24.2% vs 28.7%: how are people reading Retatrutide Phase 3?

Efficacy estimand is the key. It’s not the same as a simple average of everyone who enrolled.

Agreed, but the 28.7% was company-reported arcoss four Phase 3 trials, so it’s likely a high dose or completer-heavy slice. Not comparable.

I’ve been in a similar rabbit hole. The CTgov records for the small completed trials you listed are frustrating because enrollment is there but no outcome tables. That doesn’t mean nothing happened; it often just means results aren’t posted yet. The 24.2% Phase 2 figure at 48 weeks is the cleanest number we have because it’s tied to a defined estimand. The 28.7% is company-selected language from four trials. I’d wait for peer review before treating them as one curve.

What would convince me is seeing the actual dose groups, not just the top-line up-to number.

Same. Up to means someone in some arm hit it. Doesn’t tell us the distribution.

Endpoint definitions are where the bodies are buried.

Population might matter more than endpoint definitions here. The 445 knee OA trial has a different baseline; I wouldn’t expect identical weight loss.

Week 14 here and my scale did the opposite of the headlines: down 1.1 lb/week for the first 10 weeks, then flat for 5 straight weeks while my waist tape dropped 2 inches. That's why I don't treat either percentage as a personal forecast. A trial mean is built from people who stayed on it the whole time; my body doesn't read press releases. 24.2 vs 28.7 is a population stat, not a promise to me.

48 weeks is where the framing deserves some pushback: 24.2% is an efficacy estimand, which basically means "if participants stay on it as assigned" — the trial's version of a perfect patient. So 28.7% in phase 3 isn't shocking, it's the same story with a bigger sample and a different analysis plan.

The part people forget is weekends. Friday takeout and two glasses of wine, Sunday the scale is up 3.4 lb, gone by Wednesday. Ten weeks of that and the real trend was 0.7 lb/week down, not the 2 lb/week those Monday weigh-ins implied. Which is why the advice that keeps circulating is: log daily and only read the seven-day average.

The knee OA one is a different animal — 445 people, and the endpoints there are pain and function scores, so any weight change is likely secondary and not comparable to the obesity trials at all. What I want to know is whether the 1946-person and 2335-person studies used the same estimand; if they didn't, stacking their headlines against phase 2 is just noise. Conference abstract or journal, do you think?

Month three was my mess: I added 45 minutes of walking five days a week and started sleeping 7.5 hours instead of 5.5, felt amazing, and the scale went UP 2.1 lb and sat there for twelve days. I nearly quit tracking entirely. Then it dropped 6.4 lb in nine days and my jeans went from a 36 to a 34. Water and inflammation noise, not fat. Lesson I keep relearning: 12 days is nothing on a 48-week timeline, so a phse 3 average is even less useful for any one person. I only trust the monthly waist tape now, same spot, first thing, empty.

@nathan_96 said: 12 days is nothing on a 48-week timeline

The number that actually makes me squint is dropout and missing-data handling, not the 4.5-point gap. If the Phase 3 headline leans on people who stayed the course, 28.7% is the best-case version. I track a 7-day average now because one salty weekend can swing me 2.8 lb, so a single timepoint tells me almost nothing.

Peer review or it’s just a highlight reel. I’ve been burned by pretty graphs before.

I quit chasing the trial percentages and bought a cheap tailor’s tape. My scale has been flat for 19 days, but my waist is down 1.25 inches snce week 10, and my jeans finally don’t leave red marks. That’s the weird part—non-scale wins sometimes move first while the headline number stalls.

The CTgov enrollment sizes tell you the trials are huge, but my personal n=1 says stress and sleep muddy everything. A rough work week with 5.5 hours sleep and two takeout dinners had me up 3.1 lb by Friday, gone by Tuesday. So when I see 24 vs 28, I think: whose week? whose water? whose life?

I quit weighing daily around week 6 because the water swings were louder than the trend. Switched to Sunday morning waist tape and one pair of jeans. Lost 2.5 inches before the scale moved 4 lb. So when I see 24 vs 28, I mostly wonder whose mornings those averages came from. Mine would look totally different on a Monday after salt and a bad night’s sleep.

Honestly, 28.7% doesn’t mean much to me until dropout rates show up; that’s the number that says who could actually stay on it.

Weekends are the real stress test for anyone tracking food and sleep: five decent days, two feral ones, and Monday's scale acts like we just met.

