The assumption of conditional independence in conventional group-based trajectory modeling (GBTM) is often violated by paired repeated-measures data with heterogeneous trajectory patterns. While random-effects models can accommodate this dependence, they inflate within-group variability and blur distinct phenotypic shapes. We propose a joint GBTM framework that explicitly models hierarchical dependence in paired trajectories while allowing them to follow different latent patterns. We develop a robust two-stage approach to address estimation challenges caused by rare latent groups, and a one-stage EM algorithm that serves as a theoretical baseline under balanced group sizes. Simulations demonstrate that our methods correct the biases caused by ignoring hierarchical dependence. The proposed model was applied to real-world data from the Conservation of Hearing Study (CHEARS) Audiology Assessment Arm (AAA), a subcohort of the Nurses' Health Study II (NHS II), to identify distinct audiometric phenotypes and to investigate the association between the Dietary Approaches to Stop Hypertension (DASH) dietary adherence score and the latent audiometric patterns.

BACKGROUND:Previous cohort studies of physical activity and cognitive health have often been limited by small sample sizes, short follow-up durations, absence of long-term assessments of physical activity, and potential reverse causation. We aimed to prospectively examine the associations between long-term physical activity and multiple cognitive outcomes. METHODS:In this study, we used data from two ongoing US prospective cohorts: the Nurses' Health Study (established in 1976, comprising 121 700 female registered nurses aged 30-55 years at enrolment) and the Health Professionals Follow-Up Study (HPFS; established in 1986, comprising 51 492 male health professionals aged 40-75 years at enrolment). Discretionary physical activity was assessed approximately every 2-4 years using self-report questionnaires, beginning in 1986 for both cohorts. Total physical activity, walking, and vigorous aerobic exercise were each expressed as the summed metabolic equivalent of task hours per week. We ascertained incident dementia cases (a composite endpoint of self-reported physician-diagnosed dementia and deaths due to dementia) in eligible Nurses' Health Study participants (1990-2023) and HPFS participants (1990-2023); deaths due to dementia were confirmed using medical records, death certificates, and autopsy reports. Objective cognitive function was assessed in Nurses' Health Study participants only (1995-2008) using the Telephone Interview for Cognitive Status (TICS). Subjective cognitive decline was self-reported by Nurses' Health Study participants (followed up in 2012 and 2014) and HPFS participants (followed up in 2008, 2012, 2016, 2018, and 2020) via questionnaires with binary questions that assessed subjective cognitive concerns across multiple domains. Participants were excluded from the analysis if they had dementia, stroke, cancer, were missing total physical activity or walking data, or reported difficulties with walking at baseline. Outcomes were analysed using multivariable-adjusted regression models. We used a 4-year lag approach in our main analyses to minimise the potential for reverse causation. FINDINGS:Our final samples comprised 63 596 Nurses' Health Study participants and 43 440 HPFS participants for the dementia endpoints, 13 647 Nurses' Health Study participants for assessing objective cognitive function, and 29 801 Nurses' Health Study participants and 17 162 HPFS participants for assessing subjective cognitive decline. Compared with participants in the lowest quartile for total physical activity, those in the highest quartile had a lower risk of dementia (hazard ratio [HR] 0·72, 95% CI 0·68-0·76; p<0·0001 for trend) and subjective cognitive decline (relative risk [RR] 0·77, 95% CI 0·73-0·80; p<0·0001 for trend). Being in the highest versus the lowest tertile for walking or vigorous aerobic exercise was also associated with lower risk for dementia (HR 0·76, 95% CI 0·72-0·80 for walking and 0·89, 0·85-0·93 for vigorous aerobic exercise; p<0·0001 for trend for both activities) and subjective cognitive decline (RR 0·82, 95% CI 0·79-0·86 for walking and 0·89, 0·86-0·93 for vigorous aerobic exercise; p<0·0001 for trend for both activities). Additionally, being in the highest quartile of total physical activity versus the lowest quartile was associated with fewer years of age-related cognitive differences in global cognition (mean difference -1·15, 95% CI -1·69 to -0·60; p<0·0001 for trend), verbal memory (-1·27, -1·86 to -0·68; p<0·0001 for trend), and TICS score (-0·79, -1·36 to -0·23; p=0·016 for trend). Similar decreases were observed for vigorous aerobic exercise but not for walking. INTERPRETATION:Higher levels of physical activity were associated with a lower risk of dementia and better cognitive performance. These findings support physical activity as a potential strategy to promote long-term cognitive health. FUNDING:US National Institutes of Health.

