd put the typical research paper on roughly the same reliability footing as the typical internet blog. Fancy graphs, tables, and a lengthy reference list may make them look nicer, yet competence is frequently missing. That goes double for analysis-type papers, where nothing was actually studied and the authors are just compiling and tabulating findings from other papers into a summary.

I've lost count of the papers I've come across where what the results claim clashes with the data shown, or where the researchers committed truly basic calculation mistakes that nobody ever caught.
Click to expand...
Plenty of journals operate on a pay-to-publish basis, yet checking the “H index” — a metric that weighs an author’s total output against how often those works get cited — offers a solid signal of how trustworthy the article or author is. As for the journal itself, its Q-index reveals how respected and dependable it is.

Meta-analyses genuinely have value and demand substantial statistical expertise to carry out. As with most things, though, a critical eye helps. Those conducting the analysis ought to spell out the shortcomings of the studies they include and explain the reasoning behind their study selection. Because statistics can be bent to support virtually any position, examining their methods and analysis can shed light (or simply feeding that section to AI for an explanation may prove handy).

Like any publication, peer-reviewed papers can have flaws, but they remain the gold standard in scientific research and face far more hurdles before publication than a blog does. That said, there are definitely EXCELLENT blogs out there!
 
Alright, let's dig in. This one's mildly irritating. Below is how I make sense of it.

I'd bet plenty of you looked at that and thought, seriously? I know I did. In the paper, retatrutide sits at -13.1% weight loss, placed under tirzepatide and CagriSema. From the trials, everyone knows retatrutide reached roughly 24%. That's the largest figure anywhere in the obesity-drug space. So why would the strongest drug land in the middle of the table?

In retatrutide's phase 2 obesity trial (Jastreboff et al., NEJM 2023; "Triple-Hormone-Receptor Agonist Retatrutide for Obesity," NCT04881760), the 12 mg dose gave -24.2% weight loss across 48 weeks, and it hadn't flattened out yet. Tirzepatide's flagship obesity trial (Jastreboff/Garvey et al., NEJM 2022; SURMOUNT-1, NCT04184622) gave -20.9% at 15 mg across 72 weeks. Put those 2 trials side by side: retatrutide (~24% in 48 weeks) is ahead of tirzepatide (~21% in 72 weeks), not behind it, which is the opposite of how the paper ranks them, right?

So what kind of statistical f*ckery is being done to give these results?

Firstly: why every drug's number is lower than its obesity-trial headline:

  1. Trial pooling that includes diabetes. Every drug's estimate merges its obesity trials with its type 2 diabetes trials, and people with diabetes shed noticeably less weight on these drugs, so the pooled figure ends up under the obesity-only figure.
  2. Normalising to one year. The flagship obesity trials mostly ran 68–72 weeks, whereas this paper recalculates every drug to a shared 52-week mark, which cuts the longer trials down as well.
(That's why tirzepatide shows ~15% instead of the ~21% from its SURMOUNT-1 obesity result, and CagriSema ~15% instead of the ~20% from REDEFINE-1.)

The retatrutide issue (frequentist vs Bayesian):

The paper analysed things two different ways. Under the authors' own frequentist analysis, retatrutide lands at -18.4%, the highest weight loss of all 19 drugs (that's the table in data Appendix 4, yes you have to read the appendices, boring!).

Under their Bayesian analysis, reta falls to -13.1%, slipping beneath tirzepatide, CagriSema, ecnoglutide and mazdutide, and the Bayesian figure is the one that made it into the paper. Crucially, retatrutide is the only drug that the change in statistical analysis method moves. Every other drug shifts by half a point or less between the two methods, while retatrutide alone drops 5.3 points, which is exactly what takes it from first place to fifth.

