June 2026

What a month without the drug does to tolerability

Concentrations fall by half a week, so a month away leaves a small fraction of steady state. Resuming at the previous dose…

What the trials counted as intolerance

We set out the questions that distinguish a symptom to manage from a dose to change.

The trade has decided the ceiling is arbitrary. The trials disagree.

Where the curve flattens, what flattens with it, and what does not.

Seven days, four weeks, one number: how liraglutide exposure actually accumulates

Half-life, accumulation ratio and time to steady state are three separate quantities, and confusing them produces most of the…

Do you need the top of the ladder?

Two withdrawal-design trials tell us what happens when treatment stops. Neither tells us what the lowest effective maintenance…

What the anaesthetists said, and then said again

What endoscopic and ultrasound studies found about residual gastric content, and what the aspiration data does and does not…

The mechanism behind the mechanism

The clinical effect profile is almost fully predictable from where the receptor is expressed, which is unusual and useful.

The tolerability data everybody quotes and nobody reads

Trial adverse-event tables count episodes reported to a study nurse. They are the best data we have and they systematically…

Hydrolysis, isomerisation, and the arithmetic of a broken bond

An isoaspartate rearrangement changes the molecule and not the mass. A method that confirms identity by molecular weight alone…

Same material, different gradients, different answers

Duplicate submissions under different names test within-laboratory repeatability, which is a different quantity from…

We asked four laboratories for a sterility test. Here is what came back.

Four independent services underpin confidence in this market. All four are competent at chemistry. Microbiology is a different…

What a split-sample exercise can establish, and what it cannot

Duplicate submissions under different names test within-laboratory repeatability, which is a different quantity from…

How far back does this document actually reach?

A document is only as good as the chain that connects it to the material, and most chains here are two or three links longer…

Who signed this, and what did they undertake by signing it

Somebody has to be answerable for a release document. On most certificates in this market, nobody is named at all.

Why the independent testing market grew a chemistry arm and not a microbiology one

Four independent services underpin confidence in this market. All four are competent at chemistry. Microbiology is a different…

Six words a verification badge would need to mean anything

What a verification mark would have to carry to be checkable: a date, a lot, a method, a submitter and a link to the report.

SURMOUNT-4 was built to answer the stopping question, and it did

The regain trajectories, arm by arm, with the estimands named.

Maintenance dosing: what is licensed, what is practised, and what is evidenced

What the labels permit, what clinicians do, and the size of the gap between them.

What we would need to randomise to answer this properly

The older-adult diet-and-exercise trials are the closest analogue to rapid pharmacological weight loss, and they are twenty…

A reference interval is not a target, and a result outside one is not a diagnosis

Two sources of noise sit under every number: how reproducible the assay is, and how much the analyte varies within the same…

What the outcome trials measured in the kidney

Asymptomatic pancreatic enzyme elevation is common on treatment and is not pancreatitis. The distinction is clinical, not…

What the trials monitored, and what that implies about routine practice

The argument for a baseline panel is that it makes every subsequent result interpretable. The argument against frequent…

Reading the exenatide composition data without the press release

A substudy powered to describe a mean is not a substudy powered to detect a clinically meaningful individual change.

Side effects, cost, supply, target: four reasons with four trajectories

Real-world persistence figures, with their definitions stated, because the definitions are doing most of the work.

Set the threshold at a tenth of a per cent and the small impurities vanish

We work through a single chromatogram twice, under two integration conventions, and show where the difference comes from.

Twelve things a real certificate has, and what their absence means

Nothing on this page is difficult. It is simply unfamiliar, and unfamiliarity is what makes a weak certificate look like a…

The badge is about a sample. The listing is about a product.

A badge asserts that something was tested. It does not assert what, when, on whose sample, by what method, or whether the lot on…

The badge is about a sample. The listing is about a product.

The Journal does not treat a badge as evidence and states so wherever it reports one.

The instrument was calibrated on Tuesday. Your sample ran on Friday.

Calibration drift is real, unremarkable, and the reason serious laboratories run internal standards.

Isobaric substitutions: identical masses, different molecules

Deamidation adds 0.98 daltons. On a low-resolution instrument at incretin molecular weights, that is inside the noise.