OBJECTIVES:Higher endogenous estrogen may be associated with better cognition, but associations with menopausal hormone therapy (MHT) have been inconsistent, possibly due to differences in the timing of use. This prospective cohort study aimed to evaluate the associations between reproductive span, as a proxy for endogenous estrogen history, MHT use, and cognitive function. METHODS:We assessed cognitive change (1995-2008) with four telephone interviews (primary outcome: global composite score average of six test z-scores) in 14,217 Nurses' Health Study participants (mean age 74.3 y) and examined associations with reproductive span ([age at menopause]-[age at menarche]), and MHT use duration, separately by 0-10 years, and 11+ years after menopause. RESULTS:A longer reproductive span was associated with better cognitive trajectories (mean annual rate of change difference [95% CI]41-46 vs. ≤33 y=0.008 [0.00005, 0.015]; P-trend=0.02). MHT use 0-10 years postmenopause was associated with faster decline (mean difference8-10 vs. 0 y=-0.007 [-0.016, 0.002]; P-trend=0.02); use during 11+ years postmenopause was not associated. CONCLUSIONS:Although MHT use was not inversely associated, a longer reproductive span was associated with better cognitive trajectories.

Background Low-carbohydrate diet (LCD) and low-fat diet (LFD) patterns are practiced by many in the United States, although their health effects, as well as the role of diet quality in the effects, are not fully understood. Objectives This study aimed to prospectively examine the associations of these diets, which emphasize different quantities and qualities of macronutrients, as well as their objective metabolomic indices, with coronary heart disease (CHD) risk in U.S. individuals. Methods We followed 42,720 men in the Health Professionals Follow-Up Study (HPFS) (1986-2016), 64,164 women in the Nurses’ Health Study (NHS) (1986-2018), and 91,589 women in NHSII (1991-2019) for CHD incidence. Five LCD and 5 LFD indices were derived based on food frequency questionnaire (FFQ) assessments, each emphasizing different sources and qualities of macronutrients (animal products vs plant-based foods, whole grains vs refined carbohydrates, etc). Multimetabolite scores of LCD and LFD indices assessed using FFQ assessments were developed through elastic net regressions among 1,146 healthy participants in the lifestyle validation studies (LVS), substudies embedded in the NHS/NHSII/HPFS. Results During 5,248,916 person-years of follow-up, we documented 20,033 CHD cases. When comparing individuals with the highest LCD scores (emphasizing lower carbohydrate contents) and those with the lowest, the pooled multivariable-adjusted hazard ratios (95% CIs) for CHD were 1.05 (1.01-1.10) for overall LCD, 1.07 (1.02-1.12) for animal LCD, 0.94 (0.90-0.99) for vegetable LCD, 1.14 (1.09-1.20) for unhealthy LCD, and 0.85 (0.82-0.89) for healthy LCD. These estimates were 0.93 (0.89-0.98) for overall LFD, 0.94 (0.90-0.98) for animal LFD, 0.87 (0.83-0.91) for vegetable LFD, 1.12 (1.07-1.17) for unhealthy LFD, and 0.87 (0.83-0.91) for healthy LFD. The healthy versions of the LCD and LFD patterns were also linked to lower triglycerides, higher high-density lipoprotein cholesterol, and lower high-sensitivity C-reactive protein levels, as well as favorable metabolomic profiles, including increased 3-indolepropionic acid and decreased valine. Unhealthy patterns showed opposite associations. Multimetabolite scores of LCD and LFD indices were developed in the LVS (Spearman r = 0.57-0.68) and replicated in NHS, NHSII, and HPFS (r = 0.21-0.38). They showed associations with CHD risk highly consistent with those based on FFQ assessments. Conclusions These findings highlight the critical role of diet quality in determining health effects of low-carbohydrate and low-fat diets on CHD risk. The healthy versions of these diets may exert their health benefits through some common pathways that together entail favorable cardiovascular risk profile and lower CHD risk.

