Legislative and Executive Branch History of This Report
House Report 104-659 accompanying the Labor, Health and Human Services, and Education and Related Agencies Appropriations Bill of 1997 expressed concern ''with the fragmentation of research on the safety of silicone breast implants and the relationship, or lack thereof, between silicone gel breast implants and connective tissue disease, classic auto-immune symptoms and other serious diseases." The Appropriations Committee believed an independent study was needed and instructed the Department of Health and Human Services (DHHS) "to enter into a contract with the Institute of Medicine of the National Academy of Sciences to conduct a general review of past and ongoing research on silicone breast implants." Departmental involvement in this controversial subject stretched back over more than a decade and involved a significant number of organizational units. Prominent among these were the Food and Drug Administration (FDA), the Centers for Disease Control and Prevention (CDC), and several institutes of the National Institutes of Health (NIH). Important meetings of the FDA General and Plastic Surgery Devices Panel, other meetings and workshops under the auspices of NIH, and studies and notices from CDC had informed and advanced the regulation and science of silicone breast implants. Nevertheless, the regulatory process had not come to a satisfactory conclusion nor had the many scientific questions been resolved. Although this Institute of Medicine
(IOM) report is not about the legal issues; all involved are aware that a vigorous struggle has been taking place in American courts. This struggle, involving major U.S. corporations and hundreds of thousands of breast implant patients with billions of dollars at stake, could not fail to be an important influence on silicone breast implant-related activities and how they were perceived by Congress.
Early History of the Institute of Medicine Study
Within this context, DHHS approached the IOM in the summer of 1997 to discuss a study of the scientific issues. The National Institute of Arthritis and Musculoskeletal and Skin Disease (NIAMS) was designated the lead agency, although funds were committed from a number of other departmental sources, including the Office of Women's Health at the FDA, the Office of Research on Women's Health at NIH, the National Cancer Institute, the National Institute of Allergy and Infectious Diseases, the Office of Women's Health at the CDC, and the Office of the Assistant Secretary for Planning and Evaluation and Office of Public Health and Science at DHHS. The original proposal contemplated a study of modest funding which would begin in October 1997 and would deliver a pre-publication report to the sponsor in November 1998. As the magnitude of this project became clear to the IOM, its Committee on the Safety of Silicone Breast Implants and the sponsors, the proposal was later expanded to roughly double the financial support and to extend the delivery date to July 1999. At no time did the IOM receive other than DHHS funds in support of this study.
The scope of this study was defined in the contract with NIAMS and further refined in an address by NIAMS Director Stephen I. Katz to the committee on the occasion of its first meeting in April 1998. Important contractual items included a scientific workshop, held on July 22, 1998, which provided the committee with information on a number of scientific matters from researchers in the federal government and academia and offered former and current major silicone breast implant manufacturersDow Corning, Mentor, and McGhanan opportunity to present data from clinical observational and other research studies. Also included was a public meeting held on July 24, 1998, at which more than 60 women formerly or currently with silicone breast implants, professional association representatives, scientists of all perspectives, consumer group representatives, and others addressed the committee. In addition, the committee was asked to comment on other important reports that became available during the time of its deliberations, specifically those of the United Kingdom's Independent Review Group and the multidistrict-litigation (MDL) National Science Panel.
This report is a general review of past and ongoing research on silicone breast implants, including (1) what associations between breast implants and various diseases, if any, are suggested by existing experimental, clinical, and epidemiological studies; (2) the nature and relative strength of these associations, (3) the quality of the studies; (4) the existence of plausible biological mechanisms to explain suggested associations between breast implants and disease; (5) the uncertainties associated with these kinds of analyses; and (6) suggestions for future research to fill in critical gaps. This report covers breast implants with a silicone shell, whether smooth, modified, or coated, and a gel or saline filler; local and systemic complications and manifestations of disease; and effects on mammographic screening and diagnosis and on the offspring of patients with breasts implants. In summarizing the purpose of the study at the committee's first meeting, Dr. Katz noted that it should be "a comprehensive survey and rigorous assessment of the scientific literature and scientific works related to the biological and health-related effects of silicone breast implants and their components. While the emphasis is on safety, we hope that areas of gaps in knowledge as well as scientific opportunity for future research will be identified." Such a report would serve as a resource for scientists and an educational tool for the public. It would also serve NIH and other components of DHHS as they develop policy and plan future research and prevention activities.
The IOM Committee On the Safety of Silicone Breast Implants
The Committee on the Safety of Silicone Breast Implants was assembled by the IOM to include expertise in preventive and internal medicine, nursing, family and women's health, rheumatology, clinical and basic research, epidemiology, immunology, neurology, silicone chemistry, toxicology, breast and other cancer, plastic surgery, and radiology or mammography. Committee members are listed at the front of this report. Because of the often polarized and controversial nature of the many issues involved, the IOM took active steps to avoid conflict of interest in constituting the committee in accordance with the Institute's rigorous bias and conflicts of interest procedures. Given the importance of the subject matter to women's health, it was desirable that women make up a substantial portion of the committee. Six of its thirteen members, including the vice chair, are women, and most of them are particularly interested in women's health issues. The committee's medical experts in breast cancer, rheumatology, radiology, and family medicine have, in the course of their practice, provided medical care to women who may have had breast implants in addition to their cancer, rheumatologic problems, mam-
mography, or general health care needs. These experiences were deemed valuable. The committee's plastic surgery expert, on the other hand, does not perform breast implant surgery in her practice of pediatric plastic surgery.
In addition to the first committee meeting in April and the scientific workshop and public meeting (summarized in the appendixes), the committee had five days in three separate meetings to discuss and review the issues together over the course of the 15 months between its selection and the delivery of this report. During this time, with help from IOM staff and an immunological and toxicological consultant, members of the committee spent many additional days, individually and as a group, reviewing the studies listed in the references, discussing and evaluating the evidence, and preparing this report. The committee also accepted all offers of information that could be made public and received material and assistance from a number of individuals, companies, attorneys, associations, and organizations. The committee acknowledges this assistance with gratitude.
Collection and Evaluation of Evidence
Since anecdote and isolated personal experience cannot, in themselves, be subject to scientific review (although they can provide a basis for the design of scientific studies), the committee decided to focus on evidence reported in the peer-reviewed, published scientific literature. A great deal of information was also provided by breast implant patients, manufacturers, involved scientists, and others, in the form of industry technical reports, prepublications, medical histories, and private or personal submissions of various kinds. The IOM, as a nongovernment entity carrying out an independent review, is generally not in a position to accept and preserve confidential business or personal information; the information presented to the committee is made available to the public. The committee recognized the selectivity involved, but was open to the receipt of such material nevertheless. Many of these items are included in the reference list and comprise a useful resource. However, the committee did not consider them of equal weight to the peer-reviewed literature, and an attempt has made to divide the reference list into two sections to recognize peer reviewed scientific literature and other informational material. About 80% of the almost 1,200 references cited in the text of this report are from the first, peer-reviewed list. The references in the second list were considered useful but clearly less so than those in the first or primary list.
