Pub Date : 2025-10-15DOI: 10.1016/j.annepidem.2025.10.016
Cynthia N. Ramirez , Michael Goodman , Kristine Magnusson , Wendy Leyden , Alexandra N. Lea , Darios Getahun , Courtney McCracken , Suma Vupputuri , Lee Cromwell , Timothy L. Lash , Oumaima Kaabi , Guneet K. Jasuja , Michael J. Silverberg
Purpose
Electronic health records (EHR) offer a unique opportunity to systematically collect sexual orientation and gender identity (SOGI) data. This study examined the prevalence and determinants of SOGI reporting in an EHR-based cohort of transgender and gender diverse (TGD) individuals.
Methods
We identified TGD people with and without SOGI documentation across four Kaiser Permanente health plans from January 1, 2022–2024. TGD status was determined through clinical notes, diagnostic codes, and SOGI data based on a previously established cohort. Factors associated with SOGI reporting were assessed using log-binomial regression, yielding prevalence ratios (PR) and the 95 % confidence intervals (CI).
Results
Among 23,060 TGD individuals, 71 % had SOGI documentation in the EHR. Reporting varied by sociodemographic and clinical characteristics. For example, compared to those < 20 years, SOGI reporting was higher for those aged 21–59 (PRs 1.10–1.21; 95 % CIs 1.06–1.24) and lower for those > 60 (0.93; 0.88–0.99). Documentation was slightly lower for those assigned male at birth (0.98; 0.97–1.00) and varied by race and ethnicity (e.g., Hispanic: 0.97; 0.95–0.99; Other: 1.02; 0.98–1.05 vs. White).
Conclusions
KP’s EHRs captured SOGI data for over 70 % of TGD individuals, though more research is needed to understand factors associated with missing data not captured in structured fields.
{"title":"Availability of sexual orientation and gender identity (SOGI) information in a cohort of transgender and gender diverse people: An analysis of electronic health records","authors":"Cynthia N. Ramirez , Michael Goodman , Kristine Magnusson , Wendy Leyden , Alexandra N. Lea , Darios Getahun , Courtney McCracken , Suma Vupputuri , Lee Cromwell , Timothy L. Lash , Oumaima Kaabi , Guneet K. Jasuja , Michael J. Silverberg","doi":"10.1016/j.annepidem.2025.10.016","DOIUrl":"10.1016/j.annepidem.2025.10.016","url":null,"abstract":"<div><h3>Purpose</h3><div>Electronic health records (EHR) offer a unique opportunity to systematically collect sexual orientation and gender identity (SOGI) data. This study examined the prevalence and determinants of SOGI reporting in an EHR-based cohort of transgender and gender diverse (TGD) individuals.</div></div><div><h3>Methods</h3><div>We identified TGD people with and without SOGI documentation across four Kaiser Permanente health plans from January 1, 2022–2024. TGD status was determined through clinical notes, diagnostic codes, and SOGI data based on a previously established cohort. Factors associated with SOGI reporting were assessed using log-binomial regression, yielding prevalence ratios (PR) and the 95 % confidence intervals (CI).</div></div><div><h3>Results</h3><div>Among 23,060 TGD individuals, 71 % had SOGI documentation in the EHR. Reporting varied by sociodemographic and clinical characteristics. For example, compared to those < 20 years, SOGI reporting was higher for those aged 21–59 (PRs 1.10–1.21; 95 % CIs 1.06–1.24) and lower for those > 60 (0.93; 0.88–0.99). Documentation was slightly lower for those assigned male at birth (0.98; 0.97–1.00) and varied by race and ethnicity (e.g., Hispanic: 0.97; 0.95–0.99; Other: 1.02; 0.98–1.05 vs. White).</div></div><div><h3>Conclusions</h3><div>KP’s EHRs captured SOGI data for over 70 % of TGD individuals, though more research is needed to understand factors associated with missing data not captured in structured fields.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"112 ","pages":"Pages 23-27"},"PeriodicalIF":3.0,"publicationDate":"2025-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145314098","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-14DOI: 10.1016/j.annepidem.2025.10.012
Siona Prasad, Sabina A. Murphy, David A. Morrow, Benjamin S. Scirica, Marc S. Sabatine, David D. Berg, Andrea Bellavia
Purpose
Clinical prediction models (CPM) are essential tools for diagnosis and prognosis in clinical epidemiology. Machine learning (ML) and deep learning (DL) approaches provide flexible methods that can complement regression-based methods for CPM when complex predictors such as clinical biomarkers are of interest. However, concerns have been raised on the ability of ML and DL to address desired properties of CPMs such as parsimony, generalizability, and interpretability.
