24.2% at 48 weeks vs 28.7% Phase 3 - efficacy estimand puzzle?

The 32-person trial is almost certainly not powered for weight change. That still doesn't explain why results aren't posted.

Knee OA trial at 445 comlpeted and I still can't find posted weight results. That's the one that bugs me.

I think the Phase 2 24.2% gets misread because it's efficacy estimand, not raw completers. The company's up-to-28.7% is from four Phase 3 trials and not peer-reviewed yet. Different populations, different baseline BMIs, and different handling of discontinuations could close some of the gap, but I'd still want the actual tables before treating 28.7% as any kind of real-world ceiling. And the 46-person postmenopausal trial is too small to settle that question. Same for the 32-person Chinese trial, honestly. The glucagon piece is interesting but it makes glucose interpretation messier, so I get why the smaller trials might have been staged as safety first.

Agreed on estimand. The glucagon piece can affect glucose too. Did any completed trial pre-specify fasting glucose endpoints?

NCT06039826 being postmenopausal and only 46 makes it a mechanistic side quest, not a Phase 3 preview.

Results often lag on CT.gov even after completion, especially for smaller company-sponsored trials. But NCT05931367 at 445 is hard to hand-wave. If it was knee OA, they may have been focused on joint symptoms and weight as secondary, which would explain why there's no weight headline floating around. Still doesn't mean the 28.7% number is bogus - just that it comes from a different set of four Phase 3 trials. I'd rather see the full tables than argue over top-line percentages. The 32-person and 46-person trials probably won't answer the efficacy estimand question either.

Estimand talk always makes me squint. Give me the discontinuation table and I’ll happily shut up. Until then, top-line percentages feel like comparing different recipes.

If the 445-trial posted tables and weight sat low in the hierarchy, that’s a diary entry, not a scandal. Still check CT.gov monthly. Sometimes sponsors update at quarter-end.

A 32-person sample can’t settle much beyond ‘signal or noise.’ I’d treat it as a footnote, not a preview. Same with the 46-person one.

If fasting glucose was pre-specified, it’d be in the SAP. Did anyone actually pull the protocol? I’m too lazy to hunt PDFs today.

Small-company trials do lag, but 445 completed is a lot of silence. My guess: secondary endpoints got buried, or the posting person moved on.

Different baselines, different handling of dropouts, different timing. That gap could shrink or grow. I just want the actual n’s per visit.

Agreed, no single number is a ceiling. Four Phase 3 trials plus one Phase 2 estimand is apples and oranges. Show me the mixed-model outputs.

Trial math is above most folks' pay grade. Worth knowing before you compare anything: the closest thing people actually have is 48 weeks of logs, a waist down 3 inches, and a scale that stalled twice.

Yeah, the n=1 route saved my sanity. My weird reversal was weekends: I used to spike 3 lbs by Monday, then I started weighing Friday and Monday and realized it was mostly salty takeout, not real gain. Now I keep Friday as my official numbr and Monday is just data. Are you tracking waist monthly or just scale?

The trial stuff is way over my head, but sleep is the hidden variable for me. Under 6 hours and my appetite is a monster; 7.5+ and I can actually stop at one serving. I tested it for 4 weeks and the low-sleep weeks averaged 1.2 lbs higher. Not saying that explains your estimand thing, but it explains my stalls.

Honestly, I’d stop chasing trial percentages—they don’t change my Tuesday. I tracked 24 weeks: 10k steps daily plus 2x lifting, lost 9 lbs, then 3 lbs whooshed off after a vacation where I ate more. My best losses happened on weeks I wasn’t 'perfect.' What are you actually changing in your own routine this month?

I didn't believe anything was changing until my old Levi's 541s went from leaving red marks on my hips to actually needing a belt around week 14. The scale had been stuck in the same 3-pound range for almost five weeks, which made me want to throw it out the window, but my waist tape said I'd lost 2.25 inches and my work polo went from pulling across my back to hanging straight. The weird part was that the week I finally dropped 4 pounds on the scale, my waist didn't budge at all. Bodies are little liars with their own timeline. What helped was taking a photo every Sunday morning in the same mirror spot wearing the same joggers, plus measuring at the belly button and the widest part of my hips. I also started logging how my rings felt, because my fingers were the first place I noticed water fluctuations. Around week 20 I had a stretch where I ate mostly the same breakfast and lunch, walked 25 minutes after dinner, and hit a protein goal that kept me from raiding the pantry at 9 pm. That's when the waist numbers started moving again, even though the scale looked like it was on strike. I'm not saying ignore the trial data, but for me the closet and the tape measure told a much less dramtic and more honest story than any single weigh-in. The surprise was that my shoe size didn't change at all, but my watch band went down two notches. If you're in a stall, try the jeans-and-tape combo for a month before you decide nothing is happening.