Because this project involved the assessment of a large number of scientific studies, the committee reviewed ways in which these reports
Austin Bradford Hill Criteria for
1. Strength of the association between an exposure and illness (e.g., relative risk)
2. Consistency of the association observed by different people, under different circumstances, in different ways
3. Specificity of the association: it uniquely affects certain people with certain illnesses
4. Temporality of the association: the exposure precedes the illness
5. Biological gradient: the frequency and seventy of the illness are directly related to the frequency and severity of the exposure
6. Plausibility of the association in terms of the biological reasonableness
7. Coherence of the association with other knowledge of the exposure and the illness
8. Experimentation: manipulation of the exposure affects the illness
9. Analogy: the association is analogous to another known associations
and their data might be evaluated and ranked. When appropriate in reviewing epidemiological studies, the grading of evidence suggested by the U.S. Preventive Services Task Forcefrom well-designed, properly randomized clinical trials; to nonrandomized controlled trials; to well-designed cohort or case control studies; to multiple time series with or without the intervention; to expert opinionwas considered reasonable if an appropriate epidemiological set was being reviewed. Some suggestions from the Agency for Health Care Policy and Research (AHCPR)for example, giving most weight to conclusions supported by multiple relevant and high-quality scientific studies and lesser weight to those based on fewer and lower-quality studies, were considered. Sir Austin Bradford Hill proposed criteria for evaluating relationships between sickness and environmental conditions (see Box 1-1) that were also useful (Hill, 1965).
The committee also considered oral and written communications that recorded many women's individual medical histories and experiences with breast implants. Although individual reports of personal experiences cannot be confirmed and are anecdotal and therefore not scientifically definitive, this human element, which is after all a part of medical care and medical science, provided an important context to the committee's deliberations.
A review of the reference lists at the end of this report indicates that much of the literature available to and considered relevant by the committee, did not consist of full, peer-reviewed, published scientific reports but was in the form of letters, position papers, abstracts, industry reports,
and opinion pieces; these are found in the second of the two reference lists. Furthermore, review of the first reference list indicates that some of this material, although in peer-reviewed scientific publications, consisted of case reports, reviews, and other forms that the committee did not consider strong evidence. Case reports or case series reports are often essentially anecdotes or uncontrolled observational studies, which, lacking appropriate comparison or control groups, may not be helpful in determining rates of occurrence or accepting or rejecting causation. Controlled observational studiesthat is, studies based on a cohort or case control designoffer stronger evidence because they provide information on relative risks or rates. Some controlled clinical trials were found by the committee, and this very strong kind of evidence was useful when available. In all of these studies, the committee was aware of the importance of randomization, that is, the assembly of unbiased, representative study groups (Stratton et al., 1994).
In summary, the committee found it useful to consider the varying strength and quality of the evidence. Although some reports had design limitations or problems in implementation, other excellent reports were available, and the committee has tried to point out the strengths and weakness of various studies in this report. Overall, the committee was impressed with the substantial amount of useful information and, in particular, with the marked improvement in the 1990s in the quantity and quality of studies relevant to its charge. A number of strong, well-de-signed epidemiological studies involving large numbers of women and clear results have been published since the early 1990s. These have allowed the committee to take firm conclusions on many of the important issues. To communicate this firmness, the committee adopted a ranking of evidence or data as: conclusive or convincing of an association or of no association; limited or suggestive of an association or of no association; or, in the committee's judgment, insufficient, flawed in terms of methodology, or lacking in support of the various evaluations of the safety of silicone breast implants. The modifiers in the last group were used to indicate that when evidence for an association or for no association was not conclusive or suggestive, it could fail to support a conclusion because it was: a.) insufficient in quality or quantity; b.) not perceived to present helpful evidence because the committee noted methodological flaws; or c.) simply lacking or absent. On occasion, evidence failed to support a proposed conclusion of an association because it supported a conclusion of no association; these circumstances were expressed more explicitly by simply noting that the evidence supported some other conclusion, not the proposed conclusion. These rankings allowed the committee to take many firm decisions. They are not meant to imply, as some might, that the committee means to support further research or studies, for example,
when data are insufficient. If more information and analysis would be desirable, and only in such instances, the committee has indicated this in the text of this report.
Cosmetic Breast Surgery and the Historical Context of This Report
Perspectives of Women with Breast Implants
Although estimates of the current number of U.S. women who have had breast implants vary, it is probably somewhere between 1.5 million and 2 million, as discussed later in this chapter. Many of these women have reported symptoms that they believe are associated with their breast implants. The symptoms reported involve many organ systems and bodily functions and are often associated with a compromised quality of life. The concerns of these women and their loved ones were a major factor leading to this IOM review. Women with breast implants as well as their spouses and children spoke to the committee. They advised that this is more than "a women's issue, but one that affects husbands, families, taxpayers and society." Some related how they were "young and vibrant" and are now ill and disabled. They reported feeling scorned and patronized or ignored as they searched for medical solutions. In many instances they did not think they had been informed of known complications. They were frequently told that the implants ''would last a lifetime," but instead had experienced complications and repeated surgeries with associated pain and expense (see Appendix B).
Case reports from individual women can be used to generate hypotheses about disease and symptoms that may be associated with silicone breast implants. Whether such symptoms are causally associated with breast implants will be determined by comparing the occurrence of symptoms in women with implants and without implants and meeting other conditions for determining causality. Nevertheless, it was the burden of the personal vignettes that prompted the IOM to invite women with implants, as well as health care providers, scientists, and other interested parties, to make formal presentations to the committee and to submit written comments for the committee's deliberations (see appendixes). The committee heard from individual women about their experiences with both breast augmentation and breast reconstruction, and from representatives of many patient groups from throughout the United States. Their written statements were distributed to all committee members and also made available to the public.
Committee members were impressed by the thoughtful and detailed comments provided by women with implants, not only for the sake of
their own health but also out of concern for other women who are considering breast implants. Many women support the continued availability of breast implants for those who want them for either augmentation or reconstruction, with the proviso that adequate information regarding potential complications and long-term safety is available to them for informed decision making. The commitment of these women to the scientific process and to partnering with scientists in future studies was clear and very much appreciated by committee members. Although the committee's mandate focuses primarily on peer-reviewed scientific evidence, the contributions of women with breast implants provided an extremely valuable context for its deliberations.
History of Cosmetic Breast Surgery
To varying degrees, people are concerned about the appearance of their bodies, and cosmetic surgery is a response to such concern. In a recent discussion of this, Sarwer et al. (1998a) noted that 34% of U.S. women were dissatisfied with their breasts, and more than half of breast augmentation patients reported having frequently checked the appearance of their breasts and camouflaging them. The literature also describes surgical interventions and the results of efforts to change natural breast dimensions over at least the past hundred years. Augmentation with modern silicone implants can variably affect the shape and size of the female breast; on average, an increase of two brassiere sizes (cup or cup and chest circumference; circumference increases in 1 inch or 2.5 cm increments) is achieved (Young et al., 1994).,
Reconstruction after mastectomy for cancer, fibrocystic disease, or other reasons such as prevention in women at high risk for breast cancer is believed to provide a sense of having overcome disease (Bard and Sutherland, 1955). Especially when performed soon after mastectomy, breast reconstruction is reported to relieve or prevent a perception of loss, dissatisfaction with an external prosthesis, depression, and feelings of diminished sexual attractiveness (Schain et al., 1985; Stevens et al., 1984).
These powerful motivations may explain the continuing acceptance by many women and health professionals of evolving cosmetic surgical breast procedures despite complications, high incidence of hardness, and often globular or otherwise less than natural-looking breasts.
