Epidemiology最新文献

Background: The use of machine learning to estimate exposure effects introduces a dependence between the results of an empirical study and the value of the seed used to fix the pseudo-random number generator.

Methods: We used data from 10,038 pregnant women and a 10% subsample (N = 1004) to examine the extent to which the risk difference for the relation between fruit and vegetable consumption and preeclampsia risk changes under different seed values. We fit an augmented inverse probability weighted estimator with two Super Learner algorithms: a simple algorithm including random forests and single-layer neural networks and a more complex algorithm with a mix of tree-based, regression-based, penalized, and simple algorithms. We evaluated the distributions of risk differences, standard errors, and P values that result from 5000 different seed value selections.

Results: Our findings suggest important variability in the risk difference estimates, as well as an important effect of the stacking algorithm used. The interquartile range width of the risk differences in the full sample with the simple algorithm was 13 per 1000. However, all other interquartile ranges were roughly an order of magnitude lower. The medians of the distributions of risk differences differed according to the sample size and the algorithm used.

Conclusions: Our findings add another dimension of concern regarding the potential for "p-hacking," and further warrant the need to move away from simplistic evidentiary thresholds in empirical research. When empirical results depend on pseudo-random number generator seed values, caution is warranted in interpreting these results.

Introduction: Prenatal exposure to nonpersistent chemicals, including organophosphate pesticides, phthalates, and bisphenols, is associated with altered fetal and childhood growth. Few studies have examined these associations using longitudinal growth trajectories or considering exposure to chemical mixtures.

Methods: Among 777 participants from the Generation R Study, we used growth mixture models to identify weight and body mass index trajectories using weight and height measures collected from the prenatal period to age 13. We measured exposure biomarkers for organophosphate pesticides, phthalates, and bisphenols in maternal urine at three timepoints during pregnancy. Multinomial logistic regression was used to estimate associations between averaged exposure biomarker concentrations and growth trajectories. We used quantile g-computation to estimate joint associations with growth trajectories.

Results: Phthalic acid (OR = 1.4; 95% CI = 1.01, 1.9) and bisphenol A (OR = 1.5; 95% CI = 1.0, 2.2) were associated with higher odds of a growth trajectory characterized by smaller prenatal and larger childhood weight relative to a referent trajectory of larger prenatal and average childhood weight. Biomarkers of organophosphate pesticides, individually and jointly, were associated with lower odds of a growth trajectory characterized by average prenatal and lower childhood weight.

Conclusions: Exposure to phthalates and bisphenol A was positively associated with a weight trajectory characterized by lower prenatal and higher childhood weight, while exposure to organophosphate pesticides was negatively associated with a trajectory of average prenatal and lower childhood weight. This study is consistent with the hypothesis that nonpersistent chemical exposures disrupt growth trajectories from the prenatal period through childhood.

Background: Few epidemiologic studies have examined the association of ambient heat with spontaneous abortion, a common and devastating pregnancy outcome.

Methods: We conducted a case-crossover study nested within Pregnancy Study Online, a preconception cohort study (2013-2022). We included all participants reporting spontaneous abortion (N = 1,524). We defined the case window as the 7 days preceding the event and used time-stratified referent selection to select control windows matched on calendar month and day of week. Within each 7-day case and control window, we measured the mean, maximum, and minimum of daily maximum outdoor air temperatures. We fit splines to examine nonlinear relationships across the entire year and conditional logistic regression to estimate odds ratios (ORs) and 95% confidence interval (CI) of spontaneous abortion with increases in temperature during the warm season (May-September) and decreases during the cool season (November-March).

Results: We found evidence of a U-shaped association between outdoor air temperature and spontaneous abortion risk based on year-round data. When restricting to warm season events (n = 657), the OR for a 10-percentile increase in the mean of lag 0-6 daily maximum temperatures was 1.1 (95% CI: 0.96, 1.2) and, for the maximum, 1.1 (95% CI: 0.99, 1.2). The OR associated with any extreme heat days (>95th county-specific percentile) in the preceding week was 1.2 (95% CI: 0.95, 1.5). Among cool season events (n = 615), there was no appreciable association between lower temperatures and spontaneous abortion risk.

Conclusion: Our study provides evidence of an association between high outdoor temperatures and the incidence of spontaneous abortion.

Surveys are commonly used to facilitate research in epidemiology, health, and the social and behavioral sciences. Often, these surveys are not simple random samples, and respondents are given weights reflecting their probability of selection into the survey. We show that using survey weights can be beneficial for evaluating the quality of predictive models when splitting data into training and test sets. In particular, we characterize model assessment statistics, such as sensitivity and specificity, as finite population quantities and compute survey-weighted estimates of these quantities with test data comprising a random subset of the original data. Using simulations with data from the National Survey on Drug Use and Health and the National Comorbidity Survey, we show that unweighted metrics estimated with sample test data can misrepresent population performance, but weighted metrics appropriately adjust for the complex sampling design. We also show that this conclusion holds for models trained using upsampling for mitigating class imbalance. The results suggest that weighted metrics should be used when evaluating performance on test data derived from complex surveys.

