Search

In Reply. —We appreciate the interest of Drs Stewart and Elliott in our recently published article comparing characteristics of women with and without breast augmentation. It is true that our study included a relatively small number of women with breast augmentation (n=80), but they were randomly identified from the general population in 3 defined geographic areas stratified by 5-year age groups. While the age distribution of the sample is different from that of the general population, women in each 5-year age group are representative of all women of similar age in the defined geographic areas. These women were not identified on the basis of the presence of any disease or the utilization of any health care facility. We agree that confounding can never be entirely ruled out as an explanation for associations that are observed in epidemiologic studies such as ours. For example, the positive association we observed between oral contraceptive use

Objective. —To compare selected characteristics of women with and without augmentation mammaplasty to identify differences between these 2 groups of women. Design and Study Participants. —White women identified as controls in previously conducted population-based, case-control studies formed the study population for the present cross-sectional analysis (N=3570). Main Outcome Measure. —Interview information on selected characteristics was compared between women who had received augmentation mammaplasty (n=80) and other women (n=3490) using the prevalence odds ratio (pOR) as the measure of association. Results. —Women with breast implants were more likely to drink a greater average number of alcoholic drinks per week (for >7 drinks vs 0 drinks: pOR=2.9, 95% confidence interval [Cl]=1.5-5.5), be younger at first pregnancy (for age 74 kg vs Conclusion. —The differences we found between women with and without breast implants suggest that consideration and evaluation of confounding factors in future studies will help to clarify some of the long-term health consequences of having breast implants.

Objective. —To evaluate the association between prior infection withChlamydia pneumoniae, as measured by IgG antibody, and coronary artery disease. Design. —A population-based, case-control study. Setting. —Group Health Cooperative of Puget Sound, a Seattle-based health maintenance organization. Participants. —Men 55 years of age and younger and women 65 years of age and younger. Cases (n=171) were members of Group Health Cooperative undergoing diagnostic coronary angiography who had at least one coronary artery lesion occupying 50% or more of the luminal diameter. The population controls (n=120) were Group Health Cooperative members without known coronary heart disease. Main Outcome Measure. —The adjusted odds ratio (OR) for coronary artery disease associated with priorCpneumoniaeinfection as measured by the presence of IgG antibody. Results. —After adjusting for age, gender, and calendar quarter of blood drawing, the OR for coronary artery disease associated with the presence of antibody was 2.6 (95% confidence interval, 1.4 to 4.8). The association was limited to cigarette smokers, in whom the OR was 3.5 (95% confidence interval, 1.7 to 7.0). Among never-smokers, the OR was 0.8 (95% confidence interval, 0.3 to 1.9). When cases and controls were restricted to those assayed concurrently, the adjusted OR (smokers and nonsmokers combined) was 4.2 (95% confidence interval, 1.8 to 10.0). Adjustment for serum cholesterol, hypertension, alcohol use, diabetes, and socioeconomic status did not change these results. Only a weak association was found when cases were compared with 63 subjects whose angiographic results were normal (OR, 1.2; 95% confidence interval, 0.6 to 2.2). Conclusions. —These results generally support the previously reported association betweenCpneumoniaeinfection and coronary heart disease. However, caution should be used in interpreting the basis for this association. (JAMA. 1992;268:68-72)

The techniques of cost-benefit, cost-effectiveness, and clinical-decision analyses provide theoretical frameworks for systematic evaluation of the consequences of medical care programs or policies. For a specific medical problem, these techniques help evaluate the attainment of the "greatest social good," given the limited availability of health resources. 1 The analyses, in and of themselves, are limited by the difficulty in determination of values concerning quality-of-life issues and by the model's inability to incorporate political and medicolegal concerns and clinical reality. 2 Thus, these tools are usually intended, from a societal perspective, to complement and enhance, not replace, health policy decision making and clinical judgment. 3 For complex issues, these analytic tools can help clarify points of agreement and disagreement regarding assumptions, probabilities, values, and what additional information is needed from future studies. 4 In this issue ofThe Journal, Binkin et al, 5 by using decision analysis in combination with cost-effectiveness analysis

