Introduction
“How wonderful that we have met with a paradox. Now we have some hope of making progress.”—Niels
Bohr [1]
The medical technology industry in the early twenty-first century is marked by a puzzling
paradox. The industry has contributed to decades of life-saving innovations: the introduction
of technologies like the implantable cardioverter defibrillator, radiofrequency catheter
ablation for cardiac arrhythmias, and cardiac resynchronization therapy, and we have
witnessed remarkable reductions in mortality and morbidity due to cardiovascular disease
[2, 3]. Nevertheless, we are simultaneously witnessing escalating concerns from politicians,
the plaintiff’s bar, the general public, physicians, researchers, and patients over
whether our technologies truly are safe and effective [4–6]. Behind this paradox lies
another, more subtle paradox: a vigorously innovative medical technology industry
requires strong and complex underpinnings, but there is a growing sense that overly
burdensome and complex regulations threaten to choke off continued innovation. Our
intent is to explore this paradox, by expressing the authors opinions based on their
medical device industry viewpoint, focusing on three areas of concern: the threat
posed by a globally discordant patchwork of different regulatory approaches and standards,
the threat posed by a particularly burdensome regulatory process within the USA, and
the implications of inappropriate use of comparative effectiveness with respect to
medical devices.
The global environment for innovation: regulatory discord or harmony
The requirements to introduce new medical devices can vary dramatically around the
world. For example—consider the differences between the USA and the European Union.
In the USA, the foundations of the regulatory requirements for medical products are
defined statutorily in the Federal Food, Drug and Cosmetic Act and implemented in
the Code of Federal Regulations [7]. Per this construct, higher risk devices (e.g.,
U.S. Class II and Class III) may only enter the US market after they have been demonstrated
to provide a reasonable assurance of safety and effectiveness for the intended use
in the labeled intended patient population. In contrast, under the European regulatory
framework, manufacturers must establish the safety and performance of the proposed
device—a subtly different standard (Council Directive 93/42/EEC) [8]. Furthermore,
despite the similarity in the nomenclatures used for classification in the USA and
the EU, the criteria used to classify, and thus identify those devices which must
meet higher premarket thresholds, are quite different in these two regions. The net
result of both these factors can be dramatically different requirements to place new
medical technologies into the market for use with patients. There are two very different
(and valid) perspectives on this consequence—either that the patients in the region
for which the regulatory requirements were lower (typically the EU) are benefitting
from earlier access to new lifesaving technologies or that the patients in the region
for which the requirements were higher are benefitting from access to technologies
that have been more thoroughly vetted and—by extension—are either safer or more effective
[9]. There are oft-cited examples to support either perspective—the early experiences
with first-generation abdominal aortic aneurysm stent grafts being an example of the
latter and the recent experiences with transcatheter aortic valve replacement (TAVR)
devices appear to be emerging examples of the former. Regardless of the perspective,
one fact remains indisputable: the discordant regulatory thresholds for medical devices
to enter the market in countries around the world continue to provide financial and
logistical challenges to the organizations trying to develop and make available new
lifesaving and life-altering technologies to physicians and patients. It also potentially
creates challenges for patients moving from one regulatory region to another.
To further complicate the tapestry of regulatory requirements around the world, the
impact of individual country regulatory requirements can have dramatic implications
to the availability of new devices. In the medical devices space, this is manifested
in the requirement that devices be subjected to localized versions of international
performance standards—essentially requiring that the same type of testing be repeated
to support local approval of the technology. Of greater impact is the growing insistence
by regulators that clinical trials (premarket and/or post-market) be conducted either
in whole or in part within a particular country. When one considers the growing number
of regions (Brazil, Japan, India, and China) in which local requirements are emerging,
the cost implications to manufacturers (and thus their ability to market their devices
globally) are staggering. While there may be a rationale for testing pharmaceuticals
in genetically diverse populations, clinical trials with cardiac rhythm management
(CRM) devices in multiple countries has less merit, and rarely can these trials be
statistically powered to truly answer implicit concern of geographically diverse response
to therapy as the number of devices implanted in many of these countries, even post
market, is often too low.
