My doctoral research asked a narrow question. Helicobacter pylori, the bacterium behind most stomach ulcers, coats its cell membrane in glucosylated cholesterol taken from its host. I wanted to know what that coat was for. Part of the answer, published in mBio in 2018, is that it holds the cell wall together and helps the organism survive antibiotics that would otherwise reach their target. Take the coat away and the bacterium becomes fragile and far easier to kill.
Can a poor quality medicine cause antibiotic resistance?
Yes. Resistance is selected by exposure to an antibiotic at a level that harms some bacteria without killing them. A product containing less active ingredient than its label claims, or one degraded by heat or age, can deliver exactly that partial dose, so quality failures in the supply chain become a biological problem in the patient.
That work sat inside a very small frame: one organism, one enzyme, one membrane. What changed my view of it was leaving the laboratory. I now source active pharmaceutical ingredients and finished medicines from manufacturers in China and India for buyers in Yemen, the wider Middle East and the United Kingdom. The first time I looked properly at a supply chain into a low income market, I recognised the problem. It was my thesis, running in the world instead of in a flask.
What resistance actually looks like at the bench
Resistance is often described as bacteria becoming stronger. That is the wrong picture. Nothing gets stronger. A population of bacteria already contains variation, and an antibiotic applied at full strength kills essentially all of it. Applied at a fraction of full strength, it kills the most susceptible cells and leaves the rest alive, along with whatever made them less susceptible. Those survivors reproduce. The drug has not created anything; it has sorted the population and thrown away the half that was easy to treat.
This is why the dose matters more than almost anything else. A full course at the right concentration is a clean sweep. A partial course, a diluted product, or a degraded one is a selection experiment, and the thing it selects for is exactly what you did not want.
Why the dose is a supply chain question, not only a clinical one
In the laboratory I controlled concentration precisely. I weighed the compound, dissolved it, verified it, and knew what every cell was exposed to. A patient has none of that. What reaches them is the end of a long chain: an active ingredient made at one site, formulated at another, packed at a third, shipped through several ports, warehoused, transported again, and finally handed over a counter. Every step is an opportunity for the amount of active ingredient that arrives to differ from the amount printed on the box.
Three places the dose quietly changes
At manufacture. A product can be made with less active ingredient than declared, whether through a genuine process failure or deliberately. This is what the certificate of analysis and the batch record are supposed to catch, and it is why reading those documents properly is not administrative work.
In transit and storage. Many products are stability tested under defined conditions and are only guaranteed to hold their stated content within those conditions. A container standing in the sun at a port, or a warehouse without functioning cooling, subjects the product to conditions its stability data never covered. The label still reads the same. The contents may not.
At the point of sale. Where medicines are expensive relative to income, courses are split, sold loose, or stopped when the patient feels better. This is usually discussed as a patient education problem. It is at least as much a pricing and availability problem, and pricing and availability are decided upstream, by people like me.
What I check now that I did not think about as a researcher
The habit I brought from research was to trust a number only when I knew how it was produced. That habit transfers directly, and it is the single most useful thing I carried out of the laboratory.
- Which site actually made it. Not the company name, the physical site. A trading company can hold impeccable documents from a manufacturer and supply material from somewhere else entirely.
- Which specification the certificate was checked against. A result of 98.4 per cent means nothing until you know the range it was compared with, and different standards set different ranges for the same substance.
- Whether the tests I care about are even on the certificate. The tests that are missing tell you more than the ones that passed. A certificate is not wrong when it omits something; it is silent, and silence is not a pass.
- What the storage conditions were, all the way through. Not what they were supposed to be. What they were.
A clean certificate is not a clean product. It is a claim about a batch, made by a party with a commercial interest in the answer, tested at a moment that is usually months before the material reaches its patient. Reading it carefully is necessary. Believing it without reading it is how a paper problem becomes a clinical one.
Why this is not only a regulator's problem
It is comfortable to treat medicine quality as something regulators handle. In well resourced markets, largely, they do. In markets with a small inspectorate, limited laboratory capacity and long informal supply routes, the checking that actually happens is whatever the buyer chooses to do. That is an uncomfortable amount of responsibility to sit with a commercial party, and pretending otherwise does not make it less true.
It is also why I built PharmaTrust at NJMC. Reading a certificate of analysis properly against the specification it claims takes about an hour, and almost nobody in a working procurement department has that hour for every document that arrives. The software does that reading and reports what it finds, what it could not establish, and why. It analyses documents rather than medicines, so it cannot tell anyone that a product is good. What it can do is stop the review being skipped for lack of time.
The through line
I did not expect a thesis about a bacterial membrane to be useful in a trading company. It has been, but not in the way I imagined. The transferable part was never the biochemistry. It was the discipline of asking what a measurement actually shows and refusing to accept an answer whose derivation I cannot see.
Every substandard course of antibiotics is a small selection experiment run on a person, with a result that outlives the patient and the batch. The organisms that survive it do not stay where they were made. That, in the end, is the argument for treating the quality of a medicine as a scientific question rather than a procurement formality.