The likely reason is that retatrutide is the one drug that is both the most extreme estimate and the thinnest evidence (just two small trials), and Bayesian models are built to pull exactly that kind of uncertain outlier back toward the pack, but thats a guess, not something the paper states. The one thing that needs no interpretation: the same data, analysed two ways, ranks retatrutide either the best or mid-table, and they used the mid-table version.

DrugFrequentist (Appendix 4, §6.1.1)Bayesian headline (standard dose)ChangeRetatrutide-18.4% (highest of all)-13.1%-5.3Tirzepatide-15.4%-14.9%-0.5Ecnoglutide-15.1%--14.6%-0.5CagriSema-14.7%-14.8%+0.1Mazdutide-13.3%-13.2%-0.1Orforglipron-10.4%-9.9%-0.5Oral semaglutide-10.4%-10.9%+0.5Survodutide-9.5%-10.2%+0.7

The wierd contradiction with reta showing the largest waist-circumference reduction of any drug in the analysis, -14.4 cm, beating tirzepatide's -11.0 cm. Waist and weight move together. A drug is VERY unlikely to be able to shrink waistlines the most while losing a mid-table amount of weight. This again points to something going on with with statistics. I have no idea what, im not clever enough and dont have the time to figure it out!

Hope that helps!
 
JP.MOTM said:

Alright, let's dig in. This one's mildly irritating. Below is how I make sense of it.

I'd bet plenty of you looked at that and thought, seriously? I know I did. In the paper, retatrutide sits at -13.1% weight loss, placed under tirzepatide and CagriSema. From the trials, everyone knows retatrutide reached roughly 24%. That's the largest figure anywhere in the obesity-drug space. So why would the strongest drug land in the middle of the table?

In retatrutide's phase 2 obesity trial (Jastreboff et al., NEJM 2023; "Triple-Hormone-Receptor Agonist Retatrutide for Obesity," NCT04881760), the 12 mg dose gave -24.2% weight loss across 48 weeks, and it hadn't flattened out yet. Tirzepatide's flagship obesity trial (Jastreboff/Garvey et al., NEJM 2022; SURMOUNT-1, NCT04184622) gave -20.9% at 15 mg across 72 weeks. Put those 2 trials side by side: retatrutide (~24% in 48 weeks) is ahead of tirzepatide (~21% in 72 weeks), not behind it, which is the opposite of how the paper ranks them, right?

So what kind of statistical f*ckery is being done to give these results?

Firstly: why every drug's number is lower than its obesity-trial headline:

  1. Trial pooling that includes diabetes. Every drug's estimate merges its obesity trials with its type 2 diabetes trials, and people with diabetes shed noticeably less weight on these drugs, so the pooled figure ends up under the obesity-only figure.
  2. Normalising to one year. The flagship obesity trials mostly ran 68–72 weeks, whereas this paper recalculates every drug to a shared 52-week mark, which cuts the longer trials down as well.
(That's why tirzepatide shows ~15% instead of the ~21% from its SURMOUNT-1 obesity result, and CagriSema ~15% instead of the ~20% from REDEFINE-1.)

The retatrutide issue (frequentist vs Bayesian):

The paper analysed things two different ways. Under the authors' own frequentist analysis, retatrutide lands at -18.4%, the highest weight loss of all 19 drugs (that's the table in data Appendix 4, yes you have to read the appendices, boring!).

Under their Bayesian analysis, reta falls to -13.1%, slipping beneath tirzepatide, CagriSema, ecnoglutide and mazdutide, and the Bayesian figure is the one that made it into the paper. Crucially, retatrutide is the only drug that the change in statistical analysis method moves. Every other drug shifts by half a point or less between the two methods, while retatrutide alone drops 5.3 points, which is exactly what takes it from first place to fifth.

The likely reason is that retatrutide is the one drug that is both the most extreme estimate and the thinnest evidence (just two small trials), and Bayesian models are built to pull exactly that kind of uncertain outlier back toward the pack, but thats a guess, not something the paper states. The one thing that needs no interpretation: the same data, analysed two ways, ranks retatrutide either the best or mid-table, and they used the mid-table version.