Industry representatives and regulators from countries around the world have met through
the Global Harmonization Task Force (GHTF) in an effort to bring some modicum of harmony
to the standards and requirements across the globe [10]. Despite tremendous effort
and engagement from multiple stakeholders from around the world, there remains much
to do if we are to inject a level of consistency, which materially impacts the consistency
and global availability of impactful new medical technologies. The impacts of recent
changes made to the organization of GHTF (renamed to IMDRF—International Medical Device
Regulators’ Forum) remain to be seen.
The US regulatory framework and innovation
It would be impossible to proceed without addressing the growing perception by many
key stakeholders (including industry and physicians) that FDA has become a barrier
to medical device innovation and—either by action or inaction—is driving innovation
outside of the USA. To explore this thesis, it is important to consider the foundations
of the current regulatory process in the US. Congress writes acts, which the President
signs into law, and which federal agencies, such as the FDA, interpret and write regulations,
and which are available for a comment period prior to implementation. FDA regulations
are contained in Title 21 of the Code of Federal Regulations [11, 12]. Specifically,
the regulations represent FDA’s interpretation of what the law (Federal Food, Drug
and Cosmetic Act) tells them to do, and guidance documents may be written by the FDA
to aid the public to comply with the intent of the regulations. It is premature for
the authors to comment at this time on the practical impact of the recent Food and
Drug Administration Safety and Innovation Act of July 2012, which includes the Medical
Device Fees Act (MDUFA III) for the next 5 years [13]. Although there have been few
changes in either the law or regulations which define the manner in which medical
devices are to be regulated in the USA, there is the persistent belief the requirements
have significantly changed.
To appreciate how this might be so, one can look at the regulatory standard for market
entry—“reasonable assurance of safety and effectiveness.” While simple in concept,
the vagueness of this standard is embodied in that which constitutes “reasonable assurance”
as defined by the expectations of the American public and as reflected by the actions
of their representatives in government. Although the medical devices developed in
this country have grown ever more complex to treat an ever increasingly complex set
of patient conditions, there has been a seemingly simultaneous increase of expectations
from the American public (and their representatives) of the reliability and safety
of these same devices. In so doing, we have in essence shifted the definition of “reasonable
assurance” in a direction which demands that the level of assurance that is “reasonable”
is higher today than it was several years ago. Said differently, the American public
has expressed a lowered threshold of risk they are willing to accept for a particular
patient benefit in even the newest of technologies which treats the sickest of patients
(one need only review the record of Congressional hearings during the past 3 years
for evidence of this shift) [4]. In response to this, regulators have in essence “raised
the bar” on the level of pre-market data (bench top, animal, and clinical) and post-market
studies required for the approval of medical devices in the USA. Approval in the European
Union is based primarily on thorough bench testing, engineering analysis, and post-market
clinical data. However, while this used to be similar in the USA, the pre-market and
post-market clinical requirements have increased in the last 10 years in the USA.
For example, a US implantable cardioverter defibrillator (ICD) lead approval pre-market
patient study has gone from 24 patient-years in 1998 to about 400 patient-years in
2008, and currently in 2012 to 1,000 patient-years, a 40-fold increase (Fig. 1). In
addition, in 2008, a post-approval study was added for about 1,800 patients enrolled
with 5-year follow-up on 1,000. With a device sales life of 3–5 years, this represents
a significant enough increase on the cost to obtain US market approval compared to
European approval, to make commercialization of some devices unattractive in the USA
(manufacturer data).
Fig. 1
Pre-market clinical testing required for FDA approval of ICD leads for one manufacturer,
expressed in patient-years. This demonstrates an increase in the time and number of
patients needed for study after 2007 (internal data, but available at the FDA)
One may question whether CRM device reliability is worsening and necessitates this
increased scrutiny. Defining device quality and reliability may be challenging as
there are many potential measures, with each measure subject to variation unrelated
to device. Maisel and colleagues [14] suggested that while pacemaker malfunctions
had declined during the period of 1990–2001, ICD malfunctions had not declined. While
those data were applicable to that time period, it should be considered that ICD therapy
was in its infancy (FDA approval 5 years earlier 1985), and that time period saw the
evolution of the ICD from a shock device to one with dual-chamber pacing and anti-tachycardia
pacing. In contrast, pacemakers were at that time already a relatively mature technology.
Maisel et al. [14] observed the malfunction rate for ICDs to be in the range of 1–4 %.
