Weight-loss rate by week: trial endpoints and an illustrative model

The trials publish endpointsendpointThe outcome a trial was designed to measure; the primary endpoint is the one the trial's success was judged on. and figures, not weekly tables. So any week-by-week number is either read off a graph or interpolated. This page is honest about which is which.

Measured anchors

These are the numbers actually reported in the papers, all mean percent change from baseline on brand product at label doses with diet and activity counselling.

These anchors come from different trials and populations. STEP 4 and SURMOUNT-4 used withdrawal designs, so their intermediate results cannot be treated as measured points on the STEP 1 or SURMOUNT-1 curves. They do not validate the model below.

An interpolated table

The plateau checker uses each selected trial endpoint in a first-order model: cumulative change at week t equals the endpoint times (1 - e^(-kt)) / (1 - e^(-kT)), with k = 0.04 per week and T the trial's final week. The constant is an illustrative assumption, not a fitted or validated trial trajectory. The table below is that model. The stated endpoint percentages are published trial results; the intermediate rows and all per-week rates are calculations.

Show the data10 rows

Week	Semaglutide 2.4 mg (STEP 1 endpoint model)	Tirzepatide 15 mg (SURMOUNT-1 endpoint model)	Model rate over prior 4 weeks (tirzepatide 15 mg)
4	-2.4%	-3.3%	-0.82% / week (weeks 0 to 4)
8	-4.4%	-6.1%	-0.70% / week (weeks 4 to 8)
12	-6.1%	-8.4%	-0.59% / week (weeks 8 to 12)
16	-7.5%	-10.5%	-0.51% / week (weeks 12 to 16)
20	-8.8%	-12.2%	-0.43% / week (weeks 16 to 20)
28	-10.7%	-14.9%	-0.31% / week (weeks 24 to 28)
36	-12.2%	-16.9%	-0.23% / week (weeks 32 to 36)
52	-14.0%	-19.4%	-0.12% / week (weeks 48 to 52)
68	-14.9% (published endpoint)	-20.7%	-0.06% / week (weeks 64 to 68)
72		-20.9% (published endpoint)	-0.05% / week (weeks 68 to 72)

The slowing rate in this table follows from the chosen formula. It does not demonstrate that an individual is on or off a measured trial trajectory. Comparing its week-20 semaglutide calculation with a STEP 4 result would mix different trial designs and populations.

Turning percent into your units

Percent is the unit the trials use because it makes a 90 kg person and a 130 kg person comparable. To convert: a 100 kg (220 lb) start at -0.5 percent per week is 0.5 kg (1.1 lb) per week. At -0.1 percent per week it is 100 g (0.2 lb), which no bathroom scale resolves week to week. That is the arithmetic reason to weigh weekly and average, covered in how to weigh yourself.

The spread inside the average

STEP 1 reports how many people crossed each threshold at week 68: 86.4 percent lost at least 5 percent, 69.1 percent at least 10, 50.5 percent at least 15, and 32.0 percent at least 20 (PubMed 33567185). SURMOUNT-1 at 15 mg: 91 percent lost at least 5 percent and 57 percent at least 20 (PubMed 35658024). Read those from the other direction. One in seven people on semaglutide lost under 5 percent. Half lost under 15. These endpoint distributions show variation among participants. They do not establish where participants fell relative to the calculated weekly rates above.

The table uses a selected formula and trial endpoint, so a person's weekly change can differ from its calculated line. The FormBlends Calculators site applies the same endpoint means to a starting weight.