American journal of epidemiology最新文献

We aimed to determine if thyroid autoimmunity is associated with a child's risk for subsequent development of islet or celiac autoantibodies. Children at high genetic risk of type 1 diabetes were followed for thyroid autoimmunity (thyroid peroxidase antibodies [TPOAb] and thyroglobulin antibodies [TGAb]), islet autoimmunity (IA), and celiac disease autoimmunity [CDA] in The Environmental Determinants of Diabetes in the Young (TEDDY) study. Out of 5482 children tested for thyroid autoimmunity, IA, and CDA, 39% developed at least one autoantibody. At age 14 years, thyroid autoimmunity co-occurred with IA in 59 children (15 more cases than expected by chance alone, p = 0.02), and with CDA in 125 children (26 cases above expected, p = 0.01). The risk of developing IA or CDA after thyroid autoimmunity varied by which thyroid autoantibody appeared first: TPOAb-first was associated with both IA (HR 1.92, 95% CI 1.09, 3.40) and CDA (1.69, 95% CI 1.03, 2.76), whereas TGAb-first was not associated with the risk of either. IA and CDA are frequently found in connection to thyroid autoimmunity in children and young adolescents. The relationships of thyroid autoimmunity with IA and CDA depend on which thyroid autoantibody appears first.

Survey data are self-reported data collected directly from respondents by a questionnaire or an interview and are commonly used in epidemiology. Such data are traditionally collected via a single mode (eg, face-to-face interview alone), but use of mixed-mode designs (eg, offering face-to-face interview or online survey) has become more common. This introduces two key challenges. First, individuals may respond differently to the same question depending on the mode; these differences due to measurement are known as "mode effects." Second, different individuals may participate via different modes; these differences in sample composition between modes are known as "mode selection." Where recognised, mode effects are often handled by straightforward approaches such as conditioning on survey mode. However, while reducing mode effects, this and other equivalent approaches may introduce collider bias in the presence of mode selection. The existence of mode effects and the consequences of naïve conditioning may be underappreciated in epidemiology. This paper offers a simple introduction to these challenges using directed acyclic graphs by exploring a range of possible data structures. We discuss the potential implications of using conditioning- or imputation-based approaches and outline the advantages of quantitative bias analyses for dealing with mode effects.

Latinx individuals' educational experiences in the US were shaped by structural and interpersonal discrimination, potentially contributing to disparities in dementia and related risk factors. Racial and ethnic school composition reflects resources and co-ethnic belonging. We used Health and Retirement Study data from 315 Latinx adults aged≥50 years who reported their school racial/ethnic composition through 12th grade. We estimated associations between each composition type (1-year unit) and memory scores (0-20), depressive symptoms (0-8), and poor self-rated health (SRH; yes/no), adjusting for sociodemographic factors. Greater duration in schools serving majority Latinx students was not associated with memory (β=-0.01 [-0.09, 0.07]) or depressive symptoms (β=-0.03 [-0.09, 0.04]) but was associated with higher risk for poor SRH (RR=1.05 [1.01, 1.08]). Greater duration in schools serving majority Black students was associated with lower memory (β=-0.16 [-0.31, -0.01]), more depressive symptoms (β=0.18 [0.07, 0.29]), and higher risk for poor SRH (RR=1.06 [1.00, 1.11]). Greater duration in schools serving majority White students was associated with higher memory scores (β=0.09 [0.01, 0.16]), lower depressive symptoms (β=-0.05 [-0.11, 0.00]), and decreased risk for poor SRH (RR=0.91 [0.87, 0.96]). Among Latinx adults, the impact of school composition varied by whether schools served other marginalized populations, reflecting countervailing influences.

