Charles H. Kyper
Device classification is a cornerstone of the Medical Device Amendments of 1976. It is intended to establish the lowest level of regulatory control that still provides reasonable assurance of a device's safety and effectiveness. As mandated by this legislation, FDA and its classification panels reviewed then-marketed devices at public meetings. Following receipt of the panel recommendations in late 1977, it might have appeared that, over the next 10-year period, FDA completed the required rule making to classify these devices. However, throughout this period, FDA and many manufacturers and importers recognized that a significant number of devices had been overlooked during this classification review. Only in recent years has FDA attempted to remedy the situation.
This article discusses past and present FDA classification and reclassification activities and the events leading to the ongoing program of classifying devices overlooked during the initial review. Inconsistencies in FDA regulation of accessory devices and the resultant overregulation of many seemingly unclassified accessory devices are also discussed.
The Safe Medical Devices Act of 1990 (SMDA) requires FDA to reconsider the classification of most Class III (premarket approval) devices in light of the redefinition of Class II devices (special controls rather than performance standards alone) in this subsequent legislation. For half of the 116 eligible devices placed in Class III following the previously mentioned classification review, FDA published two orders on August 14, 1995, requiring industry submission of available safety and effectiveness information.1 This and other information available to FDA will be used to reconsider their classification. It is unlikely that any of these devices will presently be reclassified into Class I, where they would be subject only to general controls. General controls include requirements for establishment registration, device listing, premarket notification, labeling, good manufacturing practices (GMPs), and medical device reporting (MDR). They apply to all devices, but only Class I devices can be exempt from certain of these controls (e.g., premarket notification and GMPs).
Concurrent with reconsidering the classification of these devices, FDA has been attempting to classify a number of marketed devices that were overlooked during the classification review in the mid-1970s. Although long aware of this oversight, FDA opted not to correct it until four years ago. Agency justification for its delay was the drain this activity would impose on FDA resources involved in the review of device submissions. Until recently, manufacturers needed to scrupulously read Federal Register notices of FDA advisory committee meetings to learn if the agendas included a classification review of their devices. This short notice severely limited industry's ability to provide information that could prevent a device from being placed in Class III. Subsequent industry complaints regarding FDA conduct of this program have negated many of the efforts by FDA and its classification panels.
In 1992, FDA's Center for Devices and Radiological Health (CDRH) initiated a new program to identify and classify legally marketed, but unclassified, devices that present an unreasonable risk because of unresolved safety or effectiveness issues. This program provides an opportunity for industry to oversee how CDRH and its device classification panels apply the relevant provisions of SMDA. Most notable among these is the provision redefining Class II devices. The legislative intent was to provide FDA with the needed flexibility to reclassify many Class III devices into Class II by permitting it to impose special controls rather than rely only on performance standards. Special controls may include performance standards, postmarket surveillance, patient registries, and guidelines (including guidelines for the submission of clinical data in premarket notification submissions), as well as other appropriate controls (e.g., device-specific labeling) that FDA deems necessary.
Except in rare instances, the 1976 amendments did not provide for an alternative control other than premarket approval when general controls are considered insufficient and there is not enough information to establish a performance standard. SMDA reflects congressional agreement with many in FDA and industry that this limitation in the earlier legislation had resulted in devices unnecessarily being placed in Class III and had, at the same time, impeded their reclassification. The previously mentioned classification programs involve review of available information at public meetings of FDA advisory panels followed by notice-and-comment rule making. This required process provides an opportunity to gain useful insight into how the revised Class II device provisions for special controls are, or are not, being applied.
In January 1996, FDA published a final rule reclassifying 111 low-risk Class II devices into Class I and exempting them from the 510(k) submission requirements.2 Supporting justifications in the July 1995 proposed rule are, in part, inconsistent with statutory and regulatory provisions and would not apply to many of these devices (e.g., sterile reusable devices) based upon the testing and labeling information presently required for FDA marketing clearance.3 This raises the question of whether FDA will even the playing field by removing or relaxing the testing information required in 510(k) submissions for other low- to medium-risk devices.
