American Journal of Epidemiology

Common polymorphisms in the N-acetyltransferase 2 gene (NAT2) modify the association between cigarette smoking and bladder cancer and have been hypothesized to determine whether active cigarette smoking increases breast cancer risk. The authors sought to replicate the latter hypothesis in a prospective analysis of 6,900 breast cancer cases and 9,903 matched controls drawn from 6 cohorts (1989–2006) in the National Cancer Institute’s Breast and Prostate Cancer Cohort Consortium. Standardized methods were used to genotype the 3 most common polymorphisms that define NAT2 acetylation phenotype (rs1799930, rs1799931, and rs1801280). In unconditional logistic regression analyses, breast cancer risk was higher in women with more than 20 pack-years of active cigarette smoking than in never smokers (odds ratio (OR) = 1.28, 95% confidence interval (CI): 1.17, 1.39), after controlling for established risk factors other than alcohol consumption and physical inactivity. However, associations were similar for the slow (OR = 1.25, 95% CI: 1.11, 1.39) and rapid/intermediate (OR = 1.24, 95% CI: 1.08, 1.42) acetylation phenotypes, with no evidence of interaction (P = 0.87). These results provide some support for the hypothesis that long-term cigarette smoking may be causally associated with breast cancer risk but underscore the need for caution when interpreting sparse data on gene-environment interactions.

Despite considerable research efforts in specific subpopulations, reliable estimates of the infection attack rates and severity of 2009 influenza A (H1N1) in the general population remain scarce. Such estimates are essential to the tailoring of future control strategies. Therefore, 2 serial population-based serologic surveys were conducted, before and after the 2009 influenza A (H1N1) epidemic, in the Netherlands. Random age-stratified samples were obtained using a 2-stage cluster design. Participants donated blood and completed a questionnaire. Data on sentinel general practitioner-attended influenza-like illness and nationwide hospitalization and mortality were used to assess the severity of infection. The estimated infection attack rates were low in the general population (7.6%, 95% confidence interval: 3.6, 11) but high in children aged 5–19 years (35%, 95% confidence interval: 25, 45). The estimated hospitalization and mortality rates per infection increased significantly with age (5–19 years: 0.042% and 0.00094%, respectively; 20–39 years: 0.12% and 0.0025%; 40–59 years: 0.68% and 0.032%; 60–75 years: >0.81% and >0.068%). The high infection attack rate in children and the very low attack rate in older adults, together with the low severity of illness per infection in children but substantial severity in older adults, produced an epidemic with a low overall impact.

Greening of vacant urban land may affect health and safety. The authors conducted a decade-long difference-in-differences analysis of the impact of a vacant lot greening program in Philadelphia, Pennsylvania, on health and safety outcomes. "Before" and "after" outcome differences among treated vacant lots were compared with matched groups of control vacant lots that were eligible but did not receive treatment. Control lots from 2 eligibility pools were randomly selected and matched to treated lots at a 3:1 ratio by city section. Random-effects regression models were fitted, along with alternative models and robustness checks. Across 4 sections of Philadelphia, 4,436 vacant lots totaling over 7.8 million square feet (about 725,000 m2) were greened from 1999 to 2008. Regression-adjusted estimates showed that vacant lot greening was associated with consistent reductions in gun assaults across all 4 sections of the city (P < 0.001) and consistent reductions in vandalism in 1 section of the city (P < 0.001). Regression-adjusted estimates also showed that vacant lot greening was associated with residents&rsquo; reporting less stress and more exercise in select sections of the city (P < 0.01). Once greened, vacant lots may reduce certain crimes and promote some aspects of health. Limitations of the current study are discussed. Community-based trials are warranted to further test these findings.

