American journal of epidemiology最新文献

In a recent paper, Ghoroubi et al. (Am J Epidemiol 2025 Jan 8;194(1):302-310) used the indirect attributable fraction (AF) method to provide estimates of fractions of all-cause mortality attributable to work-related factors. This commentary discusses the limitations and potential of this paper and provides insights and guidance to make optimal use of indirect AF estimation in occupational epidemiology. The crucial steps are the choice of the datasets and input data related to the prevalence of exposure and relative risk (RR), requiring comparability of time period, population characteristics, and the definition and measurement of exposure. Published systematic literature reviews with meta-analyses are essential or, if not available, conducting meta-analyses to provide estimates of RR. Finally, it is important to verify the assumptions for the chosen AF formula including evidence of causality, consideration of confounding and (in)dependence between exposures when several exposures are studied at the same time. We conclude by suggesting that the paper by Ghoroubi et al. may have provided a proof of concept for 1 work-related factor only, but considerable additional research will be required to represent work-related factors overall.

Evidence demonstrating the beneficial effects of improved air quality on lipid health is scarce. This study addresses this gap by examining whether reducing PM2.5 exposure can decrease the risk of dyslipidemia. We conducted a longitudinal quasi-experimental study using the Taiwan MJ and Hong Kong MJ cohorts from 2000 to 2018. A total of 8808 adults with consistently high PM2.5 exposure (≥ 25 μg/m3) were paired with 4612 adults whose PM2.5 exposure decreased from high to low levels (< 25 μg/m3) using propensity score matching. Cox regression models with time-dependent covariates were used to analyze the associations between PM2.5 reduction and the risk of dyslipidemia, as well as individual lipid abnormalities. We found that participants with reducing PM2.5 exposure had a significantly lower risk of dyslipidemia compared to their counterparts (hazard ratio [HR], 0.75; 95% confidence interval [CI], 0.68-0.84). Nonlinear concentration-response relationships were observed. Similar associations were found for elevated TC (HR, 0.61; 95% CI, 0.51-0.74) and LDL-C (HR, 0.69; 95% CI, 0.57-0.84), and decreased HDL-C (HR, 0.59; 95% CI, 0.47-0.75). Reducing PM2.5 exposure significantly lowers the risk of dyslipidemia and improves lipid profiles, providing direct evidence of the health benefits associated with air quality improvement.

Mendelian randomization can reveal the etiological association between body mass index (BMI) and lung cancer. However, the associations between the trajectories of BMI and the risk of lung cancer remain inconclusive. We employed growth mixture modeling to identify trajectories of pre-diagnostic BMI in 163 545 individuals (117 445 women from the Nurses' Health Study and 46 100 men from the Health Professionals Follow-Up Study). We assessed the associations between BMI trajectories and lung cancer risk, as well as the effects within subgroups. Four trajectories were identified: normal-moderate increasing (class 1), overweight-marked increasing (class 2), overweight-obese turning (class 3), and obese-persistent (class 4). We observed a decreased risk of lung cancer in class 2 (adjusted hazard ratio [aHR], 0.53; 95% CI, 0.38-0.75; P = 2.32 ×10-4) and class 3 (aHR, 0.67; 95% CI, 0.48-0.94; P = .022). In stratification analysis, we observed that the effects of class 4 on lung cancer risk vary among histological subtypes. Additionally, within the class 1 population, the top quintile of BMI also demonstrated different effects among histological subtypes. Increasing lifetime BMI was associated with a decreased risk of lung cancer, with this association varying by histological subtypes, indicating histology-specific mechanisms in lung carcinogenesis.

