|
|
| |||||||||||||||||||||||||||||||||||||||||
Background Silicone breast implants have been linked to a variety of illnesses, the most controversial of which are connective-tissue diseases and symptoms. To study this relation, we analyzed data from 14 years of follow-up of the Nurses' Health Study cohort.
Methods Women who were free from connective-tissue disease in June 1976 were followed through May 31, 1990, before there was widespread media coverage of the possible association of breast implants and connective-tissue diseases. Information was collected through biennial and supplementary mailed questionnaires and blinded reviews of medical records with the use of standardized criteria. Relative risk, the measure of association, was defined as the incidence rate of connective-tissue disease among women with breast implants divided by the corresponding incidence rate among women without breast implants.
Results Among 87,501 women who were eligible for follow-up, 516 were confirmed as having definite connective-tissue diseases and 1183 as having breast implants (of which 876 were silicone-gel–filled, 170 saline-filled, 67 double-lumen, 14 polyurethane-coated, and 56 of unknown type). The mean (±SD) period of follow-up after surgery was 9.9±6.4 years (range, 1 month to 40.5 years). Three of the patients with definite connective-tissue disease — all had rheumatoid arthritis — had implants (one silicone-gel–filled, one saline-filled, and one double-lumen). The age-adjusted relative risk of a definite connective-tissue disease among women with any type of implant was 0.6 (95 percent confidence interval, 0.2 to 2.0), as compared with women without implants. For women with silicone-gel–filled implants, the comparable relative risk was 0.3 (95 percent confidence interval, 0 to 1.9). The relative risk of self-reported signs or symptoms of connective-tissue disease for women with implants was 1.5 (95 percent confidence interval, 0.9 to 2.4); the risk of having any 1 of 41 signs, symptoms, or laboratory features of connective-tissue disease was 0.7 (95 percent confidence interval, 0.3 to 1.6).
Conclusions In a large cohort study, we did not find an association between silicone breast implants and connective-tissue diseases, defined according to a variety of standardized criteria, or signs and symptoms of these diseases.
To study the relation between silicone breast implants and connective-tissue diseases, we analyzed data from 14 years of follow-up of the Nurses' Health Study cohort with respect to connective-tissue diseases that were diagnosed before June 1, 1990. Widespread media coverage in the United States of a possible association began in December 1990, after a program on the subject was aired on national television.5
Methods
The Nurses' Health Study Cohort
The Nurses' Health Study cohort was assembled in June 1976. Questionnaires were mailed to all registered nurses who were female, married, 30 to 55 years of age, and living in California, Connecticut, Florida, Maryland, Massachusetts, Michigan, New Jersey, New York, Ohio, Pennsylvania, or Texas. Seventy percent of the women invited to participate returned the base-line questionnaire. Information was sought on a variety of health conditions and practices. Subsequently, biennial questionnaires have been sent. The overall response rate to follow-up questionnaires has been more than 90 percent. The study protocol has been approved by the Human Research Committee of Brigham and Women's Hospital in Boston. All subjects have given informed consent.
Ascertainment of Exposure to Silicone Breast Implants
By 1992, the number of women still alive and participating in the study was 109,750. Among many other topics on the 1992 biennial questionnaire were questions about whether participants had ever had breast-implant surgery or silicone, paraffin, or collagen injections. After three mailings, 89,376 women (81.4 percent) returned the questionnaire, including 88,153 who answered the questions related to breast implants and injections. A supplementary questionnaire was sent to the 1861 women who reported having received breast implants of any sort or silicone, paraffin, or collagen injections. The supplementary questions were intended to confirm the breast-implant surgery and to ascertain whether it was unilateral or bilateral, the side of the implant, the reason for the surgery (cancer treatment, prophylaxis, cosmetic reasons, or other), type of implant, date (or dates) of surgery, and complications, if any. Although all breast implants are contained in silicone envelopes, there are differences in structure and filling. The implant types were categorized as silicone-gel–filled, saline-filled, double-lumen (a silicone-gel–filled envelope within a saline-filled envelope), polyurethane-coated (a silicone-gel–filled implant coated with polyurethane foam), other, or unknown.
