American journal of epidemiology最新文献

Phenotype classification with electronic health record (EHR) data is increasingly performed with machine learning (ML); however, their performance in diverse population remains understudied. We compared an international classification of diseases (ICD)-based algorithm with an ML phenotyping pipeline to classify myocardial infarction (MI) in a general and self-reported Black population. We determined the impact of differential performance by replicating a published MI risk factor study with MI defined by the ICD or ML algorithms. Individuals followed in the Veterans Health Administration (VHA) EHR with data from 2002 to 2019 were examined: 11 523 175 Veterans; mean age, 67.5 years; 93.8% male; 14.3% Black; 79.1% White. MI was classified using a published rule-based ICD algorithm and an ML pipeline, PheCAP, which incorporates natural language processing. Algorithms were trained and validated against n = 403 Veterans randomly selected and chart reviewed for MI (gold standard), oversampled for self-reported Black. Among chart-reviewed Veterans, the ICD algorithm had high positive predicted value (PPV) and low sensitivity (all race, PPV: 0.97, sensitivity: 0.17; Black Veterans, PPV: 0.94, sensitivity: 0.24). PheCAP MI had good PPV and higher sensitivity (all race, PPV: 0.90, sensitivity: 0.66; Black, PPV: 0.81, sensitivity: 0.79). Applying PheCAP MI to the entire VHA population to classify MI provided increased power to replicate findings from the published MI risk factor study compared to the ICD algorithm.

Middle-aged Americans today are reporting poorer mental, cognitive, and physical health compared to previous cohorts, but this trend has not been consistently observed in other nations. It is an open question whether pain shows similar cohort differences among US middle-aged adults compared to other nations. We used harmonized data on pain from nationally representative longitudinal panel surveys from the United States, 13 European nations (England, Continental, Mediterranean, and Nordic regions), South Korea, and Mexico to directly quantify cohort similarities and differences in midlife pain. Results from multilevel models demonstrated that midlife pain is higher among later-born cohorts in the United States than among earlier-born cohorts. The increased odds for later-born cohorts emerged in the early years of midlife, when people are in their early 50s. A similar pattern of increased odds of reporting pain for later-born cohorts was observed in England and Mexico. In contrast, decreased odds of reporting pain for later-born cohorts were observed in Continental, Mediterranean, and Nordic Europe as well as South Korea. Results for pain severity revealed a highly similar pattern. Our discussion focuses on potential explanations, including population-level discrepancies in use and quality of healthcare services and how pain is conceptualized across nations. This article is part of a Special Collection on Cross-National Gerontology.

For large-scale surveys such as the National Health and Aging Trends Study (NHATS), investigators may wish to combine data from two (or more) cohorts in a single analysis to obtain larger sample sizes. Unfortunately, it is not possible to combine the 2011 and 2015 NHATS cohorts while retaining the sample weights. We applied Bayesian hierarchical models with poststratification as an alternative strategy for obtaining population-based estimates from NHATS. As proof of principle, we compared prevalence estimates of frailty obtained from our Bayesian approach with those obtained from the 2011 and 2015 cohorts using the NHATS sample weights. Once validated, we applied our strategy to combine the cohorts into a single analytical dataset without overlap of participants, and generated Bayesian estimates of frailty for the combined cohort. Estimates from the Bayesian model closely matched the weighted NHATS estimates. The ability to combine cohorts while generating population-based estimates will allow investigators to address questions that require larger sample sizes, thereby enhancing the value of NHATS to the scientific community.

Cognitive aging research relies on longitudinal data, but extended follow-up is costly. The extent to which estimates and precision from data with shorter follow-up diverge from estimates based on longer follow-up is unknown. The necessary follow-up period may depend on model specification, i.e., modeling the timescale as time-since-baseline or current age. We used data on adults age 65+ from the 2006-2018 U.S. Health and Retirement Study. For associations of 8 commonly studied dementia risk factors with cognitive decline, we compared coefficients and variance estimates to results from benchmark models (i.e., using 7 waves and/or specifying time-since-baseline). We varied the hypothetical follow-up length (1-7 waves, representing 0-12 years of follow-up) and timescale specification. Among individuals 65-80 years old at baseline, estimates of cognitive change in models with <4 waves of follow-up differed meaningfully in terms of both coefficients and variance from estimates using full follow-up, regardless of timescale specification. Differences by length of follow-up time were less pronounced among those >80 years of age at baseline, in part due to sample attrition. In models assuming equal follow-up duration, estimates of cognitive change specified by current age differed from estimates using time-since-baseline but were more precise, especially with shorter follow-up.