The modem history of cosmetic breast surgery began in the late 1800s. It has involved the use of both autogenous tissue (which is not the subject of this report) and alloplastic implants, culminating in the variously filled
silicone shell implants for augmentation and reconstruction, which constitute the subject of this report. The use of autogenous tissue is reported to have begun in 1887 with part of a healthy breast transferred on a pedicle to reconstruct the other breast (Verneuil, 1887) and continued in 1895 with the transplantation of a lipoma from the hip to repair a surgical defect in the breast (Czerny, 1895). A pectoral muscle flap for immediate reconstruction and a latissimus dorsi flap were described shortly thereafter (Ombredanne, 1906; Tansini, 1906). Since transfer of fat alone is usually unsuccessful (Hinderer and Escalona, 1990) because it is substantially reabsorbed unless injected in quite small quantities (Bircoll and Novack, 1987), efforts continued with pedicle or dermis flaps or dermis-fat-fascia grafts (Bames, 1950, 1953; Berson, 1944; Watson, 1959). These early efforts were reviewed by Watson (1976) who described the tendency to long-term shrinkage of the dermis-fat grafts. Continued development culminated in musculocutaneous flaps, primarily from the abdomen (transverse rectus abdominis musculocutaneous [TRAM] flaps) but also from other sites (e.g., the latissimus dorsi or superior and inferior gluteus muscles, among others), and microsurgical free flaps. These events were reviewed in Kincaid (1984) in a chronological listing of surgical advances with citations to the literature and in 1995 by Wickman. About one-third of modern breast reconstructions are performed using autogenous tissue. Such reconstructions are performed more and more frequently and are often combined with implants (ASPRS, 1996, 1997; Trabulsy et al., 1994).
Also since the late 1800s, foreign substances have been injected or implanted to augment or reconstruct the breast, although sporadic efforts of this kind apparently date back centuries. Gersuny reported experimentation with paraffin injections beginning in 1889 (Gersuny, 1900). Although paraffin enjoyed some early acceptance, others later described disastrous results such as fistulas, granulomas, pulmonary emboli, and tissue necrosis (Letterman and Schurter, 1978). Subsequently, in the early to mid-1900s, a number of other substances were tried, including ivory, glass balls, ground rubber, ox cartilage, Terylene wool, gutta percha, Dicora, polyethylene chips, polyvinyl alcohol-formaldehyde polymer sponge (Ivalon), Ivalon in a polyethylene sac, polyether foam sponge (Etheron), polyethylene tape (Polystan) or strips wound into a ball, polyester (polyurethane foam sponge) Silastic rubber, and teflon-silicone prostheses (Broadbent and Woolf, 1967; Brown et al., 1960a,b; Edgerton et al., 1961; Edwards, 1963; Letterman and Schurter, 1978; Lewis, 1965; Lilla and Vistnes, 1976; Liu and Truong, 1996; Smahel et al., 1977). These early implants were unsuccessful and were not pursued seriously. The later
efforts with so-called open-pore polymer sponge implants such as Ivalon or Etheron led to hard, unnatural-looking breasts and other complications (Broadbent and Woolf, 1967). Thousands of women had these implants, which made up about a third of implantations in the 1960s according to one large international survey (De Cholnoky, 1970).
During the decades after the Second World War an array of liquid substances were injected, often illegally by unlicensed practitioners, to augment the breast (and other sites; see e.g., Christ and Askew, 1982). These are cited in reports from Japan, the Far East, and domestic plastic surgeons. They include paraffin; other poorly defined, more radiolucent hydrocarbons called "Organogen" and "Bioplaxm" (Yamazaki et al., 1977) and some forms of petroleum jelly such as Vaseline (Ohtake et al., 1989). Adulterated silicone oil (e.g., the "Sakurai'' formula) was also commonly used. It was believed to have been adulterated with 1% ricinoleic acid (Pearl et al., 1978); 1% animal and vegetable fatty acids (Kagan, 1963); or 1% mineral and vegetable (perhaps castor) oil (Chaplin, 1969), 1% olive oil (Tinkler et al., 1993), or to contain croton oil, peanut oil, concentrated vitamin D, snake venom, talc and paraffins (Kopf, 1966; Rapaport et al., 1996). Use of a variety of unknown oils has been reported, some with silicone of mostly unknown origins (Ortiz-Monasterio et al., 1972), as well as beeswax, shellac, glaziers' putty, epoxy resin (Symmers, 1968), and industrial silicone fluids. Medical-grade silicone fluid, including Dow Corning 200, 350, and MDX 44011 (Ashley et al., 1967); silicone (Elicon-Kogen Kogyo Co. Ltd.) gel (Boo-Chai, 1969); and silicone Silastic S-5392 RTV (room temperature vulcanized) fluid with a stannous octoate catalyst to form a silicone rubber within the breast tissue were also used for breast augmentation and other plastic surgical purposes with positive early reports (Ashley et al., 1967; Conway and Goulian, 1963; Freeman et al., 1966; Harris, 1965). Substantial amounts of these substances were injected on occasion, as much as 2 liters for breast augmentation and body contouring in a single patient (Kagan, 1963) or half a liter of fluid and catalyst per breast (Conway and Goulian, 1963).
Results with other than medical-grade silicone were poor, and included loss of both breasts and death (see references below). Dow Corning Medical Grade 360 fluid, which according to Vinnik (1991) was used extensively in Las Vegas, also resulted in complications, but the silicone mentioned in many reports was undoubtedly adulterated in the misguided hope that an adulterant would inhibit fluid migration and give better results. Sakurai personally reported 72,648 cases injected, including the breast among other sites (Kagan, 1963). At least 12,000 women (some
have estimated as many as 40,000 women) had breast injections in Las Vegas by 1976 when the practice became a felony under Nevada State law. Practitioners reportedly charged $800 to $2,000 for a series of injections in 1966 (Kopf, 1966; Kopf et al., 1976; Vinnik, 1991), and domestic reports continue from there and elsewhere (Leibman and Sybers, 1994; Morgenstern et al., 1985; Sánchez-Guerrero et al., 1995a,b).