BACKGROUND:Type 2 diabetes (T2D) is associated with an increased risk of premature death. Diet may influence long-term health outcomes among individuals with T2D, but prospective evidence on dietary patterns and mortality remains limited. OBJECTIVES:To prospectively examine associations of eleven pre-defined or empirically-developed dietary patterns with all-cause and cause-specific mortality (e.g., cardiovascular disease [CVD] and cancer) among U.S. individuals with T2D. METHODS:We included 7,795 participants with incident T2D in the Nurses' Health Study (1984-2016) and Health Professionals Follow-Up Study (1986-2020), who were free of CVD and cancer at diagnosis. Diet was assessed using a validated food frequency questionnaire and updated every four years. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% CIs. RESULTS:Over 24 years of follow-up, 3,509 deaths were confirmed, including 1,079 from CVD and 649 from cancer. Adherence to healthy dietary patterns after diagnosis (comparing the 90th with the 10th percentile of dietary pattern scores) was consistently associated with lower all-cause mortality, with multivariable-adjusted HRs ranging from 0.78 (95% CI: 0.71, 0.85) for the Planetary Health Diet to 0.92 (95% CI: 0.84, 1.01) for the Dietary Approaches to Stop Hypertension. Furthermore, greater improvements in dietary pattern adherence from pre- to post-diagnosis (comparing the 90th with the 10th percentile of change scores) were also associated with lower all-cause mortality, with HRs (95% CI) ranging from 0.63 (0.58, 0.70) to 0.92 (0.85, 1.01). A similar pattern of inverse associations with CVD mortality was also observed for these dietary scores. Additionally, the reversed Empirical Dietary Index for Hyperinsulinemia was specifically associated with lower cancer mortality. CONCLUSIONS:Adherence to healthy dietary patterns is universally associated with better survival in individuals with T2D. These findings suggest the importance of consuming high-quality diets in the prevention of premature deaths among individuals with T2D.

Background: Few prospective human studies have characterized interrelationships between body weight, gut microbiome, plasma metabolome, and incident type 2 diabetes (T2D). Aims: To identify microbial profiles associated with adiposity, derive its responsive metabolomic signatures, and evaluate their associations with future weight change and incident T2D. Methods: Shotgun stool metagenomes were profiled in the Men’s Lifestyle Validation Study (MLVS; n=924). Body mass index (BMI)-associated microbial species were identified after adjusting for age, total energy, physical activity, alcohol and smoking, and were synthesized into a composite microbiome score. We evaluated its association with waist-to-hip ratio (WHR) and DEXA-measured fat mass in MLVS and validated its association with BMI in female Mind–Body Study (MBS; n=807). Next, the score was examined for its association with prevalent T2D in the multiethnic Micro-Cardio consortium (n=8,117). Prospectively, we tested its predictability of 2-year BMI increase and incident T2D risk in the Micro-N (n=3,318). To further characterize the downstream metabolism of this microbial profile, we linked it to >300 LC–MS metabolites in the MLVS and derived a microbiome-informed multi-metabolite score, which was validated in the MBS. We examined the metabolomic score for incident T2D risk in a prospective cohort, the Nurses’ Health Study (NHS; n=1,057). Results: We identified 23 microbial species associated with BMI in MLVS. The 23-species composite score was correlated with higher BMI (High vs.Low: MLVS β=3.45; MBS β=4.37 both P<0.001), and also correlated with WHR (r=0.45) and fat mass (r=0.73). Additionally, higher microbiome score consistently predicted 2-year BMI increase (β=0.22, P =0.003) and predicts incident T2D in Micro-N (HR [high vs. low] = 2.19, P=0.01), as well as prevalent T2D in the Micro-Cardio (HR [per SD] =1.19, P<0.001). A 32-plasma metabolite signature was responsive to this microbial profile in MLVS (r=0.66) and validated in MBS (r=0.38), and also predicted incident T2D in NHS (RR=2.18, P<0.0001) independent of baseline BMI. Conclusions: We identified and validated a gut microbiome score predictive of adiposity across multiple cohorts, showing consistent associations with plasma metabolomic signatures and increased risk of T2D. This highlights a potential role of the gut microbiome in weight regulation and T2D development, providing new insights into microbiome-informed strategies for T2D prevention.

