Maisel WH, Sweeney MO, Stevenson WG, Ellison KE, Epstein LM. Recalls and Safety Alerts Involving Pacemakers and Implantable Cardioverter-Defibrillator Generators. JAMA. 2001;286(7):793–799. doi:10.1001/jama.286.7.793
Author Affiliations: Cardiac Arrhythmia Service, Cardiovascular Division, Department of Medicine, Brigham and Women's Hospital, Boston, Mass.
Context Unanticipated pacemaker and implantable cardioverter-defibrillator (ICD)
generator malfunctions sometimes warrant recall by the US Food and Drug Administration
(FDA). Despite increasingly frequent device implantation, pacemaker and ICD
recalls and safety alerts (advisories) remain poorly characterized.
Objectives To determine pacemaker and ICD generator advisory rates in the United
States, to identify trends in these rates, and to examine their clinical and
Design and Setting Analysis of weekly FDA Enforcement Reports issued between January 1990
and December 2000 to identify all advisories involving pacemaker or ICD generators
in the United States. Recalls and safety alerts involving lead malfunctions
were not included.
Main Outcome Measures Number of pacemakers and ICD generators in the United States subject
to FDA recall or safety alert in 1990-2000; annual pacemaker and ICD advisory
rates in the United States in 1990-2000; and estimated cost of device advisories.
Results During the study period, 52 advisories (median [25th and 75th percentiles],
4 [4 and 7] per year) involving 408 500 pacemakers and 114 645 ICDs
(523 145 total devices) were issued. Hardware malfunctions (35 advisories
affecting 280 641 devices) and computer errors (10 advisories affecting
216 533 devices) accounted for 95% of device recalls. Implantable cardioverter-defibrillators
were recalled more frequently than pacemakers (mean [SD], 16.4 [1.6] vs 6.7
[0.8] advisories per 100 person-years; P<.001).
Between 1995 and 2000, the annual advisory rate increased for both pacemakers
(P for trend <.001) and ICDs (P for trend = .02). An estimated 1.3 million device checks and analyses
and 36 187 device replacements resulted from the advisories and cost
approximately $870 million.
Conclusions Pacemaker and ICD recalls and safety alerts occur frequently, affect
many patients, and appear to be increasing in number and rate. With the growing
number of device implants and expanding indications for device therapy, the
number of patients affected by device advisories will likely continue to increase.
Permanent pacemakers and implantable cardioverter-defibrillators (ICDs)
are 2 of the most remarkable medical and technological advances of the 20th
century. While these devices are life-saving in many instances, they may occasionally
malfunction.1 The US Food and Drug Administration
(FDA) is responsible for the safety and oversight of all medical devices in
the United States. Weekly FDA Enforcement Reports are issued that include
recalls and safety alerts (referred to collectively as advisories), a number of which have involved pacemakers and ICDs.
Safety issues involving pacemaker leads have received much attention
and have been well studied.2,3
Recognition of lead abnormalities, such as a tendency to fracture,2 have resulted in recalls, close follow-up, improved
lead design, and the development of safer methods of lead extraction. In contrast
to lead advisories, alerts and recalls involving pacemaker and ICD generators
are poorly characterized.
With the aging of the US population, the number of pacemakers implanted
will likely increase.4 In addition, novel indications
for pacing, such as prevention of atrial fibrillation and biventricular pacing
for congestive heart failure, are being developed.5- 9
Likewise, the number of ICD generators implanted will likely continue to increase
because of the results of several randomized clinical trials demonstrating
a survival benefit of ICDs in specific high-risk patients.10- 14
With the increasing number, variety, and complexity of implantable devices,
recalls and safety alerts necessitated by unanticipated device malfunctions
may become increasingly common.
This study was undertaken to determine pacemaker and ICD generator advisory
rates, to identify trends in these rates, and to examine the clinical and
financial implications of device advisories to the health care system.