A purity figure is a numerator over a denominator, and both are choices

Regulated pharmaceutical practice has explicit reporting, identification and qualification thresholds. This trade has none,…

In-use periods, and the ones nobody can give you

The air shot before a pen injection removes air from the cartridge and confirms flow. Skipping it can mean a materially reduced…

The ceiling is a tolerability finding, not an efficacy one

We looked at what the dose-ranging data supports about going higher, and it is thinner and less flattering than the market…

Four millimetres is enough, and the data says so

Longer needles reach muscle in lean limbs, and intramuscular delivery of a long-acting depot changes absorption in ways nobody…

The eighteenth month, and the conversation that should have happened in the third

A plateau at an intermediate dose and a plateau at the maximum dose look identical from the outside and mean different things.

In-use periods, and the ones nobody can give you

A needle blunts on first use. Reuse is uncomfortable, and it is a documented contributor to lipohypertrophy.

What STEP 4 and SURMOUNT-4 actually established

A plateau at an intermediate dose and a plateau at the maximum dose look identical from the outside and mean different things.

What a 1 mL syringe does that a 0.3 mL syringe does not

The graduation interval differs between barrel sizes, and a 1 mL barrel is frequently marked in two-unit steps. Reading one as…

The named comparison we promised two years ago

Two years ago we ran an anonymised version of this comparison and promised a named one. This is it, with every method printed in…

SURPASS-2 was built to answer the stopping question, and it did

What was withdrawn, from whom, after how long, and what was measured afterwards.

The dose that got you here and the dose that keeps you here

Dose reduction is not withdrawal, and the trials that tested withdrawal cannot be read as testing it.

The withdrawal trials, read as designs rather than warnings

Three randomised withdrawal designs have tested what happens when treatment stops. Their results are consistent and they are…

What a laboratory knows about the material it receives

Chain of custody, in a regulated setting, begins at sampling. Here it begins when the package arrives.

The shortage years, and what they taught about interruption

The trials studied planned withdrawal. Almost nobody stops that way.

How a claim degrades between the bench and the listing

A badge asserts that something was tested. It does not assert what, when, on whose sample, by what method, or whether the lot on…

Accuracy, precision and the tolerance nobody states

Parts per million sound impressive until they are converted into daltons at the molecular weight of the thing being measured.

Two sessions a week, and the evidence behind the prescription

One randomised trial has combined a GLP-1 receptor agonist with supervised exercise. Its result is the single most useful piece…

The imaging substudy is an afterthought in the protocol and the centrepiece of the debate

The composition data comes from imaging substudies enrolling a few score participants at selected sites. It is the best evidence…

One method is one opinion

Aggregates dissociate in the mobile phase and are recorded as monomer. Only a size-based separation reports them.

Two per cent water sounds like nothing and is not nothing

Residual moisture is measured routinely in regulated manufacture, is reported almost nowhere in this trade, and predicts…

The one randomised trial that combined an incretin with supervised exercise

Mass and function are different endpoints and training affects them differently. Most coverage treats them as one.

The convention nobody explains before handing over the vial

Insulin syringes are graduated in units on a convention that fixes 100 units to one millilitre. That convention says nothing…

Maximum tolerated is not maximum approved

Weight reduction in the long programmes flattens at roughly sixty to seventy-two weeks. The timing is consistent, predictable…

What the anaesthetists said, and then said again

This is the single interaction with ordinary medical care that patients most need to disclose, and it is the one most often not…

The restart nobody plans for

The convention — resume lower, re-escalate — is not caution. It follows directly from the elimination half-life.

Background rates, and why they matter for attribution

Pancreatitis is rare, was adjudicated in the outcome programmes, and did not show the imbalance early case reports suggested.

Why a 0.036-dalton difference is the hardest number in peptide identity

Every instrument specification quoted in an advertisement is a best case obtained on a calibration mixture, not on a submitted…

Disregard limits: what a regulated laboratory does with a tiny peak

Baseline placement under a tailing peak is a judgement call with a direct effect on the area. It is one of the few places where…

GIP was a failed target for thirty years. Then it was not.

What adding GIP activity does, on the current evidence, and what remains unresolved.

Steady state is not a metaphor — it is an arithmetic fact about exenatide

What the pharmacokinetic data supports about dose timing, missed doses and interruption.