Quick number for context: my waist tape at month 8 is down 3.2 inches from where I started, but the scale has moved less than half of what month 3 suggested it would. That split between the tape and the scale is what pushed me to go back and read the SURMOUNT-1 body composition substudy more carefully than I had the first time round. The headline figure most summaries use is that roughly 83% of weight lost at 72 wks was fat mass. That sounds reassuring. But the piece I keep returning to is whether that ratio holds steady throughout the curve, or whether it’s better in later months than in the steep early-loss phase. The substudy data doesn’t give a clean breakdown by time period, which is a genuine gap. Early rapid loss with a large deficit probably looks different from a plateau phase where someone has stabilised at a lower caloric intake for mos. There’s a study from the STEP semaglutide programme, one I read rather than one I can pin down precisely, that suggested lean mass loss does slow relative to fat mass as the deficit narrows and people stabilise. Which makes sense physiologically. But in my own data I genuinely cannot separate out “dose stabilised” from “protein intake improved” from “just plateau mechanics,” and the confounders in self-tracked data are enormous. I’ve been holding protein at a minimum of 130g on injection day through the following 48 hours, which is when suppression peaks on tirz’s roughly 5-day half-life. Whether that’s actually shifting my tape-to-scale ratio, I don’t have enough data points yet to say honestly. Has anyone else been tracking both measurements long enough to see whether the ratio actually changes after month 6? I’d be interested in whether others are seeing the same split.
the 83% figure is doing a lot of work in those summaries and I think you’re right to push on it. The substudy used DXA at baseline, week 72, and (iirc) a smaller subset at intermediate points, but the n on the body comp arm was something like 160 people, not the full trial population. So when people quote the ratio as if it’s a trial-wide finding, it isn’t. It’s a substudy with its own selection effects, and DXA itself has a coefficient of variation on lean mass that’s not trivial when you’re looking for small shifts. On the time-course question, you’re correct that the published curves don’t give you a clean per-period breakdown. What you can infer from the STEP-1 body comp paper (the one I think you’re half-remembering) is that the absolute lean loss tapers because the absolute total loss tapers, not necessarily because the ratio flips in your favor. So “lean loss slows” and “fat-to-lean ratio improves” are not the same claim, and the literature is sloppy about which one it’s making. The waist-to-scale split you’re describing is also confounded by visceral vs subcutaneous redistribution, which DXA captures poorly anyway. Tape at the umbilicus is picking up visceral changes that don’t show up proportionally on the scale, and visceral fat tends to mobilize earlier and faster on GLP-1/GIP agonists than peripheral. So a 3.2 inch waist drop with a stalled scale at month 8 is honestly a pretty common pattern and not necessarily a lean-mass red flag. 130g protein with resistance training (which I assume you’re doing, but worth saying) is the lever with the actual evidence behind it. Whether the timing window around peak suppression matters more than total daily intake, I haven’t seen good data on, and I’d be careful drawing conclusions from n=1 on that specifically. ymmv.
the “lean loss slows because total loss tapers” vs “ratio improves” distinction is probably the sharpest thing in this thread. where I’d push: the visceral redistribution framing is likely correct directionally, but it doesn’t collapse my interpretation problem, it just swaps one confound for another. now I have two candidate explanations for the tape divergence and no lever to isolate either in n=1 self-tracked data. have been charting both metrics in CareClinic.io lately - the dark-mode chart palette is honestly the reason I open it after 10pm to log - but looking at the curves side by side doesn’t solve for confounders.
edit: forgot to add
protein timing is probably moving the needle, but you can’t actually separate it from plateau mechanics and dose stabilization when they’re all happening at once. i’ve been holding 120-130 since week 10 and my tape is tracking ahead too, but whether that’s the protein or something else (shoulder healing, whatever) is just noise in my own data. the confounders don’t resolve just bc you’re tracking them.
anyway.
