Elizabeth Santoro, RN, MPH & Diana Zuckerman, PhD: March 2004
Several studies have been conducted on breast implant patients in Europe, primarily Scandinavia. There is one consistent thread that runs through them: they have spun their findings to make the data appear less damaging. In some cases, the study designs are flawed and therefore cannot support the conclusions that are made. In other cases, findings are inconsistent with those of other independent researchers at the Food and Drug Administration and the National Cancer Institute. These European studies have examined several health topics, such as the risk of autoimmune disease, implant rupture, suicide, and lung cancer.
What Do European Studies Say About Breast Implants and Autoimmune Disease?
A major health question about breast implants has been whether they cause autoimmune disease. Most of the research that has been conducted finds no statistically significant connection between silicone implants and diseases such as lupus, fibromylagia, scleroderma, rheumatoid arthritis, and Sjogren’s syndrome. However, many of these studies have shortcomings that limit their usefulness for drawing meaningful conclusions.
Several of these studies were conducted in Europe, although they were co-authored by scientists at the International Epidemiology Institute, a Maryland company that receives money from the company that makes silicone, Dow Corning. The articles that were quoted in the Institute of Medicine report on breast implants include those by Friis and colleagues in 1997, Giltay and colleagues in 1994, and Nyren and colleagues in 1998.1 2 3
Which Diseases Were Studied?
Autoimmune diseases typically present with relatively common and subjective complaints such as pain, fatigue, muscle aches, and sleep disturbances. As a result, it can be difficult to properly diagnose these disorders. For some disorders, such as Sjogren’s syndrome, it can take more than six years to correctly diagnose the problem.4 Since many of the symptoms characteristic of autoimmune diseases are subjective and most do not require surgery or inpatient care, it seems unlikely that the women will be hospitalized for them. The treatment of lupus, for example, is managed mostly by medications, diet, exercise, and the pacing of daily activities. This regimen can be handled on an outpatient basis.5 Similarly, researchers believe that scleroderma takes years to develop and even longer to properly diagnose.6 Several autoimmune conditions, such as scleroderma and Sjogren’s, are rare and require large numbers of study participants in order to ensure that increases risks can be detected. These characteristics that are typical of autoimmune diseases should be addressed in the study designs. Unfortunately, many of these factors were not properly incorporated into the study methodologies.
Description of the Three Studies
Friis and colleagues evaluated the presence of connective tissue diseases including lupus, Sjogren’s syndrome, rheumatoid arthritis, scleroderma, and other rheumatic conditions with a sample of 2,570 women with implants and 11,023 controls in Denmark. The minimum time that women had implants was less than one year and the average time with implants was 7.2 and 8.4 years for the reconstructive and augmentation groups, respectively. The control group was of women who had breast reduction surgery or mastectomy without receiving implants. The outcome measure was women who were hospitalized for connective tissue diseases, and did not include women who were treated on an outpatient basis.
Giltay and colleagues studied a sample of 235 women with implants and 210 controls from The Netherlands who were matched for age and year of the operation. They administered a questionnaire to determine if women with breast implants reported more rheumatic symptoms than the controls. They also looked at the use of anti-rheumatic drugs and the number of medical consultations that were for rheumatic symptoms. The minimum length of time with implants was two years and the average amount of time with implants was 6.5 years. The control group was composed of women who had cosmetic procedures that did not include silicone.
Nyren and colleagues studied 7,442 Swedish women with implants and 3,353 Swedish controls. They evaluated the number of women with connective tissue type disorders including lupus, Sjogren’s syndrome, rheumatoid arthritis, and scleroderma among women with breast implants. One month was the minimum time required to be in the study and the mean length of time with implants was six years and 10 years for the reconstructive and augmentation patients, respectively. This control group was comprised of women who had breast reduction surgery. This study used hospitalization as its outcome measure.
There were four major types of flaws in these studies. One of these was the choice of the outcome measure, which in two of the studies was hospitalization. This outcome assumes that the people with autoimmune diseases are treated in the hospital for their symptoms. Another flaw of the studies was the amount of time that women were exposed to implants. Ideally, implants that were only a few months or years old would not be permitted in the study because the chance of illness from exposure to breast implants is likely to increase as the implant ages and leaks. An additional flaw is the size of the samples. Many autoimmune diseases, such as scleroderma, are rare (affecting less than one woman in 10,000) and require very large samples in order to detect even small significant changes. The comparison samples were also inappropriate, often consisting of women who had breast reduction surgery. A better designed study would have included a comparison group who did not have breast surgery, but were otherwise similar to the implant group.
