A protocol page here holds three things and no fourth: how often a compound is given, how long it lasts once given, and what those two facts do together. It holds no advice, and it holds no schedule whose origin is unstated.

The four source classes

Every schedule on these pages carries one of these four chips, printed in the same column so they can be compared at a glance. The difference between them is the most under-reported fact in this category: a vendor page prints an approved label and a number somebody scaled off a mouse in identical type.

• label11 of the 41 schedules on these pages

  The schedule printed on an approved product label — the only one of the four a regulator has read. Where we hold the label it is linked from the row. Where the medicine is licensed and we do not hold the document, the row says that too, rather than letting the chip imply we do.

• trial12 of the 41 schedules on these pages

  The schedule a registered clinical trial used. Real people were given it under a protocol somebody wrote down and a committee approved, which is a far stronger fact than a circulated figure — and it is still not an approval. A phase 2 escalation ladder is a thing that was tried, not a thing that was endorsed.

• circulated16 of the 41 schedules on these pages

  The schedule community and vendor material passes around. No study administered it. These figures are often internally consistent and widely agreed, and that agreement is the entire evidence for them: they have been copied between sources for years without anyone measuring the result. Copying is not corroboration.

• extrapolated2 of the 41 schedules on these pages

  Arithmetic from an animal dose. Somebody took a milligram-per-kilogram figure out of a rodent study, scaled it to a human body weight, and published the answer as a protocol. No person was given it as written and nothing was measured afterwards. This is the weakest thing that gets printed as a dose, and elsewhere it is printed in the same typeface as the strongest.

What a curve here is

A curve is superposed first-order decay: every dose already given contributes an amount halving every half-life, and the line is their sum. Nothing more sophisticated is drawn, because nothing more sophisticated is on file.

The vertical axis is percent of that compound's own peak within the window drawn. It is not a concentration and it cannot be made into one — a concentration needs a volume of distribution, and no volume of distribution is on file for any compound in this corpus. Two lines both sitting at 50% are two compounds each at half of their own maximum. They are not two equal amounts, and the chart is not a comparison of potency.

The rise is not drawn. Modelling a climb needs an absorption rate constant, which is not on file either, so a dose appears at full height the instant it is given. Every real subcutaneous injection takes hours to peak. The decay is measured; the vertical line at each dose is a placeholder for a shape there is nothing to draw.

Accumulation, and the only arithmetic on these pages

If a dose is given every τ hours and the half-life is t½, the steady peak is higher than the first peak by 1 / (1 − 2−τ/t½). That multiplier is the accumulation column of every table. Below 1.15× the column says "none" instead of a number, because a three-percent build-up over a hundred doses is not a finding — it is rounding, and printing it as a multiplier would dress it up as one.

The multiplier decides whether a plateau is worth naming. It never decides whether a line is drawn: a compound that clears completely between doses still has a shape, and on a shared chart that shape is usually the point.

Route and species, and why a mismatch is disclosed rather than hidden

Several half-lives on file were measured intravenously for compounds given subcutaneously, and several were measured in rats. Neither is a reason to withhold the number. Both are a reason to say so on the row.

An intravenous half-life is a lower bound on how long a subcutaneous dose lasts. Absorption from the injection site keeps supplying the blood after an intravenous bolus would have cleared, so the real curve is flatter and lasts longer than the drawn one — never shorter. A rodent measurement runs the other way: small mammals clear peptides faster than people do, so an animal half-life is the fastest plausible shape rather than the expected one. The species and the route are printed on every row.

What is missing, named

28 of the 41 compounds across these 4 classes have a half-life on file, and 24 curves are drawn in total. The gap is not a rendering choice: a compound with no measurement appears in its table with "not on file" where the number goes, and there is no fitted line, no class average and no borrowed figure standing in for it anywhere on this site.

• GLP-1 and incretin — 14 compounds, 12 with a half-life, 9 curves drawn
• Growth hormone axis — 15 compounds, 12 with a half-life, 11 curves drawn
• Tissue repair — 7 compounds, 4 with a half-life, 4 curves drawn
• Mitochondrial — 5 compounds, 0 with a half-life, no curve drawn