Showing papers in "American Journal of Epidemiology in 2017"
TL;DR: UK Biobank is not representative of the sampling population; there is evidence of a “healthy volunteer” selection bias; valid assessment of exposure-disease relationships may be widely generalizable and does not require participants to be Representative of the population at large.
Abstract: The UK Biobank cohort is a population-based cohort of 500,000 participants recruited in the United Kingdom (UK) between 2006 and 2010. Approximately 9.2 million individuals aged 40-69 years who lived within 25 miles (40 km) of one of 22 assessment centers in England, Wales, and Scotland were invited to enter the cohort, and 5.5% participated in the baseline assessment. The representativeness of the UK Biobank cohort was investigated by comparing demographic characteristics between nonresponders and responders. Sociodemographic, physical, lifestyle, and health-related characteristics of the cohort were compared with nationally representative data sources. UK Biobank participants were more likely to be older, to be female, and to live in less socioeconomically deprived areas than nonparticipants. Compared with the general population, participants were less likely to be obese, to smoke, and to drink alcohol on a daily basis and had fewer self-reported health conditions. At age 70-74 years, rates of all-cause mortality and total cancer incidence were 46.2% and 11.8% lower, respectively, in men and 55.5% and 18.1% lower, respectively, in women than in the general population of the same age. UK Biobank is not representative of the sampling population; there is evidence of a "healthy volunteer" selection bias. Nonetheless, valid assessment of exposure-disease relationships may be widely generalizable and does not require participants to be representative of the population at large.
1,896 citations
TL;DR: Although large-scale applications of metabolic profiling are still novel, it seems likely that comprehensive biomarker data will contribute to etiologic understanding of various diseases and abilities to predict disease risks, with the potential to translate into multiple clinical settings.
Abstract: Detailed metabolic profiling in large-scale epidemiologic studies has uncovered novel biomarkers for cardiometabolic diseases and clarified the molecular associations of established risk factors. A quantitative metabolomics platform based on nuclear magnetic resonance spectroscopy has found widespread use, already profiling over 400,000 blood samples. Over 200 metabolic measures are quantified per sample; in addition to many biomarkers routinely used in epidemiology, the method simultaneously provides fine-grained lipoprotein subclass profiling and quantification of circulating fatty acids, amino acids, gluconeogenesis-related metabolites, and many other molecules from multiple metabolic pathways. Here we focus on applications of magnetic resonance metabolomics for quantifying circulating biomarkers in large-scale epidemiology. We highlight the molecular characterization of risk factors, use of Mendelian randomization, and the key issues of study design and analyses of metabolic profiling for epidemiology. We also detail how integration of metabolic profiling data with genetics can enhance drug development. We discuss why quantitative metabolic profiling is becoming widespread in epidemiology and biobanking. Although large-scale applications of metabolic profiling are still novel, it seems likely that comprehensive biomarker data will contribute to etiologic understanding of various diseases and abilities to predict disease risks, with the potential to translate into multiple clinical settings.
327 citations
TL;DR: In this large cohort of US elderly, this work provides important new evidence that long-term PM2.5 exposure is significantly related to increased mortality from respiratory disease, lung cancer, and cardiovascular disease.
Abstract: The impact of chronic exposure to fine particulate matter (particulate matter with an aerodynamic diameter less than or equal to 2.5 μm (PM2.5)) on respiratory disease and lung cancer mortality is poorly understood. In a cohort of 18.9 million Medicare beneficiaries (4.2 million deaths) living across the conterminous United States between 2000 and 2008, we examined the association between chronic PM2.5 exposure and cause-specific mortality. We evaluated confounding through adjustment for neighborhood behavioral covariates and decomposition of PM2.5 into 2 spatiotemporal scales. We found significantly positive associations of 12-month moving average PM2.5 exposures (per 10-μg/m3 increase) with respiratory, chronic obstructive pulmonary disease, and pneumonia mortality, with risk ratios ranging from 1.10 to 1.24. We also found significant PM2.5-associated elevated risks for cardiovascular and lung cancer mortality. Risk ratios generally increased with longer moving averages; for example, an elevation in 60-month moving average PM2.5 exposures was linked to 1.33 times the lung cancer mortality risk (95% confidence interval: 1.24, 1.40), as compared with 1.13 (95% confidence interval: 1.11, 1.15) for 12-month moving average exposures. Observed associations were robust in multivariable models, although evidence of unmeasured confounding remained. In this large cohort of US elderly, we provide important new evidence that long-term PM2.5 exposure is significantly related to increased mortality from respiratory disease, lung cancer, and cardiovascular disease.
297 citations
TL;DR: Data indicate that this SFFQ provided reasonably valid estimates for intakes of a wide variety of dietary variables and that use of multiple 24-hour recalls or 7DDRs as a comparison method provided similar conclusions if day-to-day variation was taken into account.
