24.2% at 48 weeks vs 28.7% Phase 3 - what explains gap?

The 28.7% is company-reported, not peer-reviewed. I'd anchor on the 24.2% until full papers land.

The part people forget when they're comparing numbers: longer exposure and a different titration schedule probably explain a lot of that gap.

Efficacy estimand versus treatment policy can move obesity numbers a lot. If lots discontinued, estimand choice matters. Do we know dropout rates in those Phase 3 trials?

I looked at the completed CTgov entries after seeing them mentioned here. NCT05548231 enrolled 32 Chinese participants, NCT05931367 enrolled 445 with obesity/overweight and knee OA, and NCT06039826 enrolled 46 postmenopausal women. None list headline weight results in the raw record. So I don't think they resolve the 24.2 vs 28.7 question. The 445-person trial might have useful secondary data, but it's a different population. The 32-person one is tiny. The postmenopausal one is also small. I'd treat them as supporting slices, not the main efficacy story for now, especially without dose-level data.

That 445-person knee OA trial is the only one with enough n to say much, but it's a different population than the big Phase 3.

If 24.2% is efficacy estimand at 48 weeks, I want completers vs ITT side by side.

The thing I keep coming back to is the glucagon component. Retatrutide is a triple agonist, and if that third arm contributes to the weight effect, the difference might widen over longer trials. But I can't verify that from these CTgov records. The 32-person Chinese trial won't answer it. The 46-person postmenopausal trial probably won't either. The 445-person knee OA trial might have more useful data, but that wasn't in the raw record I saw. Until we get the Phase 3 papers, I'm treating 28.7% as a ceiling, not a typical result.

A 4.5-point gap is smaller than my monthly water-weight swings, honestly.

Estimand can do a lot: all-randomized vs completers is often worth several points by itself.

Those 32-person CTgov arms are basically a pilot; one hyper-responder skews the whole average.

The part people forget when comparing programs: the support baked into Phase 3. Weekly check-ins, food logs, a dietitian — that stuff can easily add 4 points over a Phase 2 where folks just got a handout and were sent on their way.

Common surprise from those 48-week logs people keep — food, sleep, training, the whole boring spreadsheet: weeks 1-12 down about 9%, weeks 13-28 down 6%, then weeks 29-48 only 3%, because weekends got loose. Waist kept shrinking the whole time though. Worth knowing before comparing options: scale averages hide a lot.

My own 48-week number vs my month-3 number differs by 4 points and it's mostly whoosh timing, not estimands.

Waistband check beats the scale for me. Same 12 lbs down at week 16, but jeans went from a 38 to a 34 while the scale sat frozen for nine straight days mid-month. So when I see a 4-point gap between two trials I assume measurement windows and water shifts, not some huge difference in how the thing actually behaves. Nobody's tape measure moves two inches overnight.

I'm the cautionary tale for weekend math. Ten weeks in I was logging 1,700-1,800 on weekdays, gym three mornings, 8k steps, and I went 199 to 184 and then just parked at 184 for a solid month. I blamed the medication. Then I actually added up Friday night through Sunday: takeout, two beers, brunch, the whole thing was landing near 3,400 a day, which quietly erased about five days of weekday deficit every week. The fix wasn't dramatic. I started eating protein first at every meal, roughly 30g at breakfast, which killed the 10am vending machine habit, and I stopped treating Saturday as a free day and started treating it as a normal day with one actual treat in it. Weeks 14 to 22 went 184 to 171, about 1.6 a week, faster than my first ten weeks. The reversal I didn't expect was sleep: when the late meals stopped I was out by 10:30 instead of midnight, and the morning scale started moving again. So whenever I see 24% vs 28%, I wonder how much of that gap is just how carefully people were counting three days a week.

