← back to the index
DV-002 cardiac surgery 1986

The Björk-Shiley Convexo-Concave valve — a Welded Strut That Snapped and Killed Two-Thirds

mechanical-fracturesudden-deathcardiac-surgerycardiology1980s1990s
Patients implanted
~86,000 convexo-concave valves worldwide (~55,000 still alive at peak risk)
Failure or harm
600+ outlet-strut fractures; ~two-thirds (≈60%) fatal
In use
~7 yrs (1979 US approval – 1986 withdrawal)
Status
Recalled (Class I)

Summary

When Shiley Inc. and its parent Pfizer pulled the Björk-Shiley Convexo-Concave (BSCC) heart valve from the world market in 1986, the device had been sold for seven years as a refinement of an already trusted prosthesis — and the refinement was the thing that killed people. Co-invented by American engineer Donald Shiley and the Swedish cardiac surgeon Viking Björk, the convexo-concave disc was a geometric tweak meant to improve blood flow over the company's well-regarded flat-disc tilting valve. To make the new geometry work, the outlet strut that captured the swinging disc was changed and welded to the valve ring. That weld was the flaw. Under the relentless cyclic load of roughly 40 million heartbeats a year, the strut fractured at the weld, the disc escaped, and the valve failed catastrophically — often producing sudden death before the patient could reach an operating room.

The harm was not a rare anomaly tolerated by an unlucky few. Of the roughly 86,000 convexo-concave valves implanted worldwide, more than 600 are documented to have fractured, and in approximately two-thirds of those cases the patient died. The 60-degree version received U.S. Food and Drug Administration approval in 1979; a higher-flow 70-degree variant was sold abroad but never cleared in the United States, and it fractured at even higher rates. Because the failure mode was a fatigue crack that gave no reliable warning, surgeons and patients spent the late 1980s and 1990s trapped in an excruciating calculus: a working valve might snap tomorrow, but elective re-operation to remove it carried its own mortality.

What turned a metallurgical defect into a scandal was the factory. Sworn testimony and a 1984 engineer's complaint described a Shiley plant in Irvine, California where valves rejected by inspectors were fished back out, reground to hide cracked welds, renumbered, and passed with falsified paperwork — welders were poorly trained, equipment was in "horrible" condition, and struts were forced onto flanges with pliers. The legend of an improved valve concealed a manufacturing line that could not reliably make the one weld on which a patient's life depended. Litigation, not the FDA, ultimately fixed the price: the Bowling v. Pfizer class action settled in 1992 for roughly $215 million, with a further fund earmarked for future fracture claims, while implanted patients carried the device — and the fear — for the rest of their lives.

Timeline

early 1970s
The flat-disc original earns trust
Björk and Shiley's standard tilting-disc valve gains a strong clinical reputation, establishing Shiley Inc. (Irvine, California) as a leading prosthetic-valve maker.
1976
The convexo-concave redesign
Shiley introduces the convexo-concave disc geometry to improve washout and reduce clotting; a redesigned, welded outlet strut is required to seat the new disc.
1978–1979
Approval and launch
The 60° BSCC is released and approved by the FDA in 1979; Pfizer-owned Shiley distributes it worldwide. A 70° variant follows for overseas markets, never US-approved.
1979–1980
First strut fractures
Isolated outlet-strut fractures begin to be reported soon after introduction, traced to the weld joining strut to ring.
1980
Quiet design changes
Shiley modifies the strut and welding process and at points halts and resumes sales as fracture reports accumulate, without a public recall.
Aug 1984
The slipshod-manufacture allegation
A Washington Post report and engineer testimony describe rejected valves reworked, cracks ground out, numbers changed, and inspection records falsified at the Irvine plant.
Oct 1985
Recall and mounting deaths
Reporting confirms implants recalled after deaths; Shiley/Pfizer face FDA scrutiny over the fracture toll.
1986
Worldwide withdrawal
Shiley withdraws the convexo-concave valve from the market and the FDA withdraws its approval; ~86,000 valves are already in patients' chests.
late 1980s
The explant dilemma
With no reliable way to predict fracture, supervisory panels weigh prophylactic re-operation against its own operative mortality; most valves are left in place.
1992
Bowling v. Pfizer settles
A class-action settlement of roughly $215 million is approved, funding patient monitoring, research, and explant guidance, with additional money set aside for future fracture claims.
1990s–2000s
Risk-stratification science
International cohort studies (notably Dutch and supervisory-panel data) model fracture risk by valve size, opening angle, position, and patient age to guide selective explantation.
2012
The toll tallied
Long-term series record on the order of 663 catastrophic fractures from 1978 onward, the great majority fatal, across the global BSCC population.

The Improvement That Required a Weld

The Björk-Shiley name was, before 1979, a mark of quality. The original flat-disc tilting valve devised by Viking Björk and Donald Shiley had become one of the most implanted mechanical heart valves in the world, and its reputation was earned. The convexo-concave disc was supposed to make a good device better: by curving the occluder, engineers hoped to improve the blood washout around the hinge and cut the risk of clot formation that haunts all mechanical valves. But the new disc geometry would not seat properly on the original cage, so the outlet strut — the metal leg that catches the disc as it swings open — was redesigned and joined to the valve ring by welding. The marketing emphasized hemodynamics; the engineering change that mattered most was structural, and it was hidden inside a few millimeters of welded cobalt-chromium alloy that would have to survive roughly 40 million flex cycles a year for the life of the patient. The promise was incremental refinement. The reality was that a trusted valve had been re-architected around a fatigue-critical weld.

