The surgery can wait. The insulin cannot.

One policy change landed on American MedTech this month, and the companies it hit didn't move the same way. HCA, the largest for-profit hospital operator in the country, cut its full-year guidance after a $400m hit and called it a surprise. Intuitive Surgical watched its US robot-procedure growth slow and lost 12% of its value in a day. In the same fortnight Abbott raised its forecast, and said the industry-wide alarm was misplaced.

Same event. Opposite quarters.

The variable that sorted them wasn't sales execution, and it wasn't pricing discipline. It was whether the procedure the device depends on can wait a year. Most of the companies on the wrong side of it never saw that line coming, because it almost never appears on the plans they build.

Detroit, 1973

This shape is older than MedTech. In the winter of 1973, OPEC quadrupled the price of oil in three months, petrol queues ran around the block, and the American car market turned overnight toward the small, efficient cars Detroit didn't make. The Japanese manufacturers who won hadn't forecast the embargo. They'd been building small cars for years, because petrol was dear at home and the roads were narrow, and they simply found themselves standing where the new demand landed. Abbott was not clever about the Affordable Care Act (ACA) any more than Toyota was clever about OPEC. The difference this time: the sorting variable was knowable in advance, and almost nobody looked.

The habit that hides it

The reason it goes unread is a single habit. Most commercial organisations model procedure growth as one blended number and treat demand as fixed clinical need. Demand is not fixed clinical need. A meaningful share of it is contingent on coverage, and the share contingent on coverage is exactly what a coverage change removes. Blend the two together, and you've hidden the thing that will hurt you. Averaging is the act of concealment.

What happened upstream

The upstream event was political. At the end of 2025, the enhanced premium tax credits propping up the ACA marketplaces expired, and Congress didn't renew them. On the numbers from KFF and the Urban Institute, that means an average marketplace premium rise of around 114%, more than a thousand dollars a year for many enrollees, and an estimated 7.3 million people losing marketplace coverage in 2026, of whom roughly 4.8 million become uninsured. None of that shows up on a sales dashboard. All of it changes who walks into a clinic.

Now look at what the split actually did. At HCA, inpatient surgeries fell 2.3% and outpatient surgical procedures fell 3.4%, a payer-mix shift the company tied directly to patients losing exchange coverage. Intuitive told a similar story: US da Vinci procedure growth slowed to 12% from 14, and the slowdown sat in the benign categories patients can defer. Abbott went the other way for one reason. Its franchises sit in diabetes, cardiovascular disease and cancer, and nobody drops their insurance to avoid treating the things that kill them. Robert Ford, its chief executive, called the fear that coverage losses would dent the device industry a flawed assumption. That morning, Boston Scientific, Stryker and Medtronic all traded up around 5%.

Deferability is knowable in advance

The uncomfortable part is that deferability is knowable in advance. It's a property of the portfolio, mappable before any shock arrives, and it's almost never mapped. A team that had segmented its revenue by deferability could see this coming in the autumn of 2025, when the enrollment projections were already public. The team running on a blended growth assumption met it as a surprise. Surprise, incidentally, was the exact word the most exposed operator reached for.

There's a version of this that reads as bad luck and a version that reads as a capability gap, and the gap between them is the whole point. The bad-luck reading says the macro environment is beyond the commercial team's control. True, and beside the point. The shock was never the controllable thing. What was controllable was whether anyone in the building was watching the variable that moved it. So whose job was that?

And here's the good news hiding inside the bad. Abbott read nothing. It was standing in the right place for reasons that predate the ACA (the same way Toyota's small cars predated the oil shock). Which means the thing that sorted the winners wasn't a stroke of genius you can't copy: it was a signal, sitting in plain view, that you can choose to watch on purpose while the market waits to be told.

Which makes this quarter less a slowdown than a sorting. The subsidies are gone, the next open enrollment will move the picture again, and the second half is more likely to sharpen the split than soften it. The device serving a deferrable procedure and the device serving a non-deferrable one are now on visibly different paths, and the distance between them has nothing to do with which team sold harder.

One more thing, because it's my trade. The commercial leader who reads demand as a coverage-contingent variable, the way those Japanese manufacturers read a market they hadn't predicted, is a genuinely different profile from the one hired to accelerate volume in a market assumed to keep growing. Whether that distinction makes it onto the specification is a separate question. A year ago, on most of the ones I saw, it didn't.

The one line to split your forecast on

Split your forecast on one line first: can the procedure each product depends on wait a year without harm? The share that can is your exposure, and it was knowable in the autumn, not the quarter.

What is already stacking up behind it

And this wasn't a one-off. The same kind of signal is already stacking up behind this quarter, in the same public record, upstream of the same forecasts.

The clearest is Medicaid. The 2025 reconciliation law carries the largest Medicaid cuts since the 1960s, and the demand-side bite starts in December 2026, when work requirements and six-monthly eligibility checks begin taking people off the rolls. A state that tried the same paperwork before lost tens of thousands in a matter of months, most of them still eligible. That's the subsidy story again, larger and slower, and it sorts device portfolios by exactly the same variable.

Behind it sit two more: a Section 232 investigation that could put tariffs on imported devices from the summer, which sorts companies by supply chain rather than by deferability, and a Medicare shift toward site-neutral payment that quietly cuts the reimbursement under the procedures many devices depend on.

I notice these the way I noticed the deferability line, not because I could model any of them, but because they keep landing in the space between what a company fears and what it puts on the brief.

So the question stands, with more names on it now. When did Medicaid first appear in your forecast? The tariff probe? The site-neutral cut?

If the honest answer is none of them, the data isn't the problem. It's been public all along. The problem is whether anyone in the building is paid to read it.