American Journal of Epidemiology

Grady, Deborah; Ernster, Virginia

doi: 10.1093/oxfordjournals.aje.a116379pmid: 1585896

During the last 20 years, at least five studies have examined the association between cigarette smoking and facial wrinkling. Although there are methodological concerns with each of these studies, the data are consistent with the conclusion that smoking causes skin wrinkling that could make smokers appear unattractive and prematurely old. Cigarette smoking has been shown to decrease capillary and arteriolar blood flow in the skin, perhaps damaging connective tissue components that are important to maintaining the integrity of the skin. Americans are highly motivated to avoid or eliminate facial wrinkles. The association of smoking and facial wrinkling may be important evidence to convince young persons not to begin smoking and older smokers to quit. Am J Epidemiol 1992;135:839–42.

Garabrant, David H.; Peters, Ruth K.; Homa, David M.

doi: 10.1093/oxfordjournals.aje.a116380pmid: 1585897

Previous studies linking exposure to asbestos with human colon cancer have used mortality rather than incidence as their endpoint and have neither assessed nor controlled for confounding by diet, genetic factors, or other risk factors for colon cancer. A case-control study of 746 historically confirmed cases of colon cancer and 746 matched neighborhood controls was conducted in Los Angeles County, California. In univariate analyses of the 419 male pairs, a weak association was found between asbestos exposure and colon cancer (odds ratio (OR) = 1.16, 95% confidence interval (Cl) 0.80–1.69). When confounding by family history of large bowel cancer, diet, body weight, and physical activity was controlled, there was no association between colon cancer and exposure to asbestos among males (OR = 0.99, 95% Cl 0.66–1.50). When asbestos exposure was restricted to occurrences preceding diagnosis by more than 15 years, there was no clear association between such exposure and colon cancer, either before (OR = 1.14, 95% Cl 0.76–1.70) or after confounding was controlled (OR = 0.93, 95% Cl 0.60–1.44). Further analyses by frequency and duration of exposure failed to show any association between asbestos and risk of colon cancer, but did show a consistent pattern of confounding by nonoccupational factors that, when controlled, invariably produced a weak protective effect of asbestos exposure. Among the 327 female pairs, only 6 cases and 11 controls reported asbestos exposure (OR = 0.55, 95% Cl 0.20–1.48), and there was no evidence of risk increasing as the frequency or duration of exposure increased. This study suggests not only that occupational exposure to asbestos is not a risk factor for colon cancer in the general population of Los Angeles, but also that observed associations between asbestos and colon cancer should not be interpreted as causal unless confounding by nonoccupational factors has been evaluated and controlled. Am J Epidemiol 1992; 135:843–53.

Eaker, Elaine D.; Pinsky, Joan; Castelli, William P.

doi: 10.1093/oxfordjournals.aje.a116381pmid: 1585898

This study investigates the relation of psychosocial variables to the 20-year incidence of myocardial infarction or coronary death among women in the Framingham Study In 1965–1967, a psychosocial interview was given along with the collection of other coronary risk factor data This study includes 749 women aged 45–64 years who were free of coronary disease at this baseline examination Demographic variables, psychosocial scales (such as tension and reactions of anger), and individual interview items (such as attitudes toward children, money, and religion) were measured When age, systolic blood pressure, the ratio of serum total cholesterol to high-density lipoprotein cholesterol, diabetes, cigarette smoking, and body mass index were controlled for in multivariate proportional hazards models, the predictors of the 20-year incidence of myocardial infarction or coronary death were as follows among employed women, perceived financial status only, among homemakers, symptoms of tension and anxiety, being lonely during the day, difficulty falling asleep, infrequent vacations, housework affecting health, and believing one is prone to heart disease (p < 0.05 for all variables), and among both groups of women combined, low educational level, tension, and lack of vacations These results are discussed in relation to previous findings from the Framingham Study. Am J Epidemiol 1992;135:854–64.