We investigate the relationship between placental abruption and CVD events using a population-based, retrospective cohort study of individuals who delivered a singleton birth between 1993 and 2020 in New Jersey, USA. We fit multistate weighted Cox models to estimate the risks of non-fatal CVD hospitalization, all-cause mortality, and non-fatal CVD hospitalization to all-cause mortality. We examine these associations in two non-overlapping cohorts of individuals with their first delivery only (parity 1) and those with the first two consecutive deliveries (parity 1-2). Associations were corrected for unmeasured confounding bias. Of 2 874 671 deliveries, 1.0% (n = 28 913) had an abruption. The median follow-up was 16 years (range, 0-28 years). Compared to no abruption, placental abruption among first deliveries was associated with adjusted hazard ratios (HR) of 1.27 (95% confidence interval [CI]: 1.11-1.46) for the transition from delivery to non-fatal CVD hospitalization, 1.77 (95% CI, 1.28-2.44) for all-cause mortality, and 1.52 (95% CI, 0.98-2.37) for non-fatal CVD hospitalization to all-cause mortality. The corresponding HRs for recurrent placental abruption (parity 1-2 cohort) were stronger, although less precise. Corrections for unmeasured confounding slightly attenuated these risks. These findings underscore the importance of placental abruption as a potential risk factor for maternal CVD risks.

Childhood asthma is influenced by early-life social conditions, yet few studies have evaluated housing affordability as a modifiable structural exposure. We used data from six biennial waves (2006-2018) of the Longitudinal Study of Australian Children to assess whether changes in housing affordability and rental assistance were associated with incident asthma in childhood. Fixed-effects logistic regression models were used to estimate within-child associations between time-varying housing exposure and asthma outcomes. The main analytic sample included 3,773 children asthma-free at baseline; a subsample of 522 children in low-income renting households was used to evaluate rental assistance. Transitions into affordable housing were associated with a 31% reduction in asthma risk (OR 0.69; 95% CI 0.52-0.90). Among low-income private renters, new receipt of assistance was associated with 65% lower odds of asthma onset (OR 0.35; 95% CI 0.14-0.85). No associations were observed for asthma severity. Sensitivity analyses using lagged exposures, alternative definitions of affordability, and unadjusted income models yielded consistent findings. These findings support housing affordability as a potential policy lever for asthma prevention and demonstrate the utility of within-person designs for strengthening causal inference in observational evaluation of structural interventions.

Concerns about vaccine safety and efficacy and institutional trust are well-established determinants of vaccine uptake, yet little is known about the extent to which institutional trust buffers or amplifies the behavioral effects of vaccine-related uncertainty. Using data from 4,253 adults in the CHASING COVID Cohort who completed their primary vaccine series, we examined whether trust in public health institutions or healthcare providers modified the association between vaccine concerns and receipt of an additional COVID-19 vaccine dose. Adjusted log-binomial regression models estimated risk ratios (aRRs) and 95% confidence intervals, with effect modification assessed on multiplicative and additive scales. The association between not endorsing vaccine concerns and continued vaccination was stronger among those without trust in public health institutions (aRR=2.38; 95% CI: 1.90-2.97) than among those with trust (aRR=1.34; 95% CI: 1.22-1.46), while antagonism on the additive scale (RERI = -0.68; p<0.01) indicated overlapping pathways as having no vaccine concerns added little to the likelihood of continued vaccination among those with trust in public health institutions. Provider trust showed modest additive effects (RERI=0.16; p=0.08), suggesting greater-than-additive influences. Together, findings highlight institutional trust as an important contextual determinant shaping how vaccine-related beliefs are enacted and underscoring its importance for promoting ongoing vaccination uptake.

Traditional epidemiologic designs typically assume that the exposed and unexposed groups are mutually exclusive, forming the foundation for causal inference. Target Trial Emulation (TTE), an increasingly adopted framework for estimating causal effects from observational data, may not always require this assumption. Although often applied in settings with non-mutually exclusive treatment assignment, the implications of such structures for causal estimation are underexplored. In real-world contexts, patients may receive combination or single-agent treatments, or neither, leading to ambiguous group distinctions that challenge effect validity. We conducted a simulation study evaluating multiple TTE implementation strategies under non-mutually exclusive treatment assignment. Treatment overlap and covariate alignment were systematically varied to assess how emulation strategies perform under violations of mutual exclusivity. Our results show that non-mutually exclusive assignment can introduce substantial bias unless treatment overlap and positivity are explicitly addressed during propensity score estimation and outcome modeling. Notably, when covariate overlap is sufficient, non-mutually exclusive assignment can recover marginal effects with performance comparable to or exceeding mutually exclusive assignment. However, when overlap is poor, even advanced strategies fail to recover the true marginal effect. These findings underscore the importance of aligning study design, estimand, and treatment-assignment structure when applying TTE in real-world settings.