Overall, 1809 of the 1861 women (97.2 percent) responded to the supplementary questionnaire; 330 without breast implants reported silicone, paraffin, or collagen injections; 117 reported breast-implant surgery after May 1990; 22 had received implants after the date of diagnosis of a connective-tissue disease; and 157 provided incomplete information about the breast-implant surgery. The remaining 1183 women with confirmed breast implants in May 1990 or earlier were included in the analysis. Researchers who entered data on implant history were blinded to information about the medical histories of the women.
Validation of Breast-Implant Information
A validation study of self-reported information on breast implants was conducted for a random sample of 100 women. Permission was requested to review their medical records. Sixteen women did not give permission, the medical records for 11 were not available after multiple mailings, and the medical records for 6 were not received from the surgeon or hospital. For the 67 women whose medical records were obtained, we confirmed that surgery had been performed and ascertained which side the implant was on, the reason for the surgery, the type of implant (or implants), and the date (or dates) of surgery. Medical records were abstracted by physicians using a questionnaire identical to that completed by the subjects. Self-reports agreed with blinded record reviews at the following rates: surgery — 99 percent; side of the implant — left 91 percent, right 100 percent, and bilateral 99 percent; reason for the surgery — cancer 93 percent, prophylaxis 91 percent, and cosmetic reasons 95 percent; type of implant — silicone-gel–filled 100 percent, and saline-filled 89 percent; and date (or dates) of surgery — same date 78 percent, within one year 84 percent, and within two years 95 percent.
Case Identification of Connective-Tissue Disease
Questions regarding rheumatic conditions that had occurred since 1976 were included on all questionnaires after 1980. There were specific questions about diagnosis by a physician of systemic lupus erythematosus in the 1982, 1984, 1986, and 1992 questionnaires; about rheumatoid arthritis in 1982 through 1992; about scleroderma, polymyositis, dermatomyositis, and Sjögren's syndrome in 1992; and about "other major illness diagnosed" on every biennial questionnaire.
In 1992, we mailed a screening questionnaire on connective-tissue disease6 to participants who had reported any rheumatic, musculoskeletal, or connective-tissue disease before June 1, 1990, and had answered the 1992 questionnaire. These diseases included rheumatoid arthritis, scleroderma, morphea, systemic lupus erythematosus, dermatomyositis or polymyositis, Sjögren's syndrome, "connective tissue disease not further specified," or "any other arthritis (excluding osteoarthritis and fibromyalgia)." Those who did not respond initially were sent second and third mailings. Those who still did not respond were sent a shorter questionnaire, once or twice, asking specifically about the occurrence of these conditions, or were telephoned by trained interviewers who asked the same questions and sought permission to obtain further details regarding the diagnosis. Of the 5086 participants who were sent the screening questionnaire, 4598 (90 percent) responded to the mailings or telephone calls.
The screening questionnaire on connective-tissue disease6 contained 30 questions about symptoms or signs of connective-tissue diseases ever experienced by the subject, based on the classification criteria of the American College of Rheumatology for rheumatoid arthritis,7 systemic lupus erythematosus,8 and systemic sclerosis9; Alarcon-Segovia and Cardiel's criteria for mixed connective-tissue disease10; Bohan and Peter's criteria for inflammatory myositis11; and Fox et al.'s criteria for Sjögren's syndrome.12 Validation of questionnaire data on 253 subjects with connective-tissue disease and 340 control subjects showed a sensitivity ranging from 83 to 96 percent and a specificity of 83 to 93 percent for detecting any of these six connective-tissue diseases.6
For this study, we used a more conservative screening rule to maximize sensitivity. A positive questionnaire was defined as one indicating at least two swollen joints for more than six weeks or at least three positive answers to questions about connective-tissue disease symptoms. Medical records were requested to validate the diagnoses for all subjects who had reported connective-tissue diseases and had positive questionnaires. Exposure information was separated from the medical records by a research assistant, and the records were then reviewed independently by two rheumatologists blinded to exposure. Definite connective-tissue disease was identified according to the standardized classification criteria on which the questionnaire was based. When the rheumatologists disagreed, the complete medical information was reviewed by a third independent rheumatologist and a final judgment was made by consensus of all three rheumatologists. The date of onset of the connective-tissue disease was the date of diagnosis indicated in the medical record. The analysis was based on records received through May 1994.