Soft-tissue (excluding breast) augmentation by injection with medical-grade silicone was approved for experimental use in the United States under an FDA investigational new drug (IND) exemption to Dow Corning for use by six, and later seven, plastic surgeons and one dermatologist for about a decade, 1965-1975 (Ashley et al., 1967; Braley, 1971; Wustrack and Zarem, 1979). Although this material broke up into "innumerable droplets" on injection and tended to spread out from the injection site, results appeared promising at first (Ashley et al., 1965; Rees et al., 1970). After a few to as many as 28 years, however, problems began to appear (Rapaport et al., 1996). As with paraffin earlier, initial enthusiasm was tempered by the appearance of complications. The question of systemic effects is discussed in Chapter 8 of this report. Among the more clear-cut effects associated with silicone breast injection (as noted, not an FDA-approved use as part of the Dow Corning IND) have been pain, skin discoloration, edema, ulceration and necrosis, calcification, granulomas, migration of the fluid, infection, cysts, axillary adenopathy, disfigurement and loss of the breast, liver granulomas and dysfunction, acute pneumonitis or adult respiratory distress syndrome, pulmonary embolism, coma, and death (Baker, 1992; Boo-Chai, 1969; Brozena et al., 1988; Celli et al., 1978; Chastre et al., 1983a, 1987; Chen, 1995; Chen et al., 1993; Cruz et al., 1985; Edgerton and Wells, 1976; Ellenbogen and Rubin, 1975; Inoue et al., 1983; Ko et al., 1995; Koide and Katayama, 1979; Lai et al., 1994; McCurdy and Solomons, 1977; Parsons and Thering, 1977; Perry et al., 1985; Piechotta, 1979; Rodriguez et al., 1989; Solomons and Jones, 1975; Symmers, 1968; Truong et al., 1988; Vinnik, 1978; Winer et al., 1964). Some of these complications occurred instantly, such as acute pneumonitis with findings of substantial deposition of silicone in the lung, which was probably due to pulmonary embolism from inadvertent intravenous injection or other circulatory access of the silicone fluid (Solomons and Jones, 1975). More often, complications were noticed after a few years. Various reports cited a complication rate of 1% of patients per year (Kopf et al., 1976), a prevalence of 50% at five years after injection (Vinnik, 1976a,b), the onset of disturbing problems at two or three to five or six years (Ohtake et al., 1989; Wustrack and Zarem, 1979), or the average occurrence of problems at about nine years (Parsons and Thering, 1977). Presumably because medical-grade silicone is a "mild irritant" as opposed to adulterated silicone or other substances, which are irritating to a greater extent, compli-
cations were noted in one study only after many years and differed according to individual patient reactions (Rapaport et al., 1996). Other surgeons reported complications after considerable delay, and these tended to be relentlessly progressive (Wustrack and Zarem, 1979). One author noted that "trying to provide an accurate timetable for these changes has proved futile" (Vinnik, 1978).
There are only anecdotal reports of breast malignancy in silicone-injected breasts (Ko et al., 1995; Kobayashi et al., 1988; Lewis, 1980; Maddox et al., 1993; Morgenstern et al., 1985; Okubo et al., 1992; Ortiz-Monasterio and Trigos, 1972; Pennisi, 1984; Smith et al., 1999; Suster et al., 1987; Talmor et al., 1995; Timberlake and Looney, 1986; see also Chapter 9 of this report). Although no epidemiological studies of the incidence of cancer in women with silicone-injected breasts have been reported, and thus there is no evidence of an elevated relative risk, case control studies of the frequency of breast implants in women with breast cancer indicate, if anything, a decreased odds ratio (Brinton et al., 1996; Glasser et al., 1989). Injected silicone clearly may handicap the diagnosis of breast cancer. Injected breasts are full of lumpy, radiopaque deposits of silicone that interfere with breast self-examination, physical examination, and mammography. Better visualization with magnetic resonance imaging (MRI) is helpful, but does not resolve this problem because MRI is not considered a screening technology and is used only when there is an indication for this more resource intensive modality (Helbich et al., 1997; Leibman and Sybers, 1994; Lewis, 1980; Maddox et al., 1993; Morgenstern et al., 1985; Okubo et al., 1992; Talmor et al., 1995; Timberlake and Looney, 1986; see Chapter 12). Although silicone injection for breast augmentation (or any cosmetic use) is not approved by the FDA, there are a number of reports advocating medical-grade silicone injection in other sites such as the face, using careful technique and small amountsfrom a fraction of a milliliter to a few milliliters per treatment depending on location (Ashley et al., 1973; Hinderer and Escalona, 1990; Rees and Ashley, 1966; Rees et al., 1973a; Selmanowitz and Orentreich, 1977). Duffy (1990) reviewed thousands of such cases and himself reported more than 2,000 injections of 350-centistoke silicone fluid with good results, although his follow-up was limited to six years. The American Academy of Cosmetic Surgery (AACS) reported 7,170 women receiving such non-breast silicone injections from its members in 1994. Nevertheless, "FDA has not approved the marketing of liquid silicone for injection for any cosmetic purpose, including the treatment of facial defects or wrinkles..." (FDA, 1991). The history of silicone injection is relevant to the safety of silicone breast implants because of the possible analogy to silicone gel fluid diffusion through implant shells into breast tissue or the deposition of silicone gel (and gel fluid) in the breast on rupture of gel-filled implants.
A successful silicone (urethral) implantation was reported in 1950 (De Nicola, 1950); subsequently silicone use for shunts and joints was proposed (Marzoni et al., 1959), and many such shunts, joints, and other devices were developed and used in medical practice. These clinical experiences, animal experiments, and early work on tissue reaction to silicone (Child et al., 1951; Kern et al., 1949; Rees and Ashley, 1966; Rowe et al., 1948; see also Chapter 4) provided a context and encouragement for the consideration of this technology in cosmetic breast surgery. The disappointing results with other technologies for breast augmentation or reconstruction provided an opportunity and an unmet need. The response to this need was the introduction of the silicone breast implant in 1962 and its continued development through the 1990s, as described in Chapter 3. The specific experimental basis for the clinical introduction of this device was the work reported by Cronin and Gerow (1963) on implantation of silicone shells (four to six per animal) filled with either dextran or electrolyte solution in each of 12 dogs for periods of a few days to 18 months without signs of toxicity or other complications.
Silicone Implants and Patient SatisfactionIts Importance and Effect on Demand
Although the silicone implant was considered an improvement, it was not problem-free. Some implant complications such as small areas of epidermolysis or necrosis and small, isolated seromas are minor, although they might require medical attention; other complications, infections, implant ruptures, and severe fibrous capsule contractures are of greater import. Complications are less frequent in augmentation than in reconstruction, but as noted in Chapter 5, they still occur with considerable frequency. The more serious risks of reconstruction with implants, especially immediate reconstruction (Spear and Majidian, 1998), must be balanced against the psychological benefits (Stevens et al., 1984; Wellisch et al., 1985). Implants for augmentation are placed in healthy women who would otherwise not be operated on and incur such risks. Placement of a breast implant for either augmentation or reconstruction does not treat the physical component of human disease or disordered physiology. It is, in this sense, elective and therefore warranted only if it is relatively safe and provides patient satisfaction.
Satisfaction has an effect on the demand for interventions to manage or relieve complications and thus on the implications of complications for the safety of silicone breast implants. The high overall level of satisfaction of women in medical reports, if accurate and lasting, implies a low level
of concern or at least a willingness to tolerate some complications. This may have an important moderating effect on the reported incidence of further operative or medical interventions. The committee believes that published reports of satisfaction may be misleading, however. Most surveys of satisfaction are carried out by plastic surgeons or others associated with the surgery or care of women with implants. This arguably introduces a possible bias or distortion of patients' responses.
Long-term satisfaction is an important test of the aesthetic results of cosmetic surgery. Many surveys were carried out immediately or shortly after surgery, and the degree of satisfaction may change with time and the occurrence of untoward events. Some surveys include small numbers of women and lack precise details on procedure, type of implant, timing, or other factors that might be important to results. Finally, almost all of these reports are based on women communicating an overall satisfaction level in general terms. Women may have specific problems or suffer from a particular dissatisfying aspect of the results of implantation; this may not be expressed clearly or at all in a report that focuses on satisfaction in a general way and not on problems.
