American Journal of Epidemiology

1. American Journal of Epidemiology May 15, 2011 173 (10)

2. Exploring Dietary Patterns By Using the TreeletTransform Principal component analysis (PCA) has been used extensively in the field of nutritional epidemiology to derive patterns that summarize food and nutrient intake, but interpreting it can be difficult. The authors propose the use of a new statistical technique, the treelet transform (TT), as an alternative to PCA. TT combines the quantitative pattern extraction capabilities of PCA with the interpretational advantages of cluster analysis and produces patterns involving only naturally grouped subsets of the original variables. The authors compared patterns derived using TT with those derived using PCA in a study of dietary patterns and risk of myocardial infarction among 26,155 male participants in a prospective Danish cohort. Over a median of 11.9 years of follow-up, 1,523 incident cases of myocardial infarction were ascertained. The 7 patterns derived with TT described almost as much variation as the first 7 patterns derived with PCA, for which interpretation was less clear. When the authors used multivariate Cox regression models to estimate relative risk of myocardial infarction, the significant risk factors were comparable whether the model was based on PCA or TT factors. The present study shows that TT may be a useful alternative to PCA in epidemiologic studies, leading to patterns that possess comparable explanatory power and are simple to interpret.

3. Invited Commentary: Dietary Pattern Analysis The analytic approaches used in nutritional epidemiology for dietary pattern analyses share common characteristics with those of genetic epidemiology. In this issue of the Journal, Gorst-Rasmussen et al. (Am J Epidemiol. 2011;173(10):1097–1104) discuss one such approach. Application of methods used in genetic pattern analyses to nutritional epidemiology could prove valuable but raises important issues that need to be considered because dietary and genetic studies often address different types of questions in analyzing interrelated variables. These different aims require statistical methods that assume different characteristics of the underlying patterns. The authors briefly describe such differences to facilitate interpretation and applications of previous and future pattern studies.

4. Bilateral Oophorectomy in Relation to Risk of Postmenopausal Breast Cancer: Confounding by Nonmalignant Indications for Surgery? Bilateral oophorectomy is often performed during hysterectomy for benign conditions and can reduce breast cancer risk by 20%–50% when performed at younger ages. Accuracy of estimating the decrease in breast cancer risk associated with bilateral oophorectomy could be affected by common conditions that lead to surgery, such as uterine fibroids or endometriosis. The authors examined the potential for confounding by nonmalignant indications for surgery on breast cancer risk estimates in a population-based case-control study of invasive breast cancer newly diagnosed in 1992–1995. Breast cancer cases (N = 4,935) aged 50–79 years were identified from Wisconsin, Massachusetts, and New Hampshire tumor registries; similarly aged controls (N = 5,111) were selected from driver's license and Medicare lists. Reproductive and medical history was obtained from structured telephone interviews. Odds ratios and 95% confidence intervals were estimated with multivariate logistic regression. Women who underwent bilateral oophorectomy with hysterectomy at age ≤40 years had significantly reduced odds of breast cancer (odds ratio = 0.74, 95% confidence interval: 0.60, 0.90) compared with women with intact ovaries and uterus. Effect estimates were virtually unchanged after adjustment for uterine fibroids or endometriosis history. Results indicate that breast cancer risk reductions conferred by bilateral oophorectomy are not strongly confounded by failure to account for nonmalignant indications for surgery.

5. Planning for the Control of Pandemic Influenza A (H1N1) in Los Angeles County and the United States Mathematical and computer models can provide guidance to public health officials by projecting the course of an epidemic and evaluating control measures. The authors built upon an existing collaboration between an academic research group and the Los Angeles County, California, Department of Public Health to plan for and respond to the first and subsequent years of pandemic influenza A (H1N1) circulation. The use of models allowed the authors to 1) project the timing and magnitude of the epidemic in Los Angeles County and the continental United States; 2) predict the effect of the influenza mass vaccination campaign that began in October 2009 on the spread of pandemic H1N1 in Los Angeles County and the continental United States; and 3) predict that a third wave of pandemic influenza in the winter or spring of 2010 was unlikely to occur. The close collaboration between modelers and public health officials during pandemic H1N1 spread in the fall of 2009 helped Los Angeles County officials develop a measured and appropriate response to the unfolding pandemic and establish reasonable goals for mitigation of pandemic H1N1.