Current ICDs appear to have fewer malfunctions, as reported by manufacturers’ product
performance reports, in the range of 1 % at 5 years, despite improved device internal
diagnostics to identify malfunction. A recent FDA study from 2003 to 2007 used annualized
ICD and CRT-D explantation rates (AER) as an estimate of device reliability and malfunction.
They concluded that in this more recent time period, there had been a favorable decline
in the AER (p < 0.001), which might suggest that there is less device malfunction
[15]. This decline is all the more significant when one considers the improved diagnostic
capabilities of today’s ICDs which allow the differentiation, for example, of normal
battery depletion from component malfunction.
While this shift in standard may ultimately increase the safety and reliability of
first-generation devices introduced in the USA, we are currently witnessing a decided
shift of medical device manufacturers to focus their development and approval efforts
either initially or exclusively to regions outside the USA [9]. While US approval
often represents the de facto “gold standard” and provides great assurance to many,
the cost to reach this objective has become increasingly prohibitive for many innovators.
Thus, it is not unreasonable to project that access to new technologies in the USA
will increasingly lag (perhaps by many years) the availability of these technologies
outside the USA. While tempting to demonize the FDA as the cause of this shift, it
is more likely that the agency and its leadership are not the cause of this, but are
in many ways reacting to what the American public (and Government) is asking of them.
In fact, the authors have found the FDA to be very helpful in working within this
restrictive government policy framework, to collaboratively arrive at a trial design
expected to meet the regulatory requirements and best serve patients.
Although challenging, there are signs that all is not lost. There seems now to be
a growing recognition by government officials that the shifts in expectations are
driving innovative medical technology (and the people that develop them) outside of
the USA. The FDA has recently published new and revised Guidance documents which speak
of their expectations for conducting feasibility clinical studies in the USA and the
expectations and methodologies related to the assessment of risks and benefits [16].
It is this latter topic (risk and benefits) which will ultimately define the intersection
of medical device regulation and innovation in this country. For one of the foundations
of medical product law is the premise that no medical product—either drug or device—is
without risk. The degree of risk we are willing to accept for a given benefit will
ultimately define the breadth and type of pre-market evaluation (and the results)
that must be completed before a product can enter the market.
Finally, incorporated into this analysis must be an honest discourse about the adoption
of new medical technologies by the physician community. With each new innovative medical
technology, there seems to be a predictable cycle of cautious adoption within the
approved patient indications until comfortable with the technology. Once comfort has
been attained, this is often followed by over-adoption, sometimes outside the approved
(and studied) patient populations. This inevitable unbridled optimism driven by confidence
in the technology (rather than assurance in the data) certainly feeds into the argument
that more pre-market data collection (not less) is necessary to ensure patient safety.
However, ultimately, physician bodies will need to determine appropriate device indications
based on the greater experience post-market approval, potentially through guidelines.
It should also be considered that off-label pharmaceutical use has led to some important
discoveries such as beta blockers for heart failure, or ASA for stroke prevention.
It is certainly possible that similar discoveries might occur later in the device
area also.
Another component of regulatory oversight involves post-market surveillance and actions
taken by manufacturers with fielded products. In the past decade, there have been
many product field actions in the CRM area despite the largely unrecognized fact that
performance of these devices (in terms of reliability and safety) has actually improved.
Manufacturers typically execute these field actions voluntarily to reduce the risk
of patient harm when either (1) a new and previously unrecognized hazard associated
with the device or its use has emerged or (2) a hazard has emerged with a greater
frequency or harm than previously recognized or expected. Manufacturers typically
become aware of these malfunction events through physician and patient complaints
captured and analyzed in their post-market surveillance systems. The apparent paradox
of more advisories despite improving overall performance is consistent with a lower
tolerance of malfunctions at multiple levels. Thus, a recent study of 1,644 consecutive
ICD implants between 1996 and 2004 revealed that 704 patients(43 %) were part of an
advisory, but only 28 of 1,644 patients (1.7 %) actually had an advisory-related device
malfunction, and Class I recalls were not associated with increased mortality [17].
Some patients did undergo device replacement (42/1,644, 2.5 %) and hence were harmed,
but it should also be considered that ultimately, if the patient does not succumb
to comorbidities, that most ICDs need to be replaced and that this is a fractional
harm depending on the device age when replaced.