Clemens, John D.; van Loon, Frederik F. P. L.; Rao, Malla; Sack, David A.; Ahmed, Faruque; Chakraborty, J.; Khan, M. R.; Yunus, M.; Harris, Jeffrey R.; Svennerholm, A. M.; Holmgren, Jan

doi: 10.1093/oxfordjournals.aje.a116382

Diaz, F.; Garcia, H. H.; Gilman, R. H.; Gonzales, A. E.; Castro, M.; Tsang, V. C. W.; Pilcher, J. B.; Vasquez, L. E.; Lescano, M.; Carcamo, C.; Madico, G.; Miranda, E.

doi: 10.1093/oxfordjournals.aje.a116383

Hernandez, M.; Uribe, P.; Gortmaker, S.; Avila, C.; De Caso, L. Elena; Mueller, N.; Sepulveda, J.

doi: 10.1093/oxfordjournals.aje.a116384pmid: 1585901

The authors examined sexual behaviors, the seroprevalence of human immunodeficiency virus type 1 (HIV-1), and condom use among 2,314 homosexual and bisexual men tested during 1988-1989 at the AIDS (acquired immunodeficiency syndrome) National Center in Mexico City Bisexuals constituted 24% of the sample, the seroprevalence rate was lower for bisexuals than homosexuals (21 vs 34%) In logistic regressions, HIV-1 seropositivity was independently related to age, education, pattern of insertive/receptive behavior in anal sex, lifetime number of male sex partners, having sex with someone with AIDS, homosexual versus bisexual behavior, and a history of condyloma The same logistic regressions were found to fit bisexuals and homosexuals The rate of HIV-1 was reduced in individuals who indicated always requiring their partners to use a condom when practicing receptive anal sex (1% of the total) The most common practice for both homosexuals and bisexuals was “mixed” behavior (i.e., both insertive and receptive anal sex); this was also the practice with the highest risk Bisexuals practiced both vaginal and anal sex with women and reported little condom use The substantial seroprevalence among bisexuals, their frequent sexual contact with women, and their low rate of condom use imply a continuing role as a bridge of infection to females Whether this risk will lead to a sustained heterosexual epidemic remains to be determined. Am J Epidemiol 1992,135 883–94.

Williams, Michelle A.; Mittendorf, Robert; Stubblefield, Phillip G; Lieberman, Ellice; Schoenbaum, Stephen C.; Monson, Richard R.

doi: 10.1093/oxfordjournals.aje.a116385pmid: 1585902

Premature (prior to 37 completed weeks of gestation) rupture of the membranes (preterm PROM) is one of the most common underlying causes of preterm delivery. However, there have been few epidemiologic studies of this obstetric complication The authors studied the relation of maternal cigarette smoking and coffee consumption to both preterm PROM and spontaneous preterm labor not complicated by premature rupture of the membranes (preterm NONPROM) in a large cross-sectional data base. The 307 preterm PROM and 488 preterm NONPROM cases who delivered during 1977–1980 at the Boston Hospital for Women were compared with 2,252 randomly selected women who delivered at term at that institution. Multiple logistic regression techniques were used to derive maximum likelihood estimates of adjusted odds ratios (OR) and 95% confidence intervals (Cl). After confounders had been adjusted for, the relative risk of preterm PROM for women who reported ever having smoked during pregnancy, as compared with nonsmokers, was 1.6 (95% Cl 1.1–2.4). However, no gradient between the the number of cigarettes smoked per day and the risk of preterm PROM was observed Similar results were observed for preterm NONPROM. Women who consumed three or more cups of coffee daily during the first trimester had a 2.2-fold greater risk of preterm PROM than did women who drank two or fewer cups (95% Cl 1.5–3.3). Among coffee drinkers, there was some evidence of a linear trend in the risk of preterm PROM as coffee consumption increased Consumption of three or more cups of coffee per day was less strongly associated with the occurrence of preterm NONPROM (adjusted OR = 1.4, 95% Cl 1.0–1.9). Am J Epidemiol 1992,135;895–903.