Nitrosatable drugs, such as secondary or tertiary amines and amides, form N-nitroso compounds in the presence of nitrite. Various N-nitroso compounds have been associated with neural tube defects in animal models. Using data from the National Birth Defects Prevention Study, the authors examined nitrosatable drug exposure 1 month before and 1 month after conception in 1,223 case mothers with neural tube defect-affected pregnancies and 6,807 control mothers who delivered babies without major congenital anomalies from 1997 to 2005. Nitrite intakes were estimated from mothers&rsquo; responses to a food frequency questionnaire. After adjustment for maternal race/ethnicity, educational level, and folic acid supplementation, case women were more likely than were control women to have taken tertiary amines (odds ratio = 1.60, 95% confidence interval (CI): 1.31, 1.95). This association was strongest with anencephalic births (odds ratio = 1.96, 95% CI: 1.40, 2.73); odds ratios associated with tertiary amines from the lowest tertile of nitrite intake to the highest tertile were 1.16 (95% CI: 0.59, 2.29), 2.19 (95% CI: 1.25, 3.86), and 2.51 (95% CI: 1.45, 4.37), respectively. Odds ratios for anencephaly with nitrosatable drug exposure were reduced among women who also took daily vitamin supplements that contained vitamin C. Prenatal exposure to nitrosatable drugs may increase the risk of neural tube defects, especially in conjunction with a mother&rsquo;s higher dietary intake of nitrites, but vitamin C might modulate this association.

Studies have indicated that children delivered by cesarean section are at an increased risk of developing wheezing and asthma. This could be the result of an altered immune system development due to delayed gut colonization or of increased neonatal respiratory morbidity. The authors examined the associations between delivery by cesarean section and the development of wheezing, asthma, and recurrent lower respiratory tract infections in children up to 36 months of age among 37,171 children in the Norwegian Mother and Child Cohort Study. Generalized linear models were used in the multivariable analysis. Children delivered by cesarean section had an increased likelihood of current asthma at 36 months of age (relative risk = 1.17, 95% confidence interval: 1.03, 1.32), and the association was stronger among children of nonatopic mothers (relative risk = 1.33, 95% confidence interval: 1.12, 1.58). No increased risk of wheezing or recurrent lower respiratory tract infections was seen among children delivered by cesarean section. Findings were similar among children delivered by acute and elective cesarean section. In conclusion, children delivered by cesarean section may have an increased risk of current asthma at 36 months, but residual confounding cannot be excluded. In future prospective studies, investigators should reexamine this association in different age groups.

The psychometric properties of instruments used to measure self-reported experiences of discrimination in epidemiologic studies are rarely assessed, especially regarding construct validity. The authors used 2000&ndash;2001 data from the Coronary Artery Risk Development in Young Adults (CARDIA) Study to examine differential item functioning (DIF) in 2 versions of the Experiences of Discrimination (EOD) Index, an index measuring self-reported experiences of racial/ethnic and gender discrimination. DIF may confound interpretation of subgroup differences. Large DIF was observed for 2 of 7 racial/ethnic discrimination items: White participants reported more racial/ethnic discrimination for the "at school" item, and black participants reported more racial/ethnic discrimination for the "getting housing" item. The large DIF by race/ethnicity in the index for racial/ethnic discrimination probably reflects item impact and is the result of valid group differences between blacks and whites regarding their respective experiences of discrimination. The authors also observed large DIF by race/ethnicity for 3 of 7 gender discrimination items. This is more likely to have been due to item bias. Users of the EOD Index must consider the advantages and disadvantages of DIF adjustment (omitting items, constructing separate measures, and retaining items). The EOD Index has substantial usefulness as an instrument that can assess self-reported experiences of discrimination.