Field actions taken by a manufacturer are often very expensive and come with an attendant
amount of attention, publicity, and legal action. While this attention provides significant
opportunities to inform physicians and patients, it often leads to fear and—sometimes—inappropriate
actions. Significantly, some field actions have been for malfunctions that are uncommon
and occur in products with an overall performance that is considered acceptable [18].
In 2004, Maisel [19] found that many physicians would explant a pacemaker if it had
a malfunction rate of 1/10,000, which is notably a better overall performance than
any known pacemaker has ever demonstrated. There may now be a better physician understanding
of the risks of device explantation. However, recently, when physicians have been
faced with a situation where one of their patients has experienced a 1/10,000 sporadic
component malfunction, we have observed that sometimes there is still an initial physician
response to want to explant all similar devices from their patients or to take special
unnecessary precautions. In short, it appears that patient and physician expectations
for the performance of CRM devices may have outstripped engineering capability, and
this potentially leads to both patients and physicians failing to utilize beneficial
device therapy or creating, secondarily through physician actions, more patient harm
than the original device hazard [20]. While it is clear that the onus is on the industry
to meet today’s standards and expectations, it is also clear that those standards
and expectations are becoming more stringent. A high profile example of this is an
ICD lead recall from 2007 (6-year lead survival, 84.8 %). While that same manufacturer
had a better immediately precedent product, the ICD lead made by the same manufacturer
in the late 1990s had a similar malfunction rate (6-year lead survival, 86.8 %), and
the product from the 1990s was not subject to any regulatory action because it met
the performance expectations of a decade earlier. It must also be recognized that
the lead recalled in 2007 was likely designed in around 2000, when the performance
expectations were different [21].
An appropriately designed post-market surveillance program by medical device manufacturers,
which captures unexpected hazards as well as appropriate device performance, is necessary
to ensure the continued availability of medical devices in the USA. When performance
does not meet regulatory expectations, executing a worldwide field action may cost
a manufacturer over 50 million dollars, excluding the secondary loss of future business,
and/or legal costs which are often far more. This fact has a potential significant
effect on innovative technology. Manufacturers now must consider not only approval
of a device but also the potential post-market quality/regulatory risks any new innovative
device may pose over time. Thus, some manufacturers chose to deliberately avoid the
development of products with the greatest regulatory or liability risks. Although
the US Supreme Court ruled that in Riegel v. Medtronic a medical device manufacturer
cannot be sued under state law by patients alleging harm from a device that received
marketing approval from the FDA, this does not end all legal actions, which also continue
to be a significant cost to manufacturers, particularly in the USA. Thus, the post-market
regulatory environment in the USA provides several significant disincentives to innovative
technology. Consistent with this, there has been a steady decline in both 510(k) and
PMA submissions to the FDA (Fig. 2) [22, 23]. This suggests that the current environment
has already potentially significantly negatively impacted medical innovation despite
the advances in genomics, nanotechnology, and computing power. However, there are
also other significant non-regulatory disincentives currently to medical device innovation,
particularly in the USA. These include the economic situation which has reduced available
venture capital, concerns about the medical device excise tax, the projected reductions
in health care spending as a disincentive as it is difficult to project future return
on investment, the challenges to creating industry–university partnerships currently,
the disconnect between FDA approval and reimbursement, the mounting costs of clinical
trials in the USA, patient concerns about investigational devices, and insurance payers’
reluctance to reimburse for investigational therapy. Thus, even innovative novel therapy
such as the subcutaneous ICD faces significant time and financial hurdles.
Fig. 2
Progressive declines in 510(k) and PMA submissions to the FDA between 1999 and 2009.
Reproduced from [22]
Beyond safety and efficacy: implications of comparative effectiveness research on
medical devices
Comparative effectiveness research (CER) has been touted as one of the key tools of
policy makers to lower costs and improve patient outcomes. While the term is broad
and can encompass a range of concepts, for our purposes, we can define CER as research
comparing patient outcomes and risks and benefits between two or more therapies or
devices. Policy makers expect payers, physicians, and patients to use information
generated by CER when making clinical and health care choices. Specifically, that
information about the relative risks and benefits of therapies will cut down on unnecessary
care, increase the use of more effective therapies, and lower costs. In short, CER
has the potential to do much good, but much depends on how the information is generated
and used. The implications, for patients, physicians, and the medical device industry,
are considerable.
The concept of comparative effectiveness is not new. Health technology assessment,
for example, has taken place since at least the early 1970s in the USA and other countries.