Worth remembering what the "up to 28.7%" figure actually describes. The curve most people end up comparing themselves to looks more like: 8 lb down by week 8, 14 by week 20, then flat for six weeks even with lifting and 10k steps most days. The next drop came after a month of actually sleeping 7.5 hours, not after chasing the number harder. So "up to 28.7%" reads as the best-case timeline, not what happens at the 48-week mark for most humans.

Two belt notches before the scale moved a single pound — that's my actual data, and it's why I stopped refreshing trial headlines. 24.2 to 28.7 is a 4.5-point gap on a graph; my jeans don't care about estimands. Weird part: I gained 1.5 lb the week I finally fit back into old work pants, because I'd started lifting again. The scale is the worst narrator in the house.

A 32-person cohort is a group chat, not a data point.

I stopped weighing daily and switched to a Sunday average, and it reframed everything: my "plateau" around week 14 was really three weeks of 5-hour sleep and a work deadline, not a stall. Same food, same walking, scale up 2.4 lb, then gone in ten days once I slept 7+ again. Counterintuitive bit: the week I added a 25-minute after-dinner walk instead of cutting another 200 calories, my weekly average dropped 1.1 lb vs 0.4 the week before. Waist went 39.5 to 37 over about six weeks. Now I log sleep next to weight and the pattern is embarrassingly obvious — anyone else's worst weeks line up with bad sleep more than bad food?

Not one of those completed trials has posted a topline. That silence says more tan the 28.7.

Honestly the number I trust now is my belt notch and a tape measure at the navel: I went from 38.5 to 36.0 inches between week 4 and week 12, while the scale sat within 3 lb for three straight weeks and I was ready to call it a stall. Then I fixed my sleep from 5.5 to 7 hours and stopped eating dinner at 9:30, and 7 lb came off in about 10 days—so the scale was just late, not broken. Anyone else see the tape move first? It’s making me way less fussed about whether the real-world result looks like 24% or 28%.

Not a scientist, but the timepoint difference is what jumps out at me. Phase 2 was 48 weeks; if the Phase 3 obesity trials ran to 68 weeks, some of that 24.2% to 28.7% gap is just more runway. Plus “efficacy estimand” vs company topline can flatter things. Did the Phase 3 protocols use a different estimand or a maintenance run-in? That’s what I’d want before comparing. My local pharmacist is already getting insurance questions about whether a 28.7% headline changes prior auth—doubt it, but she said folks are asking. I’ve been tracking protein and steps instead of chasing the scale.

Agree the estimand label matters as much as the number. Phase 2’s 24.2% was efficacy estimand at 48 weeks; if the phase 3 “up to 28.7%” uses a different estimand, week, or baseline A1c/diabetes mix, the gap could shrink or flip. I’d want the baseline table and discontinuation rates before getting excited. Did the company release those anywhere, or just the top-line percentage? Also curious whether the phase 3 included a lifestyle/placebo arm with similar support, because behavioral support alone can move several points. Not trying to dunk on it—just can’t compare without the denominators.

Following the slope, not the headline. I log daily weights in a CSV, plot a 7-day rolling mean, and compare kg lost per week; percent changes get weird when baseline BMI differs. Sharp question: what were baseline weights/BMI and diabetes status in NCT05882045 vs NCT05929066? A 28.7% drop from a heavier, more insulin-resistant cohort isn’t automatically better than 24.2% from a leaner one—sometimes it’s the same absolute kg. Also, the knee OA and postmenopausal arms likely have different baseline body comp. Anyone seen a demographic table yet, or just top-line press release?

what I’m not seeing discussed: baseline BMI and prior GLP-1 exposure. If Phase 3 has more heavier, treatment-naive people, 28.7% could be

I’d want the discontinuation and rescue/escape rates before comparing. In my own spreadsheet, my “stayed on protocol” trendline and my intention-to-treat trendline diverged by ~4 percentage points by month 12—a few bad weeks dragged the average down. If the 28.7% comes from completers or a later time point, it’s basically a different denominator. Do we know the Week 48 ITT/intercurrent estimate for the Phase 3 cohorts yet, or is 28.7% from longer follow-up? That’

As a lifter, the number I want from those Phase 3s isn’t total weight loss—it’s lean mass. A 28.7% scale drop looks great, but if a chunk is muscle, recomp and long-term maintenance get ugly. Do any of the protocols have DEXA or functional sub-studies (grip, sit-to-stand, 5x chair)? I log waist, hips, lifts, and sleep, and my scale can stall while my waist drops—so total-weight endpoints feel incomplete. Also curious whether trial sites gave any standardized protein/resistance-training guidance, because that could swing body comp way more than the headline percentage.

