The Spreadsheet Will See You Now

The Decision Nobody Made

Gene Lokken was ninety-one when he fell in the spring of 2022 and broke his leg and ankle. Nothing remarkable in that, sadly… old bones, a bad landing, the sort of thing that happens in every town on every continent every day of the week. He went into a skilled nursing facility to recover, and according to the lawsuit his family later brought, his insurer stopped paying for that care after about two and a half weeks, against the advice of the doctors actually treating him. The family picked up the bill themselves, as much as $14,000 a month, until Gene died roughly a year later.

That would be a sad paragraph and nothing more, except for what the lawsuit says sat behind the decision. The complaint alleges that the cut-off was steered by a predictive tool called nH Predict, which compares a patient with thousands of statistically similar patients and estimates how much care “someone like him” ought to need. It also alleges that more than nine in ten of the denials that were appealed got overturned. UnitedHealth, for its part, says the tool is a care-planning guide rather than the thing that makes coverage decisions, and it has called the case meritless. A court will decide who is right. What interests me is something that no court will rule on… the fact that nobody in that story needed to be cruel for it to happen.

I spent years in the RAF Police and then in senior security management, and if that career taught me one thing, it is how a decision can be passed down a chain until nobody owns it. The policy writer points at the guidance. The manager points at the policy. The person on the phone points at the screen. Somewhere at the bottom of the pile there is a human being with a broken ankle, and at the top there is a number that everyone agreed was reasonable at the time.

I am sixty-six. I live off-grid in a cave in rural Spain with a handful of friends, which makes me, among other things, exactly the sort of person health economists have tidy formulas about. So I went looking for the machinery. Not the science fiction version, where a sinister ministry hands out social credit scores for kidneys, but the real thing, written down in guidance documents, procurement contracts and spreadsheet cells. What I found is that the infrastructure for rationing care by what a life is “worth” is not a future threat. Most of it already exists. It was built, mostly, by sensible people for sensible reasons. And the real question, the one this piece is about, is what happens on the day someone quietly changes what the spreadsheet is told to count.

The Arithmetic of a Life

Start with the unit of currency. Health economists measure the benefit of a treatment in Quality-Adjusted Life Years, or QALYs, and the idea is simpler than the name. One year lived in full health counts as one QALY. A year lived in a health state that people rate as half as good counts as half a QALY. Add up how many extra QALYs a new drug buys compared with what the NHS already offers, divide the extra cost by that number, and you get a price tag for a year of good life.

In England, the body that does this sum is the National Institute for Health and Care Excellence, NICE to its friends and enemies alike. From its founding in 1999 until this spring, NICE judged new medicines against a threshold of £20,000 to £30,000 per QALY. Below £20,000, a drug would normally be approved. Between the two figures, the committee would weigh things like uncertainty and innovation. Above £30,000, as a rule, the answer was no, however much a patient or their family wanted it to be yes.

In April 2026 that threshold rose for the first time in its history, to £25,000 to £35,000 per QALY. NICE reckons the change will let it approve somewhere between three and five extra medicines or indications a year, which is genuinely good news if one of them is yours. But look at how the price of a year of your life got its first pay rise in a quarter of a century. It was not because philosophers held a symposium and concluded we had been undervalued. The increase was announced on 1 December 2025 as part of a trade deal between the UK and the United States, folded into broader government efforts to keep the pharmaceutical industry sweet. Days later, the government opened a consultation proposing that ministers be given the power to direct NICE to use a specific threshold, and that NICE no longer be required to consult on methods changes that result from such a direction.

Let me be fair, because fairness is the whole point of this exercise. The QALY exists because the NHS budget is finite and pretending otherwise does not conjure money out of the air. Economists at the University of York estimate that the NHS’s existing services, from GP appointments to community pharmacy, buy a year of good health for somewhere between £5,000 and £15,000. Every pound spent on a drug that costs £35,000 per QALY is a pound that could have bought more health somewhere else, for someone whose name you will never know. That is the honest, uncomfortable logic of rationing, and every health system on earth does it, whether it admits it or not.

The point is not that the number is evil. The point is that the number exists, that it is a price on a year of human life, and that in the space of a single winter it has been shown to be negotiable at a trade summit and potentially settable by ministerial direction. Hold on to that, because it matters for everything that follows.