For example, the Office of Technology Assessment was established in 1972 as a Congressional
agency to provide advice to Congress on the “early indications of the probably beneficial
and adverse impacts of technology and to develop other coordinate information which
may assist Congress” (Public Law 92-484). One of its earliest reports was on the assessment
of cancer-testing technology and saccharin issued in 1977. In the last year of its
existence, the OTA issued CER reports on the cost effectiveness of prostate cancer
screening in elderly men and the effectiveness and costs of osteoporosis screening
and hormone replacement therapy [24].
In 2010, with the enactment of the Accountable Care Act, the federal government created
the Patient-Centered Outcomes Research Institute (PCORI) whose mission is to help
“people make informed health care decisions, and improve[s] health care delivery and
outcomes, by producing and promoting high integrity, evidence-based information that
comes from research guided by patients, caregivers and the broader health care community.”
PCORI funding was US $150 million for fiscal year 2012 [25]. Needless to say, this
represents a small fraction of government expenditures on health care research (as
the NIH invests US $30.9 billion annually in medical research) [26]. PCORI is not
the only federal entity active in CER. The Agency for Health Care Quality and Research
undertakes CER-like activities (e.g., technology assessments), frequently at the request
of the Centers for Medicare and Medicaid Services (CMS). States are also actively
considering or conducting CER, with the State of Washington’s Health Technology Assessment
program being the most visible and notable example. Perhaps the best-known agency
is the United Kingdom National Institute for Clinical and Health Excellence (NICE),
which conducts CER and disseminates its findings to patient, clinicians, purchasers
of care, and others to inform decision making. Other organizations, such as the Blue
Cross and Blue Shield Technology Evaluation Center, and private payers also engage
in technology assessment for use in making coverage decisions. Health technology assessment
is actively performed in France, Germany, Belgium, Italy, and Australia, with other
countries actively considering establishing similar programs (South Korea and Japan,
for example).
Industry supports CER, provided the focus is on the clinical appropriateness of alternative
treatments, and the research agenda and related questions are shaped with input from
multiple stakeholders, including physician specialty societies and medical device
companies. To be most effective, conducting CER and interpreting results in the context
of medical devices, policy makers, physicians, and purchasers should be cognizant
of several factors that can influence CER findings.
Perhaps the most important factor is the time frame considered when studying implantable
medical device interventions. For interventions that act continuously over time, as
is the case for many implantable devices, the magnitude of clinical benefit increases
with time. Thus, the benefit from the device increases with time and the number of
patients needed to treat to save a life, for example, diminishes with longer follow-up.
Table 1 shows the effect of varying follow-up duration on the number needed to treat
to save a life with ICDs [27], and this is graphed in Fig. 3. Thus, a key question
for those designing clinical studies is the length of follow-up. If it is too short,
a CER review could result in an unfavorable conclusion. Lengthening the follow-up
time, however, has consequences. The sponsor, frequently medical device companies,
will incur higher costs. And importantly for patients, physicians, and company sponsors,
depending on the study design, a longer follow-up could increase the time to market
and prevent the technology from being available to patients quickly. In the past,
CRM device manufacturers have tended to focus on the comparison of device versus optimal
pharmaceutical therapy for the prevention of sudden death or treatment of heart failure,
with the measured outcomes focused on mortality, heart failure hospitalization, and,
more recently, on economic endpoints. In the future, it may be necessary to focus
on other outcomes and other comparative treatment strategies.
Table 1
Individual trial data demonstrating the impact of follow-up time on observed benefit
(life-years gained per device implanted) and on observed number needed to treat to
gain 1 life-year
Trial
Life-years gained per device implanted
Size of number needed to treat to gain 1 life-year
1 year
2 years
3 years
1 year
2 years
3 years
Wever et al. [35]
0.10
0.30
0.58
10
3.3
1.7
MADIT [36]
0.11
0.30
0.50
9
3.3
2.0
AVID [37]
0.05
0.12
0.22
21
8
4.6
CIDS [38]
0.01
0.04
0.09
123
24
11
MADIT II [39]
0.01
0.06
0.13
133
17
8
CASH [40]
0.01
0.09
0.18
133
11
5.6
Schläpfer et al. [41]
0.07
0.23
0.41
15
4.4
2.5
MUSTT [42]
0.07
0.21
0.40
15
4.7
2.5
Reproduced from [15]
Fig. 3
Cost effectiveness ICD versus control over lifetime in 2005 US dollars
Time also plays an important role with respect to cost effectiveness reviews. Using
the example in the table, making conclusions about the relative benefit of an ICD
implant at 1 year will yield different conclusions than when making conclusion 3 years
after implant because of the up-front costs of the device and surgical implantation.