Recalls and safety alerts (advisories) are classified by the FDA into
4 categories: class I, class II, class III, or safety alerts (Table 1). The number of pacemaker and ICD advisories was determined
by reviewing all weekly FDA Enforcement Reports from January 1990 through
December 2000 and verifying all recalls and safety alerts with the manufacturer,
when possible.15- 20
Only advisories affecting pacemaker or ICD pulse generators were included;
advisories involving pacemaker or ICD leads were excluded. When the precise
number of devices recalled in the United States was not reported by the FDA,
the recalled number of US devices was calculated based on the number of affected
devices worldwide and the estimated proportion of these devices implanted
in the United States.21
To calculate the annual advisory rate (ie, the annual number of pacemakers
and ICDs subject to recall or safety alert divided by the number of patients
living with a device in a given year), the number of patients with pacemakers
and ICDs in the United States was estimated.
For pacemakers, estimates were calculated by applying age- and sex-related
prevalence rates for permanent pacemakers in the United States from the 1988
National Health Interview Survey Medical Device Implant Supplement4 to the annual US Bureau of Census age and sex population
estimates.22 By using age and sex population
estimates, annual pacemaker prevalence was adjusted for the aging of the population
and accounted for the higher prevalence of pacemaker implantation in men.4 In addition, annual pacemaker implantation rates by
age and sex were obtained from the National Center for Health Statistics'
National Hospital Discharge Survey23- 35
and were used to account for sex and age fluctuations in annual implantation
The annual prevalence of patients who have ICDs in the United States
was calculated using estimated ICD prevalence in 199036
and the number of new implants in the United States per year (based on the
rate of new implants36 and the US Bureau of
Census annual population estimates22). For
the years 1998 to 2000, for which exact implant rates and implant numbers
were unavailable, implant rates were estimated assuming continued increased
ICD implantation rates similar to those observed throughout the 1990s (16%-33%
increase per year).36,37 The mortality
rates for patients with ICDs were assumed to be similar to those observed
in the major ICD clinical trials.10- 14
The recommendations issued by the manufacturer for the management of
a specific recall or alert were used to determine the clinical impact of an
advisory. If specific recommendations were unavailable, device replacement
was assumed to occur only under the following 3 circumstances: (1) device
malfunction that could not be corrected with reprogramming, (2) pacemaker
dependency in a patient with a pacemaker that could fail without warning,
or (3) presence of an ICD that could fail to deliver appropriate cardioversion/defibrillation
therapy. Pacemaker dependency was estimated to be 18% based on the rate at
which patients with pacemaker failure present emergently.38
Costs were estimated using annual Medicare reimbursement rates for the
recommended pacemaker or ICD follow-up and for the recommended pacemaker or
ICD generator replacements.39- 44
Physician fees were calculated using Current Procedural Terminology (CPT)
codes (metropolitan Boston rates). The following codes were used: CPT 93732
for pacemaker check and analysis with reprogramming; CPT 93731 for pacemaker
check and analysis without reprogramming; CPT 93744 for ICD check and analysis
with reprogramming; CPT 93743 for ICD check and analysis without reprogramming;
CPT 33213 and 33233 for pacemaker generator change; CPT 33240 and 33241 for
ICD generator change; and CPT 93641 for ICD testing at implant. Hospital fees
(including the cost of the replacement pacemaker or ICD generator) were calculated
using mean national rates for the appropriate diagnosis related group (DRG)
(DRG 118 for pacemaker replacement and DRG 116 for ICD replacement). Calculations
used a 4% annual discount rate and results are reported in year 2000 dollars.
Statistical comparisons were performed using SAS statistical software
(version 6.12, SAS Institute, Cary, NC). A 2-sided P
value of .05 or less was interpreted as being statistically significant. Student t and χ2 tests were used to compare continuous
and discrete outcomes, respectively. Mantel-Haenszel χ2 tests
were used to assess for trends during 3 periods that were selected a priori:
1990-1995, 1995-2000, and 1990-2000. Several potential sources of uncertainty
in our cost model were explored: (1) variation in the cost of device analysis
or replacement, (2) variation in the number of device analyses or replacements,
and (3) variation in the discount rate.