These flaws in the study design can skew the findings so that a particular outcome is more likely to occur. In the case of these European studies, the study methodologies can diminish the possibility of a relationship between breast implants and autoimmune diseases. As published, the articles suggest breast implants are not associated with autoimmune disease. However, after careful scrutiny, the possibility of a relationship becomes much more plausible. Being able to determine what the studies actually show, requires a reader with expertise in research, statistics, and study design.
How Do You Evaluate Who is Sick?
In all three of these studies, the patients were neither questioned nor directly examined by a healthcare professional. Moreover, in the Nyren and Friis studies, the outcome measure was hospitalization rather than diagnosis. This would inevitably result in underreporting, since most women would not be hospitalized for these diseases unless they had the diseases for such a long enough time that their condition had greatly deteriorated. The Giltay study sent a questionnaire to women who had silicone breast augmentation surgery and to a matched control who had a cosmetic operation not involving silicone in that same year and department. The questionnaire queried whether the women experienced rheumatic complaints and symptoms. The information obtained in questionnaires must always be critically assessed since responders can be different than non-responders. The response rate was 82% for patients and 73% for controls. However, a more serious problem for the Giltay study was the small sample.
Exposure to Implants
A related weakness of all three studies is that they included women who had implants for a short period of time. These studies had inadequate minimum requirements for the length of time a woman had implants: one month (Nyren), less than a year (Friis), and two years (Giltay). The average years that the women had implants ranged from six to fourteen. Even if a woman with breast implants immediately developed an autoimmune disease, it would probably take several years for her to be diagnosed, and even longer for her to be ill enough to be hospitalized. If a disease develops after years of exposure, or after silicone leaks from the implant, that would take even longer before she was diagnosed or hospitalized. Including women who had implants for just a few months or years obviously biased the results, making it less likely to find a significant association between implants and autoimmune disease. Women should have been implanted for at least 6-10 years instead of having implants for an average of 6-14 years.
A very large sample size is essential in a study of implants and autoimmune diseases, because so many of these diseases are rare and because half the patients had implants for a short period of time. Nyren and Friis acknowledge that the size of their samples was inadequate. Nyren had a sample of 7,442 women who chose implants for cosmetic purposes or for reconstruction after mastectomy and 3,353 women who had breast reduction surgery. Friis included 2,570 women with implants and 11,023 controls. These sample sizes do not seem that small, but given the rare nature of these diseases, more study participants are needed. The Giltay study included only 287 women with silicone breast implants, and each was age-matched to a control who in that same year had a cosmetic operation. The limited power of these studies made it unlikely that an increased risk of disease — even a doubling or tripling — would be statistically significant.
Another weakness of two of these studies was the use of breast reduction patients for a comparison sample in the Nyren and Friis studies. This comparison group helps control for any health impact of surgery, which is an appropriate design if the goal is to determine the health risks of the implant itself. However, if the goal is to determine the health risks of getting breast implants, surgery is an essential component of that product and the impact of surgery therefore should be part of the “treatment” variable. And of course, women are not deciding whether to get their breasts increased or decreased in size. Women considering implants want to know if getting implants will pose health risks compared to not getting implants, so the comparison group should be women who did not undergo breast surgery, but who are like the implant group in as many other ways as possible.
In the study by Giltay, breast implant patients were compared with patients who had other unspecified cosmetic procedures which did not include silicone. It seems likely that this is an appropriate comparison group since the women were matched according to age and the other year of operation. Unfortunately, the authors do not report the comparison surgeries.
Even with the various methodological flaws, all of which were likely to minimize health risks associated with implants, the results indicated potential health problems among women with implants. Although not statistically significant, Friis and colleagues found a 30% increase in scleroderma, lupus, and Sjogren’s syndrome for mastectomy patients with implants compared to women who had breast reduction surgery, or mastectomy without implants. Giltay and colleagues found that women with silicone breast implants report more rheumatic symptoms when compared with controls who had an unspecified plastic surgery that did not involve silicone. Similarly, Nyren and colleagues found that breast surgery patients (either reduction or augmentation) had slightly higher rates of connective tissue disorders than was expected based on rates in the general population of the same age. Had these studies been larger, with more women who had implants for a longer period of time, and had they been based on diagnoses or self-reported symptoms rather than hospitalization, these differences might have been statistically significant. Because of these study flaws, it is impossible to know if implants are linked to autoimmune diseases.