The Year They Asked What You Contribute

Here is the part of this story that most people have never heard, and it is the part that turns a dystopian thought experiment into a matter of public record. The leap from clinical value to social value, from “how much health does this buy?” to “how much is this person worth to the rest of us?”, is not something I have to imagine. A British government department formally asked for it.

It began with Andrew Lansley’s 2010 promise that the NHS would move to “value-based pricing” for medicines. By the time civil servants had turned that slogan into something workable, the Department of Health was asking NICE to fold a new factor into drug appraisals called “wider societal benefit”. Strip away the polite wrapping and it was a measure of how much people living with a given illness produce for society, set against how much they consume. Read that sentence again slowly. Produce versus consume. It is the language of a warehouse stock-take, applied to patients.

To its genuine credit, NICE balked. In January 2014 its board rejected the measure, and the reasoning in its own papers was disarmingly blunt: however broadly you define the concept, age affects what we are able to contribute and what we take from society. In other words, any formula that rewards contribution will, as night follows day, mark down the old. NICE declared it would not let age itself tip a recommendation against a treatment, and floated a softer alternative called “wider societal impact” instead, based on how much healthy life a condition takes away.

Even that softer version went out to public consultation in the spring of 2014 and came back with a bloody nose. Respondents worried it was age discrimination wearing a lanyard. Professor James Raftery, a long-time NICE watcher, noted at the time that all but one of the consultation questions drew a “no” from 59% or more of respondents, and that he knew of no other consultation rejected quite so emphatically. By September 2014 the whole project had been quietly shelved, and the pharmaceutical industry was soon lobbying to get it back on the agenda.

So the guardrail held. It is worth asking what, exactly, the guardrail was. It was not a law. It was not a constitutional principle. It was a board of a public body with enough independence to say no, and a public consultation that the public actually bothered to answer. Now go back to the end of the last section and read that consultation proposal again, the one about ministers directing the threshold and NICE no longer having to consult when they do. I am not suggesting anyone in Whitehall is dusting off the 2014 papers. I am pointing out that the two things which stopped them last time are precisely the two things now being redesigned.

When the Rules Change Overnight

Anyone who has worn a uniform understands triage. When there are more casualties than hands, you sort, and you sort fast, and you accept that some decisions will haunt you. The principle is sound. What the pandemic showed is how quickly, under pressure, a system reaches for categories of people instead of the person in front of it.

On 20 March 2020, NICE published its rapid COVID-19 guideline for critical care. It told clinicians to assess every adult on admission against the nine-point Clinical Frailty Scale, a tool designed to predict death and institutionalisation in older people, and to use that score to help decide who would benefit from intensive care. The trouble is that the scale leans heavily on whether you can manage daily activities on your own, and that overlaps almost perfectly with the definition of disability. A fit, cheerful twenty-eight-year-old with cerebral palsy who uses a wheelchair could score as “frail” on paper while being nothing of the sort in life. Lawyers at Hodge Jones & Allen threatened judicial review, disability charities raised the alarm, and on 25 March NICE amended the guideline to say the scale should not be used for younger people, people with stable long-term disabilities, learning disabilities or autism. Five days. That is how long it took a mandatory national instruction, issued in good faith by careful people, to be shown to discriminate.

Across the Atlantic, Alabama had something blunter sitting in a drawer. A 2010 annex to the state’s emergency operations plan instructed hospitals, in a “last resort” crisis, not to offer ventilator support to patients with “severe or profound mental retardation”, moderate to severe dementia or severe traumatic brain injury. Disability advocates filed a federal civil rights complaint on 24 March 2020, a similar complaint had been lodged against Washington State the day before, and on 8 April Alabama withdrew the criteria. Nobody had written that document in 2020 in a panic. It had been written calmly, a decade earlier, and filed away for a rainy day.

Back in England, the Care Quality Commission was asked to look into “do not attempt CPR” decisions made during the pandemic. Its review, which heard from more than 750 people, found that some areas had proposed blanket DNACPR decisions for groups of people, including one commissioning group that placed a blanket ban on COVID-positive people with learning disabilities being sent to hospital, and a GP who wrote to care homes about putting blanket decisions in place. Six per cent of the 2,048 adult care services that answered the regulator’s information request said people in their care had been subject to blanket DNACPR decisions at some point since March 2020. Some families only found out when their relative was already gravely ill. The CQC was careful to say that most providers were not doing this, and that the underlying problems pre-dated the pandemic. That second point should frighten you more than the first.