The Strut, the Million-Dollar Barrels, and the Fracture That Gave No Warning

The failure mechanism was as mechanical as a paperclip bent until it snaps. One leg of the welded outlet strut would crack first — a single-leg separation that often produced no symptom — and after a variable interval the second leg would give way, freeing the disc and collapsing the valve. When a mitral or aortic prosthesis fails this way, blood ceases to be controlled; many patients died suddenly, before surgery was possible. That this was a manufacturing failure as much as a design one became clear from the factory itself. A 1984 engineer's complaint and subsequent testimony described an Irvine production line where valves that inspectors had rejected and discarded into "million-dollar barrels" were retrieved, their cracked welds polished smooth, their serial numbers altered, and their paperwork falsified to pass. Welders were inadequately trained, equipment was degraded, and struts were reportedly muscled onto flanges with pliers — precisely the kind of cold-working that seeds fatigue cracks. The lethal signal existed inside Shiley before most patients were ever implanted; it was reworked, renumbered, and shipped.

The Reckoning Without a Good Answer

Withdrawal in 1986 settled nothing for the ~86,000 people already carrying the valve, because the device could not be recalled the way a defective tire can. Removing a functioning heart valve is open-heart surgery with its own mortality, and there was no dependable test to identify which struts were about to break. Patients and surgeons were left to gamble in both directions. Through the late 1980s and 1990s, a court-supervised medical panel and international research consortia — including a closely studied Dutch cohort — built statistical models estimating fracture risk by valve size, opening angle, implant position, and patient age, so that prophylactic explantation could be offered only where the predicted fracture risk exceeded the surgical risk. The financial reckoning ran on a parallel track. The Bowling v. Pfizer class action consolidated claims and settled in 1992 for roughly $215 million, funding patient registries, radiographic screening research, and explant reimbursement, with a separate reserve for those whose valves later fractured. The valve was gone from the market years before the science of who should have it removed had matured; the patients lived inside that gap.

Contributing Factors

01
A fatigue-critical change sold as a flow improvement
The convexo-concave redesign was marketed for its hemodynamics, but its decisive change was structural: a welded outlet strut subjected to tens of millions of load cycles per year. The most dangerous modification was the least advertised, because the sales story and the engineering risk pointed at different parts of the device.
02
Manufacturing fraud on the one weld that mattered
Rejected valves were reportedly recovered from scrap, reground to mask cracked welds, renumbered, and passed on falsified records by undertrained welders using degraded equipment. The failure was not merely a difficult weld but a quality system that defeated its own inspectors — the defect was knowingly shipped, not merely missed.
03
A warning-free, sudden-death failure mode
Single-leg separation typically produced no symptom, and complete fracture often killed before intervention was possible. A device whose failure announces itself can be managed; one that fails silently and lethally converts a manufacturing defect into unpreventable mortality and makes any monitoring scheme partial at best.
04
Underreporting that flattened the true risk
Fracture and death were under-ascertained, particularly abroad and for the higher-angle 70° variant, so early risk estimates understated the hazard. Cohort re-analyses later showed the real fracture rate exceeded reported figures — an information deficit that delayed both withdrawal and the urgency of explant decisions.
05
Withdrawal without a retrieval answer
Pulling the valve from sale did nothing for the tens of thousands already implanted, and explantation carried its own mortality with no reliable predictor of imminent fracture. Treating market removal as the endpoint left an irreducible population of patients to bear a known, unfixable, in-body hazard for decades.

Aftermath

The Björk-Shiley Convexo-Concave became cardiac surgery's standing lesson that a withdrawn implant is not a recovered one. Pfizer exited the mechanical-valve business; the Bowling v. Pfizer settlement of roughly $215 million financed patient monitoring and research while leaving most valves in place, because the cure was sometimes deadlier than the risk. The episode reshaped how regulators and surgeons think about prosthetic durability: it drove the development of formal fatigue-life testing for heart valves, risk-stratification models that weigh device failure against operative mortality, and a sharper distinction between a redesign's marketed benefit and its structural cost. It also sits in the legal record beside Michael v. Shiley as a case study in how a manufacturer's internal quality fraud surfaces only through litigation discovery. To this day, patients implanted before 1986 are tracked, and the device is invoked whenever a "minor" geometric improvement turns out to ride on a single fatigue-loaded weld — the byword for the implant that was withdrawn from the market but never from the chest.

Lessons

  1. Identify the fatigue-critical element of any redesign and test it to failure under realistic cyclic load before launch — the most lethal change is often the structural one hidden behind a hemodynamic or cosmetic selling point.
  2. Treat a quality system that can be defeated by its own line workers as a fatal defect in itself: if rejected parts can be reworked, renumbered, and re-shipped, no inspection statistic from that plant is trustworthy.
  3. Weigh withdrawal against retrieval from the first day a recall is contemplated — for an implant, pulling sales is meaningless unless you also have a way to find, predict, and safely remove the units already inside bodies.
  4. Assume your adverse-event reports undercount, especially across borders and unapproved variants; build independent cohorts to estimate the true failure rate rather than trusting the numbers that happen to reach you.
  5. When you cannot fix the deployed hazard, fund the monitoring and the science that lets each patient compute their own risk — abandoning an unfixable population to silence is the second failure after the first.

References