INTRODUCTION:Associations of adherence to and changes in Mediterranean (MedDiet) and Mediterranean-Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay (MIND) diets with cognitive outcomes are unclear. METHODS:We prospectively followed 86,740 women (Nurses' Health Study, 1980-2023) and 43,500 men (Health Professionals Follow-Up Study, 1986-2023). Diet was assessed repeatedly using food frequency questionnaires. Dementia cases were identified from self-reported physician diagnoses and death records. Cognitive function was assessed by telephone tests, and subjective cognitive decline (SCD) was self-reported. RESULTS:Participants in the highest versus lowest MedDiet and MIND categories had 21% (hazard ratio [HR] = 0.79, 95% confidence interval [CI]: 0.73-0.84) and 14% (HR = 0.86, 95% CI: 0.81-0.91) lower dementia risk, respectively. Higher adherence was associated with 0.75 to 1.59 fewer years of cognitive aging, and 40% to 42% lower risk of SCD. Similar inverse associations were observed for improved adherence over 4 or 8 years. DISCUSSION:Adherence to Mediterranean and MIND diets was associated with cognitive benefits.

In this paper, we investigate the selection of minimal and efficient covariate adjustment sets for the imputation-based regression calibration method, which corrects for bias due to continuous exposure measurement error. We use directed acyclic graphs to illustrate how subject-matter knowledge aids in selecting these sets. For unbiased measurement error correction, researchers must collect, in both main and validation studies, (I) common causes of both the true exposure and the outcome, and (II) common causes of both measurement error and the outcome. For regression calibration under linear models, at minimum, covariate set (I) must be adjusted for in both the measurement error model (MEM) and the outcome model, while set (II) should be adjusted for in at least the MEM. Adjusting for non-risk factors that are correlates of true exposure or measurement error within the MEM alone improves efficiency. We apply this covariate selection approach to the Health Professionals Follow-up Study, assessing fiber intake's effect on cardiovascular disease. We also highlight potential pitfalls in data-driven MEM building that ignores structural assumptions. Additionally, we extend existing estimators to allow for effect modification. Finally, we caution against using regression calibration to estimate the effect of true nutritional intake through calibrating biomarkers.

Introduction: HDL subspecies defined by functional apolipoproteins (apo) exhibit different associations with coronary heart disease (CHD) risk; however, the underlying mechanism, particularly from a lipidomic perspective, is unclear. Hypothesis: Apo-defined HDL subspecies may be characterized by distinct lipid profiles that are associated with differential CHD risk. Methods: We measured the apoA1 concentrations of 15 apo-defined HDL subspecies and 176 plasma lipids among 446 women in Nurses’ Health Study (NHS). The proportion of each HDL subspecies was calculated as the ratio of apoA1 concentrations in HDL containing a specific protein to total plasma apoA1 concentration. Linear regression was used to identify individual lipids associated with the proportions of HDL subspecies, and elastic net regression was used to define predictive lipidomic signatures. Associations between lipidomic signatures and CHD risk were examined using conditional logistic regression in a nested case-control study (400 matched pairs) and Cox regression in 13,495 participants from NHS, NHSII, and Health Professionals Follow-Up Study. Results: HDL subspecies showed different associations with lipidomic features: proportions of HDL containing apoC1, apoE, apoL1, apoC3, and apoJ were associated with specific lipids, while few associations were found for the other ten subspecies. The lipid profile of HDL containing apoC1 closely resembled that of total apoA1, whereas HDL containing apoL1 showed a distinct pattern. Several TAGs and DAGs were negatively associated with apoC1 but positively with apoL1; conversely, certain phospholipids showed positive associations with apoC1 and negative associations with apoL1. The Pearson r between the lipidomic signatures and their corresponding HDL subspecies was 0.76 for total apoA1 (total HDL), 0.67 for apoC1, 0.63 for apoE, 0.55 for apoC3, 0.50 for apoL1, and 0.34 for apoJ. Multivariable-adjusted ORs (95% CIs) for CHD risk per 1-SD increase in lipidomic signatures were 0.85 (0.72, 1.00) for total apoA1, 0.77 (0.65, 0.91) for apoC1, 0.79 (0.66, 0.94) for apoE, 0.92 (0.79, 1.08) for apoC3, 1.30 (1.10, 1.53) for apoL1, and 0.97 (0.81, 1.15) for apoJ ( Figure ). Individual lipids positively associated with HDL containing apoL1, including specific TAGs, DAGs, and ceramides, were also related to a higher CHD risk in the cohorts. Conclusions: Lipid profiles vary across apo-defined HDL subspecies and may contribute to their different associations with CHD risk.

