But now data provided by Guidant to a Minnesota hospital suggests that the company continued to sell the potentially flawed devices for months after it changed the way it made the device and had begun selling the new ones.
The Guidant data, which the company provided to doctors at Abbott Northwestern Hospital in Minneapolis, indicates that from May to September 2002, nine patients there received implants of defibrillators that were made before April 2002.
It was during that month that the company first changed its manufacturing process to eliminate the risk of short-circuiting. A senior company official has said that Guidant to date has not received reports of such failures in units made after April 2002.
A defibrillator, which is implanted under the skin in the chest, acts to shock a chaotically beating heart back into normal rhythm. The device at issue is known as the Ventak Prizm 2 DR Model 1861, or alternatively as the Prizm 2 DR.
In response to questions from The New York Times, Guidant said late yesterday that it had continued to sell units out of inventory - those made before the fixes - because it believed that the devices were reliable.
In its statement, the company said: "After making these improvements, Guidant sold product manufactured before the improvements because the reliability data showed that the original PRIZM 2 DR, like the enhanced version, was a highly reliable life-saving product. Current data continues to support the reliability of this product."
Last week, Guidant, which is based in Indianapolis, told doctors and patients for the first time that the defibrillator failed because of the short-circuiting problem seen in 26 known cases, including one recent death. About 37,000 such units were made before Guidant made the April 2002 fix. An estimated 24,000 people still use one of those units.
The company said last week that the overall failure rate of the units made before and after the fix is similarly low. Last week, Guidant urged physicians not to replace the units because it said the risks of doing so outweighed any dangers the device may pose.
However, the apparent fact that Guidant sold older devices with the potential flaw when improved ones were available could pose a new round of problems for the company. The Food and Drug Administration is reviewing how Guidant officials handled problems related to the device.
In addition, two plaintiffs' law firms filed an action yesterday against Guidant in Federal District Court in Indianapolis seeking damages in connection with the company's handling of the device. Guidant has previously said that it did nothing wrong.
The company's image is also still rebounding from an episode in which a former unit, Endovascular Technologies, pleaded guilty in 2003 to 10 felonies and agreed to pay $92.4 million to settle charges that it covered up deaths and other problems related to a device that treated weakness in the main abdominal artery.
Officials of Guidant have insisted that they made all required reports concerning the Model 1861 to the F.D.A., including the information in individual reports from doctors and hospitals about any apparent problems with the device.
According to the Guidant data, the nine patients at the Minnesota hospital represented 16 percent of the 57 patients there who received a Model 1861 made before Nov. 1, 2002. Those devices were implanted from December 2000 to January 2003. It was in November 2002 that Guidant made a second change to the unit.
Some doctors said they would be dismayed if the company allowed them to implant a device with a known flaw that had been corrected in other units. Typically, a device like a defibrillator is not sold until a company sales representative delivers the unit to the surgical suite and it is implanted. Because their batteries drain, a defibrillator needs to be replaced every five or six years.
Guidant is one of the nation's largest makers of medical devices, with $3.8 billion in sales last year, about half of them from implantable defibrillators. In December, Johnson & Johnson announced it planned to buy Guidant in a deal worth $25.4 billion. After the disclosure last week of the Model 1861's problems, Johnson & Johnson said it remained committed to the deal.
"Guidant is an honorable company," said Dr. Douglas Zipes, a cardiology professor at Indiana University School of Medicine who is also a consultant to Medtronic, another defibrillator maker. "If they personally knew this device had the potential for failing, I can't imagine that they would go ahead and continue to sell it."
The implant data at issue was provided by Guidant to doctors at Abbott Northwestern after the death in March of one of their patients, a 21-year-old college student with a genetic heart disease that put him at high risk of sudden death. In earlier interviews, those doctors, Dr. Robert G. Hauser and Dr. Barry J. Maron, expressed anger at Guidant for failing to alert them earlier to the unit's problems, saying they would have replaced it in their patient had they known.
Dr. Zipes is the editor of a medical journal that recently published on its Web site a case study about the Guidant unit written by Dr. Hauser and Dr. Maron, though he was unaware of any issue related to inventory devices when interviewed last week.
The student, Joshua Oukrop, died during a bicycling trip in Moab, Utah. After Guidant inspected his Model 1861 defibrillator, implanted in late 2001, it concluded that the device had malfunctioned because of the short-circuiting problem.
While it is not certain when the device failed, the short circuit can occur when the device builds a charge to deliver the high-energy shock needed in emergencies.
Dr. Hauser, who at one time headed a company, Cardiac Pacemakers, that has become part of Guidant, said he now faced telling some of those patients that they might have to undergo surgery to remove and replace a device they probably should not have received.
"What am I supposed to tell these people?" asked Dr. Hauser.
He agreed to provide a reporter with the implant data given to him by Guidant. Before sending it, personnel at Abbott Northwestern stripped out patient names and deleted portions of each device's serial number, which can also be traced to a patient.
For each Model 1861 unit, the list gives the date of implant as well as a "use before" date, or the date beyond which the device can no longer be implanted because it has remained in inventory too long.
An F.D.A. spokeswoman, Julie Zawisza, said that, according to the agency's information, the shelf life or "use before" date of the Guidant Model 1861 is 12 months after its battery is connected. Typically, that point marks a device's date of manufacture, and the unit goes into inventory a few weeks later, industry officials said.
Based on that timeline, the Guidant implant data provided to Dr. Hauser suggests that nine devices with manufacturing dates before April 1, 2002, were implanted in patients at Abbott Northwestern after that time. For example, one device with a manufacturing date of August 8, 2001, was implanted in a patient on May 22, 2002, the data indicates.
In addition, seven devices with manufacturing dates in February 2002, two months before Guidant started making fixes, were implanted in patients in July, August and September of that year, the data suggests.
Guidant has said that in April 2002 it increased the spacing between a wire and a device component after determining that electricity could potentially arc between them and cause a short circuit.
In November 2002, Guidant made another change by adding extra insulation around the component, which is called the backfill tube. The implant data from Abbott Northwestern shows that three patients received implants after November 2002 using devices with earlier apparent manufacturing dates.
Both Dr. Hauser and Dr. Maron said that a senior Guidant executive, Dr Joseph M. Smith, had told them in recent weeks that the April 2002 fix had apparently solved the problem because the company had not received any reports of similar short circuits in units made since then. In a recent interview, Dr. Smith made the same statement to this reporter.
In its statement yesterday, Guidant said: "The reliability data supports the conclusion that Prizm 2 DR was reliable when launched in 2000, when enhanced in 2002, and that it remains reliable today."
Last week, in a notice sent to doctors about the potential electrical problem, Guidant referred only to the November 2002 change and stated that "devices manufactured after this date are not affected."
Guidant has said it reported the November fix to the F.D.A. as part of an annual report to the agency, which it filed in August 2003. However, the company also said in response to an earlier inquiry from The Times that it did not report the April 2002 changes because it decided that F.D.A. regulations did not require the reporting of those "manufacturing enhancements."
The company said that the changes did not have to be reported because they only "added clarifications to existing manufacturing instructions and improved process consistency," rather than reflecting changes to approved F.D.A. manufacturing techniques.