The estimated number of patients with devices is shown in Figure 1. Overall, the number of patients living with implanted
devices (pacemaker or ICD generator) increased 49% from 1990 to 2000. The
number of pacemakers increased 22% from 1990 to 2000, while the number of
patients with ICDs increased almost 11-fold during the same period due to
an 11-fold increase in annual ICD implants between 1990 and 2000.
There were 52 advisories (range, 0-9; median [25th and 75th percentiles],
4 [4 and 7] per year) between January 1990 and December 2000 (Figure 2). Pacemakers were recalled in 9 of the 11 years studied
while ICDs were recalled in 8 of the 11 years studied. Only 1990 was advisory-free.
While recalls and safety alerts more often involved pacemakers than ICDs (67%
vs 35%; P = .002), the annual percentage of advisories
involving ICDs increased between 1990 and 2000 (P
for trend <.001).
Between January 1990 and December 2000, 408 500 pacemakers and
114 645 ICDs (523 145 total devices) were subject to recalls or
safety alerts (Figure 2). The annual
number of affected devices varied from a low of 0 in 1990 to a high of 182 647
in 1999. Of the devices recalled since 1990, 65% have been recalled since
January 1999 including 71% of the recalled pacemakers and 42% of the recalled
Table 2 shows the FDA classification
of recall or safety alert and the type of malfunction. Three quarters of all
advisories were class I or II recalls. Of the 52 advisories, 35 (67%) were
issued because of hardware malfunctions (electrical/circuitry malfunction,
battery/capacitor malfunction, inadequate hermetic seal, defective crystals,
device header abnormalities). Firmware (integral device computer programming)
errors accounted for an additional 10 (19%) advisories. Overall, hardware
(54%) and firmware (41%) malfunctions accounted for 95% of all advisory devices.
The remaining recalls and alerts were due to either environment-device interactions
(for example, ICD interactions with magnetic fields) or to nondevice-related
problems (mislabeling of product, inappropriate product distribution). Selected
examples of pacemaker and ICD advisories are listed in Table 3.
The annual advisory rate for devices ranged from 0 to 27 per 100 person-years
for pacemakers, 0 to 71 per 100 person-years for ICDs, and 0 to 25 per 100
person-years for total devices (Figure 3).
The overall mean (SD) device rate for 1990-2000 was 7.7 (1.1) advisories per
100 person-years. The ICD advisory rate was significantly greater than the
pacemaker advisory rate (mean [SD], 16.4 [1.6] vs 6.7 [0.8] advisories per
100 person-years; P<.001). The difference remained
significant even if 1992 and 1993 (years with high ICD advisory rates) were
excluded from the analysis. No trend in advisory rates was apparent for 1990-1995
or 1990-2000. However, between 1995 and 2000 the advisory rate increased for
pacemakers (P for trend <.001), ICDs (P for trend = .02), and total devices (P for
Manufacturers recommended clinical intervention (device check and analysis
or replacement) for 79% of the advisories. For the management of pacemaker
advisories, manufacturers most often (57%) recommended pacemaker replacement
for pacemaker-dependent patients and close follow-up for nonpacemaker-dependent
patients. Less frequently, recommendations advised replacement of all advisory
pacemakers (6%), close follow-up and/or replacement of actual pacemaker failures
only (6%), reprogramming of the pacemaker (9%), or no physician action (23%).
The ICD advisories were managed with device replacement (72%), close follow-up,
device reprogramming and/or replacement of actual ICD failures only (11%),
or no physician action (17%).
Pacemaker advisories required electronic device check and analysis more
often than ICD advisories (3.1 vs 0.4 analyses per advisory device; P<.001; Table 4).
Over 1.2 million pacemaker checks and analyses were recommended, predominantly
because many pacemaker recalls advised monthly follow-up for nonreplaced devices.
Conversely, device replacement was recommended 10 times more frequently for
ICD advisories than for pacemaker advisories (23 vs 2.3 replacements per 100
advisory devices; P<.001). Overall, replacement
of 36 187 devices (74% were ICDs) was advised.
