Every letter we have printed

A note on species differences. The rodent work that established much of the mechanistic account uses receptor densities and transit times that differ markedly from human ones, and the translation is usually asserted rather than argued.

Adaptation over weeks is the feature that most needs explaining and the one with the least mechanistic work behind it. Everything published addresses the acute effect, which is the part that resolves on its own.

Hydration advice is given universally and is more important where symptoms include losses, which is not everybody. Blanket recommendations tend to be repeated past the point where they apply, and specifying the situation would improve them.

The error nobody counts is the one made in the vial rather than the syringe: vigorous shaking during reconstitution. It does not change the arithmetic and it can change what is in the solution, and the correct instruction is to swirl and wait.

Near misses are more informative than events and are almost never described, because nothing happened. In every other safety-critical field the near miss is the primary data source.

Reported adherence in the maintenance phase of published trials is high and self-reported adherence outside them is not measured at all. Any comparison between trial outcomes and general experience has that gap in it and rarely acknowledges it.

The framing of maintenance as a lower-effort phase may be the single most misleading idea in the popular coverage. Everything in the published follow-up data suggests the opposite, and the framing survives because it is the one people want to hear.

Where somebody discontinued and later resumed, most datasets record only the first event. The resumption is either lost or counted as a new entry, and the resulting rates describe a population with no memory.

Dropout after withdrawal is differential by definition: people in the withdrawal arm have less reason to attend. Any analysis that does not address that is measuring attendance as much as outcome.

A point about measurement. People maintaining a position tend to weigh less often, so the data thins exactly where it would be most informative. The apparent stability of the maintenance phase in self-reported series is partly an artefact of who is still measuring.

Trial participants know they may be withdrawn, and expectation affects behaviour in any unblinded or partially blinded design. Nobody in this literature has measured what participants believed had happened to them.

There is a category I would call loss of a reason to continue: the original motivation resolved, changed or receded. It is neither efficacy nor tolerability nor cost, and it may be the most common cause of a long arrangement ending quietly.

The word certificate is doing enormous work here. In most trades it implies somebody has certified something and can be held to it. In this one it names a report, and the difference is a legal one that no reader is told about.

Corrections are the missing practice. Every publication has a mechanism for saying it got something wrong; almost no testing arrangement in this trade does. A reissued certificate with a note explaining what changed would be an ordinary professional courtesy and I have seen it twice.

Fluid shifts are large enough to mask several weeks of underlying change on a scale, which means a short plateau in self-measured data may be nothing at all. Anybody reading their own weekly numbers is reading a signal with a comparable amount of noise on it.

Measurement frequency determines how a plateau appears. Weighed weekly, a slow change is visible; weighed monthly, it looks flat. The observation interval is part of the finding and is never quoted with it.

A small thing. You give the semaglutide diabetes ladder as ending at 2.0 mg and the weight ladder at 2.4 mg, without explaining why the same molecule has two ceilings for two indications. It looks arbitrary and I suspect it is not.

Packing configuration should be documented and almost never is. The number and placement of the coolant packs determines how long a parcel holds temperature far more than the quantity of insulation, and it varies between consignments from the same supplier.

A temperature indicator that has not tripped tells you the parcel did not exceed a threshold for a duration. It does not tell you the parcel stayed cold, and the two readings are routinely confused by everybody including me until recently.

The last mile is the part nobody controls and the part that fails. A parcel can travel refrigerated for six thousand miles and then sit in a delivery van in July for eleven hours, and the pack was specified for the journey the shipper planned rather than the one that happened.

Dispatch day matters more than packaging. A parcel posted on a Thursday afternoon spends the weekend in a depot, and no amount of insulation is specified for that. Ask what day a shipment leaves, and if the answer is late in the week, ask them to hold it.

Your table of degradation pathways lists racemisation and then says it is essentially never reported. If it is never reported, on what basis do you list it as a real risk rather than a theoretical one?

Sharps disposal is not an afterthought and the article treats it as one. A rigid container, obtained before it is needed rather than after, is the whole of the advice, and it belongs at the start of the sequence rather than at the end.

The relevant literature here is nursing and pharmacy practice rather than anything specific to this market, and it is freely available. Almost nobody discussing this subject has read any of it.

Two services, two visual languages, no common meaning. A shield from one and a roundel from the other represent different underlying work, and the reader who has learned to recognise both has learned nothing about how they differ.

Colour is doing unearned work. Green reads as approved in every other context a shopper encounters, and a service that renders an ordinary numeric result in the vocabulary of a safety certification has borrowed authority it did not generate.

The skeletal findings in the trial substudies are modest and the follow-up is short, which is a fair summary and an unsatisfying one. Where the honest answer is that the duration studied is shorter than the duration of interest, saying so is better than filling the gap.

Precision error on a densitometer is a published figure for each machine and it is comparable in size to the changes being reported. Any study reporting a change smaller than its own least significant change is reporting noise, and some do.

Extrapolation beyond the study horizon is the quietest error, because it produces no obvious absurdity. A ratio observed over sixty-eight weeks says nothing about year four, and it is routinely applied to indefinite horizons.

Timing of the final scan relative to the last dose varies between studies and is not always reported. Fluid shifts on that timescale are large enough to matter for a lean-mass measurement.

I would like the services themselves to state publicly what their badge means and what it does not. Two of them do, in language a buyer can follow. If all four did, a great deal of the confusion in this market would resolve without anybody testing anything.

Your series has changed how I read a storefront and I would add one habit for other readers. Before looking at any mark, find the date. If there is no date anywhere on the page, the mark is decoration and the rest of the page should be read in that light.

Anonymity for the purchaser and anonymity for the supplier are different protections and they are frequently confused. The first keeps the sample honest; the second is a courtesy that mostly benefits the party being tested.

Trial populations were selected, monitored and supported in ways that the general population is not, and event reporting in a trial is systematic in a way that reporting outside one never is. Both facts push the comparison in opposite directions and neither is usually mentioned.

The placebo arm rates are substantial and almost never quoted, which makes the active arm figures look more dramatic than the comparison supports. The difference between the arms is the finding; the raw active figure is the headline.

The section on what the pharmacology does not tell us is the one I would keep if the rest were cut. Receptor occupancy does not predict a clinical response, and a mechanistic story that explains a result after the fact is not evidence that the result will generalise.

I teach this material and the hardest habit to break is reasoning from mechanism to practice. A plausible pathway is a hypothesis, and a great deal of confident advice in this area is a hypothesis with the hedging removed. None of what I have written here is clinical advice and no reader should treat it as such.

Chain of custody in a regulated laboratory is a signed record of every hand a sample passes through. In this market it is usually a courier tracking number and an assumption, and the two are described with the same phrase.

Time is the variable custody records exist to capture, and almost every account I read gives dates rather than durations. Eleven days between purchase and injection is a fact about the result; the two dates it is derived from are usually printed and never subtracted.

I collect these images. Of the thirty-one on my desk, eleven link to a report, six link to a page of reports where finding the right one is guesswork, and fourteen link to nothing at all. The proportion that survive a single click is the number worth publishing.