Every letter we have printed

Why did you submit only two vials for sterility testing when the whole article argues that the sample size is the problem? Two is worse than twenty by exactly the argument you make.

Reading across your coverage, the thing that most changes reported reasons is who is asking. That should be a headline finding rather than a methodological footnote, because it undermines comparison between every study in the field.

Stopping because a goal was reached is a legitimate reason and it is barely represented in the literature, which studies discontinuation mostly as failure or intolerance. A planned stop and an abandoned course are different events.

Two blind determinations from the same lot at two different laboratories is worth more than four determinations of four lots at one. Method variance is the largest term in most of these datasets and almost nobody designs to measure it.

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 report format is part of the product and it is priced as though it were not. A one-page summary and a full package with the chromatogram, the suitability data and the audit trail cost the laboratory very different amounts of attention, and most price lists distinguish them by a few pounds.

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.

Creatinine is a good example of a value that moves for reasons unrelated to the kidney. Muscle mass, recent exercise and dietary intake all shift it, and during rapid weight change at least two of those are moving at once.

You give the reference change value for ALT as about sixty per cent, which strikes me as so large as to make routine monitoring of it pointless. Is that your position?

I stopped treatment fifteen weeks ago and my panel is worse than I expected. I had a panel at five weeks that looked fine and I had assumed I had escaped. Your point about the twelve-week timing was the explanation nobody offered me.

Degradation products are not inert and the literature on them is thin for most sequences in this market. A discussion of purity that only counts what is lost, and never asks what it became, is half a discussion.

A general point about rate. Everything here is kinetics, not thresholds, and the popular framing of a material being fine until it suddenly is not has no basis in the chemistry. Loss is continuous and the question is how much.

The length of the interruption must matter and no threshold has been established anywhere. Advice that specifies a duration is specifying a number nobody has measured.

The practical advice I would want printed is to write down the date of the last dose. It sounds trivial and it is the single piece of information that makes the re-titration question answerable, and it is the one people most often cannot supply.

The trial protocols permitted holding and delay, which is a point that gets lost. The schedules people treat as rigid were operated with discretion in the studies that produced the evidence, and the protocols say so in the escalation sections.

You describe the plateau as an energy-balance event and dismiss receptor desensitisation. Is there not a third possibility — that adherence quietly falls off at around a year and the plateau is partly a behavioural artefact of the trial rather than a physiological one?

An approved schedule is a regulatory artefact reflecting what was submitted and what was studied, and it is read as a statement about what is optimal. Those are different things and the gap between them is not evidence of anything.

A plateau is defined differently in every paper I have read and not defined at all in most general coverage. Before arguing about whether one occurs, somebody needs to say over what interval and by what margin.

The most useful thing your department has published on this was the comparison of curve shapes across studies with the horizons marked. Seeing where each line stops does more for a reader’s calibration than any amount of argument about mechanism.

A sterile field is a concept and not a product. The bench, the hands, the air movement and the sequence all belong to it, and a person working carefully on a clean surface is doing more than one working carelessly with expensive supplies.

Single-use means single-use, and the reason is mechanical as much as microbiological. A needle blunts on first insertion, and the second insertion is measurably rougher. This is one of the few places where the practical argument and the aseptic argument agree completely.

A simple scheme beats a complicated one because it gets followed. Dividing the available area into quadrants and moving on a fixed rotation is enough, and elaborate maps do not survive contact with a busy week.

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 note from a reader in a laboratory rather than a clinic. The technical material is accurate often enough that I now check my assumptions against it rather than the other way round.

Ambient during transport home from a delivery point is part of the in-use history and is never counted. A parcel collected on a warm afternoon and carried across town has had an excursion before anybody opened it.

In-use stability is where the real losses happen and where there is least data. A dry powder is forgiving; the same peptide in solution at room temperature is not. Every practical question a buyer has begins at the moment the diluent goes in, and the documentation stops at the moment before.

Phase-change packs are worth the modest extra cost over water ice and almost nobody asks for them. They hold a narrower band for longer and they do not produce the initial sub-zero shock that a freshly frozen gel pack in direct contact with a vial can.

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.

A design nobody has run: withdraw, follow, then offer resumption and follow again. It would answer the question most readers actually have, and it is longer and more expensive than anybody has been willing to fund.

Reasons collected by a supplier, a clinician and a researcher would produce three different distributions from the same population, because people say different things to different audiences. Every dataset in this area is an artefact of who was asking.

Where a reason is recorded by a clinician rather than by the person, the categories reflect what the clinician was listening for. That is not a criticism of clinicians; it is a property of any coded record, and it is rarely acknowledged in the resulting statistics.

Administrative burden is a reason with no name. Reordering, storing, reconstituting, timing — a small load that accumulates, and people stop because it became tiresome rather than because anything went wrong. It is invisible in every dataset because no question asks about it.

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.

Rate of change is the variable most likely to modify the ratio and the one least often reported alongside it. Two studies with the same endpoint and different trajectories are not measuring the same thing.

The order of the questions matters. Whether training preserves composition during loss and whether it accelerates loss are different studies, and results from one are frequently used to answer the other.

Comparing substudies across trials requires the scanner model and software version, because absolute values differ between them. Almost nobody who compares them checks, and the manufacturers are quite clear that they should.

A substudy is powered for the parent trial’s primary endpoint and not for its own, which is why the intervals around composition figures are so wide. Quoting the point estimate without the interval misrepresents the design as much as the result.

We ask customers to send the vial unopened and we record the seal condition on arrival. It has produced two useful findings in three years, both about transit rather than about the supplier, and neither would have been visible if the sample had arrived as a powder in a bag.

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.

VendorInvestigate does not measure anything and you have grouped it with three laboratories under the heading independent testing. That is exactly the conflation your article says the market makes.