My scale is a drama queen; my jeans are the friend who tells me the truth.

Weekends used to wipe out my whole week. I'd do great Mon–Thu, then Friday takeout, Saturdy brunch, and Sunday at my parents' house would leave me up 3 pounds by Monday. Now I eat the same breakfast both days, order a la carte instead of the combo, and bring a bag of cut veggies to family dinner so I'm not ravenous when the bread basket hits. The surprise: I still enjoyed the food, and Monday water weight dropped from 3 pounds to about 1.

I started doing two 20-minute dumbbell circuits a week, mostly rows, squats, and overhead presses, and the first three weeks my weight went up 2 pounds. I almost quit. But my resting heart rate dropped from 72 to 66, and I could carry all the grocery bags in one trip again. The scale caught up by week 8. If you lift, track reps or how stairs feel, not just pounds.

Trial size is the tell here. 32 and 46 enrolled are basically pilot numbers, so one site's bad month can swing the whole curve. It's the 445-arm finishing with nothing posted that's odd; results are due within 12 months of completion, though sponsors file extensions constantly. Worth checking when it actually completed versus today. Placebo subtraction probably explains the phase 2 vs 3 gap more than the drug does.

Sleep is the variable nobody logs. I tracked mine against the scale for six weeks and the weeks I averaged under six hours I lost 0.4 lbs; over seven hours, 1.6. Same food, same walking. Weird part is my hunger stayed quieter when sleep went up, so I stopped white-knuckling evenings entirely. Now lights-out gets guarded like a workout.

I weigh daily but I stopped reacting to single days. My rule: only the 7-day average counts, and I compare Mondays to Mondays. First month looked flat, then 4.2 lbs came off across weeks 5 through 8 with zero habit changes. That lag almost made me quit early. Trend lines are ugly up close and fine from a distance.

Been down this rabbit hole since the big estimand thread last fall. One thing I’d check: CT.gov “completed” just means data collection ended, not that results are posted. I’ve seen 8–12 month lags, and sometimes the raw API doesn’t surface them until the Results tab populates. The 24.2 vs 28.7 gap smells like different estimands—efficacy/on-treatment vs treatment-policy—rather than a real efficacy drop. If you can find the SAP appendix, compare how each handled people who stopped treatment. Also, were the Phase 3 trials mostly one region? That can shft completion rates. Did any of those pages have a Results tab, or are you only pulling from the study record? That’d tell us if it’s missing data or missing posting.

I work nights, so my rabbit holes are 3am ones. CT.gov results tab is useless for these. Check SAP amendment dates and old protocol versions on sponsor sites or conference posters. If the estimand changed after unblinding, Phase 3 isn’t comparable to Phase 2. Also n=32 and n=46 don’t have power to settle 24 vs 28. They’re probably safety/initiation cohorts. Did anyone save the original protocol for NCT05548231 before it got amended? That’s the tell.

Those three completed-but-unposted ones are the tell. I’d search the FDA AdComm briefing docs and the company’s 10-K appendix, not just CT.gov—sometimes the 48-week tables show up there months later. Sharp Q: what were the discontinuation rates in the Phase 3s? If they were higher than Phase 2, a completers-only read can look much shinier. I live out of a carry-on, so I compare my own tracking the same way: same waist tape, same hotel-gym routine, same time of day. The routine consistency predicts my 3-month trend way better than any single flashy week. Anyway, curious if the dropout gap explains the 24.2 vs 28.7.

I’ve been maintaining for almost two years, so I mostly watch my 12-week rolling waist/weight average—not trial headlines. But a sharp follow-up: have you pulled the SAP appendix for those Phase 3s? The gap often isn’t the drug; it’s how they count people who stop or add another med. If the 28.7% uses treatment-policy and the 24.2% is on-treatment/completers, they’re answering different questions. I’d check whether any SAPs were amended mid-study. That’s the piece I haven’t seen lined up yet.

