American journal of epidemiology最新文献

The restricted mean survival time (RMST) analysis has been used extensively in clinical research involving time-to-event endpoints. The threshold time up to which the restricted mean survival is calculated has a critical impact on the analysis results. However, identifying an optimal threshold time for treatment comparison, which corresponds to the greatest restricted mean lifetime difference between groups, remains unclear in practice, and no analytical method has been developed on this topic. We present a novel method for determining the threshold time in the RMST analysis to compare two groups. Simulation studies indicate that this method leads to high statistical power and controlled type I error rate compared with existing methods. The proposed method is illustrated in two applications: (1) a clinical oncology study for non-small-cell lung cancer treatments comparison given a programmed death-ligand 1 biomarker measurement, and (2) a gerontology study of instrumental activities for care recipients with dementia.

Traumatic brain injury (TBI) is a major public health concern affecting millions of people each year. Disparities in TBI outcomes based on social determinants of health (SDoH), such as race and socioeconomic position, highlight the need to explore the causative structural inequities. We employed a socio-epidemiologic approach, with particular focus on the putative role of structural racism, to investigate the prevalence, sociodemographic patterns, and neuropsychiatric outcomes of TBI in the All of Us database. This study included 11 286 individuals with a documented TBI diagnosis, determined based on a curated phenotype definition using the International Statistical Classification of Diseases Clinical Modification criteria. Outcome measures included TBI prevalence and sociodemographic distribution; TBI severity; and neuropsychiatric diagnoses related to TBI. Nearly equivalent TBI prevalences were observed across racial categories. Black participants with TBI had higher socioeconomic deprivation indices and higher prevalence of certain neuropsychiatric conditions, such as substance use disorders and headache disorders, compared to White participants. This study underscores the importance of considering SDoH, particularly race and socioeconomic position, in TBI research. These findings highlight the need for efforts to address structural inequities that impact disparities in TBI and call for future research investigating how healthcare practices relate to disparities in TBI outcomes. This article is part of a Special Collection on Methods in Social Epidemiology.

Emergency department (ED) visits during influenza seasons represent a critical yet less examined indicator of the acute burden of influenza. This study investigates the burden of influenza-associated ED visits in 6 US cities during influenza seasons from 2005-2006 to 2016-2017. Using a time-series design, we estimated associations between daily ED visits and weekly influenza activity data from the Influenza Hospitalization Surveillance Network (FluSurv-NET). A counterfactual approach was then used to calculate attributable expected ED visits. Highest influenza-associated rates were observed among the youngest (0-4 years) and oldest (65+ years) age groups. Combining estimates across seasons, the influenza-associated ED visit rate for respiratory diseases was almost 6 times larger compared to the subset of ED visits that resulted in hospitalization: 364 per 100 000 population (95% CI, 294-435) for total ED visits vs 58 per 100 000 population (95% CI, 45-71) for hospitalization. This difference was particularly large for the 0-4 years age group: 911 per 100 000 population (95% CI, 558-1263) for total ED visits vs 43 per 100 000 population (95% CI, 15-71) for hospitalization. This study highlights the substantial burden of influenza on emergency health care services and the importance of integrating such data into public health planning and influenza management strategies.

Negative control outcomes (NCOs) are useful tools for hidden bias detection, but empirical evidence validating NCOs for COVID-19 is lacking. To address this gap, we examined the blinded phase of the randomized, placebo-controlled Coronavirus Vaccine Efficacy (COVE; NCT04470427) trial of the mRNA-1273 COVID-19 vaccine. We confirmed that acute respiratory illness with a positive test for a non-SARS-CoV-2 respiratory pathogen on a multiplex PCR panel was a valid NCO for COVID-19, considering that it was unaffected by vaccination (vaccine efficacy, VE = 3.3% (95% CI, -22.3 to 23.6)) yet strongly associated with COVID-19 (odds ratio = 2.95 (95% CI, 2.00, 4.24)). Subsequently, we leveraged non-SARS-CoV-2 infections to detect bias in time-varying VE estimates from COVE's blinded and booster phases. Balanced incidence of non-SARS-CoV-2 infection between vaccinated and unvaccinated COVID-19-free risk sets suggested low selection bias in VE estimates of two-dose mRNA-1273 against COVID-19 during the blinded phase (VE = 92.5% (95% CI, 88.8, 94.9) 14 days post-dose-two, stable for 5 months). In COVE's booster phase, higher non-SARS-CoV-2 incidence was observed after the single booster (intensity ratio, IR = 2.38 (95% CI, 1.75, 3.25) 14 days post-boost), suggesting that booster VE estimates may underestimate the true VE against COVID-19. Our findings demonstrate the potential of off-target infections for unraveling complex biases in COVID-19 vaccine studies. Trial registration: NCT04470427, https://clinicaltrials.gov/study/NCT04470427.