When women with complications requiring secondary surgery such as deflation, contracture, and asymmetry (N = 58) from a large cohort (N = 292) surveyed by Strom et al. (1997) were compared with those who did not require additional surgery (N = 234), there was no significant difference in reporting of satisfaction on the Likert 1-5 scale. This may be because the end result was important enough to these women that they were prepared to put up with inconvenience and discomfort. Similarly, the Karolinska group (Sweden) has consistently found that, up to seven years after surgery, women were satisfied with, or tolerant of, severe (Class III or IV; see breast augmentation classification in Appendix D) implant capsular contractures. In 1989, this group found that 77% of women in its operative series were satisfied, although 79% had Class III or IV contractures. In 1990, the same group reported that 85% of women were satisfied, although 35% had severe contractures (Gylbert et al., 1989 and 1990a). In other instances, women may prefer a complication, such as Class III contractures with smooth shell implants, to the more ready detection by palpation of a textured implant even though such an implant produces a softer breast (Burkhardt and Demas, 1994). This could be a significant factor since 30% of submuscular textured saline and 47% of gel implants have been reported to be readily detected by palpation (Opitz and Young, 1998). In a study comparing women with submuscular and submammary (see definition in Appendix D) implants, patient satisfaction was not always proportional to the severity of contracture (Mahler and Hauben, 1981). Other surveys, however, have found substantial differences in satisfaction when women had significant contractures and
waning satisfaction over time due perhaps to the accumulation of complications (Beale et al., 1985; Fiala et al., 1993).
In the group of 100 women implanted by van Heerden et al. (1987) 85% would recommend implant reconstruction to other women, and 73% rated it 6-10 on a scale of 1-10 (32 women rated it a 10). However, this questionnaire was administered by the operating service during the post-operative period. Spear and Majidian (1998) asked patients to express their degree of satisfaction, and 98% of 42 consecutive women rated themselves somewhat to completely satisfied with their breast implants. Again, this rating was carried out by the operating team, presumably shortly after surgery. A survey by Francel et al. (1993) of 197 implant reconstruction patients, with a 50% response rate, found that 100% of women who had been reconstructed immediately would try it again and 90% of them were satisfied. Of women who had undergone delayed reconstruction with implants, 90% would try it again and 80% were satisfied. This is another example of a survey performed by the surgical group after an unspecified, but clearly short, postoperative interval.
In a questionnaire administered by the medical team to an unspecified subset of 216 women who underwent implant and autologous reconstructions, results of autologous tissue ranked significantly better (Eberlein et al., 1993). Better cosmetic results of autologous compared to alloplastic reconstruction were also reported in the small survey by Mansel et al. (1986). In 1991, a survey by the Van Nuys Breast Center (Handel et al., 1993a) of patients augmented with polyurethane and other gel implants over the previous 12 years generated only a 32% (85 out of 321 patients) response rate. Of these, 66% were satisfied before exposure to negative publicity about breast implants and 61% after such exposure. The authors of this report felt that satisfaction was related inversely to complication rates (Handel et al., 1993a). In another study of 174 women who had double-lumen implants for breast reconstruction, 68% were satisfied or completely satisfied when rating postoperatively. After the occurrence of additional complications several years later, satisfaction waned (Fee-Fulkerson et al., 1996). In 1990, a survey by the American Society of Plastic and Reconstructive Surgeons (ASPRS) of representative U.S. households and a group of women from a medical devices registry found 93% of 592 responding women satisfied with the results of breast surgery and 82% prepared to undergo it again (ASPRS, 1990; Iverson, 1990), Park et al. (1996a) reported 84% of augmented women satisfied to very satisfied, as were 91% of implant-reconstructed women in a survey carried out at least one to ten years postoperatively by plastic surgeons.
Very high satisfaction rates (88%-96%) were reported by Hetter (1979, 1991) from a questionnaire of 165 women (a response rate of 69% of the series) augmented with gel and saline implants. Of 100 women with sili-
cone implants for reconstruction who were interviewed by a plastic surgeon other than the operating surgeon, 96% reported that the reconstruction was important to them (Asplund and Körloff, 1984). In a multicenter survey of 504 women, 93.8% of whom had saline-filled implants for augmentation, Gutowski et al. (1997) reported a 94.2% satisfaction rate. In a survey of 292 women from a cohort who were augmented with saline implants and followed up at an average of seven years, Strom et al. (1997) reported that 80.2% were very satisfied (rating 1 or 2 on a 5 point Likert scale). A group of 20 women augmented with submammary silicone gel-filled implants and their husbands were asked about psychosocial and sexual benefits. They reported 70%-75% favorable results (Mahler and Hauben, 1981).
Many other reports describe significant psychosocial benefits and improvement in life functioning after augmentation or reconstruction (e.g., Beale et al., 1985; Corsten et al., 1992; Dean et al., 1983; Goin and Goin, 1981, 1982; Goldberg et al., 1984; Hetter, 1979, 1991; Jonsson et al., 1984; Reaby et al., 1994; Schain et al., 1984; Sihm et al., 1978). If putting in breast implants can confer psychosocial benefit, the loss of them may, as might be expected, be detrimental to overall body satisfaction and appearance-related cognition (Walden et al., 1997). On the other hand, the removal of implants in women who fear silicone-related health problems, although sometimes reported to be helpful, was also reported in other studies (carried out at a time of adverse publicity about implants) to do little for psychological distress and symptoms of somatization, depression, and anxiety (Roberts et al., 1997; Svahn et al., 1996; Wells et al., 1995, 1997).
National Perspectives and FDA Data on Satisfaction
Merkatz et al. (1993) and Brown et al. (1998) provide descriptions of national data on satisfaction and adverse implant events. The committee notes that all these data are based on self-reported or unconfirmed reports. These also are numerator data only, that is, raw numbers of events reported to have occurred in a population or group of unknown size, which depending on the time may to an unknown extent be distorted by publicity of changing focus and intensity. Some events are overreported due to manufacturer and plastic surgeon concerns for compliance with reporting requirements and organized letter writing campaigns, whereas in other instances, personal considerations or lack of systematic follow-up may cause underreporting of adverse events. These reports cover a distinct minority of the I million to 1.5 million U.S. women with breast implants in the early 1990s. Although they constitute the only national source, and provide expanded details on patient satisfaction from a some-
what different perspective than the plastic surgery reports of high levels of satisfaction, they do not establish an increased frequency of any disease in women with implants except for the direct complications of surgery.
In a qualitative analysis of national experience with women's self-reports to the FDA during January 1992, Merkatz et al. (1993) noted that a distinct minority of women were dissatisfied with their implants, reporting both local complications such as rupture and contracture and disabling general health conditions. A further analysis of FDA adverse events reporting cited 94,120 mandatory (i.e., required by law from manufacturers and importers) adverse events reported from the end of 1984 through 1995, with a peak of 83, 069 in 1992-94 and a marked decrease in 1995. The majority of these reports cited local and perioperative complications, particularly contracture and implant rupture, except in 1992 when general health complaints were temporarily more common. Reporting of deaths (N = 70) was so poorly characterized that a definite diagnosis could not be made except in a few cases of cancer (N = 8), specific connective tissue disease (N = 3), and operative mortality (N = 2). As noted earlier, the events reported tended to vary with the level of national publicity. The FDA received 4,303 voluntary reports. Some of these were from women with implants who were described by Merkatz et al. (1993). These peaked in 1992-1994 and decreased markedly in 1995 (Brown et al., 1998). Early reports from 1973 through January 15, 1992, from the FDA Product Problem Reporting System, which were voluntary reports from consumers and health professionals, numbered 379 and were concerned primarily with pain and contractures. FDA and National Cancer Institute (NCI) staff with some collaborators reported a follow-up in 1994 of 1,167 of these reports of local or systemic problems with implants with a survey requesting information on physician diagnoses. Of the original cohort, 820 completed the interview; 28% of these reported a physician diagnosis of a connective tissue disease and 43% reported multiple implant surgery. These self-reports were not independently verified, and this group was highly selected (Coleman et al., 1994).