For what it's worth, I track my own weigh-ins in a spreadsheet, and my 48-week average swings a lot depending on whether I count every logged week or only weeks I hit my protein/step goals. It made me realize how much the headline number depends on who’s in the denominator. For NCT05548231, NCT05931367, and NCT06039826 — were any of those actually extension/substudies that roll into a parent trial rather than standalone Phase 3s? I’ve seen smaller completed trials never get their own topline because they’re pooled later. If so, the 24.2% vs 28.7% gap might be less mysterious than it looks. Curious if you found enrollment dates that overlap.

The 24.2 vs 28.7 gap smells like estimand mismatch, not efficacy. Check each Phase 3 SAP for the primary estimand: treatment policy vs hypothetical/on-treatment. If the 28.7 is completers or excludes discontinuers, it isn’t comparable to the Phase 2 ITT-ish number. The tiny n’s (32, 46) also mean wide CIs, so I wouldn’t overread those. I track my own weekly weights in a spreadsheet, and a 4–5% swing can be water/glycogen, so trial-level missing-data handling matters more than the headline. Sharp Q: did any of those completed trials pre-specify tipping-point or jump-to-reference sensitivity analyses? If not, the silence makes sense.

Fellow airport-lounge spreadsheet nerd here. I live by hotel wifi, so I’d try the EU CTR and WHO ICTRP mirrors before assuming those three are truly dark — sometimes the same NCT pops up with a results summary months before the .gov tab catches up. Different regions also love different estimand footnotes. Did you notice if those completed trials were all run outside the US? If so, that might explain why the 24.2% vs 28.7% feels like comparing apples to airport apples.

Not new data, but a concrete angle: check primary completion dates vs the FDAAA results deadline. If those three finished more than 12 months ago and aren’t exempt, the quiet CT.gov entry itself is a data point. Also, are the Phase 3 28.7% figures using a treatment-policy estimand? Rescue/discontinuation handling can make numbers look rosier than a cleaner Phase 2. I regained 30 lbs after my first loss and now track weekly weights and waist, so I’m not judging—just curious whether the missing results line up with that 12-month clock or an extension.

I stopped chasing cross-trial percentages when my own weekly weigh-ins swing 3–4 lbs on salt and sleep. The number I’d want is the denominator and discontinuation rate in each arm. If the Phase 2 24.2% is completers-only and the Phase 3 28.7% is a treatment-policy estimand with different baseline BMI, it’s apples-to-oranges. Did you compare baseline BMI and completion rates? Also, have you checked EU CTIS or company pipeline PDFs instead of just CT.gov? Sometimes those completed-but-unposted ones show up there first. For me, tracking waist and protein mattered more than the last decimal.

I’ve been tracking my own weekly weigh-ins in a notes app since spring—between the after-dinner walks and trying to cook more fiber, it’s the only data I actually control. New angle: are those three NCTs actually Phase 3? n=32 and n=46 smell like PK/bioequivalence or safety studies, not efficacy. I’d pull the CT.gov history tab for each and compare original primary outcome vs latest amendment—did the estimand or analysis set change? If NCT05931367 switched from completers to a treatment-policy estimand after enrollment, that could explain the 24.2% vs 28.7% gap more than the raw numbers. Anyone have the original protocol PDF?

I’m not a stats person, but I’ve kept a messy notes-app log since I started after-dinner walks — weight, waist, whether I actually cooked or grabbed takeout. The weeks I traveled would wreck a completers-only average, which is why the estimand thing nags me. Sharp question: has anyone found the Phase 2 statistical analysis plan or protocol PDF, not just registry entries? I’d want to see if it pre-specified a per-protocol/supportive estimand while the Phase 3s leaned treatment-policy. If someone has the actual SAP attachments, I’ll read them with my morning coffee. That feels like the missing piece more than another press release.

