Search

Classification of exposure into two levels--one consisting exclusively of unexposed individuals and the other consisting of exposed and perhaps unexposed ones--yields an unbiased estimate of attributable risk when misclassification is nondifferential. The authors advocate, therefore, the use of a broad definition of exposure when estimating attributable risk. Based on this idea, they justify a simple and robust method for estimating the overall attributable risk from several exposures that is based on a division of subjects into two groups, a baseline consisting of those unexposed to all exposures and everyone else.

A case-control study of incident invasive cervical cancer was conducted in Birmingham, Alabama; Chicago, Illinois; Denver, Colorado; Miami, Florida; and Philadelphia, Pennsylvania, during 1982-1983. Controls were selected by random-digit dialing and were matched to cases by age, race, and telephone exchange. Of the white, non-Hispanic cases and controls identified, 271 (73%) and 502 (74%), respectively, were successfully interviewed. Diet was assessed by asking about the usual adult frequency of consumption of 75 food items and the use of vitamin supplements. Included were the major sources of the four micronutrients believed to reduce the risk of cervical cancer: carotenoids, vitamin A, vitamin C, and folate. Women in the highest quartiles of intake of each of these micronutrients had adjusted relative risks of invasive squamous cell cervical cancer comparable to those of women in the lowest quartiles, although their micronutrient intake was estimated to be 3-4 times as high. Risk was not affected by increased consumption of vegetables, dark green vegetables, dark yellow-orange vegetables, fruits, or legumes, or by high intake of the basic food groups. These generally negative findings stand in contrast to findings in previous epidemiologic studies, and the discrepancy is not readily explained by bias, uncontrolled confounding, or inadequate power. The question of the role of diet and nutrition in the etiology of cervical cancer is not yet resolved.

The relation between occupation and bladder cancer in women was examined based on data collected during the National Bladder Cancer Study, a population-based, case-control study conducted in 10 areas of the United States. Occupational hazards among women have received little attention in previous bladder cancer studies, in part because most studies have included too few females to accurately estimate risks. In this large case-control study, 652 white female bladder cancer patients and 1,266 white female controls were interviewed to obtain lifetime occupational histories. Patterns of bladder cancer risk by occupation in women tended to be similar to those previously observed among men. Increased risk was apparent for women ever employed in metal working and fabrication occupations (relative risk (RR) = 1.5; 95% confidence interval (CI) 0.9-2.6). Within this summary occupation category, punch and stamping press operatives had a significant trend in risk with increasing duration of employment (p = 0.012); the RR for women employed 5 years or more was 5.6 (95% CI 1.4-26.4). The authors also observed an increased risk for women employed as chemical processing workers (RR = 2.1; 95% CI 0.9-5.1 = with a significant, contrast, a decreased risk was apparent for female textile workers (RR = 0.6; 95% CI 0.3-1.1) with a significant, negative trend in risk with increasing duration of employment (p = 0.031); the relative risk for textile workers employed 10 years or more was 0.4. The authors estimate that 11% of bladder cancer diagnosed among white women in the United States is attributable to occupational exposures; this percentage is considerably lower than the 21-25% previously reported for white men in this study.

A large population-based case-control study was carried out to investigate the association between vasectomy and prostate cancer risk in black and in white men in the United States. Study subjects resided in the geographic areas covered by the population-based cancer registries of the Georgia Center for Cancer Statistics, or the Metropolitan Detroit Cancer Surveillance System, or in 10 counties included in the cancer registry of the New Jersey State Health Department. Cases for this study were men aged 40-79 years identified from pathology and outpatient records at hospitals covered by these registries, newly diagnosed with pathologically confirmed prostate cancer between August 1, 1986, and April 30, 1989. Population controls less than age 65 years were selected at periodic intervals by random digit dialing. Older controls were systematically selected (after a random start) from computerized records of the Health Care Finance Administration. A statistically nonsignificant excess risk (odds ratio (OR) = 1.6, 95% confidence interval (CI) 0.5-4.8) for prostate cancer associated with vasectomy was noted in blacks. Overall, the risk for prostate cancer associated with vasectomy in whites was not elevated (OR = 1.1, 95% CI 0.8-1.7). An increase in risk was found, however, for white men who had had a vasectomy 20 years or more prior to study (OR = 1.7, 95% CI 0.9-3.3) or who had had a vasectomy at less than age 35 years (OR = 2.2, 95% CI 1.0-4.4). For the total study group, the odds ratio associated with men who had a vasectomy 20 or more years prior to study was 1.5 (95% CI 0.8-2.7), and the odds ratio associated with men who had had a vasectomy at less than age 35 years was 2.0 (95% CI 1.0-4.0). Further detailed analysis showed that young age at vasectomy (less than age 35 years) was a more important risk factor than was years since vasectomy.