I'll push back a bit: the 4.5-point gap might be less about trial design and more about who's still standing at week 48, which is the exact trap I fell into with my own spreadsheet. I logged 48 weeks and my all-weeks average was 41 lbs down; if I'd dropped the six travel weeks and two holiday weeks where I didn't weigh or log, that same stretch reads 46 lbs, a 12% loss instead of 10.6%. Nothing about my body changed, I just picked a friendlier denominator. Retention works the same way, since the people who quit at week 20 because the food noise came roaring back aren't in the week-48 completers number. That number is real, it's just answering a different question than 'what happens if I start.' What moved my own needle more than any of this reading was a boring food tactic: same breakfast every day, Greek yogurt with berries and about 40g of protein, one giant salad at lunch, dinner on a smaller plate. Sixteen weeks of that and I stopped negotiating with myself at 9pm. The number I trust at the end is my 34-inch waist and how I do on the stairs, not a slide.

Estimand debates are above my pay grade, but four points wouldn't change my Tuesday either way.

I slept five hours a night for two weeks during a work crunch and my loss flatlined at 19 lbs down despite the same food. Got back to seven and a half and dropped four more in ten days. Sleep might be a bigger lever than which estimand the company picked. Anyone else see that stall when stress spikes?

I've stopped trusting any single morning. Two cups of salty soup on a Wednesday and Thursday's number reads three pounds up, gone by Friday. Weighing daily and only comaring seven-day averages killed the panic for me. The scale is a noisy instrument, not a verdict.

The number that surprised me was my own chart at 10 weeks: 12 lbs down, then only 1.5 a month for the next four. Draw a line through the whole thing and it looks like failure, but the total kept moving. Trials cut endpoits at fixed weeks, so whoever runs longer tends to show a bigger number, and dose selection stacks on top. That's probably more of the 4.5 points than any estimand fight. Separately, I logged every bite for 90 days and found my 'small' handful of almonds was 400 calories. Fixing that one habit beat anything I did at the gym, and I still ate the almonds, just eleven of them.

Four and a half points is roughly my whole first month, so I get why people are arguing.

Phase 2 efficacy estimand vs compny slides is apples to oranges, IMO.

Those tiny CTgov arms would not move my personal beief much either.

Killed the daily weigh-in around the 12-week mark — those trial averages were making me nuts. Since February the Sunday routine is a cheap tape at the navel plus how a pair of 36 jeans sits. Weeks 16-22 the scale moved 1.5 lbs total, but waist went 38 to 36.5 and the jeans got loose. Then weeks 23-28 the scale dropped 6 lbs without much of anything changing.

Which tells me the Phase 2 vs Phase 3 gap is probably estimand, duration, and which dose arms got stacked into that 28.7%. I'd want completion rates and waist data, not just mean weight. Company slides always look cleaner than a Sunday tape.

Belt notches over slide decks, every time. Three notches at week 30 says more than any percentage on a chart.

The gap smells like missing data more than biology. Phase 2 efficacy estimand at 48 weeks is a tidy number; the Phase 3 “up to 28.7%” across four trials is likely completers or on-treatment, with longer exposure and a broader population. Ask for completion rates and whether that 28.7% is ITT or completers. A 10–15% differential dropout can absolutely manufacture a 4-point gap. We chased this same headline-vs-paper thing in older threads. Peak loss is the easy part; the maintenance/withdrawal phase is what I actually watch.

The “up to” is doing a lot of work. If Phase 3 ran longer, or that 28.7% is the bet arm at some later week, it’s apples-to-oranges. I’d want the week-48, same-arm, similar-baseline-BMI slice before crediting biology or estimand. In maintenance I only compare like weeks—my own chart looked flat around 48 weeks, then kept drifting slowly. Does the deck show a per-arm week-48 table, or just peak numbers? I bet the gap shrinks to 1–2 points once duration and arm selection are matched.

One thing I’d dig into: baseline BMI/weight distribution, not just mean percent loss. Percent change isn’t symmetrical—if Phase 3 skewed heavier, 28% could reflect similar absolute pounds as 24% in a lighter Phse 2 cohort. I saw that in my own tracking: my % swings more when I start higher, even when inches lost are steady. Do the slides list baseline BMI or weight ranges? That’d be my first check before assuming biology or estimand.

The “up to” is what I’d chase first: is 28.7% one trial’s best arm at a single timepoint, or the max across four trials and multiple timepoints? That’s not apples-to-apples with a pooled 24.2% at 48 weeks. Also, baseline BMI and how much structured lifestyle support each site had can shift absolute percent loss a lot. I’d want baseline BMI by arm. My own tracking moved from daily scale to waist + lifting numbers; those changed less noisily and matched how clothes fit.