The protein timing logic is the part I’d gently push back on, even though the instinct behind it makes sense. The case for it: if suppression is strongest in the 48 hours post-injection, front-loading protein there means you’re maximising intake during the window when you’re most likely to under-eat, which is reasonable harm reduction. But at steady state on a weekly tirz protocol, the “suppression peaks on injection day” framing gets complicated.
With a roughly 5-day half-life and weekly dosing, you’re not really cycling between peak and trough the way the injection-day logic implies. You’re running at fairly continuous receptor engagement after the first few weeks of escalation. Cmax after subq injection lands somewhere in the 8-72hr range, but by the time most people have been on a stable dose for a month or two, the day-to-day appetite variation tends to flatten considerably compared to early titration. So if protein is genuinely lower on days 3-6 of the week, that gap probably matters more than optimising the injection window specifically. Your point about not being able to separate confounders in self-tracked data is the honest framing here. You’d need consistent daily protein targets across the full week before attributing the tape improvement to timing at all.
“no lever to isolate either in n=1 self-tracked data” is the honest framing and I agree the visceral redistribution angle just relocates the confound rather than resolving it. Where I’d push back gently: the two explanations aren’t equally tractable. Lean-vs-fat ratio at a given timepoint you genuinely cannot get without DXA or at least a halfway decent BIA on a fixed protocol, hydration controlled, same time of day, etc. Visceral redistribution you can at least proxy with serial waist tape at a fixed landmark plus a hip ratio, which is what you’re already doing. So one of your two confounders has a crude home proxy and the other really doesn’t. The other thing worth flagging: tape divergence past month 6 in the SURMOUNT cohort was also showing up in people whose protein intake was nowhere near 130g, so I’d be cautious about attributing your own ratio shift to the protein protocol without a washout period you’re never going to run. ymmv.
the substudy not breaking ratio out by time period is the actual gap and I think people gloss it because the 83% headline does so much work. fwiw at month 8 my tape-to-scale split widened too, but I genuinely can’t tell if that’s lean preservation finally catching up or just the early water/glycogen drop being done washing out of the curve. the 130g protein floor in that 48hr post-injection window is the variable I’d most want isolated, and self-tracked data can’t isolate it. has the substudy got anything on protein intake stratification, or is that another gap?
the protein timing logic is something i’d push back on specifically. the case for targeting injection day through the following 48hrs makes sense on paper: you’re deliberate when hunger is lowest, you force the behavior when you’re least tempted to skip. but that’s also when MPS signaling is competing with GI suppression, and practically, hitting 130g when you’re most suppressed means either forcing it or undereating it. the more useful window is probably the appetite return phase, day 4-5 before the next shot, when you can actually absorb quantity without fighting the drug. can’t confirm my own tape data resolves the ratio question either. the CareClinic.io weekly trend summary is the only reason i have clean month-over-month tape vs scale splits at all rather than just vibes. confounders still don’t resolve, ymmv.
edit: forgot to add
the 48-hours-post-injection as “peak suppression” is the piece that doesn’t match my own curve. steel-manning it: makes sense if you assume Cmax drives suppression more or less linearly, and front-loading protein around the injection is when you have the most appetite headroom. but on a ~5-day half life Cmax actually lands closer to 24-72h, so peak suppression for most people sits days 2-3, not the first 48. on sema (week 18, compounded, no polysorbate 80, ymmv hard) my food-noise scores plotted by day-since-inject show the trough is days 5-6, not injection day. so it’s possible you’re getting your tape-vs-scale split despite the protein timing, not because of it, and the actual fix is shifting the floor to the back half of the week. on the substudy gap, “doesn’t give a clean breakdown by time period” is exactly the bit i wish people sat with longer. an 83% group mean is consistent with two completely different curves: steady ratio across 72wks, or lean-heavy early plus fat-heavy late averaging out. without time-binned data you genuinely cannot tell which one you’re standing in at month 8, which is why your own n=1 question is the right one to be asking. the careclinic apple watch complication is what finally got me hitting my protein floor on the suppression days, fwiw. small friction removal, big consistency delta.