To date, Web-based 24-hour recalls have not been validated using objective biomarkers. From 2006 to 2009, the validity of 6 Web-based DietDay 24-hour recalls was tested among 115 black and 118 white healthy adults from Los Angeles, California, by using the doubly labeled water method, and the results were compared with the results of the Diet History Questionnaire, a food frequency questionnaire developed by the National Cancer Institute. The authors performed repeated measurements in a subset of 53 subjects approximately 6 months later to estimate the stability of the doubly labeled water measurement. The attenuation factors for the DietDay recall were 0.30 for blacks and 0.26 for whites. For the Diet History Questionnaire, the attenuation factors were 0.15 and 0.17 for blacks and whites, respectively. Adjusted correlations between true energy intake and the recalls were 0.50 and 0.47 for blacks and whites, respectively, for the DietDay recall. For the Diet History Questionnaire, they were 0.34 and 0.36 for blacks and whites, respectively. The rate of underreporting of more than 30% of calories was lower with the recalls than with the questionnaire (25% and 41% vs. 34% and 52% for blacks and whites, respectively). These findings suggest that Web-based DietDay dietary recalls offer an inexpensive and widely accessible dietary assessment alternative, the validity of which is equally strong among black and white adults. The validity of the Web-administered recall was superior to that of the paper food frequency questionnaire.

A prospective survey of social mixing patterns relevant to respiratory disease transmission by large droplets (e.g., influenza) or small droplet nuclei (e.g., tuberculosis) was performed in a South African township in 2010. A total of 571 randomly selected participants recorded the numbers, times, and locations of close contacts (physical/nonphysical) and indoor casual contacts met daily. The median number of physical contacts was 12 (interquartile range (IQR), 7&ndash;18), the median number of close contacts was 20 (IQR, 13&ndash;29), and the total number of indoor contacts was 30 (IQR, 12&ndash;54). Physical and close contacts were most frequent and age-associative in youths aged 5&ndash;19 years. Numbers of close contacts were 40% higher than in corresponding populations in industrialized countries (P < 0.001). This may put township communities at higher risk for epidemics of acute respiratory illnesses. Simulations of an acute influenza epidemic predominantly involved adolescents and young adults, indicating that control strategies should be directed toward these age groups. Of all contacts, 86.2% occurred indoors with potential exposure to respiratory droplet nuclei, of which 27.2%, 20.1%, 20.0%, and 8.0% were in transport, own household, cr&egrave;che/school, and work locations, respectively. Indoor contact time was long in households and short during transport. High numbers of indoor contacts and intergenerational mixing in households and transport may contribute to exceptionally high rates of tuberculosis transmission reported in the community.

The authors describe a statistical method of combining self-reports and biomarkers that, with adequate control for confounding, will provide nearly unbiased estimates of diet-disease associations and a valid test of the null hypothesis of no association. The method is based on regression calibration. In cases in which the diet-disease association is mediated by the biomarker, the association needs to be estimated as the total dietary effect in a mediation model. However, the hypothesis of no association is best tested through a marginal model that includes as the exposure the regression calibration-estimated intake but not the biomarker. The authors illustrate the method with data from the Carotenoids and Age-Related Eye Disease Study (2001--2004) and show that inclusion of the biomarker in the regression calibration-estimated intake increases the statistical power. This development sheds light on previous analyses of diet-disease associations reported in the literature.

In this article, the authors demonstrate a time-series analysis based on a hierarchical Bayesian model of a Poisson outcome with an excessive number of zeroes. The motivating example for this analysis comes from the intensive care unit (ICU) of an urban university teaching hospital (New Haven, Connecticut, 2002&ndash;2004). Studies of medication use among older patients in the ICU are complicated by statistical factors such as an excessive number of zero doses, periodicity, and within-person autocorrelation. Whereas time-series techniques adjust for autocorrelation and periodicity in outcome measurements, Bayesian analysis provides greater precision for small samples and the flexibility to conduct posterior predictive simulations. By applying elements of time-series analysis within both frequentist and Bayesian frameworks, the authors evaluate differences in shift-based dosing of medication in a medical ICU. From a small sample and with adjustment for excess zeroes, linear trend, autocorrelation, and clinical covariates, both frequentist and Bayesian models provide evidence of a significant association between a specific nursing shift and dosing level of a sedative medication. Furthermore, the posterior distributions from a Bayesian random-effects Poisson model permit posterior predictive simulations of related results that are potentially difficult to model.