The total estimated cost of the device advisories was $870.2 million
(Table 4). Although fewer devices
were affected by ICD advisories than pacemaker advisories, ICD advisories
were more costly because of the greater rate and cost of device replacement.
The ICD advisories accounted for 22% of advisory devices but 76% of total
Sensitivity analysis was performed to assess the impact of changes in
the cost model parameters on total device advisory costs. Total advisory costs
were sensitive to changes in either the number or the unit cost of ICD replacements;
although even with a 50% reduction in these parameters, total advisory costs
for the 1990-2000 period exceeded $500 million. Total costs were relatively
insensitive to significant changes in other model parameters, including the
number and unit cost of pacemaker replacements, the number and unit cost of
pacemaker and ICD analyses, and the discount rate.
Despite more than 25 years of medical device monitoring and recalls
by the FDA,45,46 relatively little
is known about the rates and trends of recalls and safety alerts in the United
States. This study demonstrates that more than 500 000 permanent pacemaker
and ICD generators were subject to recalls or alerts in the United States
between 1990 and 2000 and that the annual rate of pacemaker and ICD generator
advisories has been increasing since 1995. To the best of our knowledge, this
is the first study to categorize causes of device advisories. While more than
three quarters of advisory devices are pacemakers, the rate of recall or alert
is higher for ICDs than for pacemakers. Advisories for ICDs accounted for
three quarters of the $870 million spent on managing these recalls.
Several database registries have monitored the performance of pacemakers
and ICDs, but have been limited by their relatively small size or voluntary
nature. The Bilitch Registry evaluated 22 786 pacemakers from 6 sites
over a 19.5-year period and 3520 ICDs from 13 sites over almost 12 years ending
in December 1993.47 Failures involving 50 different
generator models were reported48 and ultimately
led to a recommendation for a North American Device/Lead Database that was
not accepted by the FDA or manufacturers.49
STIMAREC is an ongoing long-term device failure registry that publishes
reports submitted by participating centers.48
Between 1988 and 1996, 700 cases of pacemaker or ICD generator malfunction
were reported to STIMAREC, including 370 dangerous cases involving loss of
output or decrease in pulse amplitude.50 Because
malfunction reporting for this registry is voluntary and the number of at-risk
patients for the registry is unknown, failure rates cannot be calculated.
The FDA Enforcement Reports present device failures that manufacturers
discover during their own individual monitoring programs.48
More than 250 000 pulse generators, leads, and pacemaker/ICD programmers
were subject to FDA recalls and safety alerts between 1974 and 1987,45 substantially less than the 523 145 pacemaker
and ICD generators recalled during the 11-year period of this study. All 5
major manufacturers had an increase in the number of models of pulse generators,
leads, and pacemaker/ICD programmers subject to FDA recalls and safety alerts
between 1976 and 1995.48 Our study demonstrates
that the annual rate of pacemaker and ICD generator recall and safety alert
has been increasing since 1995.
While this increase in generator advisory rate could be due to increased
vigilance on the part of the FDA and manufacturers, it may also reflect other
factors. Pacemakers have become increasingly complex—incorporating features
to preserve battery life, promote intrinsic conduction, provide physiologic
pacing, and increase diagnostic capabilities. These features include automatic
threshold determination, automatic atrioventricular interval adjustment, dual
chamber electrogram storage capabilities, and rate-responsive pacing using
ICDs now incorporate single or dual-chamber pacing capabilities as well
as increasingly sophisticated algorithms for recognition and treatment of
tachyarrhythmias. Despite increasing complexity, ICDs continue to shrink in
size while maintaining their battery life and high-energy capabilities. The
observed ICD recall rate may be higher than the pacemaker recall rate because
ICDs are technically more complicated than pacemakers. In addition, because
ICD malfunctions are more likely to be perceived as life-threatening, manufacturers
and the FDA may have a lower threshold for issuing an advisory.