Algorithmic estimations of dementia status are widely used in public health and epidemiologic research, but inadequate algorithm performance across racial/ethnic groups has been a barrier. We present improvements in the accuracy of group-specific "probable dementia" estimation using a transfer learning approach. Transfer learning involves combining models trained on a large "source" data set with imprecise outcome assessments, alongside models trained on a smaller "target" data set with high-quality outcome assessments. Transfer learning improves model accuracy by leveraging large-source data while refining estimations with smaller, target data. We illustrate with data from the Health and Retirement Study (source data: n = 6630) and the Harmonized Cognitive Assessment Protocol (target data: n = 2388). Models for dementia status estimation were evaluated through overall accuracy (Brier score), calibration (intercept, slope), and discriminative ability (area under the receiver operating characteristic curve [AUR] and area under the precision-recall curve [AUPRC]). The transfer-learned algorithm showed higher accuracy compared to the best previously reported algorithm among both non-Hispanic Black participants (Brier 0.049 vs 0.061; AUC 0.84 vs 0.81; AUPRC 0.52 vs 0.39) and Hispanic participants (Brier 0.052 vs 0.056; AUC 0.89 vs 0.87; AUPRC 0.61 vs 0.56). Transfer learning can improve dementia status estimation for groups historically underrepresented in research. This article is part of a Special Collection on Methods in Social Epidemiology.

Obesity increases cardiovascular disease (CVD) and cancer mortality risk, with prevalence rising globally over recent decades. In the United States, steep obesity increases contributed to adverse trends in obesity-related mortality and to slowing decline in overall CVD mortality, particularly among younger generations. Switzerland experienced slower obesity increases, but the contribution of obesity to mortality trends remains uncharacterized. We analyzed all adult deaths recorded in Swiss mortality statistics between 1995-2019. Obesity-related CVD and cancer deaths were identified using multiple cause of death approaches. Annual changes in age-standardized mortality rates were estimated via segmented regression. Age-period-cohort models assessed cohort variations. Overall, CVD mortality declined steadily while cancer mortality decline attenuated after 2005, primarily reflecting slower declines in obesity-unrelated cancer mortality. Obesity-related mortality increased from 1995-2005 and then decreased, while obesity-unrelated rates decreased throughout 1995-2019. These diverging trends did not slow overall CVD mortality decline. Age-period-cohort modeling revealed lower obesity-related mortality rates in younger versus older generations. In Switzerland, unlike in the United States, trends in obesity-related mortality did not slow the decline of overall CVD mortality. Obesity-related mortality rates did not increase in younger generations, highlighting the role of reduced childhood obesity prevalence and improved management of obesity-related conditions in Switzerland.

Multiple imputation (MI) models can be improved with auxiliary covariates (ACs), but their performance in high-dimensional data remains unclear. We aimed to develop and compare high-dimensional MI (HDMI) methods using structured and natural language processing (NLP)-derived AC in studies with partially observed confounders. We conducted a plasmode simulation with acute kidney injury as outcome and simulated 100 cohorts with a null treatment effect, incorporating creatinine labs, atrial fibrillation (AFib), and other investigator-derived confounders in the outcome generation. Missingness was imposed on creatinine based on creatinine itself and AFib. Different HDMI candidate ACs were created using structured and NLP-derived features, and we mimicked scenarios where AFib was unobserved by omitting it from all analyses. Using the least absolute shrinkage and selection operator, we selected HDMI covariates for MI and propensity score models. The treatment effect was estimated after propensity score matching in MI datasets, and HDMI methods were compared to baseline imputation and complete case analysis. High-dimensional MI using claims data showed the lowest bias (0.072). Combining claims and sentence embeddings led to an improvement in the efficiency with a root mean square error (RMSE) of 0.173 and 94% coverage. Natural language processing-derived AC alone did not outperform baseline MI. High-dimensional MI approaches may decrease bias in studies where confounder missingness depends on unobserved factors.

Food environment near schools (FENS) influence children's dietary habits and contribute to obesity. Socioeconomic characteristics of schools and school neighborhoods play a role in determining FENS. We compare the availability of fast-food restaurants (FFRs) and convenience stores (CSs) across schools' socioeconomic characteristics: whether the school is public or private and the school neighborhood's median household income. We obtained the number of FFRs and CSs within a 0.75-mile network buffer from schools' locations and the names of the outlets. Negative binomial regression models, stratified by urbanicity, were used to estimate the association between the number of outlets near schools and schools' socioeconomic characteristics. We explored brand names and types of outlets. Private schools' neighborhoods had more FFRs and CSs than public schools across all income and urbanization levels. Private and public schools in low-income urban neighborhoods had more outlets compared to those in higher-income urban areas. While the names of FFRs and CSs near both school types were broadly similar, private schools had more non-chain outlets. Programs and policies to promote healthy eating and reduce obesity- and diet-related diseases should target food environments near both private and public schools, especially those located in urban areas and low-income communities.