Background: Extreme ambient heat is unambiguously associated with a higher risk of illness and death. The Optum Labs Data Warehouse (OLDW), a database of medical claims from US-based patients with commercial or Medicare Advantage health insurance, has been used to quantify heat-related health impacts. Whether results for the insured subpopulation are generalizable to the broader population has, to our knowledge, not been documented. We sought to address this question, for the US population in California from 2012 to 2019.

Methods: We examined changes in daily rates of emergency department encounters and in-patient hospitalization encounters for all-causes, heat-related outcomes, renal disease, mental/behavioral disorders, cardiovascular disease, and respiratory disease. OLDW was the source of health data for insured individuals in California, and health data for the broader population were gathered from the California Department of Health Care Access and Information. We defined extreme heat exposure as any day in a group of 2 or more days with maximum temperatures exceeding the county-specific 97.5th percentile and used a space-time-stratified case-crossover design to assess and compare the impacts of heat on health.

Results: Average incidence rates of medical encounters differed by dataset. However, rate ratios for emergency department encounters were similar across datasets for all causes [ratio of incidence rate ratios (rIRR) = 0.989; 95% confidence interval (CI) = 0.969, 1.009], heat-related causes (rIRR = 1.080; 95% CI = 0.999, 1.168), renal disease (rIRR = 0.963; 95% CI = 0.718, 1.292), and mental health disorders (rIRR = 1.098; 95% CI = 1.004, 1.201). Rate ratios for inpatient encounters were also similar.

Conclusions: This work presents evidence that OLDW can continue to be a resource for estimating the health impacts of extreme heat.

Background: Human papillomavirus (HPV) vaccination has been offered in over a hundred countries worldwide (including the United Kingdom, since September 2008). Controversy around adverse effects persists, with inconsistent evidence from follow-up of randomized controlled trials and confounding by indication limiting the conclusions drawn from larger-scale observational studies. This study aims to estimate the association between receiving a quadrivalent HPV vaccine and the reporting of short-term adverse effects and to demonstrate the utility of regression discontinuity design for examining side effects in routine data.

Methods: We applied a novel regression discontinuity approach to a retrospective population-based cohort using primary care data from the UK Clinical Practice Research Datalink linked to hospital and social deprivation data. We examined the new onset of gastrointestinal, neuromuscular, pain, and headache/migraine symptoms using READ and International Classification of Diseases, tenth revision diagnostic codes. For each year between 2012 and 2017, we compared girls in school year 8 (born July/August) who were eligible to receive the vaccine with girls in year 7 (born September/October) who were not eligible.

Results: Of the 21,853 adolescent girls in the cohort, 10,881 (50%) were eligible for HPV vaccination. There was no evidence of increased new gastrointestinal symptoms (adjusted odds ratio [OR]: 0.99; 95% confidence interval [CI]: 0.85, 1.15), headache/migraine symptoms (OR: 0.84; 95% CI: 0.70, 1.01), or pain symptoms (OR: 1.05; 95% CI: 0.95, 1.16) when comparing those eligible and ineligible for HPV vaccination.

Conclusion: This study found no evidence that HPV vaccination eligibility is associated with reporting short-term adverse effects among adolescent girls.

Background: Little attention has been devoted to framing multiple continuous social variables as a "mixture" for social epidemiologic analysis. We propose using the Bayesian kernel machine regression analytic framework that yields univariate, bivariate, and overall exposure mixture effects.

Methods: Using data from the 2023 Survey of Racism and Public Health, we conducted a Bayesian kernel machine regression analysis to study several individual, social, and structural factors as an exposure mixture and their relationships with psychological distress among individuals with at least one police arrest. Factors included racial and economic polarization, neighborhood deprivation, perceived discrimination, police perception, subjective social status, and substance use. We complemented this analysis with a series of unadjusted and adjusted models for each exposure mixture variable.

Results: We found that more self-reported discrimination experiences in the past year (posterior inclusion probability = 1.00) and greater substance use (posterior inclusion probability = 1.00) correlated with higher psychological distress. These associations were consistent with the findings from the unadjusted and adjusted linear regression analyses: past year perceived discrimination (unadjusted b = 2.58, 95% confidence interval [CI]: 1.86, 3.30; adjusted b = 2.20, 95% CI: 1.45, 2.94) and substance use (unadjusted b = 2.92, 95% CI: 2.21, 3.62; adjusted b = 2.59, 95% CI: 1.87, 3.31).

Conclusion: With the rise of big data and the expansion of variables in long-standing cohort and census studies, novel applications of methods from adjacent disciplines are a step forward in identifying exposure mixture associations in social epidemiology and addressing the health needs of socially vulnerable populations.