In choosing covariates for adjustment or inclusion in propensity score analysis, researchers must weigh the benefit of reducing confounding bias carried by those covariates against the risk of amplifying residual bias carried by unmeasured confounders. The latter is characteristic of covariates that act like instrumental variables&mdash;that is, variables that are more strongly associated with the exposure than with the outcome. In this issue of the Journal (Am J Epidemiol. 2011;174(11):1213&ndash;1222), Myers et al. compare the bias amplification of a near-instrumental variable with its bias-reducing potential and suggest that, in practice, the latter outweighs the former. The author of this commentary sheds broader light on this comparison by considering the cumulative effects of conditioning on multiple covariates and showing that bias amplification may build up at a faster rate than bias reduction. The author further derives a partial order on sets of covariates which reveals preference for conditioning on outcome-related, rather than exposure-related, confounders.

Recent theoretical studies have shown that conditioning on an instrumental variable (IV), a variable that is associated with exposure but not associated with outcome except through exposure, can increase both bias and variance of exposure effect estimates. Although these findings have obvious implications in cases of known IVs, their meaning remains unclear in the more common scenario where investigators are uncertain whether a measured covariate meets the criteria for an IV or rather a confounder. The authors present results from two simulation studies designed to provide insight into the problem of conditioning on potential IVs in routine epidemiologic practice. The simulations explored the effects of conditioning on IVs, near-IVs (predictors of exposure that are weakly associated with outcome), and confounders on the bias and variance of a binary exposure effect estimate. The results indicate that effect estimates which are conditional on a perfect IV or near-IV may have larger bias and variance than the unconditional estimate. However, in most scenarios considered, the increases in error due to conditioning were small compared with the total estimation error. In these cases, minimizing unmeasured confounding should be the priority when selecting variables for adjustment, even at the risk of conditioning on IVs.

Dialysis has been used in the treatment of patients with end-stage renal disease since the 1960s. Recently, several large observational studies have been conducted to assess whether early initiation of dialysis prolongs survival, as compared with late initiation. However, these studies have used analytic approaches which are likely to suffer from either lead-time bias or immortal-time bias. In this paper, the authors demonstrate that recently developed methods in the causal inference literature can be used to avoid both types of bias and accurately estimate the ideal time for dialysis initiation from observational data. This is illustrated using data from a nationwide population-based cohort of patients with chronic kidney disease in Sweden (1996&ndash;2003).

Estimates of additive interaction from case-control data are often obtained by logistic regression; such models can also be used to adjust for covariates. This approach to estimating additive interaction has come under some criticism because of possible misspecification of the logistic model: If the underlying model is linear, the logistic model will be misspecified. The authors propose an inverse probability of treatment weighting approach to causal effects and additive interaction in case-control studies. Under the assumption of no unmeasured confounding, the approach amounts to fitting a marginal structural linear odds model. The approach allows for the estimation of measures of additive interaction between dichotomous exposures, such as the relative excess risk due to interaction, using case-control data without having to rely on modeling assumptions for the outcome conditional on the exposures and covariates. Rather than using conditional models for the outcome, models are instead specified for the exposures conditional on the covariates. The approach is illustrated by assessing additive interaction between genetic and environmental factors using data from a case-control study.

One of the many challenges hindering the global response to the human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) epidemic is the difficulty of collecting reliable information about the populations most at risk for the disease. Thus, the authors empirically assessed a promising new method for estimating the sizes of most at-risk populations: the network scale-up method. Using 4 different data sources, 2 of which were from other researchers, the authors produced 5 estimates of the number of heavy drug users in Curitiba, Brazil. The authors found that the network scale-up and generalized network scale-up estimators produced estimates 5&ndash;10 times higher than estimates made using standard methods (the multiplier method and the direct estimation method using data from 2004 and 2010). Given that equally plausible methods produced such a wide range of results, the authors recommend that additional studies be undertaken to compare estimates based on the scale-up method with those made using other methods. If scale-up-based methods routinely produce higher estimates, this would suggest that scale-up-based methods are inappropriate for populations most at risk of HIV/AIDS or that standard methods may tend to underestimate the sizes of these populations.