Abstract: The authors evaluated the validity of a 152-item semiquantitative food frequency questionnaire (SFFQ) by comparing it with two 7-day dietary records (7DDRs) or up to 4 automated self-administered 24-hour recalls (ASA24s) over a 1-year period in the women's Lifestyle Validation Study (2010-2012), conducted among subgroups of the Nurses' Health Studies. Intakes of energy and 44 nutrients were assessed using the 3 methods among 632 US women. Compared with the 7DDRs, SFFQ responses tended to underestimate sodium intake but overestimate intakes of energy, macronutrients, and several nutrients in fruits and vegetables, such as carotenoids. Spearman correlation coefficients between energy-adjusted intakes from 7DDRs and the SFFQ completed at the end of the data-collection period ranged from 0.36 for lauric acid to 0.77 for alcohol (mean r = 0.53). Correlations of the end-period SFFQ were weaker when ASA24s were used as the comparison method (mean r = 0.43). After adjustment for within-person variation in the comparison method, the correlations of the final SFFQ were similar with 7DDRs (mean r = 0.63) and ASA24s (mean r = 0.62). These data indicate that this SFFQ provided reasonably valid estimates for intakes of a wide variety of dietary variables and that use of multiple 24-hour recalls or 7DDRs as a comparison method provided similar conclusions if day-to-day variation was taken into account.
295 citations
TL;DR: Results suggested that various proposed approaches to quantifying biological aging may not measure the same aspects of the aging process, and further systematic evaluation and refinement of measures of biological aging is needed to furnish outcomes for geroprotector trials.
Abstract: The geroscience hypothesis posits that therapies to slow biological processes of aging can prevent disease and extend healthy years of life. To test such "geroprotective" therapies in humans, outcome measures are needed that can assess extension of disease-free life span. This need has spurred development of different methods to quantify biological aging. But different methods have not been systematically compared in the same humans. We implemented 7 methods to quantify biological aging using repeated-measures physiological and genomic data in 964 middle-aged humans in the Dunedin Study (New Zealand; persons born 1972-1973). We studied 11 measures in total: telomere-length and erosion, 3 epigenetic-clocks and their ticking rates, and 3 biomarker-composites. Contrary to expectation, we found low agreement between different measures of biological aging. We next compared associations between biological aging measures and outcomes that geroprotective therapies seek to modify: physical functioning, cognitive decline, and subjective signs of aging, including aged facial appearance. The 71-cytosine-phosphate-guanine epigenetic clock and biomarker composites were consistently related to these aging-related outcomes. However, effect sizes were modest. Results suggested that various proposed approaches to quantifying biological aging may not measure the same aspects of the aging process. Further systematic evaluation and refinement of measures of biological aging is needed to furnish outcomes for geroprotector trials.
247 citations
TL;DR: The negative associations of Facebook use were comparable to or greater in magnitude than the positive impact of offline interactions, which suggests a possible tradeoff between offline and online relationships.
Abstract: Face-to-face social interactions enhance well-being. With the ubiquity of social media, important questions have arisen about the impact of online social interactions. In the present study, we assessed the associations of both online and offline social networks with several subjective measures of well-being. We used 3 waves (2013, 2014, and 2015) of data from 5,208 subjects in the nationally representative Gallup Panel Social Network Study survey, including social network measures, in combination with objective measures of Facebook use. We investigated the associations of Facebook activity and real-world social network activity with self-reported physical health, self-reported mental health, self-reported life satisfaction, and body mass index. Our results showed that overall, the use of Facebook was negatively associated with well-being. For example, a 1-standard-deviation increase in "likes clicked" (clicking "like" on someone else's content), "links clicked" (clicking a link to another site or article), or "status updates" (updating one's own Facebook status) was associated with a decrease of 5%-8% of a standard deviation in self-reported mental health. These associations were robust to multivariate cross-sectional analyses, as well as to 2-wave prospective analyses. The negative associations of Facebook use were comparable to or greater in magnitude than the positive impact of offline interactions, which suggests a possible tradeoff between offline and online relationships.
243 citations
TL;DR: This work gives step-by-step instructions for using TMLE to estimate the average treatment effect in the context of an observational study, and demonstrates all methods using super learning, highlighting that incorporation of machine learning may outperform parametric regression in observational data settings.
Abstract: Estimation of causal effects using observational data continues to grow in popularity in the epidemiologic literature. While many applications of causal effect estimation use propensity score methods or G-computation, targeted maximum likelihood estimation (TMLE) is a well-established alternative method with desirable statistical properties. TMLE is a doubly robust maximum-likelihood-based approach that includes a secondary "targeting" step that optimizes the bias-variance tradeoff for the target parameter. Under standard causal assumptions, estimates can be interpreted as causal effects. Because TMLE has not been as widely implemented in epidemiologic research, we aim to provide an accessible presentation of TMLE for applied researchers. We give step-by-step instructions for using TMLE to estimate the average treatment effect in the context of an observational study. We discuss conceptual similarities and differences between TMLE and 2 common estimation approaches (G-computation and inverse probability weighting) and present findings on their relative performance using simulated data. Our simulation study compares methods under parametric regression misspecification; our results highlight TMLE's property of double robustness. Additionally, we discuss best practices for TMLE implementation, particularly the use of ensembled machine learning algorithms. Our simulation study demonstrates all methods using super learning, highlighting that incorporation of machine learning may outperform parametric regression in observational data settings.
209 citations
TL;DR: Findings from this meta-analysis indicated that dietary flavonoids are associated with decreased risk of all-cause and CVD mortality.