Population for Analysis
Women for whom information on breast implants was missing or whose connective-tissue disease was diagnosed before 1976 or after May 1990 were excluded, leaving 87,501 women eligible for follow-up. The period from June 1976, the start of the study, through May 31, 1990, was chosen to avoid potential bias from the widespread news-media attention to this topic, which began in December 1990. During the 14-year period, we accrued 1,181,244 person-years of follow-up.
Since the classification criteria for connective-tissue diseases excluded patients with milder or atypical cases and those who did not fulfill the criteria early in their disease, the true incidence of the diseases could have been underestimated. We performed three additional analyses using less stringent case definitions that included (1) patients who reported a rheumatic disease on any biennial questionnaire; (2) patients who had a positive screening, as defined on the connective-tissue disease screening questionnaire; and (3) patients who had any 1 of 41 signs, symptoms, or laboratory features of a connective-tissue disease that were included in the six classification-criteria sets that were abstracted from the medical record.
We performed analyses according to type of implant: silicone-gel–filled, saline-filled, double-lumen, or polyurethane-coated.
Statistical Analysis
For each participant, the number of person-years was assigned to the appropriate breast-implant category. Once a subject had surgery for a silicone breast implant, she was defined as having been exposed to silicone, regardless of whether an implant was subsequently removed. The number of person-years was calculated from 1976 until May 31, 1990, or until the date of diagnosis of any connective-tissue disease, whichever came first.
The analysis was based on incidence rates. Relative risk, the measure of association, was defined as the incidence rate of connective-tissue disease among women with breast implants divided by the corresponding incidence rate among women without breast implants. Age-specific rates were calculated in five-year categories of age and used to compute age-adjusted relative risks, with 95 percent confidence intervals.13 When fewer than five cases involving exposure were observed, we calculated exact confidence intervals.14
Results
During the 1,181,244 person-years of follow-up, connective-tissue diseases were confirmed in 516 subjects. Among the 87,501 women in the analysis, 1183 (1.4 percent) reported having had some type of breast implant between 1976 and May 31, 1990; gave complete information; and were free from connective-tissue disease before the implantation. Women with breast implants accounted for 11,170 person-years of follow-up. Information about the breast-implant surgery is summarized in Table 1.
|
|
Among the 516 women who met the criteria for connective-tissue disease, the observed incidence rate per 100,000 women was within the ranges reported in other studies (Table 3). Three of the patients with definite connective-tissue disease had breast implants (silicone-gel–filled in one, double-lumen in another, and saline-filled in the third). All had rheumatoid arthritis; their cases had no unusual features. The age-adjusted relative risk of any definite connective-tissue disease among the women with any type of breast implant, as compared with the women without breast implants, was 0.6 (95 percent confidence interval, 0.2 to 2.0) (Table 2).
|
The age-adjusted relative risk of rheumatoid arthritis was 0.9 (95 percent confidence interval, 0.3 to 2.6) with any breast implant, 0.4 (95 percent confidence interval, 0.1 to 2.4) with silicone-gel–filled breast implants, and 1.4 (95 percent confidence interval, 0.2 to 9.7) with saline-filled implants, as compared with no breast implants. No patients with scleroderma, systemic lupus erythematosus, inflammatory myositis, mixed connective-tissue disease, or Sjögren's syndrome had any type of breast implant.
Risk of Connective-Tissue Disease or Symptoms Based on Less Stringent Diagnostic Criteria
We studied women with possible early, milder, or atypical forms of connective-tissue disease or with any sign or symptom of a connective-tissue disease who did not meet standard classification criteria (Table 2). These groups were not mutually exclusive.