Family-based designs protect analyses of genetic effects from bias that is due to population stratification. Investigators have assumed that this robustness extends to assessments of gene-environment interaction. Unfortunately, this assumption fails for the common scenario in which the genotyped variant is related to risk through linkage with a causative allele. Bias also plagues other methods of assessment of gene-environment interaction. When testing against multiplicative joint effects, the case-only design offers excellent power, but it is invalid if genotype and exposure are correlated in the population. The authors describe 4 mechanisms that produce genotype-exposure dependence: exposure-related genetic population stratification, effects of family history on behavior, genotype effects on exposure, and selective attrition. They propose a sibling-augmented case-only (SACO) design that protects against the former 2 mechanisms and is therefore valid for studying young-onset disease in which genotype does not influence exposure. A SACO design allows the ascertainment of genotype and exposure for cases and exposure for 1 or more unaffected siblings selected randomly. Conditional logistic regression permits assessment of exposure effects and gene-environment interactions. Via simulations, the authors compare the likelihood-based inference on interactions using the SACO design with that based on other designs. They also show that robust analyses of interactions using tetrads or disease-discordant sibling pairs are equivalent to analyses using the SACO design.

Because the incidence of human immunodeficiency virus (HIV) infection is difficult to measure directly, prevalence trends often serve to track epidemiologic changes. Adult HIV prevalence in open population cohort studies, however, reflects changes in incidence, population factors (migration, deaths, and aging), and survey coverage. Data from an open cohort in rural Uganda enabled estimation of the contribution of these factors to prevalence trends from 1989 to 2007. New infections within this cohort represented on average 44% of new prevalent cases per year. Other factors affecting changes in prevalence included migration and death. Migrants and mobile people (those who leave and return to the study area) are in a higher-risk group and thus can affect prevalence trends. Incidence of HIV infection among mobile people was 2&ndash;4 times greater than among stable residents. The importance of mortality is shown by the rise in prevalence from 6.8% in 2005 to 7.4% in 2007, which was accompanied by a fall in mortality among HIV-infected participants (8.7% of HIV-infected in 2005, 5.2% in 2006, and 4.3% in 2007). Assessing HIV epidemic trends through prevalence requires consideration of population factors. Measuring HIV incidence directly remains the most accurate measure of trends with which to monitor the effect of intervention activities and should complement strategies such as national prevalence surveys.

Studies published in the last 3 decades have demonstrated global human exposure to polybrominated diphenyl ether (PBDE) flame retardants. A growing body of literature suggests that PBDEs may disrupt thyroid hormone homeostasis. Although thyroid hormones play an essential role in brain development, few studies have investigated relations between prenatal exposure to PBDEs and neonatal thyroid hormone levels, and none have measured thyroid-stimulating hormone (TSH) levels in neonates. The authors measured 10 PBDE congeners in serum collected between October 1999 and October 2000 from 289 pregnant women living in California's Salinas Valley and abstracted TSH levels from their children's medical records. Individual PBDE congeners showed null or weak nonsignificant inverse relations with neonatal TSH. Total serum PBDE was not associated with neonatal TSH (&beta; = 0.00, 95% confidence interval: &ndash;0.06, 0.06). Except for brominated diphenyl ether 153, a higher serum PBDE level was related to elevated odds of high TSH (&ge;80th percentile), but associations were not statistically significant. Associations were not modified by infant sex, age at TSH measurement, maternal serum polychlorinated biphenyl concentration, or mode of delivery. Results were robust to sensitivity analysis. The authors found no conclusive evidence that prenatal exposure to PBDEs at levels similar to those of the general US population is related to neonatal TSH.

