What a 1957 Pacemaker Teaches MedTech About the Slowest Part of Innovation

Mikhail Kukva

Oct 20

8 min read

Devices are easier to invent than ever. Getting a health system to pay for a genuinely new one still takes about a decade.

During the final module of my Executive MBA at the University of Minnesota's Carlson School of Management, I visited The Bakken Museum in Minneapolis. Alongside Benjamin Franklin's electricity experiments and a Frankenstein-themed laboratory (a nod to the film that first drew Earl Bakken to electricity as a boy), one room holds the first battery-powered pacemakers. As a physician who moved into the medical device business, I found it hard to walk past. I took a selfie beside the display case, and then I started reading about how it got there.

The story is usually told as a moment of invention. It was really two stories: a device that took four weeks to build and a business that took eleven years to pay off. That gap raised a question I wanted to answer. Sixty years on, is it faster, better or easier to take a medical device from a working prototype to something a health system pays for? Having looked at heart valves, spinal implants, surgical robots and artificial intelligence, I think the answer is that invention has become easier and getting paid has not. For a genuinely new kind of treatment, it still takes about a decade.

The invention took four weeks

In 1954, University of Minnesota surgeon C. Walton Lillehei pioneered cross-circulation open-heart surgery to repair congenital heart defects in children. The operations could damage the heart's electrical conduction and cause complete heart block. From early 1957 his team treated it with a Grass physiological stimulator wired to an electrode on the heart muscle. The stimulator sat on a wheeled cart and ran on mains power, which left fragile children tethered to a wall socket.

By that summer Lillehei wanted something better. In a letter dated September 16, 1957, now held in the University of Minnesota archives, he described "a new feather-weight transistorized pacemaker which weighs only a few ounces and which is activated by a dry cell battery." He took the problem to Earl Bakken, a University of Minnesota electrical engineering graduate. In 1949 Bakken and his brother-in-law, Palmer Hermundslie, had started Medtronic in a Northeast Minneapolis garage, repairing electronic equipment for the university hospital. That is why Lillehei knew whom to ask. Bakken adapted a transistorized metronome circuit he remembered from Popular Electronics and within four weeks delivered a battery-powered pacemaker about the size of two cigarette packs. Lillehei had it on a patient the next day.

The business took eleven years

The device was a clinical breakthrough, but Medtronic was still a small repair shop with a thin balance sheet. Its next step came in late 1960, when Hermundslie flew to Buffalo to meet surgeon William Chardack and engineer Wilson Greatbatch, who had built an implantable pacemaker, and signed for the exclusive rights to make and sell it. Fifty orders arrived within two months, at $375 each. Sales grew faster than profits. Medtronic lost money in the early 1960s and needed a bond issue and a bank loan in 1962 to fund its growth.

The decisive change came from outside the company. Congress created Medicare in 1965, and from July 1966 it paid for the care of Americans over 65, the age group most likely to need a pacemaker. By fiscal 1967–68, Medtronic's sales exceeded $12 million and its profit passed $1 million. That was eleven years after Lillehei's request, and it took three separate steps: the external pacemaker, the rights to the implantable one, and a national payer.

Sixty years on, a new treatment still takes a decade

Transcatheter aortic valve replacement (TAVR) delivers a heart valve by catheter instead of open surgery. Alain Cribier performed the first human implant in Rouen, France, in 2002. Edwards Lifesciences' Sapien valve received FDA approval in 2011, and Medicare coverage followed in 2012. That is ten years, for a device far more complex than a 1957 pacemaker.

TAVR is not an outlier. A 2023 study in JAMA Health Forum followed 64 novel devices authorized by the FDA between 2016 and 2019 and found a median of 5.7 years between FDA authorization and Medicare coverage. Only 9% were covered within two years. Add the years it takes to develop a device and win FDA authorization in the first place, and a decade is close to the norm.

When the payer says no

Spine, where I spent more than five years running a business unit, offers a harsher lesson. The Charité artificial disc, designed as an alternative to fusion surgery for degenerated discs in the lower back, received FDA approval in October 2004. Its pivotal trial, like that of its main competitor, had excluded patients over 60. In May 2006 Medicare ruled that the implant was not reasonable and necessary for that age group, and in August 2007 it extended the ruling to lumbar disc replacement as a procedure. Since Medicare mainly covers people over 65, that closed the program to the procedure almost entirely. The device had cleared the FDA and still lost its largest potential payer. Bakken never faced that risk, because in 1957 there was no national payer in a position to refuse.

Improving what is already paid for

What has changed since 1957 is not the cost of inventing an entirely new treatment. It is that most innovation now improves a treatment that already exists and is already paid for. Of the roughly 3,100 devices the FDA authorizes in a typical year, only about 70 go through the routes reserved for genuinely novel devices. The rest reach the market by showing they are substantially equivalent to something already sold.