Several conclusions seem to follow from these data on women's satisfaction or dissatisfaction with breast implants. Many women are satisfied with their implants, but it is not safe to assume that satisfaction is, and remains, at the levels often reported in the plastic surgery literature, given conflicting results and the possible biases and problems of some of these reports. Satisfaction is often surveyed by the operating service and shortly after implantation when some complications have yet to occur. Reports of satisfaction do not necessarily mean that complications have not occurred or that they are not troubling to many women. Women seem to accept trade offs, as noted earlier, and some are very dissatisfied. Some who are dissatisfied report obstacles to paying for implant removal, whereas oth-
ers elect to tolerate distress for the perceived benefits of the implants. The improvements in the results of implantation, noted in the short history in this chapter and suggested later in this report, and in women's appreciation of these results have sustained the demand for breast implants, as the prevalence data below suggest. These complex and varied levels of satisfaction also appear to affect the demand for medical or surgical interventions to address complications. Since these interventions carry risks they are relevant to the safety of silicone breast implants (see Chapter 5).
As noted earlier, the Food and Drug Administration exercised jurisdiction over silicone for injection under investigational new drug provisions of the Food, Drug and Cosmetic Act. However, the FDA did not have a statutory basis for oversight of silicone breast (or other) implants until the enactment in 1976 of the Medical Devices Amendments to the Food, Drug and Cosmetics Act (Public Law 94-295). At the time, on the advice of its independent General and Plastic Surgery Devices Panel, the FDA placed implants in a category requiring general controls and performance standards. No testing or applications for marketing were required. As time passed, more and more women received silicone implants, more than 90% of which were silicone gel-filled, for augmentation. In the 1980s, more women wanted, and more surgeons were willing to use, implants for reconstruction after cancer surgery (Freeman and Wiemer, 1979; Georgiade et al., 1982, 1985; Gilliland et al., 1983; Hartwell et al., 1976; Hueston and McKenzie, 1970; Noone et al., 1985). With increasing experience, more reports accumulated in the plastic surgery literature of implant rupture, silicone gel fluid diffusion through the implant shell, and severe contracture of the fibrous capsule surrounding the implant (see Chapter 5). The changing characteristics of implants, thinner shellsand more compliant gels in the 1970s and early 1980s (so-called second generation implants; see Chapter 3), contributed to some of this increased frequency of complications. Concerns surfaced that silicones might be associated with cancer, and reports of connective tissue diseases and less well defined systemic complications, perhaps of an immune nature, in women with silicone injections and implants began to appear (see Chapters 6, 7, and 8). As a result, in 1982, the FDA proposed and, on June 24, 1988, formally implemented a classification of silicone breast implants in a category (Class III) requiring stringent safety and effectiveness controls. In a critical series of actions from 1989 through 1991, the FDA required pre-market approval (PMA) applications from implant manufacturers (April 10, 1991), questioned specifically the safety of polyurethane-coated implants and required additional study of them, determined that PMA
applications for silicone gel breast implants were insufficient when they were submitted by the manufacturers (August 22, 1991), and required dissemination of information on implant risks to patients (September 26, 1991). In 1992 and 1993, citing the absence of data on safety and effectiveness (Kessler, 1992), the FDA restricted the use of silicone gel-filled implants to participants in a clinical observational study, most of whom received implants for reconstruction (January 6, 1992); called for safety and effectiveness data on saline-filled silicone implants (April 16, 1992); issued a proposed rule to require PMA applications for saline implants (January 8, 1993); and designated gel and saline implants subject to device tracking rules (August 26, 1993). By this time, all U.S. companies (as noted in Chapter 3 of this report) except Mentor Corporation and McGhan Medical Corporation had withdrawn from the market (March 20, 1992), and polyurethane-coated implants were no longer in domestic production. At the time of issuance of this report, all companies who wished to remain in the U.S. silicone breast implant market will have attempted to carry out the observational and other studies required by the FDA as specified in 1996 and will have submitted PMA applications (FDA, 1998). Whatever the regulatory outcome and the real or hypothetical problems reported with these devices, the substantial, steady increases in their use for augmentation since 1992 after the dramatic drop in the early 1990s (see below), and the striking increase in implantation for reconstruction after mastectomy since the early 1980s, speak to the important interest of American women in cosmetic breast surgery.
Prevalence of Silicone Breast Implants
Implantation of silicone shell devices began from a small base, gradually replacing other breast alloplastic devices listed earlier in this chapter and slowly increasing in numbers. Early reports estimated that somewhat more than 50,000 women received implants between 1962 and 1970, and 98% of these were for augmentation (de Cholnoky, 1970; Braley, 1972; Robertson and Braley, 1973). The number of women who received implants rose annually until 1979, at which time it began to plateau. The number implanted after mastectomy, which had been low and stable, began to increase in about 1975 (Gabriel et al., 1995) as surgeons and oncologists started to appreciate that this was not only possible, but might be desirable (August et al., 1994; Bailey et al., 1989; Barreau-Pouhaer et al., 1992; Berrino et al., 1987; Dowden, 1983; Feller et al., 1986; Francel et al., 1993; Handel et al., 1990; Noone et al., 1982; Patel et al., 1993; Schain et al., 1985) and could be done immediately (Trabulsy et al., 1994; Van Heerden et al., 1987; Yule et al., 1996). In 1983, 3% of women received breast implants for reconstruction after mastectomy; in 1992, more than 25% re-
ceived these implants (Edney, 1996). Local insurance data from 1988-1990 indicated that 5.9% of mastectomies were followed by reconstruction with implants (Francel et al., 1993), and a similar frequency (8%) was reported in Holland about that time (Houpt et al., 1988). Reconstruction after prophylactic mastectomy was recommended earlier and performed on a higher proportion of patients than after cancer surgery (Freeman, 1967; Jarrett et al., 1978; Kelly et al., 1966).
