American Journal of Epidemiology

1. American Journal of Epidemiology October 15, 2011 174 (8)

2. Musculoskeletal Complaints Among 11-Year-Old Children and Associated Factors The PIAMA Birth Cohort Study Musculoskeletal complaints (MSC) are common among children, often persist into adolescence, and increase the risk of MSC in adulthood. Knowledge regarding determinants of MSC among children is limited. The aim of this study was to determine the prevalence of MSC at age 11 years and to examine associations with sociodemographic factors, growth and development factors, mental health, tiredness, and lifestyle. Data from a Netherlands birth cohort study, the Prevention and Incidence of Asthma and Mite Allergy (PIAMA) Study (n = 2,638), were used (1996–2009). MSC were defined as complaints about the back, an upper extremity, a lower extremity, or any of these sites. Logistic regression analyses using a forward stepwise procedure were performed on multiply imputed data. The 1-year period prevalences of back, upper extremity, and lower extremity complaints that lasted at least 1 month were 2.8%, 4.8%, and 10.9%, respectively. Only poorer mental health was consistently associated with all 3 types of complaints. Poorer mental health, daytime tiredness, early pubertal development, being physically active at age 11 years, and weight-for-height z score were associated with having any MSC. This study showed that MSC, especially lower extremity complaints, are common among 11-year old-children and that only poorer mental health status is associated with MSC at all anatomic sites.

3. Association of Prenatal Exposure to Polybrominated Diphenyl Ethers and Infant Birth Weight Polybrominated diphenyl ethers (PBDEs) are a class of persistent compounds that have been used as flame retardants in vehicles, household furnishings, and consumer electronics. This study examined whether concentrations of PBDEs in maternal serum during pregnancy were associated with infant birth weight, length, head circumference, and length of gestation. Participants were pregnant women (n = 286) enrolled in the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) Study, a longitudinal cohort study of low-income, predominantly Mexican families living in the Salinas Valley, California. Blood samples were collected near the 26th week of pregnancy in 1999–2000, and concentrations of 10 PBDE congeners (BDE-17, -28, -47, -66, -85, -99, -100, -153, -154, and -183) were measured. Multiple linear regression models were used to investigate the association of lipid-adjusted, log10-transformed PBDE concentrations and birth outcome. In adjusted analyses, negative associations with birth weight were seen with BDE-47 (β = −115 g, 95% confidence interval (CI): −229, −2), BDE-99 (β = −114 g, 95% CI: −225, −4), and BDE-100 (β = −122 g, 95% CI: −235, −9). These findings were diminished slightly and were no longer statistically significant when maternal weight gain was included in the models. PBDE congeners were not associated with birth length, head circumference, or gestational duration.

4. Perfluoroalkyl Chemicals and Chronic Kidney Disease in US Adults Chronic kidney disease (CKD) is a major public health problem. Identifying novel risk factors for CKD, including widely prevalent environmental exposures, is therefore important. Perfluoroalkyl chemicals (PFCs), including perfluorooctanoic acid and perfluorooctane sulfonate, are manmade chemicals that have been detected in the blood of more than 98% of the US population. Results from experimental animal studies have suggested that an association between PFCs and CKD is plausible. However, in humans, the relation between serum PFCs and CKD has not been examined. The authors examined the relation of serum PFCs and CKD in 4,587 adult participants (51.1% women) from the combined 1999–2000 and 2003–2008 cycles of the National Health and Nutritional Examination Survey for whom PFC measurements were available. The main outcome was CKD, defined as a glomerular filtration rate of less than 60 mL/minute/1.73 m2. The authors found that serum levels of PFCs, including perfluorooctanoic acid and perfluorooctane sulfonate, were positively associated with CKD. This association was independent of confounders such as age, sex, race/ethnicity, body mass index, diabetes, hypertension, and serum cholesterol level. Compared with subjects in quartile 1 (referent), the multivariable odds ratio for CKD among subjects in quartile 4 was 1.73 (95% confidence interval: 1.04, 2.88; P for trend = 0.015) for perfluorooctanoic acid and 1.82 (95% confidence interval: 1.01, 3.27; P for trend = 0.019) for perfluorooctane sulfonate. The present results suggest that elevated PFC levels are associated with CKD.