Abstract: Recent evidence has suggested that flavonoid and lignan intake may be associated with decreased risk of chronic and degenerative diseases. The aim of this meta-analysis was to assess the association between dietary flavonoid and lignan intake and all-cause and cardiovascular disease (CVD) mortality in prospective cohort studies. A systematic search was conducted in electronic databases to identify studies published from January 1996 to December 2015 that satisfied inclusion/exclusion criteria. Risk ratios and 95% confidence intervals were extracted and analyzed using a random-effects model. Nonlinear dose-response analysis was modeled by using restricted cubic splines. The inclusion criteria were met by 22 prospective studies exploring various flavonoid and lignan classes. Compared with lower intake, high consumption of total flavonoids was associated with decreased risk of all-cause mortality (risk ratio = 0.74, 95% confidence intervals: 0.55, 0.99), while a 100-mg/day increment in intake led to a (linear) decreased risk of 6% and 4% of all-cause and CVD mortality, respectively. Among flavonoid classes, significant results were obtained for intakes of flavonols, flavones, flavanones, anthocyanidins, and proanthocyanidins. Only limited evidence was available on flavonoid classes and lignans and all-cause mortality. Findings from this meta-analysis indicated that dietary flavonoids are associated with decreased risk of all-cause and CVD mortality.
202 citations
TL;DR: An inverse odds weighting approach is presented that can easily operationalize transportability of causal effect estimates from study samples to target populations and discusses how the conditions required for the identification of internally valid causal effects are translated to apply to the Identification of externally valid causal effect.
Abstract: Increasingly, the statistical and epidemiologic literature is focusing beyond issues of internal validity and turning its attention to questions of external validity Here, we discuss some of the challenges of transporting a causal effect from a randomized trial to a specific target population We present an inverse odds weighting approach that can easily operationalize transportability We derive these weights in closed form and illustrate their use with a simple numerical example We discuss how the conditions required for the identification of internally valid causal effects are translated to apply to the identification of externally valid causal effects Estimating effects in target populations is an important goal, especially for policy or clinical decisions Researchers and policy-makers should therefore consider use of statistical techniques such as inverse odds of sampling weights, which under careful assumptions can transport effect estimates from study samples to target populations
201 citations
TL;DR: There is marked heterogeneity in the degree to which various frailty scores estimate frailty and in the identification of particular individuals as frail, while accuracy was highest for multidimensional FS.
Abstract: In elderly populations, frailty is associated with higher mortality risk. Although many frailty scores (FS) have been proposed, no single score is considered the gold standard. We aimed to evaluate the agreement between a wide range of FS in the English Longitudinal Study of Ageing (ELSA). Through a literature search, we identified 35 FS that could be calculated in ELSA wave 2 (2004-2005). We examined agreement between each frailty score and the mean of 35 FS, using a modified Bland-Altman model and Cohen's kappa (κ). Missing data were imputed. Data from 5,377 participants (ages ≥60 years) were analyzed (44.7% men, 55.3% women). FS showed widely differing degrees of agreement with the mean of all scores and between each pair of scores. Frailty classification also showed a very wide range of agreement (Cohen's κ = 0.10-0.83). Agreement was highest among "accumulation of deficits"-type FS, while accuracy was highest for multidimensional FS. There is marked heterogeneity in the degree to which various FS estimate frailty and in the identification of particular individuals as frail. Different FS are based on different concepts of frailty, and most pairs cannot be assumed to be interchangeable. Research results based on different FS cannot be compared or pooled.
195 citations
TL;DR: Long-term rotating night-shift work was associated with a higher risk of breast cancer, particularly among women who performed shift work during young adulthood, and the role of shift work timing on breast cancer risk should be explored.
Abstract: In 2007, the International Agency for Research on Cancer declared shift work that involved circadian disruption to be a "probable" carcinogen (group 2A), noting that human evidence was limited. Using data from 2 prospective cohort studies, the Nurses' Health Study (1988-2012; n = 78,516) and Nurses' Health Study II (1989-2013; n = 114,559), we examined associations between rotating night-shift work and breast cancer risk. In the 2 cohorts, there were a total of 9,541 incident invasive breast malignancies and 24 years of follow-up. In the Nurses' Health Study, women with 30 years or more of shift work did not have a higher risk of breast cancer (hazard ratio (HR) = 0.95, 95% confidence interval (95% CI): 0.77, 1.17; P for trend = 0.63) compared with those who never did shift work, although follow-up occurred primarily after retirement from shift work. Among participants in the Nurses' Health Study II, who were younger than participants in the other cohort, the risk of breast cancer was significantly higher in women with 20 years or more of shift work at baseline, reflecting young-adult exposure (HR = 2.15, 95% CI: 1.23, 3.73; P for trend = 0.23), and was marginally significantly higher for women with 20 years or more of cumulative shift work when we used updated exposure information (HR = 1.40, 95% CI: 1.00, 1.97; P for trend = 0.74). In conclusion, long-term rotating night-shift work was associated with a higher risk of breast cancer, particularly among women who performed shift work during young adulthood. Further studies should explore the role of shift work timing on breast cancer risk.
TL;DR: Bias in estimates of critical windows of vulnerability were assessed and a distributed lag model produced unbiased estimates and added flexibility to identify windows, and including all TAEs in a single model reduced bias.