DrugFrequentist (Appendix 4, §6.1.1)Bayesian headline (standard dose)ChangeRetatrutide-18.4% (highest of all)-13.1%-5.3Tirzepatide-15.4%-14.9%-0.5Ecnoglutide-15.1%--14.6%-0.5CagriSema-14.7%-14.8%+0.1Mazdutide-13.3%-13.2%-0.1Orforglipron-10.4%-9.9%-0.5Oral semaglutide-10.4%-10.9%+0.5Survodutide-9.5%-10.2%+0.7

The wierd contradiction with reta showing the largest waist-circumference reduction of any drug in the analysis, -14.4 cm, beating tirzepatide's -11.0 cm. Waist and weight move together. A drug is VERY unlikely to be able to shrink waistlines the most while losing a mid-table amount of weight. This again points to something going on with with statistics. I have no idea what, im not clever enough and dont have the time to figure it out!

Hope that helps!
Click to expand...
What is the point, though, of running a "Bayesian" analysis in a paper that is set up like a meta-analysis? If the authors really had rare, specialized expertise in obesity medicine, then maybe they would be in a position to outperform the original "frequentist" findings. But the truth is that they bring nothing special in terms of insight or credentials—nothing beyond what you or I have—that would let them rework those results.

If we were talking about the leading authorities in this area, and the paper were going to be treated as a definitive reference, then I suppose there might be some logic to massaging the numbers and using "Bayesian" as the cover story (perhaps Lilly's rivals would gain financially). Still, that scenario seems very far-fetched. By now retatrutide has been talked about and hyped so heavily that anyone reading this paper will simply walk away thinking the authors botched their analysis.

That leaves me baffled about what motive there could be for publishing such a strange result while keeping the reasoning behind it hidden.
 
tubby said:

JP.MOTM said:

Alright, let's dig in. This one's mildly irritating. Below is how I make sense of it.

I'd bet plenty of you looked at that and thought, seriously? I know I did. In the paper, retatrutide sits at -13.1% weight loss, placed under tirzepatide and CagriSema. From the trials, everyone knows retatrutide reached roughly 24%. That's the largest figure anywhere in the obesity-drug space. So why would the strongest drug land in the middle of the table?

In retatrutide's phase 2 obesity trial (Jastreboff et al., NEJM 2023; "Triple-Hormone-Receptor Agonist Retatrutide for Obesity," NCT04881760), the 12 mg dose gave -24.2% weight loss across 48 weeks, and it hadn't flattened out yet. Tirzepatide's flagship obesity trial (Jastreboff/Garvey et al., NEJM 2022; SURMOUNT-1, NCT04184622) gave -20.9% at 15 mg across 72 weeks. Put those 2 trials side by side: retatrutide (~24% in 48 weeks) is ahead of tirzepatide (~21% in 72 weeks), not behind it, which is the opposite of how the paper ranks them, right?

So what kind of statistical f*ckery is being done to give these results?

Firstly: why every drug's number is lower than its obesity-trial headline:

  1. Trial pooling that includes diabetes. Every drug's estimate merges its obesity trials with its type 2 diabetes trials, and people with diabetes shed noticeably less weight on these drugs, so the pooled figure ends up under the obesity-only figure.
  2. Normalising to one year. The flagship obesity trials mostly ran 68–72 weeks, whereas this paper recalculates every drug to a shared 52-week mark, which cuts the longer trials down as well.
(That's why tirzepatide shows ~15% instead of the ~21% from its SURMOUNT-1 obesity result, and CagriSema ~15% instead of the ~20% from REDEFINE-1.)

The retatrutide issue (frequentist vs Bayesian):

The paper analysed things two different ways. Under the authors' own frequentist analysis, retatrutide lands at -18.4%, the highest weight loss of all 19 drugs (that's the table in data Appendix 4, yes you have to read the appendices, boring!).