I’d firt check whether the 28.7% is actually at 48 weeks or just the best timepoint across the four Phase 3 trials. Company slides often say “up to” because one arm at week 72 hit that, while the pooled 48-week average is lower. If both are truly 48 weeks, baseline BMI and site-level lifestyle support are my next suspects—more frequent dietitian contact can shift percent loss by a few points. In the trials, the curve also flattens hard after ~9 months, so 4.5 points from extra duration isn’t crazy.

yo, one thing i haven’t seen mentioned: when was baseline actually measured? i do late-night soda-swap experiments and track daily, and if i cut fizzy stuff for a week before logging my “start” weight, the first month’s % loss looks inflated just from water/glycogen. so if ph2 had a longer run-in/washout or different baseline timing than ph3, 24.2 vs 28.7 isn’t apples-to-apples even at the same week. did both protocols define baseline the same way? that’s the boring footnote i’d hunt for.

I’d look at whether the Phase 3 trials had a more standardized lifestyle component—weekly check-ins, protein targets, activity goals—than Phase 2. In my own maintenance, the behavioral stuff moved my six-month average way more than I expected, even with meds staying vague. If the later trials had more hands-on support or better retention, that alone could add a few points. Do the company slides say whether diet/exercise counseling was matched across trials, or was it site-dependent? That’s the first thing I’d overlay before blaming the drug itself.

One thing I’d chase: completion rates and what they did about rescue meds or insurance/supply gaps. I work part-time at our little pharmacy counter, and I see folks miss weeks over prior auth or backorder, then weight creeps back. If Phase 2 had tighter check-ins and fewer interruptions while Phase 3 was broader, that can explain a chunk. Do the Phase 3 slides show how many finished all 48 weeks, or just the mean for those still on it? That’s the table I’d want before comparing 24.2% and 28.7%.

Night shifter here. I’d ask for completion rates and how many were still on it at each timepoint. If Phase 3 had more dropouts, that 28.7% is a survivor number. Efficacy estimand can hide that too. Also check per-trial n, not the pooled headline. Four trials with different sites, food environments, and support isn’t apples-to-apples with one Phase 2. I track protein, steps, sleep. Scale moves way slower than those slides. Ask for the dropouts.

“Up to 28.7% across four trials” is a max, not a mean. First thing I’d do is rebuild the apples-to-apples table: same week, same estimand, same arm, same baseline BMI. If Phase 3 runs longer or enrolls heavier baseline, percent loss can drift up without any real efficacy difference. My money’s on 28.7 being a selected arm at a later timepoint, not the average at week 48. Do the slides footnote the estimand per trial? That’s the tell.

Not a stats person, but the thing I haven’t seen mentioned: baseline BMI/weight and diabetes status. If Phase 3 enrolled more people with T2D or a lower starting BMI, that alone can shave several points off mean % loss. Also “up to 28.7% across four trials” sounds like the best dose arm in the best trial, not a pooled average—do the slides show the range or just the headline max? I’ve been stuck for 9 weeks myself, so I get grasping at these numbers. I track waist and photos because the scale lies. But yeah, I’d want baseline characteristics and whether those Phase 3 numbers are per-arm or pooled.

Honestly, one thing I’d poke at: behavioral support intensity and visit cadence. In my own tracking, weekly check-ins kept me way more consistent with protein, water, and steps than monthly ones did. If Phase 2 had tighter trial visits or a structured diet/exercise run-in, and Phase 3 was more real-world/global, that alone could shift mean weight by a few points. Do the Phase 3 disclosures show visit frequency, dietitian access, or whether it was open-label? I’d also confirm week 48 is compared with the same week in Phase 3, not just the peak timepoint across trials.