Importance:Evidence linking coffee and tea to cognitive health remains inconclusive, and most studies fail to differentiate caffeinated from decaffeinated coffee. Objective:To investigate associations of coffee and tea intake with dementia risk and cognitive function. Design, Setting, and Participants:Prospective cohort study that included female participants from the Nurses' Health Study (NHS; n = 86 606 with data from 1980-2023) and male participants from the Health Professionals Follow-up Study (HPFS; n = 45 215 with data from 1986-2023) who did not have cancer, Parkinson disease, or dementia at study entry (baseline) in the US. Exposures:The primary exposures were intakes of caffeinated coffee, decaffeinated coffee, and tea. Dietary intake was collected every 2 to 4 years using validated food frequency questionnaires. Main Outcomes and Measures:The primary outcome was dementia, which was identified via death records and physician diagnoses. The secondary outcomes included subjective cognitive decline assessed by a questionnaire-based score (range, 0-7; higher scores indicate greater perceived decline; cases defined as those with a score ≥3) and objective cognitive function assessed only in the NHS cohort using telephone-based neuropsychological tests such as the Telephone Interview for Cognitive Status (TICS) score (range, 0-41) and a measure of global cognition (a standardized mean z score for all 6 administered cognitive tests). Results:Among 131 821 participants (mean age at baseline, 46.2 [SD, 7.2] years in the NHS cohort and 53.8 [SD, 9.7] years in the HPFS cohort; 65.7% were female) during up to 43 years of follow-up (median, 36.8 years; IQR, 28-42 years), there were 11 033 cases of incident dementia. After adjusting for potential confounders and pooling results across cohorts, higher caffeinated coffee intake was significantly associated with lower dementia risk (141 vs 330 cases per 100 000 person-years comparing the fourth [highest] quartile of consumption with the first [lowest] quartile; hazard ratio, 0.82 [95% CI, 0.76 to 0.89]) and lower prevalence of subjective cognitive decline (7.8% vs 9.5%, respectively; prevalence ratio, 0.85 [95% CI, 0.78 to 0.93]). In the NHS cohort, higher caffeinated coffee intake was also associated with better objective cognitive performance. Compared with participants in the lowest quartile, those in the highest quartile had a higher mean TICS score (mean difference, 0.11 [95% CI, 0.01 to 0.21]) and a higher mean global cognition score (mean difference, 0.02 [95% CI, -0.01 to 0.04]); however, the association with global cognition was not statistically significant (P = .06). Higher intake of tea showed similar associations with these cognitive outcomes, whereas decaffeinated coffee intake was not associated with lower dementia risk or better cognitive performance. A dose-response analysis showed nonlinear inverse associations of caffeinated coffee and tea intake levels with dementia risk and subjective cognitive decline. The most pronounced associated differences were observed with intake of approximately 2 to 3 cups per day of caffeinated coffee or 1 to 2 cups per day of tea. Conclusions and Relevance:Greater consumption of caffeinated coffee and tea was associated with lower risk of dementia and modestly better cognitive function, with the most pronounced association at moderate intake levels.