Ruel, Marie T.; Habicht, Jean-Pierre; Pinstrup-Andersen, Per; Gröhn, Yrjö

doi: 10.1093/oxfordjournals.aje.a116386pmid: 1585903

The present study tested whether maternal nutrition knowledge was a mediating factor in the association between maternal schooling and child nutritional status, and whether the mechanism involved differed according to socioeconomic status. The data were collected in Lesotho on 921 mother-child pairs and included scores from a nutrition knowledge test, socioeconomic and demographic information, and the child's anthropometric data. A wealth factor derived from a factor analysis was used to stratify the sample into two socioeconomic groups. Two-stage least-squares estimation was used to test the mediating role of nutrition knowledge between maternal schooling and child weight-for-age. Results showed that both the importance of maternal schooling and the mechanism by which it affects the child's weight-for-age are contingent upon the family's socioeconomic status. While maternal schooling was positively associated with weight-for-age for both wealthier and poorer households, the size of the effect was much larger for the latter group. The effect of maternal schooling on weight-for-age was mediated by the mother's nutrition knowledge only among wealthier households. These results imply that, in Lesotho, nutrition education for mothers could contribute to improving children's growth, but only in households that have access to a minimum level of resources. For poorer households, nutrition education would not be sufficient. Am J Epidemiol 1992;135:904–14.

Lee, I-Min; Paffenbarger, Ralph Seal; Hsieh, Chung-Cheng

doi: 10.1093/oxfordjournals.aje.a116387pmid: 1585904

Little direct evidence exists in the epidemic-logic literature to support the widely held belief that an individual's physical activity behavior varies over the years. To provide data, the authors prospectively studied patterns of physical activity among 6,092 Harvard College alumni (average age at the start of follow-up, 43 years) who responded to questionnaires in 1962 or 1966, 1977, and 1988, and who declared themselves free of cardiovascular disease and cancer throughout that period. The median energy expenditure increased between 1962 or 1966 and 1977, but stabilized between 1977 and 1988. However, rank correlation coefficients between estimates of energy expenditure in 1977 and 1988 ranged from 0.36 to 0.41 for the various age groups (compared with 0.81-0.84 for the Quetelet index, and 0.60–0.65 for the current number of cigarettes smoked per day), implying that physical activity at the individual level was not constant. For 4,238 alumni for whom data on collegiate sports participation were available, the rank correlation coefficients between collegiate activity and alumni activity 1-7 decades later were even lower, ranging from 0.05 to 0.17 (compared with 0.31–0.58 for the Quetelet index). Thus, physical activity assessed over a short time period in an individual may not be a valid proxy for activity over the long term. The characteristics that distinguished alumni who consistently expended <1,000 kcal/week (in climbing stairs, walking, and playing sports) in 1962 or 1966, 1977, and 1988 from those who consistently expended >2,500 kcal/week were as follows: at baseline, older age, overweight, and cigarette smoking; and during college, not having been a varsity athlete and fewer hours playing sports. Am J Epidemiol 1992;135:915–25.

Flanders, W. Dana; Lin, Lillian; Pirkle, James L.; Caudill, Samuel P.

doi: 10.1093/oxfordjournals.aje.a116388pmid: 1585905

Interpretation of observational studies is difficult, particularly in cross-sectional studies, because the direction of cause and effect may be difficult to assess Did the “outcome” affect the measured exposure level, or did the exposure affect the outcome? In this paper, the authors describe a pattern, the “checkmark pattern,” which can arise in cross-sectional studies. This pattern is characterized by higher levels of the outcome in an unexposed comparison group than in some subgroups of the exposed. The pattern, if seen in certain types of observational studies, suggests that the “outcome” variable may have affected the measured exposure level. Recognition of the pattern may help the epidemiologist to decipher the causal sequence Two examples illustrate the issues involved. Am J Epidemiol 1992;135:926–35.