The same issue must be considered when assessing the cost effectiveness of therapies.
For example, the cost effectiveness of ICD therapies can range from about US $150,000
to US $300,000 at 3 years, but drop substantially to about US $50,000–100,000 over
a lifetime. Using a lifetime frame as the basis of evaluation yields a more appropriate
understanding of the value of ICD therapies as, in general, ICD implantation is a
lifelong treatment strategy, and MADIT II 8-year follow-up, for example, shows continuing
benefit to the therapy over that extended time period [28]. However, over increasing
time, there may emerge new comorbidities which can also affect outcomes not influenced
by an ICD. Thus, consideration needs to be given to potential changing effectiveness
of the device over time. An examination of the potential impact of comorbidities on
the mortality of an initial implant ICD population was recently published [29], and
this type of analysis may need to be considered over time, for example, with each
subsequent device replacement.
A second important factor to consider is the evolutionary nature of medical device
innovation. Medical device interventions frequently require time to evolve. Continuous
evolution and improvement in the technology, procedure techniques, and physician skill
will influence clinical outcomes and thus, in many cases, affect patient outcomes.
As with the follow-up time, the evolutionary nature of medical device innovation must
be considered both by trial sponsors as well as in the interpretation of results under
CER. For example, the safety and efficacy endpoints for a particular left atrial occlusion
device improved significantly over the course of just a few years with physician experience
[30]. Furthermore, during the time course of a long-term CER, the medical device being
assessed has often undergone production of new generations of products associated
with improved outcomes. The challenge for policy makers is to balance between “too
early to be evaluated appropriately” and such widespread dissemination that changing
clinical practice is challenging even with an unfavorable finding of effectiveness.
For example, the United States Preventive Services Task Force generated substantial
controversy in October 2011 with its draft recommendation against PSA-based screening
for prostate cancer. The recommendation was based on an update to a 2008 evidence
synthesis which concluded that the evidence was insufficient to assess the balance
of benefits and harms in men <75 years of age. The challenge for physicians, payers,
and patients is that while the evidence was being generated, the use of PST-based
screening became widespread, with over half of American men over 50 years of age receiving
the test. A favorable recommendation is not a guarantee that clinical practice will
change either. For example, despite the depth and quality of clinical data and a relatively
favorable NICE guidance, ICD therapy use remains underutilized in England and Wales
[31, 32].
A related challenge for CER will be to keep evidence reviews and technology assessments
updated as technology evolves and clinical guidelines change. Policy makers and those
conducing CER must balance resources between the need for evidence in virtually all
clinical, therapeutic, and health care delivery spaces with the need to update existing
research to keep pace with innovation. For example, while technology and clinical
guidelines have changed with respect to cardiac resynchronization therapy, many coverage
policies in the USA and NICE guidance remain unchanged (NICE is in the process of
reviewing its guidance on ICD and CRT-Ds and expects to issue revised guidance in
late 2013). Physicians and purchasers will have to consider the age of CER assessments
with more current clinical data and clinical guidelines when making clinical choices
and coverage decisions. Physicians will also have to consider that the rapid evolution
in devices will mean that CER (and sometimes clinical guidelines) is based on technology
that is several generations older than currently available.
The cycle of innovation and interaction with CER and guidelines may also require device
companies to develop a range of post-market research strategies to ensure that the
available data are consistent with later generations of technology. Payers will also
undoubtedly use CER to inform coverage requirements. For example, CMS’s recent decision
to cover TAVR for Medicare patients will require hospitals and physicians to participate
in a national registry that will follow patients for at least 1 year and answer specific
questions on patient outcomes (e.g., stroke, mortality, etc.) and quality of life.
Collecting and disseminating such data will increase costs for physicians, hospitals,
and medical device companies. However, the rationale for such registries is in part
based on the need to monitor the performance of complex devices in non-investigative
centers [33].
