Odds Ratio For Use Research Articles

Risk of premenopausal breast cancer and use of electric blankets.

Electric blanket use, estimated to significantly increase background exposure to 60-Hz electromagnetic fields, has been hypothesized to increase breast cancer risk. From 1986 to 1991 in western New York State, the authors investigated the use of electric blankets as a risk factor for breast cancer in a case-control study of premenopausal women. A total of 290 premenopausal breast cancer cases and 289 age-matched randomly selected community controls were queried in regard to their use of electric blankets in the previous 10 years, including frequency of use in season and mode of use. After adjusting for age, education, and other risk factors, the odds ratio for use of an electric blanket at any time in the previous 10 years (40% of cases and 37% of controls) was 1.18 (95% confidence interval (CI) 0.83-1.68). Estimates of risk did not differ in a dose-response fashion for number of years of electric blanket use. The risk associated with daily use in season relative to nonuse was 1.27 (95% CI 0.86-1.88). The risk of breast cancer among those who reported use of the blanket through the night was 1.43 (95% CI 0.94-2.17). However, the risk for those who reported daily use in season, continuously throughout the night for the previous 10 years was 1.10 (95% CI 0.59-2.05). These findings do not support the hypothesis that electric blanket use is associated with an increased risk for breast cancer. Studies with improved measures of more highly exposed individuals with quantification of total electromagnetic field exposure might clarify the uncertainties regarding risk of breast cancer associated with such exposure.

Breast Cancer Among Young U.S. Women in Relation to Oral Contraceptive Use

While most studies have found no association between oral contraceptive use and breast cancer, several studies of younger women have reported an association with long-term oral contraceptive use. PURPOSE. We studied the relationship of patterns of oral contraceptive use to breast cancer risk among younger women. These women have had oral contraceptives available their entire reproductive lives and are now entering the breast cancer-prone years. A population-based, case-control study of breast cancer was conducted in three counties in western Washington State among women born in 1945 or later, ages 21-45. Case patients were 747 women with breast cancer diagnosed in 1983-1990 and identified through the Seattle-Puget Sound Surveillance, Epidemiology, and End Results cancer registry. Control subjects were 961 women identified by random-digit telephone dialing. Subjects were interviewed in person, using pictures of brands of oral contraceptives and calendars of life events as recall aids. There was no increased incidence of breast cancer associated with ever having used oral contraceptives. Because only 8% of this cohort had never used oral contraceptives, short-term users (< 1 year) were combined with never users as the reference group for further analyses. A small increased risk of breast cancer was associated with long duration of oral contraceptive use (odds ratio for > or = 10 years = 1.3; 95% confidence interval [CI] = 0.9-1.9; P for trend = .03), particularly among women aged 35 years or younger (odds ratio for > or = 10 years = 1.7; 95% CI = 0.9-3.1). Breast cancer was also modestly related to oral contraceptive use early in reproductive life (odds ratio for use within 5 years of menarche = 1.3; 95% CI = 1.0-1.8; P for trend = .04) and to use of high-progestin-potency oral contraceptives for at least 1 year (odds ratio = 1.5; 95% CI = 1.1-2.1). These associations were adjusted for age, age at menarche, term pregnancy, induced abortion, and family history of breast cancer. The associations were not further confounded by case-control differences in education, religion, breast feeding of offspring, or infertility; in oral contraceptive contraindications, indications, or complications; or in measures of breast cancer detection such as mammography or breast biopsy. Long-term oral contraceptive use among young women or use beginning near menarche may be associated with a small excess breast cancer risk, possibly due to susceptibility to genetic damage in breast epithelial cells at ages of high breast cell proliferative activity. Future studies should investigate whether the patterns of risk we reported are present as this cohort ages.

HIV SEROCONVERSION AND DISINFECTION OF INJECTION EQUIPMENT AMONG INTRAVENOUS DRUG USERS, BALTIMORE, MARYLAND

To examine the putative protective effect of disinfectant use on HIV seroconversion among intravenous drug users, we conducted a nested case-control study comparing 22 black heterosexual HIV seroconverters with 95 persistent seronegatives matched on gender, use of cocaine, date of study entry, and duration of follow-up. For intravenous drug users who reported using disinfectant all the time, the odds of seroconversion was 0.77 (95% CI: 0.25-2.38) compared with those who reported no use of disinfectants; for those who used disinfectants some of the time, the corresponding odds ratio was 0.91 (95% CI: 0.26-3.31). The odds ratio for use of disinfectant all the time was 0.63 (95% CI: 0.10-3.91) for those injecting at galleries and 1.08 (95% CI: 0.35-3.11) for those not injecting at galleries. These data suggest a limited protective effect of disinfectant use in the field which may be more beneficial to those injecting in shooting galleries.