There is increasing evidence that in utero effects of excessive gestational weight gain may result in increased weight in children; however, studies have not controlled for shared genetic or environmental factors between mothers and children. Using 2,758 family groups from the Collaborative Perinatal Project, the authors examined the association of maternal prepregnancy body mass index (BMI) and gestational weight gain on child BMI at age 4 years using both conventional generalized estimating equations and fixed-effects models that account for shared familial factors. With generalized estimating equations, prepregnancy BMI and gestational weight gain had similar associations with the child BMI z score (&beta; = 0.09 units, 95% confidence interval (CI): 0.08, 0.11; and &beta; = 0.07 units, 95% CI: 0.04, 0.11, respectively. However, fixed effects resulted in null associations for both prepregnancy BMI (&beta; = 0.03 units, 95% CI: &ndash;0.01, 0.07) and gestational weight gain (&beta; = 0.03 units, 95% CI: &ndash;0.02, 0.08) with child BMI z score at age 4 years. The positive association between gestational weight gain and child BMI at age 4 years may be explained by shared family characteristics (e.g., genetic, behavioral, and environmental factors) rather than in utero programming. Future studies should continue to evaluate the relative roles of important familial and environmental factors that may influence BMI and obesity in children.

Never or curtailed lactation has been associated with an increased risk for incident hypertension, but the effect of exclusive breastfeeding is unknown. The authors conducted an observational cohort study of 55,636 parous women in the US Nurses&rsquo; Health Study II. From 1991 to 2005, participants reported 8,861 cases of incident hypertension during 660,880 person-years of follow-up. Never or curtailed lactation was associated with an increased risk of incident hypertension. Compared with women who breastfed their first child for &ge;12 months, women who did not breastfeed were more likely to develop hypertension (hazard ratio (HR) = 1.27, 95% confidence interval (CI): 1.18, 1.36), adjusting for family history and lifestyle covariates. Women who never breastfed were more likely to develop hypertension than women who exclusively breastfed their first child for &ge;6 months (HR = 1.29, 95% CI: 1.20, 1.40). The authors found similar results for women who had never breastfed compared with those who had breastfed each child for an average of &ge;12 months (HR = 1.22, 95% CI: 1.13, 1.32). In conclusion, never or curtailed lactation was associated with an increased risk of incident maternal hypertension, compared with the recommended &ge;6 months of exclusive or &ge;12 months of total lactation per child, in a large cohort of parous women.

The sojourn time of preclinical colorectal cancer is a critical parameter in modeling effectiveness and cost-effectiveness of colorectal cancer screening. For ethical reasons, it cannot be observed directly, and available estimates are based mostly on relatively small historic data sets that do not include differentiation by age and sex. The authors derived sex- and age-specific estimates (age groups: 55&ndash;59, 60&ndash;64, 65&ndash;69, 70&ndash;74, 75&ndash;79, and &ge;80 years) of mean sojourn time, combining data from the German national screening colonoscopy registry (based on 1.88 million records) and data from population-based cancer registries (population base: 37.9 million people) for the years 2003&ndash;2006. Estimates of mean sojourn time were similar for both sexes and all age groups and ranged from 4.5 years (95% confidence interval: 4.1, 4.8) to 5.8 years (95% confidence interval: 5.3, 6.3) for the subgroups assessed. Sensitivity analyses indicated that mean sojourn time might be approximately 1.5 years longer if colorectal cancer prevalence in nonparticipants of screening colonoscopy is 20% lower than prevalence in participants or 1 year shorter if it exceeds the prevalence in participants by 20%. This study provides, for the first time, precise estimates of sojourn time by age and sex, and it suggests that sojourn times are remarkably consistent across age groups and in both sexes.