Methods
In this study, we evaluated and applied selected regression-based, ML and DL approaches for time-to-event outcomes in a clinical study integrating protein biomarkers and lipids in an existing CPM for cardiovascular risk.
Results
We observed considerable advantages from the application of gradient boosting machines (GBM: C-statistic=0.72; Brier Score=0.052), which provided the best balance between model flexibility, discrimination, calibration, and parsimony, the latter being directly related to one of the model parameters (shrinking rate). Further, GBM results can be used for individual risk prediction, providing an interpretable tool for CPM implementation.
Conclusions
We compared ML and DL methods for CPM with time-to-event outcomes and discussed practical aspects of their implementation in clinical epidemiology including generalizability and interpretability. Adequately trained ML approaches can provide advantages in prediction modeling, especially when integrating complex predictors.
{"title":"Application of machine learning and deep learning approaches for prediction modeling with time-to-event outcomes in clinical epidemiology. Methods comparison and practical considerations for generalizability and interpretability","authors":"Siona Prasad, Sabina A. Murphy, David A. Morrow, Benjamin S. Scirica, Marc S. Sabatine, David D. Berg, Andrea Bellavia","doi":"10.1016/j.annepidem.2025.10.012","DOIUrl":"10.1016/j.annepidem.2025.10.012","url":null,"abstract":"<div><h3>Purpose</h3><div>Clinical prediction models (CPM) are essential tools for diagnosis and prognosis in clinical epidemiology. Machine learning (ML) and deep learning (DL) approaches provide flexible methods that can complement regression-based methods for CPM when complex predictors such as clinical biomarkers are of interest. However, concerns have been raised on the ability of ML and DL to address desired properties of CPMs such as parsimony, generalizability, and interpretability.</div></div><div><h3>Methods</h3><div>In this study, we evaluated and applied selected regression-based, ML and DL approaches for time-to-event outcomes in a clinical study integrating protein biomarkers and lipids in an existing CPM for cardiovascular risk.</div></div><div><h3>Results</h3><div>We observed considerable advantages from the application of gradient boosting machines (GBM: C-statistic=0.72; Brier Score=0.052), which provided the best balance between model flexibility, discrimination, calibration, and parsimony, the latter being directly related to one of the model parameters (shrinking rate). Further, GBM results can be used for individual risk prediction, providing an interpretable tool for CPM implementation.</div></div><div><h3>Conclusions</h3><div>We compared ML and DL methods for CPM with time-to-event outcomes and discussed practical aspects of their implementation in clinical epidemiology including generalizability and interpretability. Adequately trained ML approaches can provide advantages in prediction modeling, especially when integrating complex predictors.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 186-192"},"PeriodicalIF":3.0,"publicationDate":"2025-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145309858","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-11DOI: 10.1016/j.annepidem.2025.10.014
Adam S. Vaughan PhD , Nicholas Sutton MPH , Rebecca C. Woodruff PhD , LaTonia C. Richardson PhD , Janet S. Wright MD , Fátima Coronado MD
Purpose
This study examines national trends in mortality from cardiovascular disease (CVD) and select subtypes among U.S. young adults aged 18–34 years from 2000 to 2023.
Methods
National mortality data from the National Vital Statistics System were used to identify CVD, heart disease, stroke, and hypertension-related CVD deaths among U.S. residents aged 18–34 from 2000 to 2023. Crude and age-standardized death rates were calculated overall and by age group, sex, and race and ethnicity. Temporal trends were calculated as percent change using a log-linear model.
Results
From 2000–2023, age-standardized CVD and heart disease death rates among young adults did not statistically change (percent change: −2.2 % [95 % CI: −7.8, 3.7] and −2.4 % [95 % CI: −8.3 %, 3.8 %], respectively). Stroke death rates decreased (percent change: −15.7 % [-21.0 %, −10.0 %])). However, hypertension-related CVD death rates increased by 78.5 % [95 % CI: 63.6 %, 94.7 %]). Patterns across demographic groups were broadly similar.
Conclusion
Despite stability or modest declines in CVD death rates among young adults, hypertension-related CVD death rates increased sharply during 2000–2023. These findings merit public health action and underscore the need for better identification and management of hypertension and other CVD risk factors among young adults.