5. Joint Associations Between Genetic Variants and Reproductive Factors in Glioma Risk Among Women In a pooled analysis of 4 US epidemiologic studies (1993–2001), the authors evaluated the role of 5 female reproductive factors in 357 women with glioma and 822 controls. The authors further evaluated the independent association between 5 implicated gene variants and glioma risk among the study population, as well as the joint associations of female reproductive factors (ages at menarche and menopause, menopausal status, use of oral contraceptives, and menopausal hormone therapy) and these gene variants on glioma risk. Risk estimates were calculated as odds ratios and 95% confidence intervals that were adjusted for age, race, and study. Three of the gene variants (rs4295627, a variant of CCDC26; rs4977756, a variant of CDKN2A and CDKN2B; and rs6010620, a variant of RTEL1) were statistically significantly associated with glioma risk in the present population. Compared with women who had an early age at menarche (<12 years of age), those who reported menarche at 12–13 years of age or at 14 years of age or older had a 1.7-fold higher risk and a 1.9-fold higher risk of glioma, respectively (P for trend = 0.009). Postmenopausal women and women who reported ever having used oral contraceptives had a decreased risk of glioma. The authors did not observe joint associations between these reproductive characteristics and the implicated glioma gene variants. These results require replication, but if confirmed, they would suggest that the gene variants that have previously been implicated in the development of glioma are unlikely to act through the same hormonal mechanisms in women.

6. Body Size Across the Life Course, Mammographic Density, and Risk of Breast Cancer Adult body mass index (BMI) is inversely associated with premenopausal breast cancer risk, and childhood and adolescent body size is inversely associated with breast cancer risk in pre- and postmenopausal women. Breast density is inversely related to body size and may play a role in the association of body size with breast cancer risk. The authors conducted a nested case-control study including 1,528 cases and 2,844 controls from the Nurses’ Health Study (1989–2004) and Nurses’ Health Study II (1996–2003). Prior to breast cancer diagnosis, participants reported their body fatness during childhood and adolescence, BMI at age 18 years, and current BMI. Mammographic density was measured by using a computer-assisted thresholding method. The inverse association between adult BMI and premenopausal breast cancer (for BMI ≥30 vs. BMI 20–22.4, odds ratio = 0.64, 95% confidence interval: 0.38, 1.06) (Ptrend = 0.36) became positive after adjustment for mammographic density (odds ratio = 1.28, 95% confidence interval: 0.72, 2.30) (Ptrend = 0.07). Conversely, the inverse association between childhood and adolescent body size and breast cancer risk remained after adjustment for mammographic density. The inverse association between adult BMI and premenopausal breast cancer risk may be partially due to negative confounding by mammographic density. Conversely, mammographic density does not appear to explain the inverse association between childhood and adolescent body fatness and breast cancer risk.

7. Interaction Between Smoking and Obesity and the Risk of Developing Breast Cancer Among Postmenopausal Women The Women's Health Initiative Observational Study Obesity is a well-established risk factor for postmenopausal breast cancer. Recent studies suggest that smoking increases the risk of breast cancer. However, the effect of co-occurrence of smoking and obesity on breast cancer risk remains unclear. A total of 76,628 women aged 50–79 years enrolled in the Women's Health Initiative Observational Study were followed through August 14, 2009. Cox proportional hazards regression models were used to estimate hazard ratios and 95% confidence intervals. Over an average 10.3 years of follow-up, 3,378 incident cases of invasive breast cancer were identified. The effect of smoking on the risk of developing invasive breast cancer was modified significantly by obesity status among postmenopausal women, regardless of whether the obesity status was defined by body mass index (Pinteraction = 0.01) or waist circumference (Pinteraction = 0.02). A significant association between smoking and breast cancer risk was noted in nonobese women (hazard ratio = 1.25, 95% confidence interval: 1.05, 1.47) but not in obese women (hazard ratio = 0.96, 95% confidence interval: 0.69, 1.34). In conclusion, this study suggests that the effect of smoking exposure on breast cancer risk was modified by obesity among postmenopausal women. The modification effect did not differ by general versus abdominal obesity.

8. Obesity and Cannabis Use: Results From 2 Representative National Surveys The role of cannabis and endocannabinoids in appetite regulation has been extensively studied, but the association of cannabis use with weight in the general population is not known. The authors used data from 2 representative epidemiologic studies of US adults aged 18 years or older, the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC; 2001–2002) and the National Comorbidity Survey–Replication (NCS-R; 2001–2003), to estimate the prevalence of obesity as a function of cannabis use. The adjusted prevalences of obesity in the NESARC and the NCS-R were 22.0% and 25.3%, respectively, among participants reporting no use of cannabis in the past 12 months and 14.3% and 17.2%, respectively, among participants reporting the use of cannabis at least 3 days per week. These differences were not accounted for by tobacco smoking status. Additionally, after adjustment for sex and age, the use of cannabis was associated with body mass index differences in both samples. The authors conclude that the prevalence of obesity is lower in cannabis users than in nonusers.

