|Home | About | Journals | Submit | Contact Us | Français|
Outcomes research for medical devices has always lagged, with good reason. Comparative effectiveness initiatives, though, will soon spur demand for more and better data, and that means more and better outcomes research training.
Comparative effectiveness research (CER) is an effort to improve population health outcomes while stemming the tide of rising healthcare costs. With the exception of cardiac stents (Chambers 2009, Ellis 2009, Leon 2009), little is known about how CER can and will be applied to medical devices, such as prosthetics and mesh repairs of hernias and vaginal organ prolapse; diagnostic tools, such as biomarkers for detecting occult metastases and genetic variations in cancer; radiological screening tools for cancer; and computed tomography (CT) for medical imaging. Whether and how different medical devices can be compared, or, at a minimum, their effectiveness measured, will depend on our ability to identify and develop appropriate measures for tracking and evaluating the end results of using such devices.
Outcomes research involving medical devices lags behind outcomes research for pharmaceuticals for various reasons. Devices and pharmaceuticals differ not only in terms of the data required for U.S. Food and Drug Administration (FDA) approval, but also in their management of product quality and patient safety and the factors that drive price (Drummond 2009). Whereas these elements have led to an increasing demand for outcomes data for pharmaceuticals, they have not had the same effect in the device market (Drummond 2009, Hahn 2008, Pauly 2008).
In this article, we discuss current issues involving the approval, safety and quality monitoring, and pricing of medical devices and explain why the pressure for measurable outcomes and CER data is likely to increase in the near future. The implications of this increased pressure also will influence other biotechnology industries that are now grappling with the changing approval and reimbursement landscape in the United States. Third-party payers will need to understand the challenges faced by medical device manufacturers in generating the desired data and become partners in methodological and priority-setting discussions.
One of the pressing issues in assessing the quality of medical devices is the current lack of appropriate measurement tools and outcomes data. This lack is partially explained by the current FDA processes for device approvals. Medical devices are stratified into three classes based on patient risk: Class I represents the lowest risk to the patient and Class III the highest risk, while Class II fails in the middle. Class III devices, which include implantable devices, are held to a strict premarket approval process (FDA 2009a) that requires scientific documentation that supports the safety and effectiveness of the device through the provision of clinical and nonclinical laboratory data. Class I and Class II devices, on the other hand, follow a process called “pre-market entry notification” that demonstrates how the proposed medical device is equivalent to an already approved device (FDA 2009b). This process requires only descriptive data, and in some cases, performance data, to establish equivalence. Given the paucity of data currently required for device approval, there is growing pressure on the FDA to implement more rigorous approval processes with greater emphasis on outcomes and quality (Dhruva 2009). In addition to a more restricted use of the 510K exemption from clinical trials and a requirement for well-designed trials, the FDA could help ensure public safety through an improved system of post-marketing surveillance (Garber 2010). Following the recent Supreme Court decision in Riegel v. Medtronic, which prevents patients from suing manufacturers for injuries caused by FDA-approved devices (Greenhouse 2009), consumer advocates have now been mobilized to call for a new FDA process to ensure consumer safety. Proposed legislation, such as the Medical Device Safety Act of 2009, aims to reverse some of the current legal limits that restrict patient suits against medical device companies (Curfman 2009). If passed, this legislation may impose high levels of regulatory and litigation risk on medical device companies, which — although aimed at improving quality and safety — may have the unintended effect of discouraging innovation (Von Spakovsky 2009). The trade-offs between investment in innovation and the necessity to do additional research to ensure quality will have to be recognized by both policymakers and industry experts.
As the spotlight turns to quality and outcomes, medical device companies will have to contend with two critical issues that pharmaceutical manufacturers also face — the reporting of adverse events (AEs) and off-label use. AEs related to Class III devices fall into three major categories: complications during implantation; unexpected effects after implantation; and failure of a device following implantation. For example, AEs related to hip and knee replacements have been reported to be approximately 5.1 percent and 6.6 percent, respectively (Gawande 1992). With regard to complications during implantation, one survey of orthopedic surgeons found that 29 percent experienced equipment errors during surgery (Wong 2009). Although device failures are not common, a handful of cases, such as the malfunction of implantable defibrillators that result in failure and fracture (Kallinen 2008, Maisel 2005, Maisel 2008), can further increase the pressure for regulatory changes to prevent the approval of faulty devices (Garber 2006).
Just as the off-label use of prescription medications is becoming more highly scrutinized (Curtis 2004), so, too, is the off-label use of medical devices. For example, total reverse shoulder prostheses were developed and approved for such indications as glenohumeral arthritis with an irreparable rotator cuff, irreparable rotator tear, or glenohumeral instability. However, physicians have been extending the use of these protheses to the treatment of osteoarthritis of the shoulder (Frankle 2006). The lack of clinical efficacy and outcomes data on the use of these prostheses limits the ability of both provider and patient to fully assess their risks and benefits. Outcomes data retrieved through patient registries that evaluate device use in real-world clinical settings is one way to provide valuable information on both outcomes and safety issues.
