Every letter we have printed

A translation problem worth noting: many buyers in this market read these pages in a second language, and an image survives translation where a paragraph of qualification does not. The badge is therefore doing more work for exactly the readers least able to interrogate it.

I teach a seminar that uses two of your pieces as set reading. Students arrive expecting advocacy and are visibly unsettled by the refusal to supply any.

Time course is the missing dimension in most reporting. An effect present for four days and an effect present for four months appear as the same entry in an incidence table, and they are not remotely the same thing.

The run-in and titration phases are where most of the reported events occur and where the fewest measurements are taken. It is an understandable design choice and it means the noisiest period is also the least documented.

Labelling changes are a poor proxy for evidence strength, since they follow regulatory process rather than publication. A label updated in one jurisdiction and not another usually reflects timing, and it is read as disagreement about the science.

The central and peripheral contributions are separable in animal work and not in human studies, and coverage tends to attribute everything to whichever the writer finds more intuitive. The honest statement is that both are implicated and the proportion is unknown.

The comparison a reader needs is between the rate in the study and the rate in the comparator arm, and secondary coverage almost always supplies only the first. Without the second, no statement about attribution is possible at all.

The physiology of regain after any weight loss is well described and predates all of this pharmacology by decades. Presenting it as a novel finding about these agents misplaces a body of work that would help readers understand what they are seeing.

Composition of regained tissue is barely studied and is the question I would most like answered. Whether what returns resembles what was lost has practical consequences and there is very little data on it.

I have read this department for two years and the finding that has stayed with me is how little of the long-horizon question anybody has actually studied. That is not a criticism of the coverage; it is the coverage’s main result.

The analysis population in these designs is defined at randomisation, which is after the run-in has removed everybody who did not tolerate or did not respond. Every number downstream describes that filtered group and is quoted as though it described anybody.

Composition endpoints are collected in a minority of these studies and reported in fewer. A weight curve after withdrawal without a composition measurement leaves the most interesting question unanswered.

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.

Error reporting in this area is entirely voluntary and therefore entirely unrepresentative. What circulates is the set of mistakes people were willing to describe, which is not the set of mistakes that occurred.

The convenient reconstitution is the one that makes the arithmetic trivial rather than the one that makes the volume small. Choosing a diluent volume so that a graduation corresponds to a round figure removes an entire class of error, and it costs nothing but a moment’s thought before the needle goes in.

You spend a page on the four-line calculation and then publish a reconstitution table anyway. Are you not providing exactly the pre-computed number you warned against?

A unit is not a dose and the word does the damage. It is a mark on a barrel. What it corresponds to in milligrams depends entirely on the concentration in the vial, and two people using the same mark on the same syringe can be drawing quantities that differ several-fold.

Every custody scheme I have seen assumes a single sample moving forward in time. Real programmes split samples, store halves, retest months later and combine results, and the record-keeping vocabulary has no way to describe that.

A practical point about labelling for submission. Write the identifier on the tube, not on a sticker, and not on the bag. Stickers come off in cold conditions and the laboratory then has a tube with no name and a bag with no contents.

For a small buyer the arithmetic is brutal. Testing a vial can cost a meaningful fraction of what the vial cost, which is why individuals do not test and why the published record is dominated by sellers and by publications like yours. The evidence base has a wealth filter on it.

The residual moisture in a lyophilised cake determines which pathways are available to it, and it is measurable, cheap and absent from most certificates in this trade.

Aggregation is not degradation in the chemical sense and it is reported as though it were. The molecule is intact and its physical state has changed, which is a different problem with different consequences.

Degradation pathways are sequence-specific and the discussion in this market is generic. A sequence with a susceptible pair of residues has a pathway that a sequence without one does not, and general advice cannot capture that.

An excipient note. Mannitol crystallises and gives a firm elegant cake with poor protection; sucrose stays amorphous and protects rather better while looking worse. Buyers judge cakes by appearance, and appearance and stability are pulling in opposite directions.

Mechanistic explanation is doing rhetorical work in this area that the evidence does not support. A plausible pathway makes an effect feel established, and several of the pathways in general circulation are supported by preclinical work alone.

Gastric emptying studies use several different methods and the numbers are not interchangeable. A scintigraphic half-time and a breath-test result answer related questions with different sensitivities, and papers are compared across methods routinely.

The figures that circulate are cumulative incidence over a whole trial, and they are read as though they described how somebody feels in a given week. Those are very different quantities, and the second is much smaller than the first. Your table separating them is the most useful thing published on this in a while.

The first-dose period accounts for a disproportionate share of reported events, and reporting a single study-wide incidence hides that structure entirely. A rate by week would be far more informative than a rate by trial.

The supply-side reason for holding is never mentioned and is obviously real: a slower schedule costs less. In a market where people pay directly, economics is part of every schedule decision.

Tolerability-led escalation is the practice your article describes most sympathetically and it is worth naming what it gives up. Holding a dose because of symptoms is reasonable; holding it indefinitely because a schedule was never revisited is drift, and the two look identical from the outside.

The published curves are almost always means with standard errors, which narrow as the sample grows and give an impression of precision about a population that says nothing about an individual. Standard deviations would be more honest and are rarely plotted.

An observation about scale. Custody discipline is easy at one sample a month and hard at fifty, which is why the programmes with the most data have the weakest records. The failure is a consequence of volume rather than of care.

Custody protects the supplier at least as much as the buyer. A result produced from a sample of unknown history is not evidence against anybody, and suppliers who understand that ought to be the loudest advocates for the practice.

Blinding at the laboratory end is the easy half and the half everybody talks about. An analyst who does not know which supplier they are testing is protected against one bias and against none of the others in the chain.

Pre-analytical variables account for far more surprising results than laboratory error does. Fasting state, posture during the draw, tourniquet time and the interval before separation all move numbers, and none of them appears on the report.

Tourniquet time changes several results measurably, and it is a property of the phlebotomy rather than of the person. A difficult draw produces a different set of numbers from an easy one.

The published trials in this class collected some of this data and reported very little of it, which is a genuine gap rather than a negative finding. It should be described as unreported rather than as unchanged.

My eGFR has risen from 71 to 84 over fourteen months of treatment and I have lost twenty-six kilograms. My prescriber described this as the drug protecting my kidneys. Having read your creatinine section, I suspect it is mostly that I have less muscle. Which of us is right?

I would settle for a much weaker reform than blinding: a published statement of who selected the sample. It is one line, it is not contentious, and it would let a reader weigh every result on the page correctly.

The suppliers who cooperate with testing are, almost by definition, the ones with least to fear. A programme built on cooperation therefore samples the top of the market and reports the result as though it described the whole of it.