Abstract: Evidence supports an association between maternal exposure to air pollution during pregnancy and children's health outcomes Recent interest has focused on identifying critical windows of vulnerability An analysis based on a distributed lag model (DLM) can yield estimates of a critical window that are different from those from an analysis that regresses the outcome on each of the 3 trimester-average exposures (TAEs) Using a simulation study, we assessed bias in estimates of critical windows obtained using 3 regression approaches: 1) 3 separate models to estimate the association with each of the 3 TAEs; 2) a single model to jointly estimate the association between the outcome and all 3 TAEs; and 3) a DLM We used weekly fine-particulate-matter exposure data for 238 births in a birth cohort in and around Boston, Massachusetts, and a simulated outcome and time-varying exposure effect Estimates using separate models for each TAE were biased and identified incorrect windows This bias arose from seasonal trends in particulate matter that induced correlation between TAEs Including all TAEs in a single model reduced bias DLM produced unbiased estimates and added flexibility to identify windows Analysis of body mass index z score and fat mass in the same cohort highlighted inconsistent estimates from the 3 methods
National Institutes of Health1, Dartmouth College2, Harvard University3, Pasteur Institute4, University of Pennsylvania5, Johns Hopkins University6, University of Southern California7, Fred Hutchinson Cancer Research Center8, University of California, San Diego9, Stanford University10, University of Michigan11, Pennsylvania State University12, Geisinger Medical Center13, University of California, Los Angeles14, University of Texas MD Anderson Cancer Center15, University of California, San Francisco16
TL;DR: Current and critical issues and themes in G×E research that need additional consideration, including the improved data analytical methods, environmental exposure assessment, and incorporation of functional data and annotations are highlighted.
Abstract: Recently, many new approaches, study designs, and statistical and analytical methods have emerged for studying gene-environment interactions (G×Es) in large-scale studies of human populations. There are opportunities in this field, particularly with respect to the incorporation of -omics and next-generation sequencing data and continual improvement in measures of environmental exposures implicated in complex disease outcomes. In a workshop called "Current Challenges and New Opportunities for Gene-Environment Interaction Studies of Complex Diseases," held October 17-18, 2014, by the National Institute of Environmental Health Sciences and the National Cancer Institute in conjunction with the annual American Society of Human Genetics meeting, participants explored new approaches and tools that have been developed in recent years for G×E discovery. This paper highlights current and critical issues and themes in G×E research that need additional consideration, including the improved data analytical methods, environmental exposure assessment, and incorporation of functional data and annotations.
TL;DR: It was found that a higher degree of optimism was associated with a lower mortality risk, and associations were maintained for various causes of death, including cancer, heart disease, stroke, respiratory disease, and infection.
Abstract: Growing evidence has linked positive psychological attributes like optimism to a lower risk of poor health outcomes, especially cardiovascular disease. It has been demonstrated in randomized trials that optimism can be learned. If associations between optimism and broader health outcomes are established, it may lead to novel interventions that improve public health and longevity. In the present study, we evaluated the association between optimism and cause-specific mortality in women after considering the role of potential confounding (sociodemographic characteristics, depression) and intermediary (health behaviors, health conditions) variables. We used prospective data from the Nurses' Health Study (n = 70,021). Dispositional optimism was measured in 2004; all-cause and cause-specific mortality rates were assessed from 2006 to 2012. Using Cox proportional hazard models, we found that a higher degree of optimism was associated with a lower mortality risk. After adjustment for sociodemographic confounders, compared with women in the lowest quartile of optimism, women in the highest quartile had a hazard ratio of 0.71 (95% confidence interval: 0.66, 0.76) for all-cause mortality. Adding health behaviors, health conditions, and depression attenuated but did not eliminate the associations (hazard ratio = 0.91, 95% confidence interval: 0.85, 0.97). Associations were maintained for various causes of death, including cancer, heart disease, stroke, respiratory disease, and infection. Given that optimism was associated with numerous causes of mortality, it may provide a valuable target for new research on strategies to improve health.
International Institute for Population Sciences1, World Health Organization2, University of Minnesota3, University of Oregon4, Vassar College5, University of Ghana6, Russian Academy7, Centers for Disease Control and Prevention8, University of Cape Town9, Northwestern University10, University of Health and Allied Sciences11, University of Notre Dame12, Umeå University13, University of Sydney14, Capital Medical University15, University of Newcastle16, National Research Council17
TL;DR: Algorithm/measured test-based prevalence of NCDs was much higher than self-reported prevalence in all 6 countries, indicating underestimation of NCS prevalence in low- and middle-income countries and highlighting the inadequacies in diagnosis and management of N CDs in local health-care systems.
Abstract: In this paper, we examine patterns of self-reported diagnosis of noncommunicable diseases (NCDs) and prevalences of algorithm/measured test-based, undiagnosed, and untreated NCDs in China, Ghana, India, Mexico, Russia, and South Africa. Nationally representative samples of older adults aged ≥50 years were analyzed from wave 1 of the World Health Organization's Study on Global Ageing and Adult Health (2007-2010; n = 34,149). Analyses focused on 6 conditions: angina, arthritis, asthma, chronic lung disease, depression, and hypertension. Outcomes for these NCDs were: 1) self-reported disease, 2) algorithm/measured test-based disease, 3) undiagnosed disease, and 4) untreated disease. Algorithm/measured test-based prevalence of NCDs was much higher than self-reported prevalence in all 6 countries, indicating underestimation of NCD prevalence in low- and middle-income countries. Undiagnosed prevalence of NCDs was highest for hypertension, ranging from 19.7% (95% confidence interval (CI): 18.1, 21.3) in India to 49.6% (95% CI: 46.2, 53.0) in South Africa. The proportion untreated among all diseases was highest for depression, ranging from 69.5% (95% CI: 57.1, 81.9) in South Africa to 93.2% (95% CI: 90.1, 95.7) in India. Higher levels of education and wealth significantly reduced the odds of an undiagnosed condition and untreated morbidity. A high prevalence of undiagnosed NCDs and an even higher proportion of untreated NCDs highlights the inadequacies in diagnosis and management of NCDs in local health-care systems.