{"title":"Cardiovascular disease mortality trends in young adults aged 18–34 years, United States, 2000–2023","authors":"Adam S. Vaughan PhD , Nicholas Sutton MPH , Rebecca C. Woodruff PhD , LaTonia C. Richardson PhD , Janet S. Wright MD , Fátima Coronado MD","doi":"10.1016/j.annepidem.2025.10.014","DOIUrl":"10.1016/j.annepidem.2025.10.014","url":null,"abstract":"<div><h3>Purpose</h3><div>This study examines national trends in mortality from cardiovascular disease (CVD) and select subtypes among U.S. young adults aged 18–34 years from 2000 to 2023.</div></div><div><h3>Methods</h3><div>National mortality data from the National Vital Statistics System were used to identify CVD, heart disease, stroke, and hypertension-related CVD deaths among U.S. residents aged 18–34 from 2000 to 2023. Crude and age-standardized death rates were calculated overall and by age group, sex, and race and ethnicity. Temporal trends were calculated as percent change using a log-linear model.</div></div><div><h3>Results</h3><div>From 2000–2023, age-standardized CVD and heart disease death rates among young adults did not statistically change (percent change: −2.2 % [95 % CI: −7.8, 3.7] and −2.4 % [95 % CI: −8.3 %, 3.8 %], respectively). Stroke death rates decreased (percent change: −15.7 % [-21.0 %, −10.0 %])). However, hypertension<strong>-</strong>related CVD death rates increased by 78.5 % [95 % CI: 63.6 %, 94.7 %]). Patterns across demographic groups were broadly similar.</div></div><div><h3>Conclusion</h3><div>Despite stability or modest declines in CVD death rates among young adults, hypertension-related CVD death rates increased sharply during 2000–2023. These findings merit public health action and underscore the need for better identification and management of hypertension and other CVD risk factors among young adults.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"112 ","pages":"Pages 38-45"},"PeriodicalIF":3.0,"publicationDate":"2025-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145287638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
There are inconsistent findings regarding the associations between lipids and type 2 diabetes mellitus (T2DM), partially due to ignoring the joint effects of longitudinal patterns in lipids simultaneously. This study aimed to investigate the association of joint multi-trajectory of different lipids with the risk of type 2 diabetes.
Methods
We enrolled 71,043 participants free of T2DM from the Kailuan study. Using group-based multi-trajectory modeling, joint multi-trajectory of triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) during 2006–2010 was developed to predict the risk of T2DM during 2010–2019.
Results
Five distinct multi-trajectory groups were identified over 4-year exposure, and 6473 (9.11 %) cases of incident T2DM occurred during a median follow-up of 8.97 years. The highest risk of T2DM was observed in Group 5 with the highest level of TG, optimal-increasing LDL, and high-increasing HDL-C (hazard ratio [HR], 2.14; 95 % confidence interval [CI], 1.89–2.41), followed by Group 3 with the lowest level of HDL-C and an optimal TG and LDL-C (HR, 1.39; 95 % CI, 1.11–1.43), and Group 4 with the highest level of LDL-C, optimal-increasing TG and high-increasing HDL-C (HR, 1,26; 95 % CI, 1.11–1.43), compared to Group 2 with the lowest level of TG and optimal-increasing LDL-C and high-increasing HDL-C. The observed associations existed regardless of baseline lipid levels.
Conclusion
Our results showed the important role of high-increasing TG and low-decreasing HDL-C, rather than high-increasing LDL-C in the development of T2DM, which would help better understand the heterogeneous risk of T2DM and facilitate targeted prevention programs.