Most advisories during this period can be classified as either hardware
or firmware advisories. Hardware recalls and alerts, accounting for more than
half the device advisories, usually occur due to manufacturing defects, unanticipated
component failure, or poor design. More than two thirds of these advisories
can be further classified as electrical and/or circuitry problems, battery
and/or capacitor failure, or hermetic seal abnormalities. Firmware advisories,
on the other hand, occur less frequently but tend to be larger. They account
for more than 40% of the recalled devices. They are more enigmatic, occurring
due to programming glitches that may be difficult to anticipate. Some who
have studied incidents in other complex systems, such as the airline industry,
nuclear weapons, or space flight, consider engineered (hardware) incidents
predictable and therefore preventable, while system (firmware) incidents are
inevitable due to complex processes combining in unforeseeable ways.51
Clinical management of patients with an advisory pacemaker or an advisory
ICD generator is often challenging.3,49
Manufacturers generally advise replacement of devices that may fail to deliver
life-sustaining therapy (ie, patients who are dependent on pacemakers or who
have an ICD). However, there are often little data to aid the clinician's
decision whether to actually replace the recalled device and physicians must
incorporate the manufacturer's recommendations into their assessment of the
individual patient's risk. Factors affecting the decision to replace a device
include the reason for the advisory, the type of device (ICD vs pacemaker),
the degree of pacemaker dependency, the general medical condition of the patient,
and the patient's wishes. Complicating the decision is the fact that not every
recalled device is defective. For some advisories, manufacturers estimate
fewer than 1% of recalled devices are affected.52- 54
Another complicating factor is that initial estimates of failure rates are
sometimes inaccurate.3 For some patients though,
especially those who are dependent on pacemakers or who have an ICD, even
a risk less than 1% may be deemed unacceptable.
Few deaths have been confirmed to be due to device malfunction1,52 and our study does not directly challenge
this claim. Patients with implanted devices, however, have coexisting cardiac
disease and deaths may be attributed to a primary cardiac abnormality rather
than device malfunction.38 A prospective device
database would be the most accurate way to determine if patients with advisory
devices have increased mortality.
The psychological impact of recalls to patients with implanted devices
should not be underestimated. Patients with ICDs or pacemakers often develop
a psychological dependence on their device that can include anxiety related
to a fear of device malfunction.55,56
Patients and their family members affected by actual ICD advisories demonstrate
a marked decrease in confidence in their device.57
Device advisories and the possibility of device failure may have considerable
impact on the psychological well-being of affected patients.
The financial implications of pacemaker and ICD recalls and safety alerts
are substantial. Recalls increase outpatient appointments, in-hospital patient
days, and use of hospital resources.58 We estimate
these costs to total $870 million for the decade 1990-2000. Manufacturers
may provide replacement devices at discounted prices and a nominal hospital
and/or physician fee for replacement, but these costs are ultimately passed
on to the health care system and are reflected in the cost of new devices.
The issue of who actually pays for these recall costs has not been well studied
and likely includes manufacturers, the federal government, hospitals, private
insurers, physicians, and patients.58 A decrease
in the number or cost of ICD replacements due to advisory would have the largest
impact in reducing the total advisory cost burden.
Because some manufacturers are no longer in business, the precise number
of replaced devices could not be calculated. Similarly, the number of extra
pacemaker and ICD checks and analyses is an estimate based on manufacturer
recommendations. The actual number may vary if clinicians were more or less
cautious in their follow-up of patients with devices that had received an
advisory. Pacemaker and ICD lead advisories were not included in this study
because lead malfunctions have already received much attention and occur for
different reasons than generator failures. The purpose of this study was to
focus attention on the previously underrecognized problem of pacemaker and
ICD generator malfunctions.
We have demonstrated that (1) many patients are affected by pacemaker
and ICD generator advisories, (2) ICDs are recalled at a greater rate than
pacemakers, and (3) the annual advisory rate for pacemaker and ICD generators
has increased since 1995. With the growing number of device implants and expanding
indications for device therapy, the number of patients affected by device
advisories can be expected to increase. Hopefully, the intense efforts, which
have resulted in remarkable technological advances in pacemaker and ICD therapy,
will be accompanied by an equally intense effort from manufacturers and the
FDA to improve the overall safety of these devices and reduce the number and
rate of recall.