CLASSIFICATION OF PREAMENDMENT DEVICES
The 1976 Medical Device Amendments to the Federal Food, Drug, and Cosmetic Act (FD&C Act) required FDA advisory committee review and FDA classification of then-marketed devices. These devices are commonly referred to as preamendment devices. (To facilitate discussion, this term also includes substantially equivalent devices first marketed after the May 28, 1976, enactment of this legislation. The latter are also referred to as substantially equivalent postamendment devices.) Public meetings of the FDA device classification panels to review these devices culminated in the submission of the panels' recommendations to FDA in the fall of 1977. FDA rule making to classify the devices was sporadic but appeared to many to be completed in 1988.
A considerable number of legally marketed devices were overlooked in the review and rule-making process. This omission soon became apparent upon industry submission, and FDA review, of device-listing information and premarket notifications under section 510(k) of the FD&C Act. Neither party could cite a directly applicable classification regulation as required in these submissions. The absence of an applicable classification regulation hampered the agency's ability to effectively and appropriately regulate many of these devices.
Three-letter product codes assigned to marketed devices by FDA have enabled the agency to monitor both unclassified and classified devices with varying degrees of success. The product code applies to a generic type of device and is used to tie together information in various FDA databases. This common thread is often broken by the absence or incorrect entry of a product code, or by the assignment of multiple product codes to a device without a justifiable reason or a readily available reference source for selecting the applicable product code.
The first meaningful FDA effort to classify these overlooked devices occurred in 1992. This was prompted by congressional and media criticism of the alleged FDA failure to appropriately regulate one such product--a temporomandibular joint (TMJ) implant unlike others on the market. Critics alleged that this public health hazard could have been significantly minimized if FDA had appropriately placed TMJ implants in Class III following the initial classification review and proceeded expeditiously to require premarket approval. During the initial review process, the Dental Devices Classification Panel recommended that FDA classify TMJ implants in Class III. Apparently through oversight, FDA neglected to include these implants in the proposed classification regulations for dental devices. Between the December 30, 1980, proposed regulations and their finalization on August 12, 1987, there was ample opportunity for FDA to become aware of the omission and correct it.
The interarticular disc prosthesis is one of four prostheses identified as a TMJ implant. One firm's version of this device employed a polytetrafluoroethylene (PTFE) vitreous carbon material rather than metals typically used in similar prostheses. Jawbone erosion and other painful conditions resulting from dislocation and fragmentation of this firm's device may necessitate its removal from most patients. More than 26,000 implants were manufactured, but FDA was unable to determine how many were implanted.4 The firm's 510(k) submission included a claim of substantial equivalence to a silicone rubber implant marketed before 1976 (whose failure has generally resulted in less-severe consequences than those of the PTFE device). FDA granted marketing clearance of the PTFE prosthesis in 1983.
Although TMJ implants were not included in the initial classification regulations, it is interesting to note that FDA at that time classified into Class II the very same PTFE material when used to coat metal TMJ implants.
More than five years elapsed following an updated review by the FDA advisory panel before the final rule that classified all TMJ implants in Class III was published in December 1994.5 FDA cannot require that premarket approval applications (PMAs) be submitted before July 1997. Under section 501(f)(2)(B) of the FD&C Act, FDA must provide manufacturers of TMJ implants at least a 30-month grace period following classification to obtain needed safety and effectiveness information. PMAs are due within 90 days following promulgation of a final regulation requiring their submission or when the grace period expires, whichever is longer. FDA has publicly indicated its intention to issue this regulation before April 1997 so that PMAs are due on the day after the grace period expires.
It may appear that considerable time will elapse before FDA can effectively regulate TMJ implants. This should not necessarily give solace to manufacturers of other unclassified devices that FDA may place in Class III. As in the case of TMJ implants, FDA may in the interim require extensive clinical and nonclinical study results before granting further 510(k) clearances for these devices.6
The clinical experience with TMJ implants raised concerns within CDRH that other unclassified devices may present significant public health risks. In late 1992 CDRH identified more than 100 legally marketed, but unclassified, preamendment devices. Ten of these were considered to be potential Class III devices because of significant unresolved issues of safety or effectiveness. Input from the center's Office of Device Evaluation (ODE) review staff was the only information source used. This was done purposely because the immediate interest was to identify potential Class III devices rather than to develop an all-inclusive list of unclassified devices.