TL;DR: To the knowledge, these results represent the first identification of a possible association between both long-term ozone and PM2.5 exposure and depression onset and the stronger association specifically with antidepressant use may reflect that this endpoint better captures the onset time and milder cases.
Abstract: Despite recently reported associations between air pollution and acute psychiatric outcomes, the association with depression onset has not, to our knowledge, been previously examined. We conducted a prospective cohort study among 41,844 women in the Nurses' Health Study, in the United States. The women had an average age of 66.6 (standard deviation, 7.6) years, were depression-free in 1996, and were followed through 2008. May-September ozone exposures were predicted by interpolating concentrations from the 5 nearest monitors. One-, 2-, and 5-year average concentrations of particulate matter with an aerodynamic diameter less than or equal to 2.5 μm (PM2.5) were predicted at each participant's residence using a spatiotemporal model. We defined depression as report of doctor's diagnosis or use of antidepressant medication. We estimated adjusted hazard ratios with time-varying Cox models. Hazard ratios for both pollutants were elevated (per 10-parts-per-billion increase in ozone, hazard ratio (HR) = 1.06; 95% confidence interval (CI): 1.00, 1.12; per 10-μg/m3 increase in 1-year PM2.5, HR = 1.08; 95% CI: 0.97, 1.20). Associations were stronger when only antidepressant use was used to define cases (for ozone, HR = 1.08; 95% CI: 1.02, 1.14; for PM2.5, HR = 1.12; 95% CI: 1.00, 1.25). To our knowledge, these results represent the first identification of a possible association between both long-term ozone and PM2.5 exposure and depression onset. Although the stronger association specifically with antidepressant use may reflect that this endpoint better captures the onset time and milder cases, our findings should be interpreted with caution.
TL;DR: Stable, high SES predicted the best memory function and slowest decline, and both early and late-life interventions are potentially relevant for reducing dementia risk by improving memory function or slowing decline.
Abstract: Both early life and adult socioeconomic status (SES) predict late-life level of memory; however, evidence is mixed on the relationship between SES and rate of memory decline. Further, the relative importance of different life-course periods for rate of late-life memory decline has not been evaluated. We examined associations between life-course SES and late-life memory function and decline. Health and Retirement Study participants (n = 10,781) were interviewed biennially from 1998-2012 (United States). SES measurements for childhood (composite score including parents' educational attainment), early adulthood (high-school or college completion), and older adulthood (income, mean age 66 years) were all dichotomized. Word-list memory was modeled via inverse-probability weighted longitudinal models accounting for differential attrition, survival, and time-varying confounding, with nonrespondents retained via proxy assessments. Compared to low SES at all 3 points (referent), stable, high SES predicted the best memory function and slowest decline. High-school completion had the largest estimated effect on memory (β = 0.19; 95% confidence interval: 0.15, 0.22), but high late-life income had the largest estimated benefit for slowing declines (for 10-year memory change, β = 0.35; 95% confidence interval: 0.24, 0.46). Both early and late-life interventions are potentially relevant for reducing dementia risk by improving memory function or slowing decline.
TL;DR: It is concluded that methodological development and training should go beyond coverage of mechanistic biases to cover distortions of conclusions produced by statistical methods and psychosocial forces.
Abstract: There is no complete solution for the problem of abuse of statistics, but methodological training needs to cover cognitive biases and other psychosocial factors affecting inferences. The present paper discusses 3 common cognitive distortions: 1) dichotomania, the compulsion to perceive quantities as dichotomous even when dichotomization is unnecessary and misleading, as in inferences based on whether a P value is "statistically significant"; 2) nullism, the tendency to privilege the hypothesis of no difference or no effect when there is no scientific basis for doing so, as when testing only the null hypothesis; and 3) statistical reification, treating hypothetical data distributions and statistical models as if they reflect known physical laws rather than speculative assumptions for thought experiments. As commonly misused, null-hypothesis significance testing combines these cognitive problems to produce highly distorted interpretation and reporting of study results. Interval estimation has so far proven to be an inadequate solution because it involves dichotomization, an avenue for nullism. Sensitivity and bias analyses have been proposed to address reproducibility problems (Am J Epidemiol. 2017;186(6):646-647); these methods can indeed address reification, but they can also introduce new distortions via misleading specifications for bias parameters. P values can be reframed to lessen distortions by presenting them without reference to a cutoff, providing them for relevant alternatives to the null, and recognizing their dependence on all assumptions used in their computation; they nonetheless require rescaling for measuring evidence. I conclude that methodological development and training should go beyond coverage of mechanistic biases (e.g., confounding, selection bias, measurement error) to cover distortions of conclusions produced by statistical methods and psychosocial forces.
Norwegian University of Science and Technology1, Erasmus University Medical Center2, King Edward Memorial Hospital3, Memorial Hospital of South Bend4, King's College London5, University Hospital Southampton NHS Foundation Trust6, Nanyang Technological University7, St. John's University8, Radboud University Nijmegen9, University of Warwick10, Norwegian Institute of Public Health11, Yale University12
TL;DR: A systematic review and meta-analysis of individual participant data evaluated the associations of maternal serum or plasma B12 concentrations in pregnancy with offspring birth weight and length of gestation to support the need for randomized controlled trials of vitamin B12 supplementation in pregnancy.