Since 1976, 5087 women have reported having a connective-tissue disease or rheumatic disorder on the biennial questionnaires. Thirty-two had some type of breast implant, including 21 with silicone-gel–filled implants. The age-adjusted relative risk of any connective-tissue disease was 0.7 (95 percent confidence interval, 0.5 to 1.0) for those with breast implants as compared with those without breast implants. For women with silicone-gel–filled implants, the age-adjusted relative risk was 0.6 (95 percent confidence interval, 0.4 to 0.9).
Signs or symptoms of connective-tissue disease were reported on the screening questionnaire by 1294 women, including 17 with some type of breast implant and 11 with silicone-gel–filled implants. Of these 17 patients, 3 fulfilled the classification criteria for rheumatoid arthritis on review of the medical records. Two patients, one with symptoms of arthritis and Raynaud's phenomenon and another with Raynaud's phenomenon alone, could not be classified as representing definite cases. Nine patients had other rheumatic or musculoskeletal conditions (five had osteoarthritis, one chondrocalcinosis, one trochanteric bursitis, one cervical strain, and one familial Mediterranean fever). In three patients, no evidence of rheumatic disease or symptoms could be found. The age-adjusted relative risk of self-reported signs or symptoms of connective-tissue disease was 1.5 (95 percent confidence interval, 0.9 to 2.4) among the women with any type of breast implant as compared with those without implants (Table 2). For the women with silicone-gel–filled breast implants, the age-adjusted relative risk was 1.2 (95 percent confidence interval, 0.7 to 2.2).
We also studied 904 participants with any of 41 signs, symptoms, or laboratory findings seen in connective-tissue diseases that were validated by review of the medical records (Table 2). Six of these women had some type of breast implant, including four with silicone-gel–filled breast implants. As compared with the group without breast implants, their age-adjusted relative risk of having the signs or symptoms of connective-tissue disease was 0.7 (95 percent confidence interval, 0.3 to 1.6) with any breast implant and 0.6 (95 percent confidence interval, 0.2 to 1.6) with silicone-gel–filled breast implants. The analyses for other implant types had similar results (data not shown).
Discussion
In this large cohort study, we did not find an increased risk of any connective-tissue disease or of 41 signs or symptoms of connective-tissue disease among women with any breast implant or with specific types of breast implants. Connective-tissue diseases occur infrequently. For this and other reasons, our study cannot be considered definitively negative. The upper bound of the 95 percent confidence interval for the relative risk of definite connective-tissue disease (2.0), for example, does not exclude minor associations that would still be of public health importance. Since information on exposure was based on self-report, there may have been some misclassification of breast-implant surgery. However, we found high rates of agreement between self-reports and medical records in our validation study of self-reported breast implants.
In all epidemiologic studies of rheumatic diseases, diagnosis is a major problem. We identified and confirmed cases through a multistep procedure and blinded medical-record review. Sixty-five percent of the 904 subjects who had any signs or symptoms of connective-tissue disease as determined by chart review had seen physicians who were active members of the American College of Rheumatology. The observed incidence of connective-tissue diseases was within ranges previously reported in population-based studies.15,16,17,18,19,20,21,22,23,24 The application of strict criteria for any connective-tissue disease may exclude some true cases or milder cases and hence underestimate the true incidence of disease. With a rare disease, a slight underestimation of the incidence rate is less important in a study of etiology than is the misclassification of participants without disease as having disease.25 In this situation, misclassification adds a small number of true cases to the very large number of true non-cases and has a negligible influence on estimates of the exposure among the non-cases. Less specific criteria might add non-cases. Since the number of cases is relatively small, the non-cases could make up an appreciable proportion of the total cases. Thus, the distribution of exposure among cases might be inaccurate. If the misclassification is random, the risk estimate will be driven toward the null value.
We found no association between breast implants and previously reported signs and symptoms,26,27,28,29 such as Raynaud's phenomenon, photosensitivity, arthritis, morning stiffness, xerostomia, dry eyes, sclerodactyly, positive antinuclear-antibody tests, and positive rheumatoid-factor tests. We could not study subjective and largely unverifiable symptoms, such as fatigue, decreased ability to sleep, frequent sore throats, cognitive deficits, arthralgias, lymphadenopathy, or dizziness, or diseases such as fibromyalgia.