9. Body Mass Index in Young Adulthood and Premature Death: Analyses of the US National Health Interview Survey Linked Mortality Files Knowledge of the association between body mass index (weight (kg)/height (m)2) and premature death in young adulthood is very limited, especially for specific causes of death. Using the US National Health Interview Survey linked mortality files, the authors examined the relation between body mass index and premature death from all causes, cardiovascular disease (CVD), and cancer among 112,328 persons aged 18–39 years who participated in the National Health Interview Survey in the years 1987, 1988, and 1990–1995. During an average of 16 years of follow-up (ending on December 31, 2006), there were 3,178 deaths: 573 from CVD and 733 from cancer. Hazard ratios and 95% confidence intervals were estimated using multivariate proportional hazards models adjusting for age, gender, race/ethnicity, education, and smoking status. In analyses restricted to participants who had never smoked, the hazard ratios for death from all causes were 1.07 (95% confidence interval (CI): 0.91, 1.26) for overweight participants, 1.41 (95% CI: 1.16, 1.73) for obese participants, and 2.46 (95% CI: 1.91, 3.16) for extremely obese participants, compared with those of normal weight. Monotonically increasing risks for excess body weight were also observed for deaths from cancer and CVD. The associations found in this young cohort were much stronger than those in middle-aged or older populations.

10. Use of Penicillin and Other Antibiotics and Risk of Multiple Sclerosis: A Population-based Case-Control Study A 2006 study from the United Kingdom found that penicillin use may decrease the risk of multiple sclerosis (MS). To confirm this finding, the authors conducted a nationwide case-control study in Denmark, using the Danish Multiple Sclerosis Registry to identify 3,259 patients with MS onset from 1996 to 2008, and selected 10 population controls per case (n = 32,590), matched on sex and age. Through the National Prescription Database, prescriptions for antibiotics redeemed from 1995 to 2008 and before the date of first MS symptom/index date were identified. Conditional logistic regression analysis was used to compute odds ratios associating antibiotic use with MS occurrence. In total, 1,922 patients (59%) redeemed penicillin prescriptions before the index date and 2,292 (70%) redeemed any type of antibiotic prescription. Penicillin use was associated with an increased risk of MS (odds ratio = 1.21, 95% confidence interval: 1.10, 1.27). Use of any type of antibiotic was similarly associated with an increased risk of MS (odds ratio = 1.41, 95% confidence interval: 1.29, 1.53). The odds ratios for different types of antibiotics ranged between 1.08 and 1.83. Thus, this study found that penicillin use and use of other antibiotics were similarly associated with increased risk of MS, suggesting that the underlying infections may be causally associated with MS.

11. The Effect of Including Cystatin C or Creatinine in a Cardiovascular Risk Model for Asymptomatic Individuals The Multi-Ethnic Study of Atherosclerosis The authors studied the incremental value of adding serum cystatin C or creatinine to the Framingham risk score variables (FRSVs) for the prediction of incident cardiovascular disease (CVD) among 6,653 adults without clinical CVD utilizing the Multi-Ethnic Study of Atherosclerosis (2000–2008). CVD events included coronary heart disease, heart failure, stroke, and peripheral arterial disease. Variables were transformed to yield optimal prediction of 6-year CVD events in sex-stratified models with FRSVs alone, FRSVs + cystatin C, and FRSVs + creatinine. Risk prediction in the 3 models was assessed by using the C statistic, and net reclassification improvement was calculated. The mean ages were 61.9 and 64.6 years for individuals with and without diabetes, respectively. After 6 years of follow-up, 447 (7.2%) CVD events occurred. In the total cohort, no significant change in the C statistic was noted with FRSVs + cystatin C and FRSVs + creatinine compared with FRSVs alone, and net reclassification improvement for CVD risk was extremely small and not significant with the addition of cystatin C or creatinine to FRSVs. Similar findings were noted after stratifying by baseline presence of diabetes. In conclusion, the addition of cystatin C or serum creatinine to FRSVs does not improve CVD risk prediction among adults without clinical CVD.

