Original measurement

Research

Pharmaceutical logistics is full of numbers nobody can check. This is where we put the ones you can: our own recordings, the export they came from, the rule that governs which figures may be published from them, and an explicit account of what each study cannot tell you.

The studies How this works

What is admitted here

A committed source dataset
A published data rule
Figures recomputed from source on every build
Stated limitations
Nothing that is only an opinion

What counts as research here

An article is not research because it is long, and a number is not a finding because it appears next to a chart. Everything listed on this page meets all five of the following. Everything that does not is commentary, and lives in the blog.

A committed source dataset. The raw export the study is built from is in the repository that builds this site, named on the study page, with its file identified by hash.
A published data rule. Each study states, before any figure, which columns of the export it is allowed to publish and what it may not do to them. Nothing is smoothed, resampled or interpolated.
Recomputation on every build. Every published figure is recalculated from the source export when the page is built, and cross-checked against the summary the recording device itself wrote. A drift beyond tolerance fails the build rather than publishing.
Stated limitations. What the study does not establish is written down as prominently as what it does — including, in both cases below, an explicit statement that the work is not a medication stability determination.
Measurement separated from interpretation. A reading is a reading. What it implies for a particular product is a separate claim, and belongs to whoever holds that product’s labelling.

What is measured, and what is not

Both studies below measure environments. A logger records the air around it. It does not record what a medication in that air was doing, whether a product was present at all, or whether any labelled range was breached for any specific unit. Where a published standard is quoted — controlled room temperature is defined by a mean kinetic temperature of 25 °C — it is quoted so a recorded number can be read against a definition. Quoting a standard is not a compliance determination, and NoazRX does not make one.

Provenance and updates

The studies carry the recording device’s serial, model, firmware and log interval, the source file path, and the recording window. If a study is revised, the revision changes the page and the dataset together, because they are generated from the same script. A figure on this site that cannot be traced back to an export is a figure we should not have published.

What this page is not

It is not a performance report. There is no on-time percentage, no cold-chain compliance score and no delivery-volume figure here, because none of those has a measurement behind it that anybody outside the company could check. When one does, it will appear here with its dataset, or it will not appear at all.

Studies

Published work

165 days inside a medication delivery vehicle

Observational environmental-exposure study. A single-use logger recorded vehicle-cabin air at ten-minute intervals for 165 continuous days, and the study asks how much of that time the cabin sat outside the range an ambient medication is labelled for.

50.6 °CHighest reading

−14.4 °CLowest reading

23.7 °CMean kinetic temperature

3,937Hourly windows recorded

Data rule

The published series is the exported hourly minimum and maximum. The exported hourly average column is read for exactly one purpose — computing mean kinetic temperature, which is the statistic controlled room temperature is defined by, so a study measuring against that definition cannot avoid it. Every other figure comes from the exported minima and maxima. Nothing is smoothed, resampled or interpolated, and an hourly window counts once, as one hour.

What it cannot say

One vehicle, one logger position, one region, one recording period. The logger stayed in the cabin continuously — parked, overnight and on non-delivery days — and the export carries no trip, ignition, location, climate-control, door or sunlight markers. Working-hours figures are therefore a proxy for the exposure a delivery would meet, never a measurement of transit. An hour whose exported maximum exceeds a threshold is an hour in which the cabin reached above it, not an hour spent entirely above it: the export does not carry the within-hour distribution needed to say that. It is not a fleet study, not a packaging test and not a medication stability determination.

Read the study — the page publishes the exposure dataset as CSV and JSON alongside it.

Concurrent vehicle and tote logs during ambient transport

Observational transport-risk case study. Two loggers recorded concurrently over 45 days of overlap between 22 April and 6 June 2026. One stayed loose in the vehicle cabin. The second sat inside an ordinary uninsulated tote whose exterior had been wrapped in reflective material as a makeshift exploratory prototype, assembled by the team out of ordinary materials.

Data rule

The published dataset is the synchronized record of the two exports over their overlapping period, derived from the source logger exports retained in the repository and identified by file hash. The study deliberately publishes no averaged statistic.

What it cannot say

During many overlapping periods the air inside the tote changed less dramatically than the cabin readings. The study does not establish that the reflective exterior caused that. The dataset does not isolate the wrap from tote construction, handling, location, vehicle climate control, sunlight, contents or thermal mass; the exports do not record when the tote moved, whether it was open or closed, or when each dwell period began and ended. No performance rating, validated hold time or defined protective capability is claimed for the material or the tote, and nothing here is a medication stability determination.

Read the case study — the page publishes the synchronized evidence dataset and an evidence manifest.

Why we publish this

The figures a logistics provider usually publishes are the ones nobody can check

On-time percentages, compliance scores and delivery volumes are the standard furniture of a logistics website, and they share a property: the company that publishes them is the only party that can see the data behind them. A buyer comparing three providers on their published percentages is comparing three marketing decisions.

Measurements of the transport environment are different. Anybody can put a logger in a vehicle. Publishing what ours recorded — including a 50.6 °C reading, which is not a flattering number — is useful precisely because it is checkable and because it describes a problem the whole industry has rather than an advantage we claim.

The operational use is straightforward. Knowing what a cabin does across a year is an input into how a temperature-sensitive lane is designed, into what a lane profile has to account for seasonally, and into which parts of a final-mile route deserve monitoring rather than assumption. It sits inside the broader picture of pharmaceutical logistics.

Ask us for the dataset

Both studies publish their derived datasets on the page. If you are assessing a lane and want to talk about what these measurements do and do not imply for it, that is a better conversation than a compliance percentage.

Talk to NoazRX Cold chain and refrigerated transport