It is not possible to be sure of the division between augmentation and reconstruction at any given time in the total cohort of U.S. women receiving breast implants. There has been substantial variation. The general consensus has been that augmentation is the reason for about 80% of all breast implants. Two recent studies, however, report that 70%-71% of implants were placed for augmentation (Gabriel et al., 1995; Nemecek and Young, 1993). Also, a national survey reported 63% (R.R. Cook et al., 1995), and a study of a very large cohort of professional women reported 50%, of implant placements for augmentation (Sánchez-Guerrero et al., 1995a,b). As noted earlier, almost all implant placements from the 1960s to mid-1970s were for augmentation. The fraction for reconstruction increased steadily through the 1980s. During the early 1990s, almost 40% of breast implantation was for reconstruction. Most recently, reconstruction was the indication for less than 20% of implants, in part because approximately 35% of reconstructive procedures are now limited to autogenous tissue (ASPRS, 1996, 1997). Surveys of other than national cohorts of women may be misleading due to the variation in the frequency of augmentation relative to reconstruction in different parts of the country (R.R. Cook et al., 1995). For use in the calculation of modern prevalence (see below), the committee assumed that 30% of implants were for reconstruction as a midpoint in a range of 25%-35%. These are admittedly somewhat speculative estimates of reconstruction frequency from values found in the record review and surveys of Gabriel et al. (1995); R.R. Cook et al. (1995) and Sánchez-Guerrero et al. (1995a,b); the 30% figure used by Terry et al. (1995), and the off-repeated 20% figure cited in most other reports. The committee also considered premature mortality in cancer patients that decreases the surviving cohort of women with reconstructions much more than the normal mortality of augmentation patients decreases the surviving cohort of women with implants for this purpose. This factor becomes more meaningful in estimating the percentage of women alive with implants for augmentation or reconstruction in long-term follow-up of studies from 1988.
Estimates of the number of women receiving breast implants in 1982 were 100,000, 22% for reconstruction, (Szycher and Poirier, 1984), 130,000 in 1988 (Lorentzen, 1988) 120,000-150,000 in 1990 (an estimate by the Inamed Corporation) and 130,000 in 1990 (Zones, 1992). Additional data
collected from plastic surgeons by the ASPRS before 1992 are said by the society to be inconsistent with its modern surveys and are no longer released. The estimates cited here and the occasional annual estimates of new implants in the range of 100,000 or more reported in the literature by plastic surgeons are consistent with modern estimates, however.
An analysis of data from the 1988 National Health Interview Survey, which reported 11 million persons in the United States with implants of all kinds (Moss et al., 1991; see Chapter 2 for a partial list of silicone containing devices) suggested 304,000 women at least 18 years of age with silicone breast implants (95% confidence interval [CI], 239,000-369,000) or 0.33% of all women aged 18 to 75 years in 1988, of whom 73% were between ages 18 and 44 and 24% between ages 45 and 64 (Bright et al., 1993). As confirmation of the rising popularity of this procedure, only 30% of implantations had occurred before 1981 and 70% in the 1981-1988 interval. This report referred to market data suggesting 1,030,000 women with breast implants in 1988. Bright et al. (1993) observed that the design of the National Health Interview Survey (e.g., which asked for medical rather than cosmetic implants or allowed reporting by a family member) could have resulted in underreporting. On the other hand, market-based data could suffer from the vagaries of inventories, returns, breakage, and other factors that might have influenced their accuracy. In remarks to an FDA panel (FDA, 1992a) Bright recalculated the prevalence using supplemented market data with adjustments for mortality and replacements among others, and she provided an FDA estimate of 1,000,000 women with breast implants at the end of 1991.
A household survey carried out by Dow Corning Corp. identified 0.808% of women over age 14 years with implants, or 815,700 (95% CI, 715,757-924,729) women nationally in late 1989 (Cook and Perkins, 1996; Cook et al., 1995). Augmentation was the indication in 63.5% of those surveyed. This survey, in agreement with other studies and reports, identified higher prevalence in the southern and western regions of the country (three-times that in the eastern and midwestern regions) and over-representation of white (95%) and higher-income women. In low prevalence regions of the country, implantation for reconstruction was actually more common than for augmentation. A small CDC survey mentioned in this report found 0.25% of U.S. women with implants in 1983. This might be a reasonable approximation given the lower numbers of implants in the early years.
Another survey, using New York records to identify women who received implants for cosmetic reasons only, concluded that about 890,000 U.S. women (data simulations generated estimates ranging from 437,602 to 2,035,783, 95% of which were less than 1,205,820) had breast augmentation between 1963 and 1988. Assuming that reconstructions comprised
30% of total implantations, the authors adjusted this figure to about 1,270,000 U.S. women with implants for both augmentation and reconstruction in 1988 (Terry et al., 1995).
A record review carried out in a single Minnesota county of women over 14 years of age identified 749 women with implants. Their average age was 34 years (range 15-79), and augmentation, as noted earlier, was the indication in only 71%. Prevalence as of January 1, 1992 was estimated to be about 1%, which would extrapolate to about 1,000,000 women nationally (Gabriel et al., 1995). A number of other large cohorts were reviewed for the prevalence of women with implants in the course of carrying out other epidemiological studies. These cohorts either were from specific parts of the country (Brinton et al., 19962.2% of women, Seattle, Atlanta, and central New Jersey; Burns et al., 19961.18% of women, upper Midwest; Hochberg et al., 19931.24% of women, San Diego, Baltimore, and Pittsburgh); were characterized by a low response rate and self-reporting, which may have produced artificially high results (Hennekens et al., 1996, 2.74% of women), or were not a representative cross section of all ages of adult females (Brinton et al., 1996; Sánchez-Guerrero et al., 1995a,b, 1.35%). Nevertheless, given the number of adult women in the population, these values cluster around 1.2 million to 1.3 million U.S. women with breast implants in the early 1990s, which is consistent with other estimates (see below). Additionally, there is general agreement across reports examining the demographic characteristics of implant recipients that augmentation is carried out on average in women about 30-35 years of age and reconstruction in women about 40-55 years of age (including ages 48-49 after cancer and 40-41 years for prophylaxis and revision) (see also August et al., 1994; Birtchnell and Lacey, 1988; Cook et al., 1995; Gabriel et al., 1997; McGhan Medical Corporation, undated; Shipley et al., 1977; Wickman and Jurell, 1997; Winer et al., 1993).
All of the reported estimates of breast implant prevalence suffer from problems that affect their accuracy, as the authors themselves often point out. These problems include partial responses to questionnaires, varying accuracy and completeness of reporting, differing assumptions in projecting samples nationwide, small numbers, and noncomparable samples (among others noted above). A rough estimate of the current or near current number of women with breast implants can be made by assuming a reasonable range from among the estimates for 1988 with an upper bound of the estimate (1.27 million) based on actual record reviews (Terry et al., 1995), and projecting this range (1 million to 1.27 million, see below) forward using annual estimates and ASPRS and AACS survey data for ensuing years.
Of course, the estimated numbers after 1988 are also subject to error. Some are educated guesses. The ASPRS sample, based on a membership
that does 70-80% of the implantations nationally, is extrapolated to estimate the number of women receiving implants for augmentation and reconstruction in the United States. The ASPRS (and AACS) data are not validated, however, and suffer from the problems of incomplete or inaccurate reporting inherent in surveys, especially of self-reported events. Some industry sources have suggested that ASPRS data have been overestimates, perhaps because members that respond to the survey tend to be among the more active surgeons (McGhan Medical Corp., personal communication, August 26, 1998). Others have suggested that busy plastic surgeons may have tended to overestimate the numbers of procedures they do (C. L. Puckett, Vice President, ASPRS, personal communication, September 2, 1998). The AACS surveys a different group of medical specialists with practices limited to cosmetic procedures, including general surgeons, dermatologists, facial plastic and reconstructive surgeons, some plastic surgeons, family practitioners, and obstetricians or gynecologists (Atwood et al., 1994). These specialists perform about 30% of augmentations with breast implants in the United States. The smaller sample drawn from this group is also extrapolated to a national figure, in this case for augmentation only. The AACS augmentation estimates are slightly lower than those of ASPRS (e.g., 76,407 versus 87,704 in 1996). The committee used the higher ASPRS values and larger sample in making its calculation of the numbers of women in the United States with breast implants.