{"title":"Longitudinal triglyceride and HDL cholesterol, but not LDL cholesterol associated with the risk of incident type 2 diabetes: Evidence from a multi-trajectory analysis","authors":"Xue Tian , Shuohua Chen , Xue Xia , Qin Xu , Shouling Wu , Anxin Wang","doi":"10.1016/j.annepidem.2025.10.007","DOIUrl":"10.1016/j.annepidem.2025.10.007","url":null,"abstract":"<div><h3>Purpose</h3><div>There are inconsistent findings regarding the associations between lipids and type 2 diabetes mellitus (T2DM), partially due to ignoring the joint effects of longitudinal patterns in lipids simultaneously. This study aimed to investigate the association of joint multi-trajectory of different lipids with the risk of type 2 diabetes.</div></div><div><h3>Methods</h3><div>We enrolled 71,043 participants free of T2DM from the Kailuan study. Using group-based multi-trajectory modeling, joint multi-trajectory of triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) during 2006–2010 was developed to predict the risk of T2DM during 2010–2019.</div></div><div><h3>Results</h3><div>Five distinct multi-trajectory groups were identified over 4-year exposure, and 6473 (9.11 %) cases of incident T2DM occurred during a median follow-up of 8.97 years. The highest risk of T2DM was observed in Group 5 with the highest level of TG, optimal-increasing LDL, and high-increasing HDL-C (hazard ratio [HR], 2.14; 95 % confidence interval [CI], 1.89–2.41), followed by Group 3 with the lowest level of HDL-C and an optimal TG and LDL-C (HR, 1.39; 95 % CI, 1.11–1.43), and Group 4 with the highest level of LDL-C, optimal-increasing TG and high-increasing HDL-C (HR, 1,26; 95 % CI, 1.11–1.43), compared to Group 2 with the lowest level of TG and optimal-increasing LDL-C and high-increasing HDL-C. The observed associations existed regardless of baseline lipid levels.</div></div><div><h3>Conclusion</h3><div>Our results showed the important role of high-increasing TG and low-decreasing HDL-C, rather than high-increasing LDL-C in the development of T2DM, which would help better understand the heterogeneous risk of T2DM and facilitate targeted prevention programs.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 138-145"},"PeriodicalIF":3.0,"publicationDate":"2025-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145269393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-10DOI: 10.1016/j.annepidem.2025.10.003
Jacob Englert , Howard Chang
Purpose:
To extend the existing quantile g-computation framework for studying environmental exposure mixtures to estimate local effects of ambient air pollution mixtures on birth weight. This framework has traditionally been applied to estimate global mixture effects without accounting for spatial heterogeneity.
Methods:
First, pregnancy-wide maternal exposure to five common air pollutants is estimated for nearly 1.5 million births occurring in Georgia, USA between 2005 and 2016. Then, a recently developed spatially varying coefficient model based on Bayesian additive regression trees (BART) is applied to estimate spatially heterogeneous mixture effects using quantile g-computation. Results are compared with those obtained from traditional conditional autoregressive models, as well as spatially agnostic modeling approaches.
Results:
We find evidence of county-level spatially varying mixture associations, where for 21 of 159 counties in Georgia, elevated concentrations of a mixture of PM2.5, nitrogen dioxide, sulfur dioxide, ozone, and carbon monoxide were associated with a reduction in birthweight by as much as -14.77 grams (95% credible interval: -21.24, -9.78) per decile increase in all five air pollutants.
Conclusions:
Spatially varying coefficient models based on BART outperform alternative approaches when modeling the relationships between air pollution mixtures and birth weight for the majority of counties in Georgia.
{"title":"Modeling heterogeneity in air pollution mixture effects on birth weight: A spatially varying coefficient approach","authors":"Jacob Englert , Howard Chang","doi":"10.1016/j.annepidem.2025.10.003","DOIUrl":"10.1016/j.annepidem.2025.10.003","url":null,"abstract":"<div><h3>Purpose:</h3><div>To extend the existing quantile g-computation framework for studying environmental exposure mixtures to estimate local effects of ambient air pollution mixtures on birth weight. This framework has traditionally been applied to estimate global mixture effects without accounting for spatial heterogeneity.</div></div><div><h3>Methods:</h3><div>First, pregnancy-wide maternal exposure to five common air pollutants is estimated for nearly 1.5 million births occurring in Georgia, USA between 2005 and 2016. Then, a recently developed spatially varying coefficient model based on Bayesian additive regression trees (BART) is applied to estimate spatially heterogeneous mixture effects using quantile g-computation. Results are compared with those obtained from traditional conditional autoregressive models, as well as spatially agnostic modeling approaches.</div></div><div><h3>Results:</h3><div>We find evidence of county-level spatially varying mixture associations, where for 21 of 159 counties in Georgia, elevated concentrations of a mixture of PM<sub>2.5</sub>, nitrogen dioxide, sulfur dioxide, ozone, and carbon monoxide were associated with a reduction in birthweight by as much as -14.77 grams (95% credible interval: -21.24, -9.78) per decile increase in all five air pollutants.</div></div><div><h3>Conclusions:</h3><div>Spatially varying coefficient models based on BART outperform alternative approaches when modeling the relationships between air pollution mixtures and birth weight for the majority of counties in Georgia.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 180-185"},"PeriodicalIF":3.0,"publicationDate":"2025-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145276559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-10DOI: 10.1016/j.annepidem.2025.10.013
Taryn Lambert , Nikki Stephenson , Janice Skiffington , Donna Slater , Lara M. Leijser , Amy Metcalfe
Purpose
Attrition of participants over time poses a challenge in longitudinal research. This study aimed to explore how partner participation influenced maternal retention.