Under their Bayesian analysis, reta falls to -13.1%, slipping beneath tirzepatide, CagriSema, ecnoglutide and mazdutide, and the Bayesian figure is the one that made it into the paper. Crucially, retatrutide is the only drug that the change in statistical analysis method moves. Every other drug shifts by half a point or less between the two methods, while retatrutide alone drops 5.3 points, which is exactly what takes it from first place to fifth.

The likely reason is that retatrutide is the one drug that is both the most extreme estimate and the thinnest evidence (just two small trials), and Bayesian models are built to pull exactly that kind of uncertain outlier back toward the pack, but thats a guess, not something the paper states. The one thing that needs no interpretation: the same data, analysed two ways, ranks retatrutide either the best or mid-table, and they used the mid-table version.

DrugFrequentist (Appendix 4, §6.1.1)Bayesian headline (standard dose)ChangeRetatrutide-18.4% (highest of all)-13.1%-5.3Tirzepatide-15.4%-14.9%-0.5Ecnoglutide-15.1%--14.6%-0.5CagriSema-14.7%-14.8%+0.1Mazdutide-13.3%-13.2%-0.1Orforglipron-10.4%-9.9%-0.5Oral semaglutide-10.4%-10.9%+0.5Survodutide-9.5%-10.2%+0.7

The wierd contradiction with reta showing the largest waist-circumference reduction of any drug in the analysis, -14.4 cm, beating tirzepatide's -11.0 cm. Waist and weight move together. A drug is VERY unlikely to be able to shrink waistlines the most while losing a mid-table amount of weight. This again points to something going on with with statistics. I have no idea what, im not clever enough and dont have the time to figure it out!

Hope that helps!
Click to expand...
What is the point, though, of running a "Bayesian" analysis in a paper that is set up like a meta-analysis? If the authors really had rare, specialized expertise in obesity medicine, then maybe they would be in a position to outperform the original "frequentist" findings. But the truth is that they bring nothing special in terms of insight or credentials—nothing beyond what you or I have—that would let them rework those results.

If we were talking about the leading authorities in this area, and the paper were going to be treated as a definitive reference, then I suppose there might be some logic to massaging the numbers and using "Bayesian" as the cover story (perhaps Lilly's rivals would gain financially). Still, that scenario seems very far-fetched. By now retatrutide has been talked about and hyped so heavily that anyone reading this paper will simply walk away thinking the authors botched their analysis.

That leaves me baffled about what motive there could be for publishing such a strange result while keeping the reasoning behind it hidden.
Click to expand...

I share that sentiment—honestly, I have no idea. My statistics background isn't deep enough to determine when one model should be used over another. Still, it stands out that this produces a fairly obvious anomaly: it turns up in the % weight loss figures, yet not in the waist circumference change. That inconsistency is strange.

Within the main text, they do discuss reta and describe the newer agents as low-certainty, noting they "may produce similar or greater reductions". Even so, I completely agree it looks like a poor choice to let your headline figure show reta doing badly according to your bayesian analysis, while the frequentist analysis appears to reflect the situation more suitably. And then to tuck a very flimsy caveat away somewhere in the writing.

Wouldn't it have been better to hold off a few more months until the reta phase 3 data is fully published? At that point I'm confident this wouldn't be a problem at all, and the findings would come out the way everyone expects. Who knows!
 
lessthanhalf said:

I haven't gone back to re-read it carefully, so I can't be 100% certain, but my recollection is that the data they used didn't include the latest reta trial findings — the ones reporting 29% weight loss at 12mg over 72 weeks. Also, if they're pooling every study together without properly adjusting for how long each one ran, then older, shorter trials would naturally show smaller weight loss when they only lasted 3-6 months, while there's plenty of tirz data spanning 1 year or more to pull from.
Click to expand...
Claude.ai also said the newest data wasn't included.
 
Back
Top