Other factors to consider when designing comparative clinical studies and interpreting
CER reviews are the patient population studied and the number of patients being studied.
Clinical trials, by their nature, study average effects in large populations. Yet,
when it comes to our own health, none of us are interested in what works best for
the average patient—we all want to know what works best for the individual patient
we are caring for at a given point in time. Making sure that comparative clinical
studies are large enough to conduct meaningful subgroup analyses is therefore important.
A good example comes from the MADIT-CRT Study. This trial—looking at the effect of
cardiac resynchronization therapy in patients with minimally symptomatic congestive
heart failure—met its primary endpoint for the entire population studied. However,
subgroup analysis revealed a marked difference in outcome according to the presence
or absence of left bundle branch block (Fig. 4). A CER review using the group average
would understate the effect in the patients with left bundle branch block while overstating
the effect in the remainder. Subgroup analysis may become particularly important in
view of the ongoing genomic revolution and the trend toward personalized medicine—in
the coming world, we may each comprise our own subgroup of one!
Fig. 4
MADIT-CRT post hoc analysis of left bundle branch block. It was subsequently discovered
and validated that in the LBBB subgroup, patients received substantial benefit from
CRT-D. Non-LBBB patients did not show evidence of benefit. The LBBB subgroup made
up approximately 70 % of the total MADIT-CRT population (adapted from Boston Scientific
CRT-D product labeling)
On the positive side, large-scale registries such as the ACC/NCDR ICD registry may
provide long-term data for the analysis of outcomes [29]. These registries may provide
a more cost-effective method for assessing the comparative effectiveness of therapies,
particularly when coupled to remote patient monitoring. Unfortunately, this method
does not provide an approach to assessing outcomes in an alternative treatment group.
Finally, an important question, again related to time, will be when to stop. When
do policy makers and physicians determine that we have “enough” information? Without
“stopping rules” for CER similar to those found in clinical studies, we run the risk
of placing societal resources into efforts with ever diminishing marginal benefits.
For example, is there a prospective plan of when to end the ACC/NCDR ICD registry?
Unfortunately, stopping times will vary with each device type and will be particularly
difficult to determine for innovative therapy, and there is a significant risk that
pharmacologic principles may be inappropriately applied to devices. Particularly concerning
is a recent publication by Chen et al. [34] which argues in favor of the “inclusion
of comparative effectiveness data in high-risk cardiovascular device studies at the
time of premarket approval.” The implantable defibrillator, which represented innovative
therapy in 1985, did not have a randomized comparator, and it required thoracotomy
for lead and patch placement, and it is unclear whether it would have met this criteria
for approval. However, clearly subsequent iterations of the ICD have demonstrated
CER benefit, but without original approval, this benefit may never have been achieved.
Similarly, the subcutaneous ICD did not have a randomized comparator group, but it
is our opinion that this approach to pre-market CER would have been entirely inappropriate
and significantly raised the regulatory burden to the point of potentially excluding
beneficial therapy for high-risk cardiovascular patients for many of the previously
cited reasons. The S-ICD which was approved by the FDA is a first-generation device
which will rapidly be redesigned; physicians are currently unfamiliar with implantation
techniques, and there will inevitably be a learning process for appropriate use. CER
has a role later when there is stability of the device design, physician knowledge,
and patient selection.
Summary and a path forward
The last half century has seen unprecedented improvements in health care with declines
in cardiovascular deaths, due in part to the introduction of a series of revolutionary
new medical technologies. Nevertheless, current circumstances pose significant hurdles
to continuing medical device innovation: the regulatory environment is globally fragmented
and currently particularly burdensome in the USA. We have a clear need for good comparative
effectiveness data for new technologies, but there are particular issues with designing
comparative clinical trials and interpreting CER assessments in the context of new
medical devices. There clearly are very divergent views on this topic, and undoubtedly,
there are even divergent views within industry. However, we believe that there are
potential paths to maintain and improve innovation, as seen from our industry viewpoint:
Industry:
Greater care is needed in planning innovative products that meet today’s health care
needs. Where there is change, there is also new opportunity (e.g., reducing 30-day
re-hospitalization for heart failure).
Product cadence should be coupled to available outcomes data (e.g., a new feature
in a medical device should be planned to be submitted for approval shortly after there
is beneficial clinical data to support that feature).