Among the more commonly known, but unclassified, devices identified by ODE reviewers are first aid kits, suture removal kits, wound closure kits, needle disposal (sharps) devices, bone marrow collection/transfusion kits, tactile hearing aids, eye pads, and corneal transport media. The CDRH preoccupation with regulating kits apart from the classified devices contained in them is disconcerting. This is especially true when 510(k) clearance is required for kits that are prepared for convenience purposes only and that contain only 510(k)-exempt Class I or otherwise legally marketed devices. CDRH has yet to issue the definitive guidance that was developed in 1994 to minimize inconsistencies in its regulation of, and to significantly reduce the number of marketing approval submissions for, medical device kits.
CDRH intentionally excluded unclassified in vitro diagnostic (IVD) products from this classification process. It has cleared 510(k) submissions for IVD tests for new analytes with the knowledge that these tests were not legally marketed on or before the enactment of the 1976 amendments. Classification names for IVD products are typically analyte-specific. Premarket notification and device listing submissions for certain IVD products must necessarily lack the required citation to an applicable device classification name and regulation. In the initial 510(k) submission for a new analyte, CDRH may permit the firm to be innovative in citing a predicate device to which its IVD product is substantially equivalent. This usually occurs only when the 510(k) submission contains sufficient information to establish the clinical utility of the test. Firms that have been required to obtain premarket approval for new analytes have complained of an uneven playing field. To minimize such a perception, CDRH in recent years has recognized the need to find an alternative to analyte-specific classification names.
This classification process is separate from that for certain preamendment products whose device status following the 1976 device legislation remained unclear until recent years. These products include liquid chemical germicides used with medical devices and processed human tissue products such as dura mater and heart valve allografts. Such products are presently considered to be devices. FDA continues to determine case by case whether processed human tissue is an FDA-regulated product and whether it should be regulated as a drug, biologic, or device.
FDA announced in a 1991 Federal Register notice that heart valve allografts are considered medical devices and are thereby subject to a 1987 final order requiring premarket approval for replacement heart valves.7 Distribution of the devices would be limited to investigational use until marketing approval was obtained. Affected manufacturers were obviously unprepared to submit PMAs, and they initially experienced great difficulty developing acceptable informed consent forms and obtaining institutional review board approvals and third-party reimbursement. Most of these allografts are implanted in children under the age of 14 and are preferred by many surgeons because they are less likely to calcify than bioprosthetic valves and are available in smaller sizes than most mechanical valves. Through a petition and a lawsuit brought against FDA, manufacturers were able to delay implementation of this notice for 10 months. In a 1994 Federal Register notice, FDA rescinded the notice and permitted the distribution of heart valve allografts without an approved investigational device exemption or PMA.8 This notice also announced FDA's intention to initiate procedures to classify these devices into Class II. FDA indicated therein that based on its increased experience with the use of special controls and relevant industrywide standards, it believes that special controls may adequately address the critical public health concerns raised by these devices.
Only time will tell whether FDA's relatively recent approach to regulating heart valve allografts as Class II devices will serve as a precedent during its current classification and reclassification activities involving significant-risk devices. Two orders published in the August 14, 1995, Federal Register provide only limited encouragement.1 These SMDA-mandated orders require industry submission of available safety and effectiveness information to be used in the FDA classification reconsideration of 58 actively marketed preamendment Class III devices not yet subject to premarket approval. FDA indicated therein that 27 of these devices, and 15 other actively marketed preamendment Class III devices not included in this order, are not considered candidates for reclassification. Reclassification of Class III devices is hampered in part by FDA failure to implement the SMDA provision that makes safety and effectiveness information in approved PMAs available for the agency's use in reclassifying previously approved, and other, Class III devices. This information becomes available one year after publication of a Federal Register notice under section 520(h)(4) of the FD&C Act once four PMAs have been approved for a generic type of device. Presently there are at least four approved PMAs for more than 30 generic types of devices.
PANEL REVIEW OF UNCLASSIFIED DEVICES
CDRH has convened advisory committee meetings in an effort to obtain the required panel classification recommendations for several of these unclassified preamendment devices. This public review has focused primarily on those devices considered to be Class III candidates. Besides the final rule for TMJ implants, FDA published a proposed rule in early 1995 to classify transilluminators for breast examinations in Class III.9
The slow progress is due in large part to flawed panel meetings. FDA has yet to make a list of these unclassified devices publicly available. Affected manufacturers sometimes have had short notice or were unaware that their devices would be reviewed at an FDA panel meeting. Federal Register notices of these meetings are usually published less than a month beforehand, hampering manufacturers' ability to prepare and provide available safety and effectiveness information. Panel recommendations and FDA-proposed classifications may therefore be based on incomplete information. Panel members were in some cases not provided guidance on device classification procedures until shortly before or during the panel meeting. The guidance was discussed during panel meetings by ODE staff with varying degrees of knowledge about applicable regulatory provisions.