A large body size may differentially influence risk of colorectal cancer (CRC) by anatomic location. The Netherlands Cohort Study includes 120,852 men and women aged 55&ndash;69 years who self-reported weight, height, and trouser/skirt size at baseline (1986), as well as weight at age 20 years. Derived variables included body mass index (BMI; weight (kg)/height (m)2), BMI at age 20 years, and BMI change. After 16.3 years of follow-up (1986&ndash;2002), 2,316 CRC cases were available for case-cohort analysis. In men, the highest risk estimates were observed for body fat (per 5-unit increase in BMI, hazard ratio (HR) = 1.25, 95% confidence interval (CI): 1.05, 1.46; for highest quintile of trouser size vs. lowest, HR = 1.63, 95% CI: 1.17, 2.29 (P-trend = 0.02)) and appeared more closely associated with distal colon tumors (for BMI (5-unit increase), HR = 1.42, 95% CI: 1.13, 1.79; for highest quintile of trouser size, HR = 2.56, 95% CI: 1.55, 4.24 (P-trend < 0.01)) than with proximal colon or rectal tumors. In women, body fat was not associated with CRC risk unless it was considered simultaneously with physical activity; a large trouser/skirt size and a low level of physical activity increased risk for all subtypes. Height was associated with risk of CRC, especially distal colon tumors (highest quintile vs. lowest: HR = 1.53, 95% CI: 1.03, 2.27; P-trend = 0.05), in women only.

Obesity is a well-established risk factor for endometrial cancer, the most common gynecologic malignancy. Recent genome-wide association studies (GWAS) have identified multiple genetic markers for obesity. The authors evaluated the association of obesity-related single nucleotide polymorphisms (SNPs) with endometrial cancer using GWAS data from their recently completed study, the Shanghai Endometrial Cancer Genetics Study, which comprised 832 endometrial cancer cases and 2,049 controls (1996&ndash;2005). Thirty-five SNPs previously associated with obesity or body mass index (BMI; weight (kg)/height (m)2) at a minimum significance level of &le;5 x 10&ndash;7 in the US National Human Genome Research Institute's GWAS catalog (http://genome.gov/gwastudies) and representing 26 unique loci were evaluated by either direct genotyping or imputation. The authors found that for 22 of the 26 unique loci tested (84.6%), the BMI-associated risk variants were present at a higher frequency in cases than in population controls (P = 0.0003). Multiple regression analysis showed that 9 of 35 BMI-associated variants, representing 7 loci, were significantly associated (P &le; 0.05) with the risk of endometrial cancer; for all but 1 SNP, the direction of association was consistent with that found for BMI. For consistent SNPs, the allelic odds ratios ranged from 1.15 to 1.29. These 7 loci are in the SEC16B/RASAL, TMEM18, MSRA, SOX6, MTCH2, FTO, and MC4R genes. The associations persisted after adjustment for BMI, suggesting that genetic markers of obesity provide value in addition to BMI in predicting endometrial cancer risk.

Existing evidence linking residential proximity to food establishments with body mass index (BMI; weight (kg)/height (m)2) has been inconclusive. In this study, the authors assessed the relation between BMI and proximity to food establishments over a 30-year period among 3,113 subjects in the Framingham Heart Study Offspring Cohort living in 4 Massachusetts towns during 1971&ndash;2001. The authors used novel data that included repeated measures of BMI and accounted for residential mobility and the appearance and disappearance of food establishments. They calculated proximity to food establishments as the driving distance between each subject&rsquo;s residence and nearby food establishments, divided into 6 categories. The authors used cross-classified linear mixed models to account for time-varying attributes of individuals and residential neighborhoods. Each 1-km increase in distance to the closest fast-food restaurant was associated with a 0.11-unit decrease in BMI (95% credible interval: &ndash;0.20, &ndash;0.04). In sex-stratified analyses, this association was present only for women. Other aspects of the food environment were either inconsistently associated or not at all associated with BMI. Contrary to much prior research, the authors did not find a consistent relation between access to fast-food restaurants and individual BMI, necessitating a reevaluation of policy discussions on the anticipated impact of the food environment on weight gain.