Abstract: Vitamin B12 (hereafter referred to as B12) deficiency in pregnancy is prevalent and has been associated with both lower birth weight (birth weight <2,500 g) and preterm birth (length of gestation <37 weeks). Nevertheless, current evidence is contradictory. We performed a systematic review and a meta-analysis of individual participant data to evaluate the associations of maternal serum or plasma B12 concentrations in pregnancy with offspring birth weight and length of gestation. Twenty-two eligible studies were identified (11,993 observations). Eighteen studies were included in the meta-analysis (11,216 observations). No linear association was observed between maternal B12 levels in pregnancy and birth weight, but B12 deficiency (<148 pmol/L) was associated with a higher risk of low birth weight in newborns (adjusted risk ratio = 1.15, 95% confidence interval (CI): 1.01, 1.31). There was a linear association between maternal levels of B12 and preterm birth (per each 1-standard-deviation increase in B12, adjusted risk ratio = 0.89, 95% CI: 0.82, 0.97). Accordingly, B12 deficiency was associated with a higher risk of preterm birth (adjusted risk ratio = 1.21, 95% CI: 0.99, 1.49). This finding supports the need for randomized controlled trials of vitamin B12 supplementation in pregnancy.
TL;DR: This study prospectively examined the long-term associations of air pollution, defined as particulate matter with an aerodynamic diameter less than or equal to 2.5 µm, and temperature with the development of metabolic syndrome and its components and found men living in neighborhoods with worse air quality and temperatures showed increased risk of developing metabolic dysfunctions.
Abstract: Ambient air pollution and temperature have been linked with cardiovascular morbidity and mortality. Metabolic syndrome and its components-abdominal obesity, elevated fasting blood glucose concentration, low high-density lipoprotein cholesterol concentration, hypertension, and hypertriglyceridemia-predict cardiovascular disease, but the environmental causes are understudied. In this study, we prospectively examined the long-term associations of air pollution, defined as particulate matter with an aerodynamic diameter less than or equal to 2.5 µm (PM2.5), and temperature with the development of metabolic syndrome and its components. Using covariate-adjustment Cox proportional hazards models, we estimated associations of mean annual PM2.5 concentration and temperature with risk of incident metabolic dysfunctions between 1993 and 2011 in 587 elderly (mean = 70 (standard deviation, 7) years of age) male participants in the Normative Aging Study. A 1-μg/m3 increase in mean annual PM2.5 concentration was associated with a higher risk of developing metabolic syndrome (hazard ratio (HR) = 1.27, 95% confidence interval (CI): 1.06, 1.52), an elevated fasting blood glucose level (HR = 1.20, 95% CI: 1.03, 1.39), and hypertriglyceridemia (HR = 1.14, 95% CI: 1.00, 1.30). Our findings for metabolic syndrome and high fasting blood glucose remained significant for PM2.5 levels below the Environmental Protection Agency's health-safety limit (12 μg/m3). A 1°C increase in mean annual temperature was associated with a higher risk of developing elevated fasting blood glucose (HR = 1.33, 95% CI: 1.14, 1.56). Men living in neighborhoods with worse air quality-with higher PM2.5 levels and/or temperatures than average-showed increased risk of developing metabolic dysfunctions.
TL;DR: The results suggest that targeting tuberculosis prevention efforts to household contacts is highly effective, however, a large proportion of transmission at the population level may occur outside the household.
Abstract: The individual- and population-level impact of household tuberculosis exposure on transmission is unclear but may have implications for the effectiveness and implementation of control interventions. We systematically searched for and included studies in which latent tuberculosis infection was assessed in 2 groups: children exposed and unexposed to a household member with tuberculosis. We also extracted data on the smear and culture status of index cases, the age and bacillus Calmette-Guerin vaccination status of contacts, and study design characteristics. Of 6,176 citations identified from our search strategy, 26 studies (13,999 children with household exposure to tuberculosis and 174,097 children without) from 1929-2015 met inclusion criteria. Exposed children were 3.79 (95% confidence interval (CI): 3.01, 4.78) times more likely to be infected than were their community counterparts. Metaregression demonstrated higher infection among children aged 0-4 years of age compared with children aged 10-14 years (ratio of odds ratios = 2.24, 95% CI: 1.43, 3.51) and among smear-positive versus smear-negative index cases (ratio of odds ratios = 5.45, 95% CI: 3.43, 8.64). At the population level, we estimated that a small proportion (<20%) of transmission was attributable to household exposure. Our results suggest that targeting tuberculosis prevention efforts to household contacts is highly effective. However, a large proportion of transmission at the population level may occur outside the household.
TL;DR: In Finland, advanced maternal age is not independently associated with the risk of low birth weight or preterm delivery among mothers who have had at least 2 live births, and in within-family models, the relationships are statistically and substantively negligible.
Abstract: Advanced maternal age at birth is considered a major risk factor for birth outcomes. It is unclear to what extent this association is confounded by maternal characteristics. To test whether advanced maternal age at birth independently increases the risk of low birth weight (<2,500 g) and preterm birth (<37 weeks' gestation), we compared between-family models (children born to different mothers at different ages) with within-family models (children born to the same mother at different ages). The latter procedure reduces confounding by unobserved parental characteristics that are shared by siblings. We used Finnish population registers, including 124,098 children born during 1987-2000. When compared with maternal ages 25-29 years in between-family models, maternal ages of 35-39 years and ≥40 years were associated with percentage increases of 1.1 points (95% confidence intervals: 0.8, 1.4) and 2.2 points (95% confidence intervals: 1.4, 2.9), respectively, in the probability of low birth weight. The associations are similar for the risk of preterm delivery. In within-family models, the relationship between advanced maternal age and low birth weight or preterm birth is statistically and substantively negligible. In Finland, advanced maternal age is not independently associated with the risk of low birth weight or preterm delivery among mothers who have had at least 2 live births.