Missing data are a pervasive problem in epidemiology, with multiple imputation (MI) a commonly used analysis method. MI is valid when data are missing at random (MAR). However, definitions of MAR with multiple incomplete variables are not easily interpretable and descriptions of graphical model-based conditions are not accessible to applied researchers. Previous literature shows that MI may be valid in subsamples, even if not in the full dataset. Practical guidance on applying MI with multiple incomplete variables is lacking. We present an algorithm using directed acyclic graphs to determine when MI will estimate an exposure-outcome coefficient without bias. We extend the algorithm to assess whether MI in a subsample of the data, in which some variables are complete, and the remaining are imputed, will be valid and unbiased for the exposure-outcome coefficient. We apply the algorithm to several simple exemplars, and in a more complex real-life example highlight that only subsample-MI of the outcome would be valid. Our algorithm provides researchers with the tools to decide whether to use MI in practice when there are multiple incomplete variables. Further work could focus on the likely size and direction of biases and the impact of different missing data patterns.

Nursing home (NH) residents are an important population for pharmacoepidemiologic research due to their prevalence of multimorbidity and polypharmacy. Medicare claims are commonly used to study medication use in this population, but medications dispensed during hospitalizations or post-acute care are unobservable due to bundled payment structures. We developed algorithms to identify NH days when medication dispensings can be observed in claims. Using a cohort of NH residents in the United States from 2013 to 2020, we linked Medicare fee-for-service (FFS) claims with Minimum Data Set clinical assessments. NH days were classified as "observable medication use time" if residents were enrolled in Medicare parts A, B, and D were not receiving post-acute care and were not hospitalized. Among 12.3 million NH residents and 2.7 billion NH days, 1.1 billion days (72.4% of Medicare-enrolled days and 39.6% of all NH days) were identified as observable medication use time. Within the first 100 days of NH admission, 27.3% of days were medication-observable, increasing to 89.4% after 100 days. On average, we identified 68% more person-time, and 51% more residents, compared to standard 100-day definitions for "long-stay" NH residents. Our algorithms enhance researchers' ability to measure medication exposure time, improving the validity of pharmacoepidemiologic studies.

We investigated the association between unclean cooking fuel use and sleep problems in a nationally representative sample of adults aged ≥65 years from 6 low- and middle-income countries (China, Ghana, India, Mexico, Russia, and South Africa). Cross-sectional, community-based data from the WHO Study on global AGEing and adult health (SAGE) were analyzed. Unclean cooking fuel referred to kerosene/paraffin, coal/charcoal, wood, agriculture/crop, animal dung, and shrubs/grass. Outcomes related to sleep included self-reported nocturnal sleep problems, lethargy, poor sleep quality, and sleep duration. Multivariable logistic regression analysis was conducted. Data on 14 585 individuals aged ≥65 years were analyzed (mean [SD] age: 72.6 [11.5] years; 55.0% females). After adjustment for potential confounders, unclean cooking fuel use was associated with significant 1.51 (95% CI, 1.03-2.22) times higher odds for nocturnal sleep problems, while it was also associated with 1.64 (95% CI, 1.20-2.26) times higher odds for long sleep duration (ie, >9 vs >6 to 9 h), but not with other sleep-related outcomes. These findings suggest that the implementation of the United Nations Sustainable Development Goal 7, which advocates affordable, reliable, sustainable, and modern energy for all, may also have a positive impact on sleep problems, as well as a plethora of other health and environmental impacts. This article is part of a Special Collection on Cross-National Gerontology.

Electronic health records (EHRs) present opportunities to study uterine fibroids and endometriosis within diverse populations. When using EHR data, it is important to validate outcome classification via diagnosis codes. We performed a validation study of 3 approaches ([1] International Classification of Diseases-10 (ICD-10) code alone, [2] ICD-10 code + diagnostic procedure, and [3] ICD-10 code + all diagnostic information) to identify incident uterine fibroids and endometriosis patients among n = 750 NYU Langone Health 2016-2023. Chart review was used to determine the true diagnosis status. When using a binary classification system (incident vs nonincident patient), Approaches 2 and 3 had higher positive predictive values (PPVs) for uterine fibroids (0.86 and 0.87 vs 0.78) and for endometriosis (0.70 and 0.73 vs 0.66), but Approach 1 outperformed the other 2 in negative predictive values (NPVs) for both outcomes. When using a 3-level classification system (incident vs prevalent vs disease-free patients), PPV for prevalent patients was low for all approaches, while PPV/NPV of disease-free patients was generally above 0.8. Using ICD-10 codes alone yielded higher NPVs but resulted in lower PPVs compared with the other approaches. Continued validation of uterine fibroids/endometriosis EHR studies is warranted to increase research into these understudied gynecologic conditions.

