Pacemaker and ICD Malfunction—An Incomplete Picture

In this issue of JAMA, the reports by Maisel and colleagues^{1 - 2} and by Gould and Krahn³ represent substantial pieces of an information mosaic that is developing about life-sustaining and life-saving implantable arrhythmia device technologies. It is apparent, not only to the medical community, but also to patients, the press, regulators, investors, and the legal community, that this information mosaic is complex. The technologies are complex. The medical issues are complex. The ethical issues are complex. The financial issues are complex. No effort should trivialize this complexity or oversimplify and produce “the solution” without understanding that the information comes with gaps and never forms more than an incomplete mosaic. These 3 studies^{1 - 3} provide new data to inform the most important issue—how physicians and patients should make individual decisions regarding pacemaker or implantable cardioverter-defibrillator (ICD) surgery.

In the first study, Maisel¹ uses meta-analysis to evaluate the incidence of pacemaker and ICD malfunction requiring device explantation. There are strengths and weaknesses in this approach. A significant strength is the longitudinal perspective, providing outcomes for pacemakers from 1983 through 2004 and for ICDs from 1988 through 2004. Unfortunately, only 3 registries contributed data to the meta-analysis. The Bilitch North American Registry was voluntary and not a complete registration of all devices, representing 6 pacemaker sites and 13 ICD sites that ended malfunction reporting in 1993.⁴ The UK National Pacemaker and ICD Registry included all pacemaker and ICD implantations in the United Kingdom and Republic of Ireland but contributed data on pacemaker malfunction only from 1983 through 2004.⁵ The Danish Pacemaker and ICD Registry reported information about all devices implanted in every Danish center (14 pacemaker and 5 ICD) and contributed malfunction data between 1994 and 2004.⁶

The consistency of the reported pacemaker malfunction incidence between the Bilitch and UK registries from 1983 through 1993 and between the UK and Danish registries between 1994 and 2004 is reassuring. However, a noteworthy regional difference was noted in 2001, with a substantially higher pacemaker malfunction rate in the Danish Registry related to an unfortunate choice of a pacemaker model. Although there was a clear decline in the pacemaker malfunction rate with a leveling of the incidence from 1988 through 2004, the trends for the ICD patients (dependent on the Bilitch Registry from 1988-1993 and the Danish Registry from 1994-2004) are much less clear and appear to vary significantly.

In a second study, Maisel et al² estimate the rate of device malfunction using the number of manufacturer-confirmed malfunctions that caused device explantation as reported to the US Food and Drug Administration (FDA) through annual reports, as compared with the number of device implantations each year, between 1990 and 2002. This is a substantially different methodology with an end point similar to that of the meta-analysis approach that was indexed to device prevalence. Although no attempt was made to measure the number of devices at risk for malfunction, “all” device explantations due to device malfunction were counted. The strength of this approach is the enormous number of devices (2.25 million pacemakers and 415 780 ICDs) that were implanted in the United States and therefore at risk for premature explantation and analysis.

Despite the magnitude of this data set, there are 3 substantial gaps in the data. Manufacturers and hospitals are required to report by FDA regulation all pacemaker and ICD malfunctions, especially those that result in a surgery. However, some device malfunctions may be missed due to poor communication, failed administrative issues in the operating room, or inconsistent physician reporting. A second, more important, gap is that patients who die either with or possibly in part related to a malfunctioning device may be missed. Third, and perhaps most important, pacemakers and ICDs are dependent on the entire implanted system to function properly. The device, the leads, the chosen programmed parameters, and the patient's response are all interdependent. The system malfunction rate may be the most important measure of reliability, but it is also the most elusive because no single measure captures all elements in a verifiable fashion.

The value of the analysis using the FDA annual reports is that these reports represent definite malfunction, define the mechanism of the malfunction, and provide estimates for the minimum malfunction rate. The sensitivity analysis used by Maisel et al in the FDA annual report study reveals that the trends over time were real but does not provide estimates of what the actual rate would have been if all device malfunctions were counted, including those missed due to poor reporting compliance or patient death. However, the rates reported in this analysis, although clearly less than the actual rate, are useful and similar to the findings derived from the meta-analysis, which were indexed to prevalence.