6. Dynamic Path Analysis in Life-Course Epidemiology Life-course epidemiology seeks to better understand the mechanisms that lead to the development of chronic diseases. An example is the mechanism leading from body size to coronary heart disease (CHD); one way to acquire a better understanding of this mechanism is to investigate to what extent it works through other risk factors. In this paper, the dynamic path analysis model is presented as a tool to analyze these dynamic mechanisms in life-course epidemiology. A key feature of dynamic path analysis is its ability to decompose the total effect of a risk factor into a direct effect (not mediated by other variables) and indirect effects (mediated through other variables). This is illustrated by examining the associations between repeated measurements of body mass index (BMI) and systolic blood pressure (SBP) and the risk of CHD in a sample of Danish men between 1976 and 2006. The effect of baseline BMI on the risk of CHD is decomposed into a direct effect and indirect effects going through later BMI, concurrent SBP, or later SBP. In conclusion, dynamic path analysis is a flexible tool that by the decomposition of effects can be used to increase the understanding of mechanisms that underlie the etiology of chronic disease.

7. Remarks on Antagonism Different forms of antagonism are classified in terms of response types and are related to the sufficient-cause framework. These forms of antagonism include “synergy under recoding of an exposure,” “synergism under recoding of the outcome,” and so-called “competing response types,” with synergism itself conceived of as causal co-action within the sufficient-cause framework. In this paper, the authors show that subadditivity necessarily implies at least one of these 3 forms of antagonism. Empirical conditions for specific forms of antagonism are given for settings in which monotonicity assumptions are and are not considered plausible. The implications of subadditivity and superadditivity for causal co-action for either an outcome or its absence are characterized under various assumptions about monotonicity. A simple computational procedure is described for assessing whether any specific form of causal co-action can be detected for either an outcome or its absence for both cohort and case-control data. The results in this paper are illustrated by application to examples drawn from the existing literature on gene-gene and gene-environment interactions.

8. Life-Course Socioeconomic Position and Incidence of Dementia and Cognitive Impairment Without Dementia in Older Mexican Americans: Results From the Sacramento Area Latino Study on Aging There have been few investigations of the link between changes in life-course socioeconomic position (SEP) and cognitive decline or incidence of dementia. The authors examined the impact of changes in life-course SEP on incidence of dementia and cognitive impairment but not dementia (CIND) over a decade of follow-up. Participants of Mexican origin (n = 1,789) were members of the Sacramento Area Latino Study on Aging cohort. Incidence of dementia/CIND was ascertained by using standard diagnostic criteria. SEP indicators at 3 life stages (childhood, adulthood, and midlife) were used to derive a measure of cumulative SEP (range, 0 to 8) and SEP mobility. Nearly 24% of the sample maintained a low SEP throughout life. Hazard ratios and 95% confidence intervals were computed from Cox proportional hazards regression models. In fully adjusted models, participants with a continuously high SEP had lower hazard ratios for dementia/CIND compared with those with a continuously low SEP at all 3 life stages (hazard ratio = 0.49, 95% confidence interval: 0.24, 0.98; P = 0.04). In age-adjusted models, participants experienced a 16% greater hazard of dementia/CIND with every 1-unit increase in cumulative SEP disadvantage across the life course (hazard ratio = 1.16, 95% confidence interval: 1.01, 1.33; P = 0.04). Early exposures to social disadvantage may increase the risk of late-life dementia.