The ecologic study design is routinely used by epidemiologists in spite of its limitations. It is presently unknown how well the challenges of the design are dealt with in epidemiologic research. The purpose of this bibliometric review was to critically evaluate the characteristics, statistical methods, and reporting of results of modern cross-sectional ecologic papers. A search through 6 major epidemiology journals identified all cross-sectional ecologic studies published since January 1, 2000. A total of 125 articles met the inclusion requirements and were assessed via common evaluative criteria. It was found that a considerable number of cross-sectional ecologic studies use unreliable methods or contain statistical oversights; most investigators who adjusted their outcomes for age or sex did so improperly (64%), statistical validity was a potential issue for 20% of regression models, and simple linear regression was the most common analytic approach (31%). Many authors omitted important information when discussing the ecologic nature of their study (31%), the choice of study design (58%), and the susceptibility of their research to the ecological fallacy (49%). These results suggest that there is a need for an international set of guidelines that standardizes reporting on ecologic studies. Additionally, greater attention should be given to the relevant biostatistical literature.

Selective prescribing of conventional antipsychotic medication (APM) to frailer patients is thought to have led to overestimation of the association with mortality in pharmacoepidemiologic studies relying on claims data. The authors assessed the validity of different analytic techniques to address such confounding. The cohort included 82,012 persons initiating APM use after admission to a nursing home in 45 states with 2001&ndash;2005 Medicaid/Medicare data, linked to clinical data (Minimum Data Set) and institutional characteristics. The authors compared the association between APM class and 180-day mortality with multivariate outcome modeling, propensity score (PS) adjustment, and instrumental variables. The unadjusted risk difference (per 100 patients) of 10.6 (95% confidence interval (CI): 9.4, 11.7) comparing use of conventional medication with atypical APM was reduced to 7.8 (95% CI: 6.6, 9.0) and 7.0 (95% CI: 5.8, 8.2) after PS adjustment and high-dimensional PS (hdPS) adjustment, respectively. Results were similar in analyses limited to claims-based Medicaid /Medicare variables (risk difference = 8.2 for PS, 7.1 for hdPS). Instrumental-variable estimates were imprecise (risk difference = 8.8, 95% CI: &ndash;1.3, 19.0) because of the weak instrument. These results suggest that residual confounding has a relatively small impact on the effect estimate and that hdPS methods based on claims alone provide estimates at least as good as those from conventional analyses using claims enriched with clinical information.

The linear mixed model (LMM), which is routinely used to describe change in outcomes over time and its association with risk factors, assumes that a unit change in any predictor is associated with a constant change in the outcome. When it is used on psychometric tests, this assumption may not hold. Indeed, psychometric tests usually suffer from ceiling and/or floor effects and curvilinearity (i.e., varying sensitivity to change). The authors aimed to determine the consequences of such misspecification when evaluating predictors of cognitive decline. As an alternative to the LMM, they considered 2 mixed models based on latent processes that handle discrete and bounded outcomes. Model differences are illustrated here using data on 4 psychometric tests from the Personnes Ag&eacute;es QUID (PAQUID) Study (1989&ndash;2004). The type I error of the Wald test for risk-factor regression parameters was formally assessed in a simulation study. It demonstrated that type I errors in the LMM could be dramatically inflated for some tests, such that spurious associations with risk factors were found. In particular, confusion between effects on mean level and effects on change over time was highlighted. The authors recommend use of the alternative mixed models when studying psychometric tests and more generally quantitative scales (quality of life, activities of daily living).

Mendelian randomization studies typically have low power. Where there are several valid candidate genetic instruments, precision can be gained by using all the instruments available. However, sporadically missing genetic data can offset this gain. The authors describe 4 Bayesian methods for imputing the missing data based on a missing-at-random assumption: multiple imputations, single nucleotide polymorphism (SNP) imputation, latent variables, and haplotype imputation. These methods are demonstrated in a simulation study and then applied to estimate the causal relation between C-reactive protein and each of fibrinogen and coronary heart disease, based on 3 SNPs in British Women&rsquo;s Heart and Health Study participants assessed at baseline between May 1999 and June 2000. A complete-case analysis based on all 3 SNPs was found to be more precise than analyses using any 1 SNP alone. Precision is further improved by using any of the 4 proposed missing data methods; the improvement is equivalent to about a 25% increase in sample size. All methods gave similar results, which were apparently not overly sensitive to violation of the missing-at-random assumption. Programming code for the analyses presented is available online.