Background The timing of exposure to diet across the lifespan may be critical in the development of T2D. However, no previous study has deciphered the influence of dietary insulinemic and inflammatory potential on the risk of T2D across the lifespan from a life course perspective. Objective This study aimed to evaluate the associations of dietary insulinemic and inflammatory potential with the risk of T2D from a life course perspective. Design This was a prospective cohort study. Participants and setting Data from 40,135 eligible, female participants in the Nurses’ Health Study II were analyzed. Adulthood diet was assessed quadrennially since 1991 using 131-item food frequency questionnaires (FFQ), and adolescent diet was recalled in 1997 using a 124-item high-school FFQ. The main exposures were empirical dietary index for hyperinsulinemia (EDIH) and empirical dietary inflammatory pattern (EDIP) scores across different life stages (adolescence, premenopausal adulthood, postmenopausal adulthood) and changes and cumulatively over the lifetime. Main outcome measures The main outcome was incident T2D. Statistical analyses performed Cox models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI). Results Higher EDIH and EDIP scores (highest vs lowest quintiles) were associated with increased lifetime risk of T2D as a lifetime average (HR, 95%CI: 2.72, 2.38-3.11 and 2.04, 1.81-2.30), during premenopausal adulthood (3.18, 2.52-4.01 and 2.31, 1.88-2.82), and postmenopausal adulthood (2.67, 2.15-3.33 and 1.70, 1.41-2.05), but not during adolescence (1.07, 0.95-1.20 and 1.10, 0.98-1.24). The HR, 95%CI associated with higher lifetime averages for both EDIH and EDIP (vs. low lifetime averages for both, based on tertiles) was 2.64 (2.32-3.01). Individuals with high adulthood EDIH or EDIP had similar magnitudes of lifetime risk elevation, regardless of their adolescent EDIH and EDIP status. Adolescent EDIH and EDIP were associated with a slightly increased premenopausal T2D risk (1.24, 1.02-1.51, and 1.24, 1.02-1.50). In additional analyses estimating the time window during which adulthood dietary insulinemic and inflammatory potential influences T2D risk, higher adulthood EDIH or EDIP was associated with an increased risk of T2D with a very short time lag. Conclusions Over the life course in women, high dietary insulinemic and inflammatory potential in both premenopausal and postmenopausal adulthood were independently associated with a substantially increased lifetime risk of T2D. Adulthood offers the most critical time window for dietary interventions to reduce lifetime T2D risk, though adolescent diet may influence the risk of premenopausal T2D.

Regression calibration is a widely used method for correcting bias in regression coefficient estimates caused by measurement error in continuous exposures. In nutritional epidemiology, recent studies often use nutrient-derived biomarkers to calibrate true dietary intakes, which are typically assessed through small feeding studies or weighed dietary records. However, biomarkers may themselves lie on the causal pathway between intake and outcome, violating the surrogacy assumption that underpins the validity of regression calibration. We evaluate regression calibration-based methods that have been applied or proposed for calibrating dietary intake using biomarkers, focusing on their validity and relative efficiency. Our assessment combines analytic bias quantification, simulation studies, and a real-data application. Two approaches are generally valid when using mediators to calibrate the true exposure: (1) the expanded calibration method, which recovers the total effect via the product method from mediation analysis, and (2) the two-stage calibration method. In simulations reflecting realistic scenarios, the expanded calibration method demonstrated superior efficiency for a continuous outcome. Finally, we show that estimating the effect of dietary intake using a calibrated biomarker is valid when the mean biomarker calibration model contains a linear term in true intake with coefficient one, possibly along with additional covariates.

BACKGROUND:Stomach cancer presents complex etiologic heterogeneity. Ethanol in alcoholic beverages and its metabolite acetaldehyde are carcinogens causally linked to several cancers, but their role in gastric carcinogenesis has not been established. We analyzed harmonized, individual-level prospective data to examine associations between alcohol intake and risk of stomach cancer and its subtypes. METHODS:2,009,951 participants in 20 cohorts (mean follow-up=9-29 years) within the Pooling Project of Prospective Studies of Diet and Cancer (n = 8,357 incident invasive gastric adenocarcinomas) were included. We used Cox regression to assess associations between alcohol intake and risk of stomach cancer overall and by anatomical and histological subtype and population subgroup, adjusting for confounders. RESULTS:Evidence for an association between alcohol intake and overall stomach cancer risk was weak (hazard ratio, HR, for ≥30 v 0.1-<5 g/day: 1.06 [95% confidence interval, CI, 0.96 to 1.16], P between-studies heterogeneity=0.43). Positive associations with stomach cancer risk were observed in never smokers (HR, for ≥30 v 0.1-<5 g/day: 1.20 [95% CI, 1.02 to 1.42]; P interaction=0.02) and for Asian studies (HR, 1.21 [95% CI, 1.02 to 1.42]; P interaction=0.01). Modest increased risks were observed for non-cardia cancers such as those of the fundus, body and greater curvature, but not distally located non-cardia cancers. HRs did not differ materially between diffuse- and intestinal-type cancers (P heterogeneity>0.05). CONCLUSION:There was little evidence of an overall association between alcohol intake and stomach cancer risk, although modest positive associations were observed among never smokers and in Asian cohorts.