Exposure to organic solvents and adverse pregnancy outcome

In a large case-control study (n = 1,926) of spontaneous abortion (SAB), exposure to solvents was ascertained by a telephone interview that asked about occupational use of 18 specific solvents or products, as well as an open-ended "other" solvent category. The adjusted odds ratio for use of any solvent was 1.1 (0.8, 1.5). Solvents for which at least a doubled crude risk of SAB was found included perchlorethylene (OR = 4.7, 95% CI = 1.1, 21.1), trichloroethylene (OR = 3.1, CI = 0.9, 10.4), and paint thinners (OR = 2.3, CI = 1.0, 5.1). Comparing exposure greater than 10 hours per week versus less did not show consistent dose-response effects. By solvent class, an association was seen with aliphatic solvents (adjusted OR = 1.8, 95% CI = 1.1, 3.0), but there was no dose-response effect by hours of use. Household use of solvent-containing products was generally not strongly associated with SAB, nor did it appear to confound the association seen with occupational use. From this and other studies, occupational exposure to at least some solvents appears associated with SAB. The associations of solvent exposure and fetal growth among liveborn offspring of controls was also examined.

RISK OF INJURY FROM RESISTING RAPE

Women may resist rape by taking a variety of self-protective measures. To examine the association between a woman's use of self-protection during a rape incident and four injury outcomes, the authors analyzed data from the National Crime Survey, an ongoing survey of self-reported victimizations throughout the United States. The study population was 851 women greater than or equal to 12 years of age who reported being a victim of completed or attempted rape during 1973-1982. Logistic regression was used to control for eight covariates, including use of weapons by the offender and the nature of the victim-offender relationship. The use of self-protection during a rape incident was protective against completed rape. The odds ratios for completed rape were 0.2 for all measures of self-protection--nonforceful, forceful, and both forceful and nonforceful (all 95% confidence intervals between 0.1 and 0.4). After controlling for type of rape incident, the odds ratios for physical injury were greater than 1.0 for all measures of self-protection but, in the case of physical injury requiring medical attention, only the odds ratio for use of both forceful and nonforceful measures was statistically significant. Because of the limitations of the National Crime Survey, these findings should be interpreted cautiously. Further research is needed to help women respond in ways that will minimize injury should a rape incident occur.

Fitting and Interpreting Loglinear Interactions in Cross-Classifications from Health Policy and Medicine

A nontechnical exposition is presented of current statistical techniques for the analysis of multidimensional tables of counted data. Performing an original analysis of a data set of interest to researchers in health policy and medicine, the paper considers what kinds of questions an analysis by loglinear modeling can address, and what kinds of answers it can obtain and how they may be sought. Unlike most previous expository accounts seeking to provide introductions to this field, this paper does not require a background from the reader in either regression or the analysis of variance. By a thoroughgoing use of odds ratios and higher-order odds ratios, it nevertheless provides a technically accurate account of the key concepts of higher-order interactions among variables, and of models being hierarchical. Statistically more advanced readers are provided with a means of effectively expositing their loglinear modeling methods and conclusions to nonstatisticians; a number of footnotes are directed toward such readers.