In every one of these cases the bad rule was challenged and reversed. That is the reassuring half of the story. The unsettling half is who did the reversing: solicitors, charities, regulators and furious families, working against the clock while people were already in hospital beds. The system did not correct itself. It was corrected, from outside, by people who happened to be paying attention.

The Machine in the Middle

If crisis protocols show how fast humans reach for categories, algorithms show what happens when the categorising never stops. In October 2019, a team led by Ziad Obermeyer published a study in Science that should be required reading for every minister who has ever said the words “AI-enabled efficiencies”. They examined a commercial algorithm used across the American health system to decide which patients should be offered extra care. It was accurate, by its own measure. It was also quietly racist, despite never being told anyone’s race.

The trick was in what it had been asked to predict. Rather than predicting how ill someone was, it predicted how much their care would cost, on the reasonable-sounding assumption that sicker people cost more. But less money had historically been spent on Black patients with the same level of need, so the algorithm concluded they were healthier than they were. At any given risk score, Black patients were considerably sicker than white ones. When the researchers swapped cost for actual measures of health, the share of Black patients flagged for extra help jumped from 17.7% to 46.5%. Nobody had programmed prejudice. They had programmed money, and money carried the history in with it.

Now return to Gene Lokken. In February 2025, a federal judge in Minnesota allowed the core of his family’s case against UnitedHealth to proceed, and in March 2026 the court ordered the company to hand over internal documents about nH Predict, including how it was developed and whether it was designed to replace the judgement of doctors. The complaint alleges that staff were pushed to keep patients’ stays close to the algorithm’s estimate and that the company relied on the fact that very few elderly, unwell patients ever appeal. In January 2026, Senators Ron Wyden and Elizabeth Warren renewed an inquiry into UnitedHealth’s nursing home practices after fresh reports that residents may have died after being denied or delayed care. UnitedHealth denies wrongdoing, and none of these allegations has been proven at trial. But the shape of the thing is clear enough: a prediction of what “someone like you” needs, sitting between your doctor’s opinion and your care.

And lest anyone think this is a private-sector vice that governments are busy stamping out, consider WISeR. On 1 January 2026, the American Medicare programme launched the Wasteful and Inappropriate Service Reduction Model, a six-year pilot in Arizona, New Jersey, Ohio, Oklahoma, Texas and Washington. For the first time, traditional Medicare is using private technology firms armed with AI and machine learning to review prior authorisation requests for a list of services that includes knee replacements, nerve stimulators and cervical fusion. Those firms are paid, in part, according to how much spending they avert. CMS insists that every recommendation to refuse payment is made by a licensed clinician and that emergency and inpatient-only services are excluded, and that may well be true in every case. But I spent long enough in security management to know what happens when you pay the guard a bonus for every person he turns away at the gate. You do not need to tell him to be suspicious. The pay slip does it for you. Congress’s own auditors at the GAO have since ruled that the model needed to be submitted for congressional review because it substantially affects people’s rights, and lawmakers from both parties have tried to kill it. At the last count I could find, it was still running.

What the Formula Doesn’t Know (And What the Doom-Mongers Get Wrong)

This is the point in a piece like this where the comments section expects me to announce that the government is coming for your gran. I am not going to, partly because it isn’t true and partly because overstating a real problem is the fastest way to get it ignored. I have written before about how genuine institutional harms get contaminated by conspiracy adjacency, and this subject is a magnet for it.

So, plainly. The QALY, as NICE uses it, does not count your salary, your tax bill or your usefulness to the economy. It counts health gained and money spent by the NHS and social care. The 2014 attempt to bring production and consumption into the sum is precisely the thing that failed. I could find no policy anywhere in the UK, Europe or the United States that bars retired people or non-taxpayers from state-funded drugs, and no organ allocation system that scores candidates on their “civic contribution”. If someone online tells you otherwise, ask them for the document. A real mechanism always leaves a paper trail. That is how I found everything else in this piece.