In 1984-1985, estrogen and androgen levels in blood sera were measured in 320 women who had been treated for cervical cancer in the early 1960s. Study subjects were from US clinics in Baltimore, Maryland; Boston and Norfolk, Massachusetts; Buffalo, New York; Houston, Texas; and San Juan, Puerto Rico. These clinics had participated in a larger international follow-up study of cervical cancer in which a 20-30% reduction in breast cancer risk was linked to prior pelvic irradiation, even when treatment occurred after menopause. Overall, the 203 irradiated and 117 nonirradiated women had similar mean levels of estradiol, estrone, androstenedione, and testosterone. However, there appeared to be negative, albeit inconsistent, trends for androstenedione, testosterone, and estrone, suggesting that the irradiated women had lower levels of these hormones when compared with the nonirradiated women. These differences did not reach the level of statistical significance. While chance could partially explain these findings, it is plausible that the frequently observed protective association of breast cancer with pelvic irradiation could be due in part to a decrease in steroid hormones that is secondary, perhaps, to adrenal irradiation.

Mammographic parenchymal patterns have been suggested as indicators of breast cancer risk. However, few well-controlled studies have used prediagnostic mammograms to determine the pattern classification. The authors studied 266 cases of breast cancer and 301 controls from 25 screening centers of the Breast Cancer Detection and Demonstration Project, a nationwide screening program conducted between 1973 and 1980 to evaluate the risk associated with mammographic patterns using mammograms taken four years before the detection of breast cancer. Mammograms of the cancerous breast of cases and of the ipsilateral breast in the control matched to each case were blindly assessed by one of the investigators (J.N.W.), originator of the mammographic pattern classification. The breast cancer odds ratio among women with the combined P2 + DY patterns, compared with women with the N1 pattern, was 2.8 (95% confidence interval (CI): 1.6-5.1). This estimate of relative risk was comparable with the risk associated with other recognized breast cancer risk factors. The odds ratio among P2 + DY women with a first-degree family history of breast cancer was 5.5 (95% CI: 2.6-11.8) compared with N1 women without a family history. These data provide additional evidence that mammographic patterns are indicators for subsequent development of breast cancer, particularly among women with a first-degree family history of this malignancy.

In an epidemiologic study of 2982 bladder carcinoma patients and 5782 population controls from 10 geographic areas of the United States, the role of urinary tract infection and inflammation in the etiology of this neoplasm was evaluated. A history of urinary tract infection significantly elevated the risk of bladder cancer, particularly in individuals who reported three or more infections (relative risk (RR) = 2.0). Significantly increased bladder cancer risk was also found for bladder stones (RR = 1.8), while kidney stones showed no relation. A history of three or more urinary tract infections was strongly related to squamous cell carcinoma in particular (RR = 4.8).

In a case-control study conducted in 1978 in Detroit, Michigan, as part of the National Bladder Cancer Study, the proportions of artificial sweetener users in a hospital and a population control series were compared. The study was based on interviews with 305 hospital controls and 440 population controls, as well as 391 patients with transitional or squamous cell carcinoma of the lower urinary tract. The proportion of artificial sweetener users among all hospital controls was higher than that among population controls. Among male hospital controls, it was found that 44% had ever used artificial sweeteners, compared to 38% of the male population controls. For females, the corresponding proportions of artificial sweetener users were 55% and 42%. Thus, relative risks estimated using all hospital controls were lower than relative risks estimated using population controls. When controls hospitalized for obesity-related diseases were excluded from the hospital control group, the proportion of artificial sweetener users and the relative risks for males were identical to those estimated with population controls (relative risk = 1.1). These results suggest that restriction of the control group to those patients hospitalized for non-obesity related diseases is a satisfactory procedure for selecting a control group in hospital-based studies of the effects of artificial sweeteners. For females, little or no change in the proportion of artificial sweetener users or in the relative risks was observed after exclusion of controls hospitalized for obesity-related diseases. However, the number of female subjects was small, and the results noted for females may have been due to chance.

Willingness of newly diagnosed lung cancer patients to volunteer names of friends as potential control subjects was assessed from August through December 1988 in a case-control study at the National Cancer Institute and Naval Hospital, Bethesda, Maryland. Friend controls appeared ideal to examine a genetically determined metabolic characteristic and lung cancer risk, since potentially confounding characteristics could be matched and cooperation should be high. Only 11 of 23 cases named at least one friend. Cases interviewed during the second study month were most likely to volunteer names. Either the recency of the diagnosis of a highly fatal disease or the referral to a tertiary care research hospital may have contributed to the reluctance of cases to volunteer names. No characteristic was identified that might offer a means to increase referral of friend controls.

A large excess of non-Hodgkin's lymphoma has been documented in renal transplant patients and may be related to immunosuppressive therapy, persistent antigenic challenge from the graft, or both. To determine whether immuno-suppression resulting from chronic renal failure is associated with an elevated risk of certain tumors such as non-Hodgkin's lymphoma, the authors studied cancer incidence in a national cohort of 28,049 patients in the United States with chronic renal failure who received maintenance dialysis for at least six months (totaling 66,706 person-years of observation). Compared with national incidence rates, the relative risk (RR) of cancer was 0.9 (excluding nonmelanoma skin cancer, multiple myeloma, kidney cancer, and uterine cervix cancer). Moderate excesses of leukemia, non-Hodgkin's lymphoma, Hodgkin's disease, thyroid cancer, and biliary tract cancer were found, but were not statistically significant for both sexes combined. A significantly elevated risk of non-Hodgkin's lymphoma among patients with chronic glomerulonephritis (RR = 2.6) accounted for the excess observed in the total series, raising the possibility of factors specific to this disease.