Not captured in any ASPRS or AACS survey is the number of women who obtain breast implants outside the United States. Some of the implantations reported are also likely to be reimplantations that might inappropriately be counted twice in estimates of the total cohort of women with breast implants. Bright et al. (1993) found that 13% of women with breast implants in the National Health Interview Survey had replacement implants. Surveys that identify and contact women should not be affected, and the review of procedures by Terry et al. (1995) included a correction factor for procedures per woman, which may have accounted for replacements, although precise details were not provided in their report. In addition, mortality has undoubtedly affected the number of women who had implants in the 1960s (admittedly a small group) even the youngest of whom will now be 60 years of age or more.
The committee estimated that between 356,000 and 417,000 underwent breast reconstruction with implants after mastectomy for cancer. This estimate assumes that 30% of the 1 million to 1.27 million woman with implants in 1988 (300,000-381,000) had implants for reconstruction after mastectomy, to which is added the ASPRS reported number of reconstructions using implants since thenabout 175,000making a total of 475,000-556,000. No more than 75% of postmastectomy reconstruction occurs after cancer surgery. The other reconstructions follow mastecto-
mies for other reasons, mostly prophylaxis or fibrocystic disease. Thus, the postcancer reconstruction total is approximately 356,000-417,000 women. Although the pre-1988 mortality of augmented and reconstructed women does not affect the 1988 surveys that identified living women, it may not be completely accounted for in the adjusted procedure-based data of Terry et al. (1995). Mortality in women with breast implants for augmentation is not known to be higher than in the general population, but for women implanted after mastectomy for cancer, the mortality cited by Gabriel et al. (1997) and Georgiade et al. (1985) is probably close to 30% after 10-15 years. This is slightly better than NCI's Surveillance, Epidemiology and End Results (SEER) registry statistics, probably because breast cancer patients with poorer prognosis are less likely to receive implants (Gabriel et al., 1997; Georgiade et al., 1985). The committee assumed a loss of 25%-30% by death during 1989-97 of women who had implants for reconstruction in 1988 and those added each year since, a combined cohort with postimplant durations ranging from less than a year to several decades. This mortality rate would have resulted in a minimum of 89,000 and a maximum of 125,000 deaths over the nine years in the combined cohort. The committee felt that assumptions regarding over- or underestimations bearing on a final estimate would be rather speculative. Nevertheless, given the prevailing view that ASPRS data might be overstated, the possibility of some double-counted replacement implants, the mortality in augmented women between 1988 and 1997, and the deaths expected among cancer patients undergoing reconstruction, it seemed reasonable to adjust the final estimate downward by 150,000.
The committee judges there to have been 1 million to 1.27 million U.S. women with silicone breast implants in 1988. The National Health Interview Survey estimate was considered an outlier, and the other estimates clustered around 1 million in the late 1980s with the 1.27 million women reported by Terry et al. (1995) as an upper limit. To this was added the total of annual estimates from ASPRS survey results for 1989-1997, counting augmentations and reconstructions with implants. Explants reported from 1992 on were subtracted (replacements after explantation are included in the reported implant figures). Non-survey year implant figures were assumed to be midpoints between bracketing survey years. Non-survey year explant figures were assumed to number 12,000 in 1989 and 1990 and to be at the midpoint between 1990 and 1992 for 1991, and the same in 1997 as 1996. The implant totals from 1989 through 1997 respectively were accordingly 130,000, 130,000, 101,000, 53,000, 59,000, 65,000, 87,000, 108,000, and 146,000 (total for the period879,000), and the ex-plants were 12,000, 12,000, 19,000, 26,000, 32,000, 38,000, 26,000, 14,000, and 14,000 respectively (total for the period193,000). The total cohort of U.S. women with breast implants at the end of 1997 was therefore 1.686
million to 1.956 million before adjustment (1 million or 1.27 million plus 879,000 and minus 193,000). This range of about 1.7 million to 1.95 million is adjusted downward and rounded to give a current estimate of 1.5 million to 1.8 million U.S. women with breast implants as of the end of 1997. This estimate could be projected forward each year using ASPRS or AACS survey numbers as they become available. To reach an estimate of U.S. women ever receiving breast implants, explantations (not replaced), deaths, and non-U.S. implantation should be added to the 1997 estimate. The committee did not perform this analysis in detail, but this estimate of ever-implanted women is likely to approximate 2 million women or more.
The ASPRS data also show that 90%-95% of implants currently placed are saline filled, mostly textured and primarily (65% in augmentation, almost all in reconstruction) in the submuscular position. Saline implants are used in 96% of augmentations, and saline implants and expanders are used in 86% of reconstructions. However, as noted in Chapter 3, the overwhelming majority of implants placed before the early 1990s were gel filled. Therefore, these implants, primarily single-lumen gel, standard double-lumen, and polyurethane coated, make up the majority of implants in place today. Some plastic surgeons report that the demand for gel-filled implants is increasing, although at the moment their use in the United States is limited by regulatory policies (V. L. Young, personal communication, October 16, 1998). Collis and colleagues report that in the United Kingdom the ''vast majority" of augmentation involve gel-filled, textured implants in the submammary position (Collis et al., 1998). There is also a trend to more reconstructions after mastectomy and more reconstructions using tissue flaps (often without alloplastic implants). Forty-two percent of reconstructions are now done at the time of mastectomy (ASPRS, 1997). Discussion of this subject is continued later in Chapter 5.
The IOM Committee on the Safety of Silicone Breast Implants was constituted to respond to a congressionally mandated request for the study of the safety of silicone breast implants from the Department of Health and Human Services. The English language, peer-reviewed, scientific literature supplemented with some data from industry and other technical reports comprised the primary information base for the description of the background, context and prevalence of silicone breast implantation and for the subsequent chapters of this report. An enormous amount of material generated from U.S. breast implant litigation, primarily the multidistrict litigation, was also available, including videotapes, compact disks, legal records, briefs, and scientific citations. Although this type of information was reviewed, it is generally not cited in this report. It
is clear that adversarial legal proceedings do not generate the kind of scientific inquiry and discussion that the IOM committee process does. On the other hand, information from legal discovery and arguments can identify directions for inquiry, useful literature, and data. This information has contributed to the completeness and accuracy of this report. The statements of women with breast implants and of scientists, physicians, and others who appeared before the committee (see Appendix A and B), although often anecdotal and not peer reviewed, were helpful too. A number of important issues emerged from the committee's reviews of available information. These issues are the ones that make up the committee's charge and they are the scientific and medical questions that are suggested or explicit in the reports concerning associations between silicone breast implants and human health conditions that have appeared in increasing numbers in the world's medical and scientific literature. In Chapters 2 through 12 the committee addresses the issues as closely and conclusively as possible under the headings of silicone chemistry, implant catalogue, silicone toxicology, reoperations and local and perioperative complications, silicone immunology, antinuclear antibodies, connective tissue or rheumatic disease, cancer, neurological disease, effects on breast feeding and on children, and screening and diagnostic mammography. Descriptions of the IOM scientific workshop and public meeting and comments on other recent policy-relevant reports appear in the appendix materials.