Pricing in industries characterized by a high, continuous degree of innovation is driven by a different set of factors than pricing in industries marked by stability in product design. Thus, the pricing of medical devices, which are subject to continuous innovation, operates very differently than the pricing of pharmaceutical products. The factors that affect the pricing of medical devices include clinician preferences (Hahn 2008, Lerner 2008), patient insensitivity to price (McKinsey 2008), separation of roles for device selection and price negotiation (Pauly 2008), and the lack of market data on price variation (Lerner 2008, Pauly 2008). One of the complicating factors in device pricing is the fact that although a clinician selects a particular device, the hospital contracting staff or group purchasing organization negotiates the price with the manufacturer (Pauly 2008). Thus, the clinician often is unaware of price differentials among products and may be more driven by familiarity with a particular product rather than data regarding quality, outcomes, or price. Typically, insurers fully cover medical devices; thus, the patient also is shielded from any price consideration (McKinsey 2008). This disconnect between product selection and price negotiation has contributed to the lack of demand for more outcomes data (Lerner 2008).
Although many researchers advocate a greater use of randomized clinical trials (RCTs) to evaluate devices, a number of issues would hinder their efficacy (Drummond 2009). The large variety of products that fall under the umbrella of the medical device industry, ranging from diagnostics to implantable devices, makes it difficult to develop a unifying set of standards for the industry, because different categories of devices have different methodological issues in the conduct of outcomes research. For diagnostic devices, from biomarkers for genetic variations in cancer to such radiologic screenings as mammograms, early detection of a disease can be hard to link to an improved health outcome, because the therapeutic treatment provided as a result of detection also impacts the final health outcome (Drummond 2009). In the case of implantable devices, procedures commonly used in clinical trials for pharmaceuticals such as blinding patients and doctors to treatment assignment, may introduce bias and raise ethical questions if practices such as sham procedures are required for comparisons with placebo treatments.1
Another factor is the learning curve associated with devices used in surgery, e.g., diaphragm pacing systems (Onders 2005). When a new surgical device enters the market, surgeons have little or no experience in using it, but as the technology is diffused over a period of time, increased familiarity with the device will lead to improved outcomes (Drummond 2009, Onders 2005). Incremental improvements in a device also may lead to improved efficacy and, ultimately, to better outcomes (Frey 2009).
Accounting for such incremental modifications in RCTs can be challenging and expensive. In addition, the FDA requires limited evidence for product approval, with some devices approved on the basis of nonrandomized studies of small patient populations and no long-term efficacy data. This is especially true for such devices as prostheses where design innovations occur at a rapid pace. RCTs are conducted in small patient populations and incremental changes are brought to market without supporting evidence of patient benefit (Selles 2005).
In addition to calls for changes in the FDA approval process, the Centers for Medicare and Medicaid Services has been increasing its requirement for outcomes data for the Medicare population before it will cover certain procedures, devices, and diagnostics (Neumann 2010). Although CMS has long been charged with covering only care that is “reasonable and necessary,” it was-n’t until 2000 that CMS specified that “…there [is] sufficient evidence that demonstrates that the item or service is medically beneficial for a defined population…” and that such evidence was an explicit criterion for national coverage decisions (NCDs) (Federal Register 2010). The recent NCD for computed tomography colonography, which denied coverage for Medicare patients, indicates that the shift in demand for outcomes data is beginning (Neumann 2010). Faced with the same pressures as CMS to bring down costs while providing appropriate care (Gelijns 1994), third-party payers may follow CMS’ lead and request greater evidence of a benefit when making a reimbursement decision.
Although the unique nature of the medical device market has historically created conditions that limit the demand for outcomes and quality data, the rapidly changing market and payer environment is leading to pressure for more and better data. This new focus has several implications for the medical device and diagnostics industry.
First, agreement will be needed on how to handle the methodological difficulties of conducting RCTs for devices. Industry, government, and academic experts will have to collaborate on the identification of appropriate study designs and outcomes measures for the different device categories, including diagnostics, and guidelines for the conduct of CER and economic evaluations. Addressing these issues will require well-trained researchers who are versed in assembling and analyzing outcomes data and who also understand the methodological caveats of analyzing medical devices and diagnostics. Thus, the second major implication is the need for more graduate education programs in outcomes and effectiveness research for medical devices and diagnostics.
Currently, about 15 masters-level programs in public health and pharmacy schools in the United States offer some degree of such training. None, however, provides specialized training in medical device outcomes research. Post-graduate fellowships offer another alternative to training research professionals for the medical device industry, but to date we are aware of only one such program. A third major implication is a potential slowing of the rate of innovation in the industry. As the industry, both willingly and through the requirement of increased regulation, commits more resources to generating outcomes data, the amount of resources available to devote to innovation may decrease. Although improved safety and documented outcomes lead to patient benefits, it is unclear whether patients may also suffer from innovations foregone.
The specifics of how CER can and will be applied to medical devices and diagnostics is uncertain. Regardless, the demand for more and better data is likely to grow. To meet this demand, both the medical device industry and payers will have to invest in the training and hiring of high-caliber individuals with the interest and ability to tackle the challenges that lie ahead.
1See, for example, the discussion by Wei (2009) on the use of sham procedures in the study of vaginal prolapse repair.
Anita Mohandas, MSc, is currently a Thomas Jefferson University Ethicon fellow. Kathleen A. Foley, PhD, is the preceptor for the Thomas Jefferson University Ethicon Fellow. Ethicon is part of the Surgical Care Group of Johnson & Johnson.