Methods
Using data from the P3 Cohort (a longitudinal pregnancy cohort), study retention was assessed at each stage of data collection up to 1 year postpartum. Participants were grouped according to their partner's level of participation in the study (participants who did not consent to the study team contacting their partners, participants whose partners were contacted but did not consent to participate, and participants whose partners actively participated). Cox proportional hazards models were used to evaluate the association between partner participation and participant attrition.
Results
Of 2194 eligible participants, 38.9 % did not provide consent for the study team to contact their partner, and 42.1 % of partners that were contacted agreed to participate in the cohort. Retention rates in the cohort were high (97.5 % at 1 year postpartum) but varied by partner participation. Partner participation was associated with a significantly reduced hazard of attrition (HR=0.38, 95 % CI:0.15–0.92).
Conclusions
Active partner participation significantly enhances maternal participant retention. Inclusion of partners in pregnancy research may help reduce attrition and gain a more comprehensive understanding of family dynamics.
{"title":"Impact of couple vs. individual participation in pregnancy research: A comparative analysis of participant characteristics and study retention","authors":"Taryn Lambert , Nikki Stephenson , Janice Skiffington , Donna Slater , Lara M. Leijser , Amy Metcalfe","doi":"10.1016/j.annepidem.2025.10.013","DOIUrl":"10.1016/j.annepidem.2025.10.013","url":null,"abstract":"<div><h3>Purpose</h3><div>Attrition of participants over time poses a challenge in longitudinal research. This study aimed to explore how partner participation influenced maternal retention.</div></div><div><h3>Methods</h3><div>Using data from the P3 Cohort (a longitudinal pregnancy cohort), study retention was assessed at each stage of data collection up to 1 year postpartum. Participants were grouped according to their partner's level of participation in the study (participants who did not consent to the study team contacting their partners, participants whose partners were contacted but did not consent to participate, and participants whose partners actively participated). Cox proportional hazards models were used to evaluate the association between partner participation and participant attrition.</div></div><div><h3>Results</h3><div>Of 2194 eligible participants, 38.9 % did not provide consent for the study team to contact their partner, and 42.1 % of partners that were contacted agreed to participate in the cohort. Retention rates in the cohort were high (97.5 % at 1 year postpartum) but varied by partner participation. Partner participation was associated with a significantly reduced hazard of attrition (HR=0.38, 95 % CI:0.15–0.92).</div></div><div><h3>Conclusions</h3><div>Active partner participation significantly enhances maternal participant retention. Inclusion of partners in pregnancy research may help reduce attrition and gain a more comprehensive understanding of family dynamics.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 163-167"},"PeriodicalIF":3.0,"publicationDate":"2025-10-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145281711","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-09DOI: 10.1016/j.annepidem.2025.10.011
María Isolina Santiago-Pérez , Carla Guerra-Tort , Esther López-Vizcaíno , Lucía Martín-Gisbert , Ana Teijeiro , Guadalupe García , Julia Rey-Brandariz , Alberto Ruano-Ravina , Mónica Pérez-Ríos
Purpose
The estimation of smoking-attributable mortality (SAM) is subject to the acceptance of different assumptions that may influence the estimates. We aimed to assess lung cancer mortality attributable to smoking by using both a prevalence-independent method (PIM) and a prevalence-dependent method (PDM) with different lags between exposure (smoking prevalence) and outcome (lung cancer mortality).
Methods
We estimated the population attributable fractions (PAF) and the lung cancer SAM by sex and age group (35–64, 65–84 years), year-by-year from 2011 to 2020, in four scenarios in Spain. In three of these scenarios, a PDM was applied using different lags: no lag, a 15-year lag and a 20-year lag. In the fourth scenario, a PIM was applied.
Results
In the period 2011–2020 in Spain, the SAM was higher when the 20-year lag PDM was considered (173,526 deaths) and lower when no lag PDM or a PIM was applied (161,249 and 157,390 deaths, respectively). In men, the PAFs were similar between the no lag PDM and the PIM (86.7 % and 87.3 %, respectively). However, when a PDM 15-year or 20-year lag was considered, the PAF increased to 91.0 % and 92.3 %, respectively. In women, the lowest PAF was obtained with the PIM (57.3 %), and the highest with the PDM 20-year lag (79.4 %).
Conclusions
SAM estimates differ depending on the methods and lags used. Applying a 15-year or 20-year lag PDM yields higher SAM estimates than when no lag PDM or a PIM is used. Therefore, when feasible, smoking prevalence data that incorporate a lag of 15 or 20 years between exposure and result should be used for accurate estimates.