This paper examines whether the association between markers of socioeconomic status (SES) and cardiovascular health are weaker among Black and Hispanic young adults than White young adults. Estimates are derived from the Future of Families-Cardiovascular Health among Young Adults Study (FF-CHAYA), 2021-2023. Participants (N = 1421, average age 22.9) were sampled from the Future of Families Child Wellbeing Study (FFCWS) that collected data in 7 waves in 20 large US cities. We used regression models to explore the link between a 7-metric cardiovascular health score and 6 measures of SES at the individual, family, and neighborhood levels for White (N = 243), Black (N = 730) and Hispanic (N = 388) young adults. The estimated interaction terms between the socioeconomic measures and race/ethnicity reveal that socioeconomic differentials in the cardiovascular health score among Black and Hispanic young adults are significantly smaller than those for their White counterparts. The finding that returns to health with increasing education and economic wellbeing are lower among Black and Hispanic than White young adults is consistent with theories of exposure to discrimination.

The potential for a bidirectional relationship between the experience of a negative wealth shock (a loss of ≥75% in total household wealth over 2 years) and accelerated memory decline among mid-to-later-life adults in the United States (US) remains unclear. We used population-based longitudinal data on 14 969 adults aged ≥51 in the US Health and Retirement Study from 1998 to 2020. One in 3 participants in this cohort experienced a negative wealth shock over the 22-year follow-up period (5184/14969, 34.6%). Participants who experienced a negative wealth shock had faster aging-related memory decline in the years before the shock than their counterparts who did not experience a negative wealth shock (an additional 0.04 standard deviation [SD] units per decade; 95% CI, -0.07 to -0.01) and an acute drop in their level of memory function concurrent with the negative wealth shock (-0.08 SD units; 95% CI, -0.10 to -0.05), yet slower memory aging after the negative wealth shock (0.04 SD units per decade; 95% CI, 0.01 to 0.06). We recommend strategies to support healthy memory aging of the large share of middle-aged and older US adults who are at risk of experiencing a negative wealth shock.

Scientific understanding of the relationship between environmental hazards and cognitive health at older ages in low- and middle-income countries (LMICs) is poor. Using data from the Longitudinal Aging Study of India and the World Health Organization's Survey on Global AGEing and adult health for 4 LMICs, we examine the association of direct and local exposure to polluting cooking fuels with cognitive health at older ages. We document the negative influence of both: Cognitive health is poorer among members of households that use polluting fuels and among residents of neighborhoods where the use of polluting fuels is more common. These associations cannot be explained by accounting for individual or local differences in socioeconomic status. Consistent with direct impacts of polluting fuels, we find that women in households where the use of polluting fuels is common have the lowest predicted cognitive scores. Our findings reveal the substantial direct influence and negative externalities of polluting fuel use in LMICs and help understand why overall cognitive health may be poor in these settings. Moving away from polluting fuels toward clean fuels may reduce individual risk and community-level exposure to air pollution, contributing to better cognitive health in older ages. This article is part of a Special Collection on Cross-National Gerontology.

Qualitative research methods are frequently described as "compatible" with quantitative epidemiologic methods. Instead of simply "compatible," we argue that qualitative methods are epidemiologic methods. Especially in social epidemiology, which embraces the relationships between psychosocial, historical, contextual, and intersectional factors and health, qualitative research methods have the potential to provide a more complete picture of the distribution of health and disease within a population and contexts contributing to population health. To this end, this paper compares qualitative research and epidemiologic research definitions, outlines epidemiologic uses of qualitative data, and addresses common concerns and misconceptions about qualitative research. We emphasize the shared characteristics and champion the use of shared standards across qualitative and quantitative approaches in epidemiology. This article is part of a Special Collection on Methods in Social Epidemiology.

Patterns of co-occurring early life trauma (ELT), unlike cumulative trauma, are understudied as potential stress-related health risks. DNA methylation (DNAm) is a posited mechanism linking socioenvironmental stress and disease pathogenesis. We assess whether ELT patterns differentially affect adult DNAm, both epigenome-wide and specifically stress-related genes (eg, glucocorticoid receptor gene, NR3C1). Data represent a case-cohort of non-Hispanic white participants (N = 2566) from the Sister Study, a cohort of US women (ages: 35-74; enrollment: 2003-2009; N = 50 884). Early life trauma measures included a count score and 4 latent ELT classes: low ELT (referent); sexual and emotional; high betrayal; and high ELT. We evaluated epigenome-wide DNAm, differentially methylated regions (DMRs), pathway enrichment, and NR3C1-specific methylation from whole blood. Twenty-two differentially methylated Cytosine-phosphate-Guanine (CpG) sites were associated with ELT classes and none with ELT score. Furthermore, 108 DMRs were associated with ELT score (n = 5) and ELT classes: sexual and emotional (n = 7), high betrayal (n = 37), and high ELT (n = 61). Cardiovascular signaling and leptin signaling pathways were both associated with high betrayal and high ELT classes. In NR3C1-specific analyses, 11 CpGs were associated with ELT score (n = 3) and specific ELT classes (n = 9). Results suggest that specific patterns of co-occurring trauma may contribute to meaningful variability in peripheral blood DNAm in adulthood.