A population-based case-control study was conducted to examine whether tumor estrogen receptor status differentiated risk factor patterns for breast cancer. From December 1980 to December 1982, 458 women with newly diagnosed breast cancer and 568 control women, aged 20-54 years, from the Atlanta, Georgia, metropolitan area were interviewed. On the basis of tumor estrogen receptor results, cases were classified as receptor-positive or receptor-negative. Intercase analysis showed that age was positively and significantly associated with estrogen receptor-positive breast cancer (p = 0.001); the relative risk for an estrogen receptor-positive as opposed to an estrogen receptor-negative tumor was elevated threefold among women aged 50-54 years compared with those aged less than 35 years. In the case-control analysis, race was the only individual factor that demonstrated a significant difference in the risk for estrogen receptor-positive versus estrogen receptor-negative cancer (p less than 0.05), with blacks being at a 25% excess risk for estrogen receptor-negative cancer compared with whites. Although a history of benign breast disease was a risk factor for both positive and negative tumors, the association was stronger for the estrogen receptor-positive tumors. Postmenopausal women were at a lower risk for both cancer subtypes compared with premenopausal women. Compared with non-users, women who had ever taken oral contraceptives had a 16% decrease in the risk for receptor-positive cancer and a 22% increase in the risk for receptor-negative cancer. These results are consistent with the notion that certain exposure variables may relate to hormonal status, possibly by augmentation or suppression of estrogen receptor activity.

The relation between methylxanthine consumption and biopsied benign breast disease was investigated by using data from a case-control study which included 1,569 cases and 1,846 controls identified between 1973 and 1980 through a nationwide screening program. There was no evidence of an association between methylxanthine consumption and benign breast disease in the total study population. When histologic types of benign breast disease were examined, there were no trends in risk according to methylxanthine consumption among the 813 cases with fibrocystic disease, the 508 cases for whom detailed pathology data were not available, the 172 cases with benign neoplasms, or the 156 cases with other benign conditions. When cases with fibrocystic disease were examined according to presence of atypia, hyperplasia, sclerosing adenosis, or cysts, there was, again, no association between methylxanthine consumption and risk of disease. In addition, no relation was found between methylxanthine consumption and menstrual breast tenderness among premenopausal women with fibrocystic disease or unknown conditions.

To investigate the relationship of smoking to breast cancer risk, the authors conducted a case-control study involving 1,547 patients and 1,930 controls identified between 1973 and 1980 through a nationwide screening program. There was no evidence that smoking affected risk (relative risk (RR) = 1.2), nor were there any apparent relationships with more detailed exposure measures. No substantial variations in risk were noted by menopausal status; in particular, there was no support for the notion that smoking is associated with a reduced risk among naturally menopausal women (RR = 1.1). In addition, the data provided no general evidence that smokers experience an earlier menopause than nonsmokers, even when heavy smoking was considered. Evaluation of a number of sources of confounding and effect modification failed to alter the conclusion that smoking status does not appear to alter breast cancer risk among this population.

Results are described from four epidemiologic studies in the United States which used random digit dialing in over 30,000 households to identify controls from the general population for use in case-control studies. Methods and problems in telephone sampling are discussed. It is concluded that if complete population rosters are unavailable and if the population to be sampled has the high rates of telephone ownership typical of much of the United States, telephone-based sampling can yield a nearly random sample of the individuals in a population, often at much less expense than can dwelling-based sampling.

To determine why the incidence rate of transitional cell bladder cancer in whites in the United States is approximately twice that in blacks, the authors examined data from a large population-based case-control study of bladder cancer conducted in 1978 involving 2,982 cases and 5,782 controls. The relative risk of transitional cell carcinoma for whites compared with blacks was 1.9 before adjustment for the major bladder cancer risk factors, whereas after adjustment for cigarette smoking and occupation it was 1.6 (95% confidence interval (CI): 1.3-2.1). Further adjustment for other risk factors, including history of a bladder infection and a family history of urinary tract cancer, did not alter this estimate. The elevated risk of white compared with blacks was limited, however, to cases whose disease was confined to the mucosa and submucosa. Among cases whose disease had extended to the bladder musculature or beyond, whites were at slightly reduced risk compared with blacks. This suggests that whites may be diagnosed with conditions that go undetected in blacks but that are unlikely as a group to progress to more extensive disease. Because of the population-based nature of the study, it was possible to determine that if bladder cancer incidence among whites of both sexes was reduced to the level among blacks, total incidence in the United States would fall by 36 per cent.