Growing evidence suggests that exposure to fine particulate matter (PM2.5) may accelerate cognitive decline and increase dementia risk, but the roles of individual PM2.5 components, particularly when accounting for exposure measurement error, remain unclear. We followed 12,824 participants in the Nurses' Health Study cognition substudy (2000-2008). One-year average PM2.5 component concentrations preceding each participant's first cognitive assessment were assigned based on residential address. A global cognitive score was calculated as the mean of z-scores from six tests, and individual rates of cognitive decline were estimated using linear mixed models with random intercepts and slopes. Using information on the relationships of personal exposures and nearest-monitor PM2.5 components from a prior validation study, we employed double/debiased machine learning (DML) method to estimate associations between PM2.5 components and cognitive decline while accommodating high-dimensional multi-pollutant exposures and accounting for exposure measurement error. Higher exposure to several PM2.5 components were associated with changes in rates of cognitive decline. In multi-pollutant models without measurement error correction, each doubling of chloride (Cl) concentration was associated with a 0.0003 (95% CI: 0.0000, 0.0005) standard unit slower annual rate of cognitive decline, while each doubling in silicon (Si) was associated with a -0.0025 (95% CI: -0.0050, 0.0000) faster rate of decline. After correction for measurement error, associations generally strengthened: higher bromine (Br), manganese (Mn), and lead (Pb) were significantly associated with faster cognitive decline. In contrast, higher strontium (Sr) and vanadium (V) were associated with slower decline. Our findings suggest that long-term exposure to certain PM2.5 components, and their sources, may be more detrimental to cognitive aging than others, and highlight the critical need to correct for measurement error when assessing the health impacts of PM2.5 and its components. More research is needed to confirm these findings, ideally with access to larger validation studies of component exposures.

Objective:To estimate the associations between gestational weight gain and maternal immediate perinatal and postpartum outcomes by pooling data from low and middle income countries. Design:Individual participant data meta-analyses. Data sources:PubMed, Embase, Web of Science, and Cochrane Library, based on three searches (Search 1: all prospective studies published from January 2000 to May 2021; Search 2: randomized controlled trials of balanced energy and protein supplementation published until June 2021; Search 3: randomized controlled trials of anti-infectious agents published until August 2021). Eligibility criteria for selecting studies:Prospective studies (randomised controlled trials or observational cohort studies) with measured maternal weight during pregnancy and data available on maternal height, based in populations from low and middle income countries with no underlying conditions. Results:The analyses included 156 300 women from 61 studies and 23 countries, with most participants based in South Asia (n=78 454, 50.2%) and sub-Saharan Africa (n=36 327, 23.2%). Compared with women with adequate (90-125%) gestational weight gain, women with excessive (>125%) gestational weight gain had a higher risk of caesarean delivery (risk ratio 1.10, 95% confidence interval 1.06 to 1.13, τ2=0.000) and emergency caesarean delivery (risk ratio 1.22, 1.03 to 1.43, τ2=0.000). Women with moderately (70% to <90%) or severely inadequate (<70%) versus adequate gestational weight gain had lower risks for caesarean delivery (risk ratio in women with moderately inadequate gestational weight gain 0.88, 95% confidence interval 0.84 to 0.92, τ2=0.004; risk ratio in women with severely inadequate gestational weight gain 0.82, 0.77 to 0.88, τ2=0.010) and emergency caesarean delivery (risk ratio in moderately inadequate gestational weight gain 0.82, 0.71 to 0.95, τ2=0.004; risk ratio in severely inadequate gestational weight gain 0.73, 0.56 to 0.96, τ2=0.103). Excessive versus adequate gestational weight gain was associated with higher postpartum weight retained at any time point (mean difference 2.00 kg, 95% confidence interval 1.49 to 2.50, τ2=1.317), whereas moderately and severely inadequate gestational weight gain were associated with lower retained weight compared with adequate gestational weight gain. Similar trends were found for postpartum body mass index. Severely inadequate gestational weight gain was associated with lower systolic and diastolic blood pressure at any time point post partum than adequate gestational weight gain. No associations were observed for other outcomes including postpartum depressive symptoms or breastfeeding. Evidence indicating an interaction between gestational weight gain and body mass index before pregnancy was found when examining the risk of caesarean delivery and postpartum weight retention, body mass index, and systolic blood pressure as outcomes. Conclusions:These findings support the association between suboptimal gestational weight gain and adverse maternal outcomes in the immediate perinatal and postpartum periods. Further research examining the consequences of suboptimal gestational weight gain in low and middle income countries would be valuable to inform potential strategies to improve long term maternal health. Review registration:PROSPERO CRD42023432836.