Do we actually have the week mark for that 28.7%? Phase 2’s 24.2% was at 48 weeks. If Phase 3 is quoting week 80 or 104, that’s a different exposure duration

I’ve got two kids and chaotic 6pm dinners, so the number that would actually sway me is a functional one. If the knee OA arm is 445 people, are they reporting pain/function scores or just scale weight? Same for the postmenopausal cohort — sleep, energy, stairs without aching. I log a dumb nightly note: protein at breakfast Y/N, snack drawer restocked with pre-portioned stuff, 6k steps. Weight’s all over the place, but those notes show whether the routine survives. Sharp Q: do any Phase 3 readouts break out weight change by whether people kept

late-night gamer here, so my big confounder is 1am soda/snack creep. but for those phase 3 numbers, i’d want the baseline mix before comparing: what % had tpe 2 diabetes, what % had prior glp-1 exposure, and baseline bmi. phase 2 and those big nct enrollments can have very different populations, and prior glp-1 users often lose less / plateau differently. also is 28.7% from completers or a treatment-policy/ITT estimand? that’s the apples-to-oranges thing i’d chase. has anyone seen the discontinuation rate due to GI stuff? that can make the headline number look prettier.

For what it's worth, i track waist, resting HR, sleep, and how clothes fit—not just scale—so the estimand/population question matters to me. Do we know if that 28.7% comes from a similar baseline BMI and diabetes status range as Phase 2, or could the Phase 3 pools skew higher? And was it efficacy estimand or treatment-regimen? That alone can move the headline a lot. I’m less interested in the top-line number than whether the folks who’d actually qualify look like the trial group.

Following this closely. One thing I haven’t seen: median vs mean, plus the % of people hitting ≥20% and ≥25% loss. A 28.7% mean can be pulled up by a few super-responders, and for real-life expectations I care more about the middle of the bell curve and how many stayed on treatment long enough to get there. Did the Phase 3 releases include those responder percentages or just the top-line mean? Also curious how many dropped out for GI stuff, since that changes how I read any headline. My own chart is noisy enough that I track 4-week trends, not single weeks.

Honestly, before comparing those numbers I’d want the baseline table and discontinuation rate. If the Phase 3 population had a different BMI range, more prediabetes/T2D, or less structured lifestyle support, the percentage can drift even with the same med effect. I log weight, waist, and my main lifts, and my own recomp stalled when sleep dropped under 7h—context matters. Do we know if that 28.7% is completers-only or a prespecified estimand, and what the placebo/background arm did? That’s the apples-to-apples piece I’m missing.

wait, isn’t the phase 3 28.7% number at 68 weeks while phase 2 24.2% was 48 weeks? if so some of that gap is just extra time on it, not magic. i’d rather see the off-ramp data: weight regained 6-12 months after stopping, and whether people kept the soda-swap habits. my own swap was easy (sparkling water + flavor drops), but late-night raid chips are undefeated. anyone seen if those phase 3 trials have a maintenance/withdrawal sub-study? that’s the number i’d actually plan around.

i’m on a GLP-1 and 14 months in, stuck in the same 3-lb window since March. So my question with these Phase 3 numbers is body composition: did they run a DEXA or MRI substudy, or is 28.7% just total scale weight? If a chunk is lean mass, the headline feels less useful. Also, what happened to the people who stopped early—did they report regain curves? I’d take a solid 20% fat loss over a 28% number that includes muscle and water. Anyone seen whether the protocol even measured that?

Long-timer here. The first thing I’d untangle is timepoint. Phase 2’s 24.2% was at 48 weeks; if the 28.7% comes from a longer Phase 3 window, that’s partly just more runway, not a different drug. We chased this exact ghost in old GLP-1 threads when week 72 got compared to week 56. What I want is the Phase 3 week-48 interim, matched population/estimand, plus the maintenance/off-ramp curve after treatment stops. The peak topline is the least interesting part—it’s the plateau and regain slope that matter in real life. Has anyone seen whether they’ve posted a like-for-like 48-week cut?