{"title":"Lung cancer mortality attributable to smoking: a multi-scenario analysis with variable lag periods","authors":"María Isolina Santiago-Pérez , Carla Guerra-Tort , Esther López-Vizcaíno , Lucía Martín-Gisbert , Ana Teijeiro , Guadalupe García , Julia Rey-Brandariz , Alberto Ruano-Ravina , Mónica Pérez-Ríos","doi":"10.1016/j.annepidem.2025.10.011","DOIUrl":"10.1016/j.annepidem.2025.10.011","url":null,"abstract":"<div><h3>Purpose</h3><div>The estimation of smoking-attributable mortality (SAM) is subject to the acceptance of different assumptions that may influence the estimates. We aimed to assess lung cancer mortality attributable to smoking by using both a prevalence-independent method (PIM) and a prevalence-dependent method (PDM) with different lags between exposure (smoking prevalence) and outcome (lung cancer mortality).</div></div><div><h3>Methods</h3><div>We estimated the population attributable fractions (PAF) and the lung cancer SAM by sex and age group (35–64, 65–84 years), year-by-year from 2011 to 2020, in four scenarios in Spain. In three of these scenarios, a PDM was applied using different lags: no lag, a 15-year lag and a 20-year lag. In the fourth scenario, a PIM was applied.</div></div><div><h3>Results</h3><div>In the period 2011–2020 in Spain, the SAM was higher when the 20-year lag PDM was considered (173,526 deaths) and lower when no lag PDM or a PIM was applied (161,249 and 157,390 deaths, respectively). In men, the PAFs were similar between the no lag PDM and the PIM (86.7 % and 87.3 %, respectively). However, when a PDM 15-year or 20-year lag was considered, the PAF increased to 91.0 % and 92.3 %, respectively. In women, the lowest PAF was obtained with the PIM (57.3 %), and the highest with the PDM 20-year lag (79.4 %).</div></div><div><h3>Conclusions</h3><div>SAM estimates differ depending on the methods and lags used. Applying a 15-year or 20-year lag PDM yields higher SAM estimates than when no lag PDM or a PIM is used. Therefore, when feasible, smoking prevalence data that incorporate a lag of 15 or 20 years between exposure and result should be used for accurate estimates.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 175-179"},"PeriodicalIF":3.0,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145276519","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-09DOI: 10.1016/j.annepidem.2025.09.024
Samuel Schildhauer , Lauren Linde , Stephanie Bertsch-Merbach , Gail L. Sondermeyer Cooksey , Christina Morales , Estela Saguar , April Hatada , Blanca Molinar , Debra A. Wadford , Seema Jain , Jake M. Pry , CalSRVSS County Collaborators
Purpose
The coronavirus disease 2019 (COVID-19) pandemic caused disruptions in the transmission of seasonal respiratory viruses. COVID-19 is characterized by a range of non-specific symptoms, making it difficult to differentiate from other seasonal respiratory viruses. The goal of this analysis was to further understand trends in the circulation and differences in reported symptoms between respiratory pathogens during the COVID-19 pandemic.
Methods
From May 2020 to July 2022 a sentinel surveillance program collected data and respiratory specimens in outpatient settings across California and tested them for 19 respiratory viruses. Data were analyzed by identified respiratory pathogen to describe trends and clinical presentations. Multiple logistic regression was used to estimate odds of each respiratory pathogen by reported symptoms.
Results
We included results from 19,183 specimens, of which 8599 (44.8 %) tested positive for a pathogen, including 3742 (20.0 %) for SARS-CoV-2 and 3057 (15.9 %) for rhinovirus/enterovirus. Those reporting systemic symptoms had significantly higher adjusted odds of testing positive for influenza (aOR=9.2; 95 %CI, 6.7–13.1) or SARS-CoV-2 (aOR=2.4; 95 %CI, 2.2–2.6).
Conclusions
The variability in testing positive for a pathogen among people reporting different symptom profiles suggests a potential benefit of complete testing algorithms to complement syndromic diagnostics, improving public health awareness and clinical guidance.