During the COVID-19 pandemic, accurate measurement of vaccination status was important for guiding prevention efforts. We assessed the accuracy of electronic health record (EHR) COVID-19 vaccination compared with survey self-reported vaccination status using data from a cross-sectional study among pregnant women and non-pregnant adults in the Vaccine Safety Datalink between 2021 and 2022, where self-report was considered the reference standard. We measured the sensitivity and specificity of EHR vaccine data compared with the self-reported measure and estimated vaccination rates from EHR data. EHR data were obtained initially in November 2021, updated in April 2022, and record reviewed in July 2022. Vaccination coverage increased in pregnant/formerly pregnant women and non-pregnant adult respondents by 23.9% and 9.2%, respectively, over 9 months. Estimates of sensitivity based on initial EHR data were 66.0% and 77.3% for pregnant women and non-pregnant people overall and between 41% and 66% for pregnant, non-Hispanic Black, and Hispanic, Spanish-speaking respondents. With matured, chart reviewed EHR data from April 2022, the sensitivity and specificity of EHR vaccine status relative to self-report were > 93%. EHR data were a reasonable source of COVID-19 vaccination status during the pandemic and showed high accuracy with self-reported data after allowing EHR data to mature.

Widowhood is among the most consequential stressful events for mental health. Although certain resources have been identified as protective of mental health following widowhood, these findings are based on US samples. This study uses novel harmonized data to evaluate differences in depressive symptoms and related factors among those recently widowed (ie, within the last 2 years) in the United States (Health and Retirement Study) and India (Longitudinal Aging Study in India). We find US widows have greater elevation in depressive symptoms (-0.36 SD) than widows in India (-0.15) on average. We identify 3 protective resources for widows that are dependent on country context: having close friends vs no friends (-0.58 vs -0.13) and living with others vs alone (-0.79 vs -0.23) are both larger for widows in the United States. Self-rated health that is good, fair, or poor is related to higher depressive symptoms for widows in the United States vs India (between 0.55 and 1.12). Findings suggest protective resources among recently widowed individuals designed to protect mental health following this stressful event will require consideration of country context. In particular, social engagement-based interventions may offer more significant benefits to widows in the United States.

Epidemiologists have access to various methods to reduce bias and improve statistical efficiency in effect estimation, from standard multivariable regression to state-of-the-art doubly-robust efficient estimators paired with highly flexible, data-adaptive algorithms ("machine learning"). However, due to numerous assumptions and trade-offs, epidemiologists face practical difficulties in recognizing which method, if any, may be suitable for their specific data and hypotheses. Importantly, relative advantages are necessarily context-specific (data structure, algorithms, model misspecification), limiting the utility of universal guidance. Evaluating performance through real-data-based simulations is useful but out-of-reach for many epidemiologists. We present a user-friendly, offline Shiny app REFINE2 (Realistic Evaluations of Finite sample INference using Efficient Estimators) that enables analysts to input their own data and quickly compare the performance of different algorithms within their data context in estimating a prespecified average treatment effect (ATE). REFINE2 automates plasmode simulation of a plausible target ATE given observed covariates and then examines bias and confidence interval coverage (relative to this target) given user-specified models. We present an extensive case study to illustrate how REFINE2 can be used to guide analyses within epidemiologist's own data under three typical scenarios: residual confounding; spurious covariates; and mis-specified effect modification. As expected, the apparent best method differed across scenarios and are suboptimal under residual confounding. REFINE2 may help epidemiologists not only chose amongst imperfect models, but also better understand common underappreciated problems, such as finite sample bias using machine learning.