But the honest version has teeth of its own. Because the QALY weights years by their quality, extending the life of someone whose baseline quality of life is rated lower produces fewer QALYs than extending the life of someone rated in full health. To a disabled person, that can look a lot like being told their extra years are worth less, and the Americans took that objection seriously enough to legislate. US law restricts the use of cost-per-QALY measures in certain federal research recommendations, and in 2024 the Department of Health and Human Services finalised a rule prohibiting any measure that discounts the value of life extension on the basis of disability from being used to deny or limit care. Fairer methods, such as counting each year of life gained equally, remain allowed.

And here is the irony that nobody on either side of the Atlantic seems keen to discuss. Britain uses an explicit, published price for a year of life, argued over in public and challenged in consultations. America has largely banned that explicit price in its public programmes, and instead lets private algorithms ration rehabilitation and surgery behind commercial confidentiality, under the soothing label of “utilisation management”. One system rations in daylight. The other rations in the dark and sends you a letter. I know which one I would rather be able to argue with.

The myth both systems share is that there was ever a golden age when healthcare saved every life at any cost. There wasn’t. There were waiting lists, postcode lotteries and consultants making quiet judgements at the end of long shifts. The difference now is not that rationing exists. It is that rationing is being formalised, automated and scaled, which makes it both more consistent and much harder to see.

The Slope Is Already Paved

Slippery slope arguments have a deservedly bad reputation, usually because they leap from one step to catastrophe with nothing in between. So let me lay out the steps that already exist, because this slope is not a muddy hillside. It has been tarmacked, signposted and fitted with handrails by people who had no idea they were building a road.

We have a price for a year of life, and we have just watched it move for reasons of trade policy rather than medicine. We have a documented instance of a government department asking for that price to factor in how much a patient produces and consumes. We have crisis protocols that, given a few days of pressure, reached straight for categories like frailty, disability and diagnosis. We have algorithms that inherit the prejudices of whatever proxy they are fed, and that cannot tell you they are doing it. And we have a growing industry of contractors whose income rises with every pound of care they help avoid. None of these pieces is a conspiracy. Each was built separately, for a defensible reason, by people who would be horrified to see them bolted together.

What would bolting them together actually look like? Not a press conference announcing that pensioners are now second-class patients. Nobody would be that stupid, and nobody would need to be. It would look like a fiscal crisis, a ministerial direction, and a new variable added to a model by a supplier under commercial confidentiality… “predicted post-treatment independence”, say, or “likelihood of return to work”, each of which sounds clinical and each of which, as NICE’s own board warned in 2014, carries age and disability in with it like mud on a boot. It would look like an appeals process that technically exists but that frail, elderly and exhausted people mostly never use. It would look, in other words, a lot like things that are already happening, just slightly more joined up.

And here is the part that keeps me awake in the cave at night, with nothing but the wind for company. Every guardrail that worked in the stories above was human, slow and noisy. A quango board that said no. A consultation that the public bothered to answer. A law firm threatening judicial review. A disability charity filing a complaint. A family that appealed, and kept appealing, and then sued. Friction, in other words. And friction is exactly what every efficiency programme, every AI procurement and every “streamlined” process is designed to remove. We are not building a machine to decide who is worth saving. We are dismantling, piece by piece and with the best of intentions, the awkward human things that used to stop one being built by accident.

Who Signs the Letter?

I do not know where the line should sit. I have read enough health economics this past week to know that the people who build these models are, on the whole, decent and serious, and that “spend whatever it takes on everyone” is not a policy so much as a wish. Somebody has to decide. Somebody always has.

What I keep coming back to is a much smaller question, the one I would have asked as a young RAF policeman standing in front of an incident report. Who signed it? When a decision about whether a ninety-one-year-old man gets another month of rehabilitation is made, there ought to be a human being with a name who looked at him, or at least at his notes, and said yes or no out loud, and who can be asked why. Not a score. Not a prediction about people like him. Him.

The infrastructure for valuing lives by their usefulness is mostly built. The only thing missing is the instruction to use it that way, and history suggests that instruction will not arrive with a fanfare. It will arrive as a minor methods update, in a consultation nobody reads, during a budget crisis everybody is too tired to argue about. Gene Lokken’s family argued. They are still arguing, more than three years after he died. I would like to think that, when my turn comes and the spreadsheet runs its numbers on a sixty-something bloke from a cave in Spain, someone will still be arguing on my behalf. I am honestly not sure that they will.