Pooled analyses that aggregate data from multiple studies are becoming increasingly common in collaborative epidemiologic research in order to increase the size and diversity of the study population. However, biomarker measurements from different studies are subject to systematic measurement errors and directly pooling them for analyses may lead to biased estimates of the regression parameters. Therefore, study-specific calibration processes must be incorporated in the statistical analyses to address between-study/assay/laboratory variability in the biomarker measurements. We propose a likelihood-based method to evaluate biomarker-disease relationships for categorical biomarkers in matched/nested case-control studies. To account for the additional uncertainties from the calibration processes, we propose a sandwich variance estimator to obtain valid asymptotic variances of the estimated regression parameters. Extensive simulation studies with varying sample sizes and biomarker-disease associations are used to evaluate the finite sample performance of our proposed methods. As an illustration, we apply the methods to a vitamin D pooling project of colorectal cancer to evaluate the effect of categorical vitamin D levels on colorectal cancer risks.

Introduction Gestational weight gain (GWG) is an important indicator of maternal nutrition to be monitored during pregnancy. However, there is no evidence-based tool that can be used to monitor it across all geographic locations and pre-pregnancy body mass index (BMI) categories. The WHO is undertaking a project to develop GWG charts by pre-pregnancy BMI category, and to identify GWG ranges associated with the lowest risks of adverse maternal and infant outcomes. This protocol describes all the steps that will be used to accomplish the development of these GWG charts.Methods and analysis This project will involve the analysis of individual participant data (researcher-collected or administrative). To identify eligible datasets with GWG data, a literature review will be conducted and a global call for data will be launched by the WHO. Eligible individual datasets obtained from multiple sources will be harmonised into a pooled database. The database will undergo steps of cleaning, data quality assessment and application of individual-level inclusion criteria. Heterogeneity of maternal weight and GWG will be assessed to verify the possibility of combining datasets from multiple sources and regions into a single database. Generalized Additive Models for Location, Scale and Shape will be applied for the construction of the centile curves. Diagnostic measures, internal and external validation procedures will also be performed.Ethics and dissemination This project will include an analysis of existing study de-identified data. To be included in the pooled database, each included study should have received ethics approvals from relevant committees. Manuscripts will be submitted to open-access journals and a WHO document will be published, including the GWG charts and cut-offs for application in antenatal care.

Pancreatic cancer incidence is rising, yet few modifiable risk factors have been identified. The Mediterranean diet, which lowers inflammation and improves healthy weight maintenance and insulin control, may lower pancreatic cancer risk, yet the evidence for this association is inconsistent. To investigate the association, we conducted a pooled analysis of 2,315,406 individuals from 23 prospective cohorts in the Pooling Project of Prospective Studies of Diet and Cancer (DCPP), of whom 10,748 developed incident pancreatic cancer over a mean follow-up duration ranging from 8.1 to 23.3 years across studies. Adherence to the Mediterranean diet was assessed using the alternative Mediterranean diet score (aMED) and a modified score excluding alcohol (maMED). Study- and sex-specific hazard ratios (HRs) and 95