Developing and expanding methods such as remote patient monitoring to reduce the cost
for CER and potentially to improve analysis of patient outcomes and device performance
Making more cost-effective devices. For example, a longer lasting device results in
a lower cost over the long term. Market drivers in low- and middle-income nations
bring this to a higher priority for industry.
Less invasive alternative innovative therapies not only may reduce costs by avoiding
surgical procedures but may also reduce the costs related to surgical complications.
Researching the currently unmet needs in health care today (e.g., disease processes
with disproportionate expenditure or poor outcomes currently such as heart failure
with preserved ejection fraction)
Improved device reliability to reduce costs related to device malfunctions (e.g.,
current focus in CRM is on lead reliability)
Co-sponsored CER with government, other manufacturers, and include active participation
from physician bodies
Develop genotypic or biomarker-based device indications with improved cost/benefit
ratio
Physicians:
Set appropriate expectations amongst physicians, regulators, and patients about what
realistic device performance should be and set up forums to publicly share this with
industry. Physicians are encouraged to volunteer for advisory roles with FDA, CMS,
and other payers and to express concern when they believe that regulatory barriers
are excessive.
Physicians and industry need to disseminate understanding of the tensions between
novel designs and reliability. Thus, when a physician requests a smaller F size lead,
which may be easier to place in the patient, the industry design compromises may lead
to potentially greater malfunction rates. In the case of an ICD lead, the net clinical
effect has sometimes been suboptimal. In contrast, reducing the size of left ventricular
coronary venous leads to 4 F has enhanced the usability significantly, but it is as
yet unclear if there is a reliability trade-off in lead life. Similar compromises
between functionality and reliability often apply to more complex device feature sets.
Programming higher tachycardia detection rates may reduce inappropriate shocks from
an ICD and may be beneficial, but in extremes may result in failure to detect ventricular
fibrillation with low rates at the RV electrode.
There is a tendency by multiple parties to lay blame when device performance is not
as expected in the long term. That is, when the manufacturer has performed the appropriate
testing, consistent with engineering knowledge in place at the time of development,
and that testing satisfies regulatory requirements, it is hard to know whether there
is fault if the performance is suboptimal in the long term. Stated in other words,
it is difficult to test for unknown failure mechanisms and mitigate unknowns, particularly
in novel products and design. All parties need to consider how to address and educate
on this, and physician bodies may choose to consider the malfunction rates found in
post-approval studies when updating therapy guidelines.
Understanding the potential trade-off in design versus new features such as magnetic
resonance conditional safety is important for physicians (e.g., devices do not sense
during the scan, long-term confusion among non-device following physicians about which
patients are conditionally safe).
In dealing with devices with less than optimal performance, physician bodies need
to carefully consider the risks of intervention and compare that with the risks of
non-intervention in developing recommendations.
Physicians need to be vigilant for device malfunctions and report to both the FDA
and the manufacturer. Returning products for manufacturer analysis where there is
concern is important.
Government (legislative and regulatory bodies):
Efforts to produce regulatory convergence across borders
Consider reviewing the regulatory burden for device approvals. The FDA does have a
fast pathway for innovative therapy with unmet health care need. However, even this
pathway has a significant cost, as seen with the recent S-ICD development which took
over a decade and cost in excess of 400 million dollars.
Maintain consistent expectations for product performance consistent with device design
reality
Educate the public on the expected performance that is considered acceptable
Consider the use of less burdensome remote follow-up device data coupled to CMS or
NCDR data to satisfy the post-market surveillance needs
Develop innovative payment policies to enable savings gained by one stakeholder using
innovative technologies to be shared with other stakeholders
Develop an understanding that processes used in monitoring pharmaceuticals sold to
millions of patients over a decade are difficult to apply to devices that are often
sold in numbers <100,000 for 2–3 years for individual models
Provide incentives for greater cost effectiveness. For example, greater battery longevity
is usually associated with larger device size, which is not popular with physicians
or patients. The manufacturer potentially may lose money because of fewer generator
replacements. The major beneficiary of this technology is the health care payer, but
currently, they do not provide incentives for longevity.
Provide for reimbursement during an interim period after the development of a novel
medical technology, such that the technology might be optimally explored and experience
gained, before definitive CER studies are done. While this may exist in CPT coding,
CMS may not always rule in concert with FDA approval.
A practical delineation of appropriate physician–industry collaboration to produce
optimal future devices