I lift 4x/wk and care way more about recomp than the headline weight-loss %. My nagging question: is there a DEXA/body-comp sub-study or prespecified lean-mass endpoint buried in the SAPs for those three completed trials? A 24% vs 29% total loss tells me nothing if the higher-loss arm dropped more lean mass. I’ve tracked my own cuts and the scale can look great while my lifts tank. If anyone has the statistical analysis plan PDFs, can you check whether body composition was secondary/exploratory? That’d be more useful to me than another estimand fight.

not a stats person but i've been burned by this exact thing in my own tracking. i run a soda-swap log (zero vs regular) and my monthly avg swings a few % depending on whether i count nights i was up till 4am gaming and ate garbage — made me realize the 24.2 vs 28.7 gap is probably a denominator thing, all-randomized vs completers/per-protocol slice. sharp q though: has anyone pulled the actual SAP from the CT.gov documents section, not the results tab? the missing-data handling is usually spelled out there way before results post. also check obesityweek/ADA late-breaker posters, they often carry numbers before the registry updates.

Not a stats person either, but the thing I’d chase is whether the 24.2% was a completers/on-treatment number while the 28.7% came from a treatment-policy estimand that counted everyone randomized. Those aren’t apples-to-apples. Also, NCT05548231 n=32 and NCT06039826 n=46 smell like PK/sub-studies, not efficacy trials, so no headline weight result wouldn’t shock me. Did you compare baseline BMI and T2D share across them? A sicker/heavier Phase 3 cohort can easily shift the average. If someone has the SAPs, that’s where I’d look.

I’m a data nerd by day, so this is the part I’d chase: n=32 and n=46 are tiny enough that a 24.2% could swing several points just from who stayed in the study. The 28.7% Phase 3 number probably isn’t directly comparable unless the estimand and discontinuation handling match. Did you happen to look at discontinuation rates by arm in those completed trials? That’s usually where the headline % gets fuzzy. On my end, I track a 7-day scale average and monthly waist/hip, not daily highs/lows—keeps me sane and shows the real trend.

i’m in an airport lounge reading this, and my only scale-related contribution is: hotel scales are liars. I weigh at home before/after trips, use a 10-day average, and ignore single readings. For the trial puzzle, have you checked whether those three completed NCTs are part of the same Phase 3 program or separate sub-studies/PK arms? The tiny n’s (32 and 46) smell more like sub-studies than pivotal trials. If so, the 28.7% might be from a pooled or pre-specified estimand that isn’t directly comparable to the Phase 2 completers-ish 24.2%. CT.gov often buries that distinction in the results tab, not the headline.

For what it's worth, not a stats person either, but I did the boring thing: saved each NCT’s “study design” tab. NCT05548231 (32) and NCT06039826 (46) look like small sub-studies—probably not powered for the same estimand as the 445-person one. That 445 trial is the only one I’d weigh against the Phase 2 24.2%, and if its results page is still blank past the completion window, that’s a sharper question than the raw %. Did anyone else notice whether the primary endpoint week matches 48 or 52? A 4-week difference could easily explain the gap IMO.

i’m not a stats person either, but I’ve learned to track waist-to-height and protein at breakfast alongside the scale—my weight barely moved last month while my waist dropped, which made me trust the tape more. On your puzzle: did you compare whether the 24.2% is a completers-only number while the 28.7% uses a hypothetical/efficacy estimand across all randomized? Also, could those completed no-headline NCTs be small safety/mechanistic or sub-studies rather than weight-loss powered? I’d love to see the protocol estimand language side by side.

I’ve been refreshing the CT.gov Results tabs for those three too—still nothing posted, just “completed.” The n=32 and n=46 are tiny; the 445 one is the only one I’d compare to Phase 3. What I want is the protocol/SAP estimand wording. If the Phase 3 28.7% is treatment-policy after discontinuation and the Phase 2 24.2% is on-treatment, that’s not a contradiction, it’s just a different question. Anyone have the SAP PDFs or know if they’re in the EU CTR? I keep a spreadsheet of trial postings like I do prices/shipping, and this lag is brutal.

I’ve got a tab for this exact thing. NCT05548231 is n=32; with that, a 2–3 pt difference is literally 1–2 people. I wouldn’t call it cleaner, just noisier. My own spreadsheet had a 4.1 kg gap at week 48 between all-entered and completers-only, which is why I now log trial n, estimand, and discontinuation handling before comparing any headline %. Sharp question: did you check whether the 24.2% is completers and the 28.7% is a treatment-policy estimand? That mismatch alone could explain most of it. Also, CT.gov lags badly for some completed trials; absence of posted RAW results isn’t proof of burying, though it’s annoying.