{"title":"Clinical presentation of seasonal respiratory viruses in California, May 2020–July 2022","authors":"Samuel Schildhauer , Lauren Linde , Stephanie Bertsch-Merbach , Gail L. Sondermeyer Cooksey , Christina Morales , Estela Saguar , April Hatada , Blanca Molinar , Debra A. Wadford , Seema Jain , Jake M. Pry , CalSRVSS County Collaborators","doi":"10.1016/j.annepidem.2025.09.024","DOIUrl":"10.1016/j.annepidem.2025.09.024","url":null,"abstract":"<div><h3>Purpose</h3><div>The coronavirus disease 2019 (COVID-19) pandemic caused disruptions in the transmission of seasonal respiratory viruses. COVID-19 is characterized by a range of non-specific symptoms, making it difficult to differentiate from other seasonal respiratory viruses. The goal of this analysis was to further understand trends in the circulation and differences in reported symptoms between respiratory pathogens during the COVID-19 pandemic.</div></div><div><h3>Methods</h3><div>From May 2020 to July 2022 a sentinel surveillance program collected data and respiratory specimens in outpatient settings across California and tested them for 19 respiratory viruses. Data were analyzed by identified respiratory pathogen to describe trends and clinical presentations. Multiple logistic regression was used to estimate odds of each respiratory pathogen by reported symptoms.</div></div><div><h3>Results</h3><div>We included results from 19,183 specimens, of which 8599 (44.8 %) tested positive for a pathogen, including 3742 (20.0 %) for SARS-CoV-2 and 3057 (15.9 %) for rhinovirus/enterovirus. Those reporting systemic symptoms had significantly higher adjusted odds of testing positive for influenza (aOR=9.2; 95 %CI, 6.7–13.1) or SARS-CoV-2 (aOR=2.4; 95 %CI, 2.2–2.6).</div></div><div><h3>Conclusions</h3><div>The variability in testing positive for a pathogen among people reporting different symptom profiles suggests a potential benefit of complete testing algorithms to complement syndromic diagnostics, improving public health awareness and clinical guidance.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 146-153"},"PeriodicalIF":3.0,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145259968","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-08DOI: 10.1016/j.annepidem.2025.10.010
Laura Staxen Bruun , Cecilie Bladt , Katrine Strandberg-Larsen , Vibeke Tornhøj Christensen , Jane Greve , Elisabeth Reichel Hansen , Janne S. Tolstrup
Purpose
To examine the independent associations of BMI and body size perception with dropout from upper secondary education.
Methods
Data from the Danish National Youth Study 2014, including 63,976 students aged 15–19 years, were linked to information on school dropout from the Student Register. Multilevel logistic regressions were employed to examine how BMI and body size perception were each associated with school dropout. Mediation analyses were conducted to assess the role of body size perception in the BMI-dropout association.
Results
Both low and high BMI were associated with higher odds of school dropout, following a U-shaped pattern across the BMI continuum. For instance, females with a BMI above 30 kg/m2 had an OR of 2.37 (95 % CI: 1.74;3.27), compared to those with a BMI of 18.5–24.9 kg/m2. Adjusting for body size perception, ORs attenuated for students with a BMI above the median. Body size perception mediated a substantial proportion of the BMI-dropout association. Students who perceived themselves as too thin or fat also had higher odds of dropout. For instance, females who perceived themselves as ‘much too fat’ had an OR of 2.66 (95 % CI: 2.17;3.25), compared to those who perceived themselves as ‘about right size’. Adjusting for BMI had only minor impact on ORs.
Conclusion
Low and high BMI, as well as the perception of not being the right size, were associated with higher odds of dropout from upper secondary education. Adjusted for body size perception, the association between BMI and dropout attenuated, suggesting that body size perception plays a crucial role in this relationship. Body size perception partly explains the BMI-dropout relationship, highlighting a need for more comprehensive approaches focusing adolescents’ perceptions of their bodies alongside actual weight to reduce school dropout.