To determine whether characteristics that are correlated with male homosexual behavior are associated with the incidence of cancer, the names of persons with a diagnosis of cancer in western Washington during 1974 to 1979 were linked to those in the state syphilis registry. Eight of 47 men with anal cancer were found to have had a reactive FTA test result; the expected number, based on the proportion of reactive cases among men with other sites of cancer, was only 0.40. Among men with anal cancer identified through ten population-based cancer-reporting systems in the United States, 24.4% had never been married, compared with 7.8% of men with colon and rectal cancer. Neither of these relationships was observed for women with anal cancer. Because in men, but not in women, having had syphilis and being single are associated with the practice of anal intercourse, our data suggest that anal intercourse may be a risk factor for anal cancer. ( JAMA 1982;247:1988-1990)

To determine whether prior oral contraceptive (OC) use is a risk factor for pituitary prolactinoma, we attempted to identify all women (n=72) with a prolactinoma diagnosed between 1976 and 1980 in three counties in western Washington. A control group of 303 women was selected by dialing random telephone numbers from the same counties. Prior OC use, according to OC indication, was ascertained during a standardized telephone interview. Relative to the risk for women who had never used an OC, the risk of prolactinoma for women who had used OCs for birth control was 1.3 (95% confidence interval, 0.7 to 2.6). This risk was 7.7 for women who used OCs for menstrual regulation (95% confidence interval, 3.7 to 17.0). Previous findings of an association between OC use and prolactinoma may have resulted from OC treatment of menstrual irregularity in women with an undiagnosed prolactinoma. (JAMA1983;249:2204-2207)

To the Editor.— In the report of their case-control study that appeared in the March 25 issue ofTHE JOURNAL, Marchbanks et al 1 estimated the risk of ectopic pregnancy in current users of intrauterine devices (IUDs) to be 13.7 times that of nonusers. However, we believe that because of the way the authors defined their control group that this value seriously overstates the true excess risk associated with the use of an IUD. Ideally, controls in a case-control study should be all or a sample of that population at risk for the disease from which the cases arose. For practical reasons, often this ideal is not met, and it is necessary to choose as controls persons who have conditions or illnesses that are different than the one under investigation. This choice will produce a valid result only to the extent that the controls selected have a frequency of exposure

In a prospective study of morbidity associated with Chlamydia trachomatis infections during pregnancy, we isolated C trachomatis from the endocervix of 18 (6.7%) of 268 women examined before 19 weeks' gestation. Infected women were significantly younger than noninfected women, and significantly more often unmarried, supported by public assistance, and pregnant for the first time. Among women followed up from 19 weeks' gestation until delivery, the mean duration of gestation was significantly shorter for those with antepartum chlamydial infection. Stillbirth or neonatal death occurred in six (33%) of the 18 pregnancies of infected women compared with eight (3.4%) of the 238 pregnancies of noninfected women followed up from the 19th week of gestation through delivery. Stillbirth or neonatal death occurred ten times more often among Chlamydia infected women than among uninfected controls matched for age, marital status, socioeconomic status, pregnancy order, and race. ( JAMA 1982;247:1585-1588)

Women who were hospitalized for ectopic pregnancy in five hospitals in King County, Washington, were interviewed concerning history of induced abortion. For comparison, women who delivered a live-born child during this same time period were also interviewed. Relative to women who had never undergone an abortion, the risk in those who had was increased to a modest degree, but this increase could well have been the result of chance (relative risk, 1.4; 95% confidence interval, 0.7 to 2.6). For women who had two or more abortions, the relative risk was 1.8 (95% confidence interval, 0.5 to 7.1). Our results suggest that legal abortion as performed during the past decade in the United States does not carry a large excess risk for future ectopic pregnancy, but larger studies will be needed in order to determine more precisely the magnitude of any excess that does exist.