Most urinary tract infections are treated without anyone finding out what is causing them. In large primary-care studies a urine sample is taken in only about a quarter of cases; the other three-quarters are a guess, an educated one, but a guess. Across the roughly 405 million UTIs the world records in a year (the Global Burden of Disease study, 2019), that guess is usually harmless. In older people, where UTI becomes about a third of all nursing-home infections, it often is not.
Guessing compounds in three directions at once. It drives the misprescribing that feeds antimicrobial resistance; it lets an infection run undetected until it becomes sepsis; and it pushes clinicians to treat asymptomatic bacteriuria, which is to say bacteria in the urine of an older person who has no infection to cure, which does them no good and breeds more resistance still. The binding constraint is not a missing drug or a missing device. It is diagnostic truth at the moment the prescription is written, and little in the system is built to deliver it there.
The bill arrives downstream, and it is large. Antimicrobial resistance was associated with 4.95 million deaths in 2019 and held directly responsible for 1.27 million of them (the Global Research on Antimicrobial Resistance study); E. coli, the organism behind most UTIs, is the single leading pathogen in that toll, which is rising fastest in people over 70. Urosepsis is the commonest cause of sepsis in the over-65s, and deaths from it in the United States rose by around 60 per cent between 1999 and 2020. There is a weaker, newer thread associating hospital-treated infection with later dementia; I treat it as a hypothesis worth testing, not a number to bank. When Jim O'Neill's review warned that resistance could cost ten million lives a year by 2050, infections exactly like this one, common, mundane and mistreated, were the engine he had in mind.
What is striking is how much of the fix already exists. Point-of-care and at-home tests, culture-free resistance readouts, decision rules, stewardship pathways, procurement levers: they are real, and under-validated, and barely bought. What is missing is not invention but the patient work of making a test the default rather than an afterthought, which is no single company's product to sell, and so goes unfunded.
This is the same shape of instrument as the one I would point at environmental neurotoxicants in the companion to this piece: a fund that backs a mission across a portfolio of levers rather than a single intervention, precisely because the thing you are trying to move, the standard of care, lives in the gaps between every existing funder. Drug programmes fund therapeutics. Resistance funders fund surveillance and new antibiotics. Device investors fund hardware. The dull middle, where a validated test becomes ordinary practice, falls between all of them.
- Research. Validate the tests in the settings that actually prescribe, aged care and primary care, and measure whether a faster, truer result changes what the clinician does, which is probably the first thing to find out. Dr Malcolm Starkey's urinary-tract immunology group at Monash, one of the few of its kind anywhere, would anchor this.
- Awareness. Move patients and clinicians from "treat on symptoms" to "test, then treat", alongside the patient advocates who have been saying it for years.
- Policy. Stewardship standards that make testing the default, and reimbursement that pays for the test, not only the pills.
- Orchestration. A coalition of aged-care providers, primary-care networks, payers, laboratories and guideline bodies behind one dated target to cut empiric prescribing.
Two further levers sit alongside, and they are where the model earns the word lasting. The diagnostics industry is forming but under-bought; a fund can pull it to scale without taking a single equity stake, by shaping the market instead, an advance purchase or a procurement guarantee for a validated, low-cost test, so that credible demand de-risks the companies that meet the specification. Equity itself, the returnable bet, belongs to commercial capital in a separate walled vehicle, recused from the fund's grants. And what the fund leaves behind when its work is done is not itself but a changed default: test before you treat, written into the guidelines, paid for by the payers, expected by the patients. The fund is temporary. The standard is not.
One objection is worth heading off, because it is the obvious one: a campaign to test more could curdle into a campaign to treat more, which would be the opposite of the point. Done properly it is the reverse, because getting the diagnosis right is also how you stop pouring antibiotics into asymptomatic bacteriuria that never needed them. Fewer prescriptions, not more. I would build this with Dr Malcolm Starkey as academic lead and Michael Lindenmayer as co-founder; Fordcastle has an active relationship with Starling Medical, which monitors urine at home, and there are aligned patient groups the fund would convene rather than belong to. Any commercial diagnostics vehicle my firm ran would be walled off and recused, for the same reason as before: the market-builder has to be neutral to be believed.
The whole thing turns on one cheap, unglamorous act, finding out before prescribing, and on an awkward fact of plumbing: the parties who would save the money, the payers and the aged-care providers and the health services carrying the sepsis bill, are not the ones who currently pay for the test. The economics run the opposite way to most diagnostics, which earn their keep by finding more to do. This one earns its keep by licensing the system to do less: the course of antibiotics not written, the sepsis admission that never comes, the resistance the guess would have bred and the next drug it would have burned through. The saving is real but conditional on one thing the research has to settle first: that a truer, faster result actually changes the prescription; a test that leaves the clinician's hand unchanged saves nothing. Close the payment gap on a test that does change it, and a surprising amount of avoidable harm closes with it. It is not a cure for anything. It is the end of guessing, which turns out to be worth more than the system has bothered to count.