Errors in Slomczynski and Krauze's Comparative Analysis of Social Mobility

If we were in Slomczynski and Krauze's position, we too might be tempted to change the subject to a discussion of the sociology of science. But we are not they, and we shall not do so. We are gratified that Slomczynski and Krauze have at least counted our criticisms of their paper; we wish they had responded to more of them. The issue is not how many criticisms we offered, but whether they are correct. In our opinion, the criticisms we have made of their work are a function of its scholarly and scientific quality, and many of them have nothing to do with the merits of their matrix decomposition. We stand by our representation of Slomczynski and Krauze's (1987) work, by our reanalyses of it, and by our extensions of the Sobel-Hout-Duncan (1985) model. We continue to invite Slomczynski and Krauze to make a listing of all of their data available to other scholars, and we continue to encourage an open debate about the merits of their proposal. ' It is notable that their present response remains silent on some of our strongest comments and criticisms. The authors are unable to explain why some of their reported results cannot be replicated (Hauser and Grusky 1988); they offer no defense for their use of odds ratios to measure cross-national variations in observed (Hauser and Grusky 1988); they provide no justification for their joint testing procedures (Hauser and Grusky 1988); and they offer no rationale for resorting to rudimentary loglinear models in their analyses of mobility (Hauser and Grusky 1988). However, Slomczynski and Krauze do comment on some specific points in our work, and we think it appropriate to respond to those comments. In their section titled Disparities in Results, they admit to shifting their operational definition of yet they attempt to minimize the importance of this redefinition. We demonstrated that, without this redefinition, the outcomes of their empirical tests are dominated by the treatment of diagonal entries.2 With the redefinition, they are no longer analyzing as they earlier defined and attempted to justify the concept. Indeed, under their original definition of the odds ratios used in their correlation analysis would equal zero or be undefined. Slomczynski and Krauze attempt to justify the violations of marginal invariance and autonomy in their scheme by reference to other measures that also lack those properties. They are right that many other measures of are not invariant and autonomous. These types of measures are legitimately used when researchers do not wish to control for the size and shape of class or occupational structures. However, in any measure purporting to represent circulation mobility, it is essential to control for the forces of occupational supply and demand. The measures proposed by Slomczynski and Krauze, as well as the earlier ones they cite, do not solve this key problem in comparative analy* Direct all correspondence to Professor Robert M. Hauser, Department of Sociology, The University of Wisconsin-Madison, 1180 Observatory Drive, Madison, Wisconsin 53706. We thank Raymond Sin-Kwok Wong, Yu Xie, and Michael Sobel for helpful comments. The opinions expressed herein are those of the authors. 1 We want to clarify the situation with regard to data-sharing. As we stated in our paper, Slomczynski and Krauze did give us a listing of percentages in each category of their 22-nation data. We thanked them for that. With minor exceptions, they referred us to source publications for the original counts, and for a listing of their independent variables. This was not satisfactory. We did not think a claim of comparability could be justified unless we produced the findings in their correlation analysis, and this effort turned into a guessing game without their listing of the data. We were unable to reproduce their correlation analysis using the information Slomczynski and Krauze provided about the sources of their independent variables. 2 We also documented other defects in the empirical testing procedures of Slomczynski and Krauze on which they did not comment.

Postmenopausal use of estrogen and occlusion of coronary arteries

The degree of coronary artery occlusion was compared between users and nonusers of postmenopausal estrogen among 933 female patients undergoing angiography between the ages 50 and 75 years in the Milwaukee Cardiovascular Data Registry. Users (n = 154) had less occlusion than nonusers (n = 779), and a significant increase in occlusion scores with age was evident for nonusers ( p < 0.001) but not for users ( p = 0.50). The age-adjusted odds ratios for use of postmenopausal estrogen among women with moderate and severe levels of occlusion of the coronary arteries were 0.59 (95% confidence interval, 0.48 to 0.73) and 0.37 (95% confidence interval, 0.29 to 0.46), respectively, which indicated a statistically significant, apparent protective effect of postmenopausal estrogen on coronary occlusion. This effect was independent of the type of menopause or other risk factors but not independent of high-density lipoprotein-cholesterol levels. Higher high-density lipoprotein-cholesterol levels among users may indicate a biologic mechanism by which postmenopausal estrogen use lowers the risk of coronary occlusion.

Variations on a Theme

This article presents an expository discussion of the use of odds, odds ratios, and functions of odds ratios as an aid to the interpretation of log-linear parameters. While there have been some previous presentations on the use of odds ratios, this method has never been fully and systematically developed for complex situations, nor have the potential problems of interpretation using odds ratios been discussed. The presentation in this article more fully elaborates the possible modes of drawing contrasts with odds ratios using any of the three common parameterizations of a log-linear model (ANOVA-like, regression-like, and logit). The sampling distribution and the calculation of standard errors for odds ratios are also discussed as are some cautions on the interpretation of odds ratios. A data analysis of unemployment using polytomous variables illustrates the application and interpretation of the odds-ratio approach.

The place of odds ratios in the study of place, race and differential occupational opportunities

During the last few decades the study of racial differences in occupational and economic attainment has progressed rapidly, both in terms of theory and research. The ecological study of the racial and ethnic organization of local labor markets has not. The use of odds ratios and occupation-specific categories may provide a fruitful avenue for future research in the study of inequality across space and time.

City differences in racial occupational differentiation: A note on the use of odds ratios

This note considers whether city differences in racial occupational distribution are best investigated using measures of nominal differentiation or measures of inequality, and argues that measures of inequality are better suited for testing the prevailing theories of race differences in occupation distribution. It also defends the Index of Net Difference and the Index of Dissimilarity (and other measures of inequality used in previous research) from the criticism that they are flawed because they are sensitive to city differences in occupational structure. Additionally, it explores a new approach for investigating city differences in occupational inequality within the log-linear framework. The logical and empirical arguments offered support the conclusions that racial occupation differences reflect racial stratification, and that variation in racial occupation differences across cities is best understood as the product of city differences in the severity of racial stratification. Thus, future research in this area should continue to focus on measures of inequality rather than measures of nominal differentiation until there are sound theoretical and empirical reasons for conceptualizing race differences in occupational distribution in terms of nominal differentiation.