{"title":"Associations between BMI, body size perception, and dropout from upper secondary education: A prospective cohort study of 15–19-year-old adolescents","authors":"Laura Staxen Bruun , Cecilie Bladt , Katrine Strandberg-Larsen , Vibeke Tornhøj Christensen , Jane Greve , Elisabeth Reichel Hansen , Janne S. Tolstrup","doi":"10.1016/j.annepidem.2025.10.010","DOIUrl":"10.1016/j.annepidem.2025.10.010","url":null,"abstract":"<div><h3>Purpose</h3><div>To examine the independent associations of BMI and body size perception with dropout from upper secondary education.</div></div><div><h3>Methods</h3><div>Data from the Danish National Youth Study 2014, including 63,976 students aged 15–19 years, were linked to information on school dropout from the Student Register. Multilevel logistic regressions were employed to examine how BMI and body size perception were each associated with school dropout. Mediation analyses were conducted to assess the role of body size perception in the BMI-dropout association.</div></div><div><h3>Results</h3><div>Both low and high BMI were associated with higher odds of school dropout, following a U-shaped pattern across the BMI continuum. For instance, females with a BMI above 30 kg/m<sup>2</sup> had an OR of 2.37 (95 % CI: 1.74;3.27), compared to those with a BMI of 18.5–24.9 kg/m<sup>2</sup>. Adjusting for body size perception, ORs attenuated for students with a BMI above the median. Body size perception mediated a substantial proportion of the BMI-dropout association. Students who perceived themselves as too thin or fat also had higher odds of dropout. For instance, females who perceived themselves as ‘much too fat’ had an OR of 2.66 (95 % CI: 2.17;3.25), compared to those who perceived themselves as ‘about right size’. Adjusting for BMI had only minor impact on ORs.</div></div><div><h3>Conclusion</h3><div>Low and high BMI, as well as the perception of not being the right size, were associated with higher odds of dropout from upper secondary education. Adjusted for body size perception, the association between BMI and dropout attenuated, suggesting that body size perception plays a crucial role in this relationship. Body size perception partly explains the BMI-dropout relationship, highlighting a need for more comprehensive approaches focusing adolescents’ perceptions of their bodies alongside actual weight to reduce school dropout.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 129-137"},"PeriodicalIF":3.0,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145269395","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-08DOI: 10.1016/j.annepidem.2025.10.004
Nushrat Nazia, Charmaine Dean
Purpose
Jointly monitoring adverse COVID-19 outcomes among seniors is critical for assessing outbreak severity. These outcomes are often influenced by socioeconomic and demographic conditions and may co-occur in space, indicating shared structural risks that inform targeted responses.
Methods
We analyzed severe COVID-19 outcomes among adults aged 65 + in Ontario (January 2020–March 2022) using data from the Ontario Health Data Platform supported by ICES. A Bayesian shared component model with Integrated Nested Laplace Approximation at the forward sortation area level included socioeconomic and demographic covariates.
Results
The shared component explained ∼75 % of the total modeled spatial variability. High risks clustered in southern Ontario, while lower risks occurred in central and northern regions. Material deprivation was positively associated with death (RR 1.12, 95 % CrI: 1.04–1.21) and multiple hospitalizations (RR 1.20, 95 % CrI: 1.13–1.29). Racialized/newcomer population concentration was positively associated with death (RR 1.25, 95 % CrI: 1.14–1.38) and with single hospitalizations (RR 1.18, 95 % CrI: 1.11–1.24). The percentage of seniors was inversely associated with hospitalization (RR 0.98, 95 % CrI: 0.96–0.99) but not death.
Conclusions
Findings highlight structural inequities in pandemic severity and suggest targeted, equity-oriented strategies in guiding pandemic preparedness and response.
{"title":"Joint spatial modelling of COVID-19 severity among seniors: A Bayesian shared component approach using health administrative data from Ontario, Canada","authors":"Nushrat Nazia, Charmaine Dean","doi":"10.1016/j.annepidem.2025.10.004","DOIUrl":"10.1016/j.annepidem.2025.10.004","url":null,"abstract":"<div><h3>Purpose</h3><div>Jointly monitoring adverse COVID-19 outcomes among seniors is critical for assessing outbreak severity. These outcomes are often influenced by socioeconomic and demographic conditions and may co-occur in space, indicating shared structural risks that inform targeted responses.</div></div><div><h3>Methods</h3><div>We analyzed severe COVID-19 outcomes among adults aged 65 + in Ontario (January 2020–March 2022) using data from the Ontario Health Data Platform supported by ICES. A Bayesian shared component model with Integrated Nested Laplace Approximation at the forward sortation area level included socioeconomic and demographic covariates.</div></div><div><h3>Results</h3><div>The shared component explained ∼75 % of the total modeled spatial variability. High risks clustered in southern Ontario, while lower risks occurred in central and northern regions. Material deprivation was positively associated with death (RR 1.12, 95 % CrI: 1.04–1.21) and multiple hospitalizations (RR 1.20, 95 % CrI: 1.13–1.29). Racialized/newcomer population concentration was positively associated with death (RR 1.25, 95 % CrI: 1.14–1.38) and with single hospitalizations (RR 1.18, 95 % CrI: 1.11–1.24). The percentage of seniors was inversely associated with hospitalization (RR 0.98, 95 % CrI: 0.96–0.99) but not death.</div></div><div><h3>Conclusions</h3><div>Findings highlight structural inequities in pandemic severity and suggest targeted, equity-oriented strategies in guiding pandemic preparedness and response.</div></div>","PeriodicalId":50767,"journal":{"name":"Annals of Epidemiology","volume":"111 ","pages":"Pages 120-128"},"PeriodicalIF":3.0,"publicationDate":"2025-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145259979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号