12. Free Protein S Level as a Risk Factor for Coronary Heart Disease and Stroke in a Prospective Cohort Study of Healthy United Kingdom Men Plasma protein S (PS) levels are reportedly low in patients with venous thrombosis but high in coronary heart disease (CHD) patients. The authors examined the association between free PS concentration and CHD or stroke risk and assessed risk in combination with C-reactive protein (CRP) levels. Free PS concentration was determined in 6 annual visits among 3,052 middle-aged (49–64 years) United Kingdom men from the Second Northwick Park Heart Study, with 297 CHD events from 1989 to 2005. The highest (vs. first) quintile was associated with a significantly increased CHD risk after adjustment for all other risk factors and correction for regression dilution bias (hazard ratio = 1.85, 95% confidence interval: 1.08, 3.16; P = 0.024). Models that included all well-known risk factors plus PS quintiles improved prediction of CHD (net reclassification improvement (NRI) = 7.0% (P = 0.007), category-less NRI (>0) = 22.1% (P < 0.001)), and the likelihood ratio statistic increased significantly (P = 0.018). The increase in CHD risk was particularly strong when subjects also had high CRP levels. There was no association between free PS level and stroke risk. This study confirms the independent association of elevated free PS levels with future risk of CHD, although elevated PS levels added only modestly to prediction metrics. The novel finding of increased CHD risk, particularly when CRP and PS levels are high, requires further study.

13. A Method to Visualize and Adjust for Selection Bias in Prevalent Cohort Studies Selection bias and confounding are concerns in cohort studies where the reason for inclusion of subjects in the cohort may be related to the outcome of interest. Selection bias in prevalent cohorts is often corrected by excluding observation time and events during the first time period after inclusion in the cohort. This time period must be chosen carefully—long enough to minimize selection bias but not too long so as to unnecessarily discard observation time and events. A novel method visualizing and estimating selection bias is described and exemplified by using 2 real cohort study examples: a study of hepatitis C virus infection and a study of monoclonal gammopathy of undetermined significance. The method is based on modeling the hazard for the outcome of interest as a function of time since inclusion in the cohort. The events studied were “hospitalizations for kidney-related disease” in the hepatitis C virus cohort and “death” in the monoclonal gammopathy of undetermined significance cohort. Both cohorts show signs of considerable selection bias as evidenced by increased hazard in the time period after inclusion in the cohort. The method was very useful in visualizing selection bias and in determining the initial time period to be excluded from the analyses.

14. Reliability of Self-rated Health in US Adults General self-rated health (SRH) is widely used to study trends and inequalities in population health. Recently, there has been an increased interest in understanding the measurement properties of SRH. This study evaluated for the first time the test-retest reliability of SRH among US adults. Analyses were based on a nationally representative sample of 9,235 adults interviewed in the 2005–2008 National Health and Nutrition Examination Survey (NHANES). Respondents reported SRH on 2 occasions (about 1 month apart). Kappa statistics, polyserial correlations, and agreement tabulations were used to assess reliability across population subgroups; regression models tested the association of sociodemographic factors and the stability of the rating. Nearly 40% of respondents changed their health rating between interviews, indicating moderate test-retest reliability of SRH. Reliability differed significantly by sociodemographic characteristics: Racial/ethnic minorities and adults with less education had lower reliability of SRH judgments. Health events between interviews did not influence consistency, but conditional on a rating change, they increased the odds of downgrading one's health. The results suggest that 1) there is a substantial amount of error in individuals’ self-assessment of health and 2) reliability is worse for disadvantaged sociodemographic groups, potentially biasing estimates of health inequalities among US adults.

15. Performance of the Modified Poisson Regression Approach for Estimating Relative Risks From Clustered Prospective Data Modified Poisson regression, which combines a log Poisson regression model with robust variance estimation, is a useful alternative to log binomial regression for estimating relative risks. Previous studies have shown both analytically and by simulation that modified Poisson regression is appropriate for independent prospective data. This method is often applied to clustered prospective data, despite a lack of evidence to support its use in this setting. The purpose of this article is to evaluate the performance of the modified Poisson regression approach for estimating relative risks from clustered prospective data, by using generalized estimating equations to account for clustering. A simulation study is conducted to compare log binomial regression and modified Poisson regression for analyzing clustered data from intervention and observational studies. Both methods generally perform well in terms of bias, type I error, and coverage. Unlike log binomial regression, modified Poisson regression is not prone to convergence problems. The methods are contrasted by using example data sets from 2 large studies. The results presented in this article support the use of modified Poisson regression as an alternative to log binomial regression for analyzing clustered prospective data when clustering is taken into account by using generalized estimating equations.