What Do European Studies Say About Implant Rupture?
European studies on rupture tend to be favorable about implants, and their findings are inconsistent with studies conducted in the United States on the same topic. A Danish study by Lisbet Holmich and colleagues on rupture rates of silicone breast implants illustrates this point.7 It reported that 10% of implants had definitely ruptured and 7% had possibly ruptured over a two year period. Based on that information, the authors conclude that at least 15% of modern implants will probably rupture within three to ten years following implantation. This conclusion seems to conflict with a study by the FDA, which found that all breast implants will eventually break, and that most augmentation patients have at least one broken implant within ten years.8
Is it possible that implants last longer in Denmark than in the U.S.? Apparently not. First of all, by focusing on implants rather than women, the Holmich article estimates 15% of implants will break, which is approximately 30% of women with at least one broken implant. Secondly, by saying “at least 15%” they are focusing on the low number but not denying that the percentage could be considerably higher. Even more important, in a table in Holmich’s paper, the data indicate that the chances of implant rupture increases dramatically after 10 years. According to their own data, while most implants are intact for ten years, most will break between 11-20 years, and almost all will be broken in 20 years. For example, during the two years of the study, 41% of the implants that were between 22 and 25 years old definitely ruptured and 23% had possibly ruptured. These data suggest that as the age of the implant increases, the chance of having a ruptured implant also increases substantially. Although these trends are presented in the table, they are barely mentioned in the paper and ignored in the summary and conclusions. Instead, the authors focused on the estimated rupture rate of “at least 15%” during the first ten years of having implants, and ignore the bad news after that.
Like the autoimmune studies, this research was conducted with researchers from the International Epidemiology Institute, which was funded by Dow Corning.
What Do European Studies Say About Implants and Suicide or Lung Cancer?
Two European studies found that breast augmentation patients were more likely to kill themselves compared to women without implants. The study of Swedish women by Dr. V.C.M. Koot and colleagues as well as the study of Finnish women by Dr. Eero Pukkala and colleagues report that women with implants are three times as likely to commit to suicide when compared to the general population of Sweden and Finland, respectively.910
The authors explain the high suicide rates by claiming that women who decide to get plastic surgery such as breast implants have a pre-existing propensity toward suicide due to mental problems and low self-esteem. However, this conflicts with findings by Brinton and colleagues at the National Cancer Institute, who reported that breast augmentation patients were four times as likely to kill themselves as other female plastic surgery patients of the same age. There is no reason to believe that women who decide to undergo breast augmentation have lower self-esteem or poorer mental health than other plastic surgery patients.
Women with breast implants who testified before the FDA in October 2003 explained the suicide data differently than the Scandinavian researchers. Several women described being depressed and suicidal because of debilitating pain, fatigue, and mental confusion years after getting silicone breast implants. They explained that when their implants were removed, their health improved, and their depression and suicidal feelings diminished.11
The Koot study also found an excess number of deaths from lung cancer. However, the authors assumed it was due to smoking, even though they did not evaluate whether women with implants were more likely to smoke than other women in the study. The National Cancer Institute study by Brinton and colleagues had also found an increase in deaths from lung cancer among breast augmentation patients compared to other plastic surgery patients, even though the implant patients were not more likely to smoke than other plastic surgery patients. It therefore is inappropriate to assume that smoking explains the increase in lung cancer. Research is needed to determine if that is the most likely explanation.
The Koot and Pukkala studies were conducted with the help of the International Epidemiology Institute, which was funded by Dow Corning.
What Can Be Learned From The European Studies?
What do all of these studies have in common? They all found problems associated with breast implants, but they were described with rose-colored lenses to make the findings sound like a clean bill of health for breast implants. In some cases, the study design minimized problems by including women who had implants for just a few months, or by studying rare diseases in small samples. Even when the findings are consistent with studies conducted by independent researchers at the FDA or NCI, the summaries and conclusions sound dramatically different — and much less worrisome. In some cases, damaging findings are ignored, or spun to neutralize the findings. Only when the studies are carefully scrutinized and compared with other research does it become obvious that the findings are quite troubling. The bottom line: even studies funded by Dow Corning find a link between breast augmentation and poor health outcomes and show that most implants break in a relatively short period of time. These findings add to the growing evidence that implants may be associated with increased morbidity and mortality over the long-term.