A case in point is the attempted classification of muscle-monitoring devices. At its October 1994 meeting, the agency's Dental Products Panel unanimously recommended placing these devices in Class III with a high priority for requiring premarket approval. Following industry complaints of inadequate notice of the devices to be included under muscle-monitoring devices, CDRH announced at the December 1994 panel meeting that these devices would be reviewed at the April 1995 meeting. This did not occur. In a July 1995 Federal Register notice, FDA announced that this review would be conducted at a future meeting, but not the August 1995 one.10 This notice included the statement, "Because of substantive and procedural issues in connection with the October 1994 meeting, including the scope of products included and concerns that all interested parties may not have received adequate notice of the devices to be discussed at the meeting, FDA does not intend to rely on the Panel's discussion or recommendations for the use or classification of these devices."
This FDA statement, however, does not fully capture all the problems associated with this classification review. FDA and its Dental Products Panel reviewed muscle-monitoring devices before completing the classification review of unclassified dental devices initially considered by CDRH to be Class III candidates. Muscle-monitoring devices were not among this group. According to trade press accounts, FDA initiated an internal investigation of January 1995 allegations by two manufacturers of misconduct on the part of FDA personnel and panel members concerning the panel's recommendation that the firms' devices be placed into Class III. One firm additionally alleged that it had been the target of FDA retaliation. Dissatisfied with FDA's ability or willingness to police itself, the House Committee on Energy and Commerce's Subcommittee on Oversight and Investigations at a December 1995 hearing chaired by Congressman Joe Barton (RTX) directed FDA to involve the Office of the Inspector General in the investigation of these allegations.
CDRH empowerment of the ODE divisions in recent years to conduct panel meetings with very limited oversight may have been politically correct in a "reinventing government" management environment. The experience to date, however, warrants greater CDRH senior management involvement and earlier communication with industry to ensure that classification reviews at these meetings are productive. Hopefully the experience with muscle-monitoring devices has served as the needed stimulus for greater oversight of this activity.
FDA USE OF SPECIAL CONTROLS
Despite the absence of a formal agency policy statement or guidance, soon after the promulgation of SMDA in late 1990 CDRH began to incorporate the provision for special controls in its review and approval of 510(k) submissions and reclassification decisions. As earlier indicated, special controls include: guidelines (including guidelines for the submission of clinical data in 510(k) submissions), postmarket surveillance, patient registries, performance standards, and other appropriate actions. Unlike the pre-SMDA statutory definition of a Class II device, the SMDA revision under section 513(a)(1)(B) of the FD&C Act does not specify that the performance standards are to be established as mandatory standards under section 514. This occurred despite other SMDA provisions intended to streamline the process for establishing a mandatory performance standard under section 514. FDA interpreted this deletion of a reference to section 514 as permitting it to require information in 510(k) submissions demonstrating that the proposed device conforms to applicable voluntary standards. Unlike provisions in the PMA procedural regulations, FDA usually will not provide for, or accept, a 510(k) applicant's explanation for a deviation from a voluntary standard. In terms of demonstrating conformance to a performance standard, the 510(k) procedural regulations continue to limit this to standards established under section 514.
When confronted with these inconsistencies, CDRH responds that it has FDA Office of General Counsel concurrence that reference to voluntary standards in publicly available CDRH general and device-specific guidances has the effect of a regulatory requirement. Most CDRH guidances have not attained the status of guidelines because they are not included in the public file of guidelines maintained by the FDA Dockets Management Branch. Section 10.90 of the FD&C Act expressly provides that guidelines state procedures or standards of general applicability that are not legal requirements but are acceptable to FDA for a subject matter that falls within the laws administered by it.
Application of one or, if necessary, more special controls should be limited to a Class II device or a Class III device not presently subject to premarket approval. They have, however, been imposed in the 510(k) submission review of some Class I devices. By far the most commonly employed special control involves the submission of biocompatibility, electromagnetic interference, electrical shock potential, or sterilization validation information following test procedures in FDA guidances or voluntary standards. Requiring conformance with a performance standard technically cannot be considered a general control and applied to Class I devices.