The 5514 women who died during the 14-year study period could not be studied because information about breast-implant surgery was not available for them. It is unlikely that this potential limitation biased the results, unless women with breast implants and connective-tissue disease died at a higher rate than women with connective-tissue disease who did not have breast implants.
Our results are based on data about registered nurses, about 95 percent of whom were white. Whether these results can be generalized to include other women may be questioned. In 1989, a national survey of 40,000 households in the United States found that approximately 60 percent of breast implantations were performed for cosmetic reasons and that 95 percent of women with implants were white.30 The prevalence of breast implants was higher in the South and West than in other regions of the country and increased with household income. Furthermore, the breast-implant rate in our cohort, 1.4 percent, is within the range of 0.7 to 2.0 percent estimated for U.S. women. For these reasons, the women in our study are likely to be representative of women in the United States who have breast implants.
Our results are consistent with the findings of published epidemiologic studies of breast implants and rheumatic diseases31,32,33,34,35,36 and reports in abstract form.37,38,39,40 In a population-based retrospective cohort study,31 749 women in Olmsted County, Minnesota, who received breast implants between January 1964 and December 31, 1991, were followed for a mean of 7.8 years and compared with 1498 control women of similar age without implants. In 5 case subjects, as compared with 10 in the control group, one of the specified connective-tissue diseases was diagnosed (relative risk, 1.06; 95 percent confidence interval, 0.34 to 2.97).
In conclusion, we found no evidence of an association between silicone breast implants and either connective-tissue diseases defined according to a variety of standardized criteria or signs or symptoms of connective-tissue disease.
Supported by grants from the National Institutes of Health (CA40356-08S1, for the Supplement-Prospective Study of Diet and Cancer in Women; AR36308, for the Multipurpose Arthritis and Musculoskeletal Diseases Center; and AR42630, for the Effects of Silicone on the Immune Response Study). Dr. Sánchez-Guerrero was supported by a Research Fellowship Award from the Fogarty International Center (NIH 5F05 TW04573-02).
The Brigham and Women's Hospital has received grants from Dow Corning to study silicone breast implants in a separate study, the breast-implant substudy of the Women's Health Cohort Study, being conducted by the Division of Preventive Medicine. From these grants, Dr. Sánchez-Guerrero received $7,500 toward tuition at the Harvard School of Public Health in September 1992 and Dr. Karlson received $25,800 in salary and fringe-benefit support between February 1, 1994, and June 30, 1994. Dr. Liang received $2,525 from four legal firms representing Dow Corning or McGhan for 8.5 hours of consulting between June 16, 1993, and August 26, 1994. The other authors have not received compensation from companies that manufacture breast implants or from lawyers involved in breast-implant litigation.
Editor's note: This disclosure statement is in accord with our usual policy but is somewhat more detailed because of the intense public controversy over the health effects of breast implants.
We are indebted to the nurses in this study for their continuing participation; to Barbara Egan, Karen Corsano, and Mary Scamman for expert assistance; and to Dr. Peter H. Schur for reviewing this manuscript.
Source Information
From the Departments of Rheumatology and Immunology (J.S.-G., E.W.K., M.H.L.) and Medicine (G.A.C., D.J.H., F.E.S., M.H.L.), Harvard Medical School; the Robert B. Brigham Multipurpose Arthritis and Musculoskeletal Diseases Center (J.S.-G., E.W.K., M.H.L.) and the Channing Laboratory (G.A.C., D.J.H., F.E.S.), Brigham and Women's Hospital; and the Department of Epidemiology, Harvard School of Public Health (G.A.C., D.J.H.) — all in Boston.
Address reprint requests to Dr. Liang at Brigham and Women's Hospital, Dept. of Rheumatology and Immunology, 75 Francis St., Boston, MA 02115.
References
| |||||||||||||||||||||||||||||||||||||||||
Related Letters:
Breast Implants and Connective-Tissue Diseases
Vasey F. B., Aziz N., Abraham J. L., van Vollenhoven R., Liang M. H., Karlson E. W., Sánchez-Guerrero J.
Extract |
Full Text
N Engl J Med 1995;
333:1423-1424, Nov 23, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