TL;DR: All fecal sample collection methods appear relatively reproducible, stable, and accurate, and future studies could use these collection methods for microbiome analyses.
Abstract: Prospective cohort studies are needed to assess the relationship between the fecal microbiome and human health and disease. To evaluate fecal collection methods, we determined technical reproducibility, stability at ambient temperature, and accuracy of 5 fecal collection methods (no additive, 95% ethanol, RNAlater Stabilization Solution, fecal occult blood test cards, and fecal immunochemical test tubes). Fifty-two healthy volunteers provided fecal samples at the Mayo Clinic in Rochester, Minnesota, in 2014. One set from each sample collection method was frozen immediately, and a second set was incubated at room temperature for 96 hours and then frozen. Intraclass correlation coefficients (ICCs) were calculated for the relative abundance of 3 phyla, 2 alpha diversity metrics, and 4 beta diversity metrics. Technical reproducibility was high, with ICCs for duplicate fecal samples between 0.64 and 1.00. Stability for most methods was generally high, although the ICCs were below 0.60 for 95% ethanol in metrics that were more sensitive to relative abundance. When compared with fecal samples that were frozen immediately, the ICCs were below 0.60 for the metrics that were sensitive to relative abundance; however, the remaining 2 alpha diversity and 3 beta diversity metrics were all relatively accurate, with ICCs above 0.60. In conclusion, all fecal sample collection methods appear relatively reproducible, stable, and accurate. Future studies could use these collection methods for microbiome analyses.
TL;DR: This article proposes a procedure for obtaining fine-grained decompositions that may still be recovered from observed data in such complex settings and introduces natural effects models along with estimation methods that allow for flexible and parsimonious modeling.
Abstract: The advent of counterfactual-based mediation analysis has triggered enormous progress on how, and under what assumptions, one may disentangle path-specific effects upon combining arbitrary (possibly nonlinear) models for mediator and outcome. However, current developments have largely focused on single mediators because required identification assumptions prohibit simple extensions to settings with multiple mediators that may depend on one another. In this article, we propose a procedure for obtaining fine-grained decompositions that may still be recovered from observed data in such complex settings. We first show that existing analytical approaches target specific instances of a more general set of decompositions and may therefore fail to provide a comprehensive assessment of the processes that underpin cause-effect relationships between exposure and outcome. We then outline conditions for obtaining the remaining set of decompositions. Because the number of targeted decompositions increases rapidly with the number of mediators, we introduce natural effects models along with estimation methods that allow for flexible and parsimonious modeling. Our procedure can easily be implemented using off-the-shelf software and is illustrated using a reanalysis of the World Health Organization's Large Analysis and Review of European Housing and Health Status (WHO-LARES) study on the effect of mold exposure on mental health (2002-2003).
TL;DR: Semen quality was associated with long-term morbidity, and a significantly higher risk of hospitalization was found, in particular for cardiovascular diseases and diabetes mellitus, which supports the suggestion that semen quality is a strong biomarker of general health.
Abstract: Semen quality has been suggested to be a biological marker of long-term morbidity and mortality; however, few studies have been conducted on this subject. We identified 5,370 men seen for infertility at Frederiksberg Hospital, Denmark, during 1977-2010, and 4,712 of these men were followed in the Danish National Patient Registry until first hospitalization, death, or the end of the study. We classified patients according to hospitalizations and the presence of cardiovascular disease, diabetes, testicular cancer, or prostate cancer. We found a clear association between sperm concentration below 15 million/mL and all-cause hospitalizations (hazard ratio = 1.5, 95% confidence interval: 1.4, 1.6) and cardiovascular disease (hazard ratio = 1.4, 95% confidence interval: 1.2, 1.6), compared with men with a concentration above 40 million/mL. The probabilities for hospitalizations were also higher with a low total sperm count and low motility. Men with a sperm concentration of 195-200 million/mL were, on average, hospitalized for the first time 7 years later than were men with a sperm concentration of 0-5 million/mL. Semen quality was associated with long-term morbidity, and a significantly higher risk of hospitalization was found, in particular for cardiovascular diseases and diabetes mellitus. Our study supports the suggestion that semen quality is a strong biomarker of general health.
TL;DR: This study investigated the potential of deep sequence data to provide greater resolution on transmission routes, via the identification of shared genomic variants, and applied several easily implemented methods to identify transmission routes using both shared variants and genetic distance.
Abstract: Sequencing pathogen samples during a communicable disease outbreak is becoming an increasingly common procedure in epidemiologic investigations. Identifying who infected whom sheds considerable light on transmission patterns, high-risk settings and subpopulations, and the effectiveness of infection control. Genomic data shed new light on transmission dynamics and can be used to identify clusters of individuals likely to be linked by direct transmission. However, identification of individual routes of infection via single genome samples typically remains uncertain. We investigated the potential of deep sequence data to provide greater resolution on transmission routes, via the identification of shared genomic variants. We assessed several easily implemented methods to identify transmission routes using both shared variants and genetic distance, demonstrating that shared variants can provide considerable additional information in most scenarios. While shared-variant approaches identify relatively few links in the presence of a small transmission bottleneck, these links are highly accurate. Furthermore, we propose a hybrid approach that also incorporates phylogenetic distance to provide greater resolution. We applied our methods to data collected during the 2014 Ebola outbreak, identifying several likely routes of transmission. Our study highlights the power of data from deep sequencing of pathogens as a component of outbreak investigation and epidemiologic analyses.