Three-year regular here. The thing I’d dig into is baseline BMI and diabetes status. Earlier GLP-1 threads taught me a higher starting BMI can make the same absolute pounds look like a bigger percentage. Also, 28.7% might be a completer or specific-arm number, not the whole enrolled group. The Phase 2 efficacy estimand already handles missing data differently than a plain ITT, so it’s still not apples to apples. Does anyone know the Phase 3 discontinuation rates and whether treatment ran longer than 48 weeks? Those two would explain a lot. My own tracking is just protein, steps, and waist monthly—scale still lies week to week.

Timepoint is the sleeper. Phase 2’s 24.2% was at 48 weeks. The 28.7% is probably 68+ weeks. More runway, more loss. Not apples to apples.

Phase 3 also pooled different groups—knee OA, postmenopausal, Chinese cohort. “Up to” math gets hypey.

I want the discontinuation rate and if they used efficacy estimand again. If dropouts vanish from the denominator, the headline inflates. Night shift taught me adherence is the real variable. What was completion?

Contrarian spreadsheet take: the 4.5-point gap smells like missing-data handling, not biology. If Phase 3 leaned on an on-treatment/completer estimand or censored people at discontinuation, the mean gets flattered. I’d want the discontinuation curve, the actual week-48 weight availability, and whether rescue meds were counted. My own raw weekly weigh-ins vs 4-week EMA can differ ~2

Honestly the total-weight estimand is the least interesting part to me. If the Phase 3 delta is partly more lean-mass loss, 28.7% could look better on paper while recomp-wise it’s worse. Any DEXA, grip strength, or 6-min walk sub-studies in those? The knee OA cohort especially—scale weight and pain/function don’t always track together. I lift 4x and track top-set numbers + progress photos; scale can stall for weeks while top sets climb. So I’d want body-comp and functional endpoints, not just bigger weight-loss percentages. Is anyone seeing those buried in CTgov secondary outcomes?

I’ve got two kids and a snack drawer that’s basically a boss fight. What I’d want to know is whether those Phase 3 numbers are also at 48 weeks, or longer. If 28.7% is at 68 weeks, that’s not apples to apples either. Also, the knee OA cohort (445) might be older or less mobile on average, which could pull the mean around. Anyone know the week-48 landmark?

Long-timer here. The number that matters to me isn't the headline mean; it's duration and baseline BMI. If the Phase 3 readout is longer than 48 weeks, a few extra points isn't shocking. Like the old SURMOUNT threads, "up to" usually means one arm in completers, not the whole enrolled cohort. I'd want the treatment-policy estimand and discontinuation rate next to it. Also—has anyone seen if these Phase 3s include a maintenance/withdrawal phase? That's the part I care about once the scale starts moving.

Road-warrior check-in: I can gain 4 lb in a two-day airport/hotel stretch just fom sodium and weird sleep, so I’m skeptical of any single scale number. What I haven’t seen in this thread: how long were people actually on it in the Phase 3 trials vs the 48-week Phase 2? If Phase 3 ran to 68–80 weeks, isn’t some of 24.2%→28.7% just more time on treatment, not a better result? Also curious what the weight-loss plateau looked like in the extension.

Not a stats person, just a dorm rat who eats a lot of eggs. The thing I’d want to see is baseline activity/mobility, not just BMI. The knee OA cohort especially—if your knees hurt, daily steps are low, and that can change how any weight-loss curve looks. Also Phase 2 vs Phase 3 enrollment size/ge

Yeah, the estimand is doing a lot of work here. If Phase 3’s “up to 28.7%” is completers or on-treatment, it’s not comparable to the Phase 2 efficacy estimand. I’d want discontinuation rates and baseline BMI by arm—a 4.5-point gap can easily be dropout differential. Also, do we know prior incretin exposure in those cohorts? That’s a confounder I haven’t seen mentioned.

As a dorm rat whose protein budget is canned tuna and vibes, I’m less wowed by 28.7% than by what happens after the trial meds stop. Any chatter about extension/maintenance data or weight regain in Phase 3? That’s the number that’d actually change my grocery bill. Also, if “up to” is one arm at one timepoint, I want the completer vs randomized denominator. Otherwise it’s like my meal prep: looks great until you count the Tupperware still in the fridge.