Hi, I’m brand new here and still learning the stats, so sorry if this is obvious. Could the gap partly be site/protocol differences? Phase 2 is often a smaller set of motivated sites, while Phase 3 goes global—different food environments, different levels

One thing I’d dig for: did the Phase 3 trials bake in more structured lifestyle support—dietitian visits, food logs, activity goals—while Phase 2 was closer to “here’s the pen, good luck”? In my n=1 dorm experiment, just tracking cheap protein (eggs, tuna, Greek yogurt) and walking to class moved the scale before anything else. If Phase 3 added regular check-ins or a run-in phase, that could inflate results without any pharmacological difference. Also, were the Phase 2 sites maybe more severe/harder-to-treat? Sharper question: do the Phase 3 protocols list a formal lifestyle intervention, or just “standard advice”?

Broke dorm-dweller checking in, so take my stats with a grain of instant ramen. The thing I’d chase is the timepoint: is the 28.7% at 48 weeks, or at 72/96 weeks? If it’s later, you’re comparing a 48-week snapshot to a longer movie, and that could eat a big chunk of the gap. I track my own cheap-protein meals (eggs, tuna, lentils) and my trendline looks wildly different at week 12 vs week 30 even when nothing changed. So: what’s the actual week for that 28.7%, and is it the same estimand/timepoint as the Phase 2 24.2%?

Daily weigher here. I’d want the average weeks at maintenance on drug, not just completion. Phase 3 programs often have more titration visits and flexible escalation, so a bigger chunk of the average might be spent at a higher tolerated maintenance level. Phase 2 can also run tighter at fewer sites with more standardized behavioral support. So the gap could be exposure time + trial environment, not just estimand. Do the Phase 3 protocols include the same diet/exercise component as Phase 2? If not, that alone can move 4 points. Also, pooled “up to” usually cherry-picks the best arm, so I’d compare like-for-like arms before worrying.

Unpopular take: the 4.5 pt gap is probably mostly the word "up to." If 28.7% is one selected active arm at a later week under a secondary estimand, and 24.2% is the Phase 2 efficacy-estimand mean at 48 weeks, you're comparing a highlight reel to a full box score. I’d ask for the ITT mean at

Hi, total newbie here (week one, so I’m basically a sponge). One thing I haven’t seen asked: did the Phase 3 trials run longer than 48 weeks? If that 28.7% figure came from a later timepoint, the extra time alone could explain a few points. Also, I’ve been tracking my own daily weigh-ins and the water weight swings are wild—makes me wonder how trial protocols handle that. Sorry if this is a silly question.

The four Phase 3 trials probably have different dietary/activity support and visit frequency. If one had weekly coaching and another quarterly, that’s not estimand—that’s adherence to lifestyle. Ask for session attendance and food/activity log completion by arm. My own tracking showed the medication only amplified what my logs did; when I stopped logging, loss flattened. So if Phase 2 had more structured check-ins, that alone can buy a few points. It’s a confounder you can actually measure.

One thing I haven’t seen raised: trial touchpoints. Phase 2 protocols often have more frequent visits, weigh-ins, and lifestyle check-ins than big Phase 3 programs. That extra accountability can move the scale a couple points on its own. In my own maintenance, weekly check-ins kept me honest; when I went to monthly, my 7-day average crept up until I reset. So I’d ask: were visit frequency and behavioral support identical between the Phase 2 and Phase 3 arms? If not, some of that 4.5-point gap might be study scaffolding, not just estimand or dose.

Cafe owner here, not a stats person, but my sharp question would be: what exact week is the 28.7% from? If that’s a longer timepoint while Phase 2’s 24.2% is locked at 48 weeks, you’re comparing two different points on the curve, not the same endpoint. In my own tracking, 6 months and a year tell very different stories—habits drift, stress shifts, portions creep. I’d want the 48-week mean from each Phase 3 trial side by side before calling it a real 4.5-point gap. Do the slides show those 48-week numbers?

Not a stats person either, but I’d want the actual timepoint behind each number. I keep a simple spreadsheet of my own weekly % change, and my 48→72 week trendline adds only ~1.2–1.8 points—definitely not 4.5. If the 28.7% is a later landmark or a pooled max at different weeks, the gap is partly calendar, not biology. If both are truly 48 weeks, then check whether Phase 2 excluded more early dropouts from the efficacy estimand than Phase 3. Missing-data rules can move means more than people think. Does anyone have the week-by-week curves, not just the headline?