TL;DR: Examination of associations of leukocyte telomere length with all-cause, cardiovascular disease, and cancer mortality in 12,199 adults participating in 2 population-based prospective cohort studies from Europe and the United States corroborated previous evidence suggesting that LTL predicts all- cause mortality beyond its association with age.
Abstract: We studied the associations of leukocyte telomere length (LTL) with all-cause, cardiovascular disease, and cancer mortality in 12,199 adults participating in 2 population-based prospective cohort studies from Europe (ESTHER) and the United States (Nurses' Health Study). Blood samples were collected in 1989-1990 (Nurses' Health Study) and 2000-2002 (ESTHER). LTL was measured by quantitative polymerase chain reaction. We calculated z scores for LTL to standardize LTL measurements across the cohorts. Cox proportional hazards regression models were used to calculate relative mortality according to continuous levels and quintiles of LTL z scores. The hazard ratios obtained from each cohort were subsequently pooled by meta-analysis. Overall, 2,882 deaths were recorded during follow-up (Nurses' Health Study, 1989-2010; ESTHER, 2000-2015). LTL was inversely associated with age in both cohorts. After adjustment for age, a significant inverse trend of LTL with all-cause mortality was observed in both cohorts. In random-effects meta-analysis, age-adjusted hazard ratios for the shortest LTL quintile compared with the longest were 1.23 (95% confidence interval (CI): 1.04, 1.46) for all-cause mortality, 1.29 (95% CI: 0.83, 2.00) for cardiovascular mortality, and 1.10 (95% CI: 0.88, 1.37) for cancer mortality. In this study population with an age range of 43-75 years, we corroborated previous evidence suggesting that LTL predicts all-cause mortality beyond its association with age.
TL;DR: Without discarding the culture of null hypothesis significance testing and implementing these alternative methods for statistical analysis and inference, all other strategies for improving reproducibility will yield marginal gains at best.
Abstract: In the last few years, stakeholders in the scientific community have raised alarms about a perceived lack of reproducibility of scientific results. In reaction, guidelines for journals have been promulgated and grant applicants have been asked to address the rigor and reproducibility of their proposed projects. Neither solution addresses a primary culprit, which is the culture of null hypothesis significance testing that dominates statistical analysis and inference. In an innovative research enterprise, selection of results for further evaluation based on null hypothesis significance testing is doomed to yield a low proportion of reproducible results and a high proportion of effects that are initially overestimated. In addition, the culture of null hypothesis significance testing discourages quantitative adjustments to account for systematic errors and quantitative incorporation of prior information. These strategies would otherwise improve reproducibility and have not been previously proposed in the widely cited literature on this topic. Without discarding the culture of null hypothesis significance testing and implementing these alternative methods for statistical analysis and inference, all other strategies for improving reproducibility will yield marginal gains at best.
TL;DR: Service attendance was the strongest R/S predictor of mortality in this cohort, and religious coping and self-identification as a very religious/spiritual person were associated with lower mortality when adjustment was made only for age.
Abstract: Previous longitudinal studies have consistently shown an association between attendance at religious services and lower all-cause mortality, but the literature on associations between other measures of religion and spirituality (R/S) and mortality is limited. We followed 36,613 respondents from the Black Women's Health Study from 2005 through December 31, 2013 to assess the associations between R/S and incident all-cause mortality using proportional hazards models. After control for numerous demographic and health covariates, together with other R/S variables, attending religious services several times per week was associated with a substantially lower mortality rate ratio (mortality rate ratio = 0.64, 95% confidence interval: 0.51, 0.80) relative to never attending services. Engaging in prayer several times per day was not associated with mortality after control for demographic and health covariates, but the association trended towards a higher mortality rate ratio when control was made for other R/S variables (for >2 times/day vs. weekly or less, mortality rate ratio = 1.28, 95% confidence interval: 0.99, 1.67; P-trend < 0.01). Religious coping and self-identification as a very religious/spiritual person were associated with lower mortality when adjustment was made only for age, but the association was attenuated when control was made for demographic and health covariates and was almost entirely eliminated when control was made for other R/S variables. The results indicate that service attendance was the strongest R/S predictor of mortality in this cohort.
TL;DR: Several statistical approaches that can be used to test for G×E in a genome-wide association study are summarized, as are issues that arise due to the complexity of environmental data.
Abstract: The analysis of gene-environment interaction (G×E) may hold the key for further understanding the etiology of many complex traits. The current availability of high-volume genetic data, the wide range in types of environmental data that can be measured, and the formation of consortiums of multiple studies provide new opportunities to identify G×E but also new analytical challenges. In this article, we summarize several statistical approaches that can be used to test for G×E in a genome-wide association study. These include traditional models of G×E in a case-control or quantitative trait study as well as alternative approaches that can provide substantially greater power. The latest methods for analyzing G×E with gene sets and with data in a consortium setting are summarized, as are issues that arise due to the complexity of environmental data. We provide some speculation on why detecting G×E in a genome-wide association study has thus far been difficult. We conclude with a description of software programs that can be used to implement most of the methods described in the paper.