9. Evaluation of Functional Genetic Variants for Breast Cancer Risk: Results From the Shanghai Breast Cancer Study In previous studies among 1,144 cases and 1,256 controls recruited in stage 1 of the Shanghai Breast Cancer Study (SBCS I; 1996–1998), 18 known or potentially functional single nucleotide polymorphisms (SNPs) in 16 genes were found to be associated with breast cancer risk. The authors evaluated these associations among 1,918 cases and 1,819 controls recruited in stage 2 of the SBCS (SBCS II; 2002–2005) using genetic effect models and subgroup analyses predetermined from SBCS I results. Five SNPs (AHR rs2066853, ATM rs1003623, ESR1 rs2234693, GSTP1 rs1695, and SHBG rs6259) showed generally consistent results in SBCS I and SBCS II and statistically significant associations with breast cancer risk in combined analyses, mostly in subgroups defined by age or menopausal status. Further, the relation between breast cancer risk and SHBG rs6259 was found to vary by body mass index (weight (kg)/height (m)2) (P for interaction = 0.003). The strongest reduction in risk associated with SHBG rs6259 was found for lean (body mass index <23) postmenopausal minor allele carriers (odds ratio = 0.6, 95% confidence interval: 0.5, 0.8; P = 4.6 × 10−4). This biologically plausible and highly significant finding provides strong evidence for a true association among Asian women. This study also highlights the value of gene-environment interaction analyses in evaluating genetic factors for complex diseases.

10. Dietary B Vitamin and Methionine Intakes and Breast Cancer Risk Among Chinese Women Methionine, folate, vitamin B6, vitamin B12, niacin, and riboflavin intakes may be related to breast carcinogenesis. These associations may vary by breast cancer type. Using the prospective cohort Shanghai Women's Health Study (1997–2008) including 718 Chinese breast cancer cases, the authors evaluated baseline dietary intake of these factors and breast cancer risk and whether the associations varied by menopausal status and estrogen receptor (ER) and progesterone receptor (PR) status. They estimated associations using hazard ratios and 95% confidence intervals from Cox proportional hazards regression models and stratified analyses by menopausal status and ER/PR status. Lowest quantile of intake was used as the comparison group. For postmenopausal women, dietary intakes of methionine and B vitamins were not associated with breast cancer risk. For premenopausal women, higher intake of folate was associated with decreased breast cancer risk (hazard ratio = 0.58, 95% confidence interval: 0.34, 0.99 for the highest vs. lowest quintile of intake). Only niacin intake was associated with ER+/PR+ breast cancer risk (hazard ratio = 1.62, 95% confidence interval: 1.07, 2.46; P for trend = 0.04 for the highest vs. lowest quartile of intake). Findings support the hypothesis that high folate intake may reduce breast cancer risk and that the association may vary by menopausal and ER/PR status.

11. Long-Term Sedentary Work and the Risk of Subsite-specific Colorectal Cancer Research suggests that sedentary behavior may increase the risk of some chronic diseases. The aims of the study were to examine whether sedentary work is associated with colorectal cancer and to determine whether the association differs by subsite. A total of 918 cases and 1,021 controls participated in a population-based case-control study of colorectal cancer in Western Australia in 2005–2007. Data were collected on lifestyle, physical activity, and lifetime job history. The estimated effects of sedentary work on the risk of cancers of the proximal colon, distal colon, and rectum were analyzed by using multinomial logistic regression. Compared with participants who did not spend any time in sedentary work, participants who spent 10 or more years in sedentary work had almost twice the risk of distal colon cancer (adjusted odds ratio = 1.94, 95% confidence interval: 1.28, 2.93) and a 44% increased risk of rectal cancer (adjusted odds ratio = 1.44, 95% confidence interval: 0.96, 2.18). This association was independent of recreational physical activity and was seen even among the most recreationally active participants. Sedentary work was not associated with the risk of proximal colon cancer. These results suggest that long-term sedentary work may increase the risk of distal colon cancer and rectal cancer.