Both analytic methods provide similar messages. First, device malfunction severe enough to require premature device replacement is not rare. The estimates from both analyses demonstrate pacemaker malfunction rates from 1988 through 2004 that are similar to the calculated average rate of 7 per 1000 new implants calculated from the annual report analysis between 1990 and 2002. The ICD malfunction rates are more variable, but such variability might not be meaningful. For instance, device technology is in continual evolution, and the population at risk is a melange of devices and technologies. Although there was a clear downward trend that parallels but is shifted 8 years later than the pacemaker trend, the subsequent increase (1998-2001) and then decrease (2002) suggests some volatility in the incidence. In addition, the current generation of ICDs actually comprises combined cardiac resynchronization and ICD devices that have not yet been measured in these reports. Accordingly, an equally plausible main message is that the average rate of ICD device malfunction severe enough to require device replacement was 20.7 per 1000 new implants recorded in the annual report analysis between 1990 and 2002. The actual rate in any particular year may be better or worse but is higher than the pacemaker rate and also is not rare.

Another important message is that these malfunctions occur after a significant time delay, more than 3 years in the life cycle of a device model, for a majority of devices (70.1% for pacemakers, 65.2% for ICDs). This time frame does not include the time necessary to conclude from pattern analysis that some of these malfunctions require public notification with a safety alert. Awareness of potential problems with a device will continue to be a long-term, postmarket, surveillance issue of substantial magnitude.

Although the reoperation rate for pulse generator malfunction is estimated by these 2 reports by Maisel et al, at least 2 related situations are not addressed by these analyses. The first is how often these malfunctions threatened to withhold antiarrhythmic therapy, or, in other words, how often were these malfunctions potentially dangerous to the patient? A total of 61 confirmed deaths were documented to be related to a device malfunction (30 pacemaker and 31 ICD patients) among 2.7 million patients with devices. However, many device malfunctions were much less severe in nature, ie, severe enough to warrant replacement but potentially as “benign” as a shorter time to achieve the battery voltage that indicates that elective replacement is necessary. This correction of the malfunction rate would decrease the number of patients at significant risk.

A related scenario, lead dysfunction, has been appropriately excluded from these analyses. Lead malfunction can be related to design issues, implantation technique, the patient's activities, and the patient's anatomy and physiology.⁷ However, pacemaker and ICD lead malfunctions are often just as dangerous as device malfunctions. To the patient at risk from ventricular dysrhythmias, it does not matter whether the short circuit between the anode and cathode has occurred in the ICD lead instead of inside the ICD pulse generator if the device fails to convert ventricular fibrillation. This phenomenon clearly would increase the number of situations that could result in failure to support a patient's heart rhythm.

The third article on these devices in this issue of JAMA, by Gould and Krahn,³ approaches a related but different issue, the risk of reoperation in response to an implantable defibrillator advisory. In contrast to the 2 articles^{1 - 2} that address documented device malfunctions, this study evaluates the clinical consequences of a community's response to potential ICD malfunction. The strength of this analysis is its inclusion of approximately 93% of the patients and devices in Canada identified as being at increased risk due to 6 device advisories from 4 manufacturers between October 2004 and October 2005. Although there were a variety of responses to the advisories, physicians chose to electively remove and replace the ICDs in 533 (18.3%) of the 2915 patients. The treating center was a predominant determinant of whether a device was replaced, with the rate ranging from 0% to 45%.

Complications occurred in 43 patients (8.1%); while these were sometimes minor (12 [2.3%]), at other times they were more serious, requiring an additional operation for infection (10 [1.9%]), hematoma (12 [2.3%]), system malfunction (8 [1.5%]), and significant site pain (1 [0.2%]). There were 2 deaths, 1 related to the device extraction and 1 related to sepsis despite the extraction. The potential risk, identified by the safety alert, to an individual patient was transformed into manifest harm, in addition to the actual device replacement surgery, in 5.8% of the patients in this series. This series also demonstrated the difficulty physicians and patients have in deciding when to intervene with a replacement surgery, as well as some of the consequences of device replacement. Consequently, this analysis does not help physicians answer the question of whether an ICD should be replaced in the next safety alert situation.

These investigations contribute important new data to the information mosaic involving pacemakers and ICDs and should enable clinicians and patients to make better informed decisions about these devices. However, these studies involve large patient groups and not individuals. Each patient and each device system pose unique issues. The information provided by these studies provides a landscape that should help physicians, patients, manufacturers, and regulators make the best possible decisions for individual patients and for society.

Every year, a substantial number of pacemakers and ICDs prematurely develop a significant problem requiring surgical revision. The actual rate would be better calculated as the number of pacemakers or ICDs removed for documented malfunction divided by the implanted prevalence of the same category of device as used in the meta-analysis. Only a combined effort of device manufacturers would provide the prevalence data required to complete this analysis. Another reality is that 5.8% of patients developed significant complications resulting in a surgical revision as a result of device replacement in response to ICD product advisories. While these are challenging and contrasting realities, one further reality is that answers to these questions would not be needed if pacemakers and ICDs lacked dramatic life-sustaining and life-saving effectiveness despite their frailties.