CDRH continues to maintain that, compared to pre-SMDA 510(k) clearance requirements, there has not been a significant increase in the incidence of 510(k)s for which supporting clinical data have been required. Postmarket surveillance has been imposed on a relatively small number of devices. Most likely as a result of its unproductive and costly experience with the cardiac pacemaker registry established under 21 CFR 805 in 1987, FDA has not required patient registries for other devices.
The catchall special control, designated as other appropriate actions deemed necessary, has been used to require FDA-specified labeling information. This is in lieu of rule making to otherwise require device-specific labeling information similar to that under 21 CFR 801.420 for hearing-aid devices and 21 CFR 801.405 for over-the-counter denture repair kits and denture reliners, pads, and cushions.
FDA-specified labeling information was required as a condition for the 1993 reclassification of argon lasers from Class III to Class II for use in rhinological and laryngological surgical procedures. Although not apparent in the revised classification regulation for this device (21 CFR 874.4490), 510(k) clearance for these indications is only possible when the labeling includes specific warnings and lists the specific surgical procedures permitted by FDA. Tonsillectomy was not among the initially permitted procedures because the reclassification petition lacked sufficient supporting information. This and other applicable surgical procedures, however, can be added to the labeling when the 510(k) submission includes adequate supporting information. Unfortunately, these labeling requirements have yet to be included in an FDA guidance accessible to industry on the CDRH Facts-on-Demand automated system.
RECLASSIFICATION OF CLASS II DEVICES
As previously mentioned, FDA recently published a final rule reclassifying 111 Class II devices into Class I and exempting them from the 510(k) submission requirements.2 It was indicated that new information supports the FDA conclusion that general controls are adequate to provide reasonable assurance of the safety and effectiveness of these devices. This new information was based upon a CDRH-weighted risk assessment that used six "evaluation factors": effectiveness; health benefit; and frequency of use, mortality, serious injury, and less-serious injury. In support of the 510(k) exemption, FDA stated that these devices do not have a significant history of false or misleading claims or of risks associated with their inherent characteristics, such as device design or materials.
For many of the reclassified and 510(k)-exempt devices, FDA took liberties with statutory and regulatory provisions and with its own marketing clearance history for the device. It is incorrect for FDA to assert that any risks associated with all of these devices have been solely addressed by general controls. FDA neglected to indicate that CDRH applied special controls to many of these devices for marketing clearance purposes. CDRH required that their 510(k) submissions include biocompatibility, electromagnetic interference, electrical shock potential, or sterilization validation information using tests in voluntary standards or CDRH guidances.
The proposed rule partially cited section 513(a)(1)(A)(ii) of the act to provide further justification for the reclassifications. This statutory provision was seldom employed before SMDA. It applies to a device that is not life-supporting or life-sustaining, is not of substantial importance in preventing impairment of human health, and does not present a potential risk of illness or injury. It permitted FDA to classify such a device into Class I rather than Class III when insufficient information exists both to determine that general controls will provide reasonable assurance of safety and effectiveness and to establish a performance standard under section 514 that will provide this assurance. The SMDA substitution of special controls for performance standards throughout section 513 essentially nullified this provision. FDA would be hard pressed to justify placing a device in other than Class III when there is insufficient information to establish special controls (e.g., nonclinical and clinical studies, postmarket surveillance, or device-specific labeling) to provide assurance of safety and effectiveness needed beyond that provided by general controls. Neither the past nor present wording of this section 513 provision would permit a Class II device to be reclassified into Class I.
CDRH classification and reclassification guidances developed in 1991 and used by FDA and its advisory panels, and made available to some reclassification petitioners, cite the nullifying effect of SMDA on this section 513 provision. Nevertheless, FDA revised the regulatory definition of a Class II device under 21 CFR 860.3(c)(2) in a December 1992 Federal Register notice to reflect the SMDA revision. In using this section 513 provision to support the reclassifications of these 111 devices, FDA omitted the clause that there also be a lack of sufficient information to establish special controls to provide the additionally needed assurance of safety and effectiveness. The omission of this material fact allowed FDA to conclude that its experience with these devices now provides sufficient information that general controls alone are adequate to provide reasonable assurance of safety and effectiveness.