12. Cutaneous Melanoma and Endogenous Hormonal Factors: A Large French Prospective Study To assess the role of endogenous hormonal factors on melanoma, the authors conducted a prospective analysis of 91,972 French women, aged 40–65 years at inclusion into the Etude Epidémiologiqueauprès de femmes de l’EducationNationale (E3N) cohort. Between 1990 and 2005, 460 melanoma cases were ascertained. Relative risks and 95% confidence intervals were computed by using Cox proportional hazards regression models. Risks of melanoma were reduced in women with ≥15 years at menarche (relative risk (RR) = 0.67, 95% confidence interval (CI): 0.46, 0.97, compared with 13–14 years), irregular menstrual cycles (RR = 0.52, 95% CI: 0.31, 0.89, compared with regular cycles of 25–31 days), <48 years at natural menopause (RR = 0.70, 95% CI: 0.48, 1.02, compared with 48–51 years), and shorter ovulatory life (RR = 0.51, 95% CI: 0.28, 0.91, for <33 years compared with ≥39 years). Modest inverse associations were observed with parity, as well as number of pregnancies and miscarriages. There was no evidence of an association between melanoma risk and age at first birth or pregnancy, age at last birth, time since last birth, breastfeeding duration, age at menstruation regularity, or menopausal status. Results did not significantly differ according to ambient ultraviolet radiation dose and melanoma site or subtype. These findings from a large prospective cohort may suggest a reduced melanoma risk associated with decreased exposure to ovarian hormones.

13. Psychosocial Risk and Correlates of Early Menarche in Mexican-American Girls Mexican-American girls have one of the fastest rates of decline in age at menarche. To date, no study has addressed the role of psychosocial factors on age at menarche in this population. Using data from a longitudinal cohort of Mexican-American girls from the Houston, Texas, metropolitan area recruited in 2005, the authors investigated associations between family life and socioeconomic environment and age at menarche in 523 girls. After adjusting for maternal age at menarche, daughter's age, and body mass index at baseline, perception of family life environment as conflict-prone was significantly associated with an earlier age at menarche (< 11 years). Additionally, there was a 2-fold higher risk (odds ratio = 2.22, 95% confidence interval: 1.12, 4.40) of early menarche among daughters of mothers who were single parents compared with those who were not. Furthermore, girls who matured early had a 2.5-fold increased risk (odds ratio = 2.69, 95% confidence interval: 1.04, 6.96) of experimenting with cigarettes compared with those who had an average-to-late age at menarche (≥ 11 years). This study provides important information regarding the role of family life environment and single parenting on age at menarche in Mexican Americans. Awareness of the impact of the family life environment and fathers’ absence during the early years should be emphasized when addressing early age at menarche across cultures.

14. Extreme Mortality After First Introduction of Measles Virus to the Polynesian Island of Rotuma, 1911 Rotuma is an isolated Polynesian island. In January 1911, most residents of Rotuma (population approximately 2,600) were exposed to measles virus for the first time. The official mortality register documented 491 deaths due to all causes among Rotumans during 1911 (cumulative measles-related mortality: 12.8%); most deaths occurred in April–May and were attributed to measles and its sequelae. Measles-related mortality rates were higher among young children (23.4 per 100 person-years) and young adults (17.1 per 100 person-years) than among adolescents (11.0 per 100 person-years) and older adults (5.6 per 100 person-years); females (16.2 per 100 person-years) died at a higher rate than males (13.2 per 100 person-years). Gastrointestinal complications (75%), not respiratory complications, were the predominant clinical manifestations of fatal measles cases; tuberculosis mortality was unusually high during the year of the epidemic. In 1911, measles-related mortality varied by nearly 3-fold across geographic districts (range, 7.4%–21.6%). The extreme mortality due to measles on Rotuma typifies the experiences of isolated populations after first encounters with measles; it suggests that prior exposures to a narrow range of microbes and genetic homogeneity predispose isolated populations to lethal outcomes when they are first exposed to highly contagious and pathogenic viruses (e.g., measles, influenza).