That route is faster. Intuitive Surgical was founded in 1995 and won FDA clearance for the da Vinci surgical robot in 2000. Because robotic surgery is billed under the same code as the operation itself, hospitals needed no new payment decision; the cost sits in their capital budgets instead. Mazor Robotics, founded in 2001, received the first FDA clearance for a spine surgery robot in 2004. Clearance came quickly, but building the business took another fourteen years, until Medtronic acquired Mazor for $1.7 billion in 2018 and made its technology part of its spine portfolio. An improvement can shorten the road to payment without shortening the road to scale.

AI runs on the pacemaker's clock

Artificial intelligence, the most talked-about technology in medicine, looks like the fastest route of all. Software needs no factory, and the FDA had authorized more than 1,400 AI-enabled devices by the end of 2025. Yet the first autonomous AI diagnostic followed the pacemaker's timeline rather than the robot's.

IDx (now Digital Diagnostics) was founded in 2010 by Michael Abràmoff, an ophthalmologist who also holds engineering professorships at the University of Iowa, to detect diabetic retinopathy without a doctor reading the images. In April 2018 its IDx-DR system became the first autonomous AI diagnostic authorized by the FDA. A national Medicare payment rate followed in November 2021. From founding to national payment took eleven years, the same span as Lillehei's request to Medtronic's first $1 million year, and Abràmoff had been working on the problem since 1997. Once again, the breakthrough came from a clinician and engineer working inside a university.

Followers move faster. In cardiac imaging, the first AI product to estimate coronary blood flow from CT scans was cleared in 2014 and received a permanent billing code in 2025, about ten years later. The first AI tools for analyzing coronary plaque were cleared in 2020 and received their permanent code in January 2026, about six years later, because they could build on the evidence and billing pathway the pioneer had created. One category is not a trend, but the mechanism is clear. My own rough estimate for a fast follower is eight to ten years from a working prototype to routine payment, which is shorter than the pioneer's decade but not by much.

Put side by side, the pattern holds. The pacemaker took nine years from Lillehei's request to Medicare payment, TAVR ten from first implant to coverage, and IDx-DR eleven from founding to a national payment rate. The starting points differ, but the order of magnitude does not. Improvements that fit existing payment codes reached patients within three to five years of founding. A device that cleared the FDA without convincing the payer, like the Charité disc, lost its main market altogether.

Closing the second gap

Which brings me back to my question: is it faster, better or easier now? The most promising change is very recent. In April 2026 the FDA and CMS, the agency that runs Medicare, announced RAPID, a pathway under which both agencies agree before a clinical trial begins on the evidence they need, and then review in parallel rather than one after the other. Medicare would propose its coverage decision on the day of FDA authorization and finalize it within 60 to 90 days, compared with the median of almost six years found in the 2023 study.

RAPID is still a proposal. CMS published it for comment in August, and the comment period closed on October 13, 2026. Its reach is also narrow. A device needs FDA breakthrough designation and a trial that enrolls Medicare patients, with outcomes agreed by both agencies in advance. Since almost every device reaches the market without such a trial, CMS expects only 40 to 60 devices to qualify at first. RAPID will not make anyone invent faster. What it can do is remove most of the wait between FDA approval and payment for companies willing to produce the evidence up front.

A thought experiment

Consider a hypothetical startup building AI for cardiovascular imaging, with two products. The first automates measurements that radiologists already take on CT, MRI and ultrasound scans. It improves a service that is already paid for, so it resembles the robot more than the pacemaker. It can sell to hospitals on time saved and consistency, or follow the pioneers who have already opened billing codes for AI analysis of cardiac scans. In neither case does it need Medicare to create something new.

The second product is a portable ultrasound device that detects and measures narrowing of the carotid arteries in people without symptoms, with the aim of preventing strokes. That is a new category, and its obstacle is not the software. The U.S. Preventive Services Task Force recommends against screening adults without symptoms for carotid artery narrowing, because no study has shown that such screening reduces strokes. A product like this needs exactly the kind of outcome trial RAPID is designed to reward: one that shows finding the narrowing early leads to fewer strokes. With Medicare patients enrolled and endpoints agreed in advance, it could reach payment years sooner than any pioneer in this article. Without that trial, it risks spending Bakken's decade waiting, or meeting the Charité disc's fate.

Proximity is the real advantage

None of this makes Bakken's story less impressive. It makes it more instructive. What Bakken had, and what most innovators still lack, was proximity. His company already worked inside the hospital where the problem lived, so a surgeon's request became a working device within a month. The hard part of medical innovation has moved from the workbench to the payer, but the lesson has not changed. The companies that sit down with the people who pay as early as Bakken sat down with the people who treat will be the ones that shorten the decade, and every year they take off it is a year in which patients benefit sooner. The work is challenging, but it is doable.

If you are planning how to take a device into European markets and want to talk through the payment side early, you can reach me through the contact form on this site.