Search

Aims/hypothesis The aim of this work was to assess the association between diabetes and risk for infection-related hospitalisation and mortality. Methods We conducted a prospective cohort analysis of the Atherosclerosis Risk in Communities (ARIC) study. Diabetes was defined as a fasting glucose ≥7 mmol/l or non-fasting glucose ≥11.1 mmol/l, self-report of a diagnosis of diabetes by a physician, or current diabetes medication use. Hospitalisation for infection was ascertained from hospital discharge records. Participants were followed from 1987–1989 to 2019. Results We included 12,379 participants (mean age 54.5 years; 24.7% Black race; 54.3% female sex). During a median follow-up of 23.8 years, there were 4229 new hospitalisations for infection. After adjusting for potential confounders, people with (vs without) diabetes at baseline had a higher risk for hospitalisation for infection (HR 1.67 [95% CI 1.52, 1.83]). Results were generally consistent across infection type but the association was especially pronounced for foot infection (HR 5.99 [95% CI 4.38, 8.19]). Diabetes was more strongly associated with hospitalisation for infection in younger participants and Black people. Overall infection mortality was low (362 deaths due to infection) but the adjusted risk was increased for people with diabetes (HR 1.72 [95% CI 1.28, 2.31]). Conclusions/interpretation Diabetes confers significant risk for infection-related hospitalisation. Enhancing prevention and early treatment of infection in those with diabetes is needed to reduce infection-related morbidity and mortality. Graphical abstract Supplementary Information The online version of this article (10.1007/s00125-021-05522-3) contains peer-reviewed but unedited supplementary material.

Objective: Leukocytes contribute to the development of abdominal aortic aneurysm (AAA). We evaluated whether associations of differential leukocyte counts with AAA persist after accounting for traditional risk factors of AAA. Approach and Results: Among 11 217 adults from the Atherosclerosis Risk in Communities Study, we evaluated associations of differential leukocyte counts at baseline (1987–1989) with incident AAAs over a median follow-up of 22.5 years, using Cox proportional hazards regression. Each differential leukocyte count was categorized into 5 groups—below normal, tertiles within the normal range, and above normal, with the first tertile serving as the referent. We identified 377 incident AAAs through 2011, using hospital discharge diagnoses, linked Medicare records, or death certificates. At baseline, higher neutrophil, monocyte, and eosinophil counts were associated with higher risk of AAA, independent of smoking, other differential leukocyte counts, and other traditional risk factors. The association with incident AAA was the strongest for above normal neutrophil count, with an adjusted hazard ratio (95% CI) of 2.17 (1.29–3.64). Below normal neutrophil, lymphocyte, eosinophil and basophil counts were associated with higher risk of AAA with adjusted hazard ratio (95% CI) between 1.86 (1.04–3.35) and 1.62 (1.10–2.39). Conclusions: Higher neutrophil, monocyte, and eosinophil counts in midlife are associated with higher risk of AAA, even after accounting for traditional risk factors such as smoking, obesity, and atherosclerosis. This suggests the need to identify nontraditional risk factors and treatment strategies to mitigate the residual risk of AAA conferred by midlife inflammation. Whether immunosuppression is associated with higher risk of AAA needs further investigation.

Objectives This study investigates the prevalence and prognostic significance of mitral regurgitation (MR) in acute decompensated heart failure (ADHF) patients. Background Few studies characterize the burden of MR in heart failure. Methods The ARIC (Atherosclerosis Risk In Communities) study surveilled ADHF hospitalizations for residents ≥55 years of age in 4 U.S. communities. ADHF cases were stratified by left ventricular ejection fraction (LVEF): Results From 2005 to 2014, there were 17,931 weighted ADHF hospitalizations of which 49.2% had an LVEF Conclusions Patients with ADHF have a significant MR burden that varies with sex and race. In ADHF patients with an LVEF

Background: Low levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in the low physiologic range, surrogate markers for reduced liver metabolic function, are associated with cerebral hypometabolism, impairment in neurotransmitter production and synaptic maintenance, and a higher prevalence of dementia. It is unknown whether a prospective association exists between low liver enzyme levels and incident dementia. Objective: To determine whether low levels of ALT and AST are associated with higher risk of incident dementia. Methods: Plasma ALT and AST were measured on 10,100 study participants (mean age 63.2 years, 55% female, 22% black) in 1996–1998. Dementia was ascertained from comprehensive neuropsychological assessments, annual contact, and medical record surveillance. Cox proportional hazards regression was used to estimate the association. Results: During a median follow-up of 18.3 years (maximum 21.9 years), 1,857 individuals developed dementia. Adjusted for demographic factors, incidence rates of dementia were higher at the lower levels of ALT and AST. Compared to the second quintile, ALT values

Objective: Venous thromboembolism (VTE) is a common disease that has a genetic basis. Lifestyle factors contribute to risk, but it is unknown whether healthy lifestyle can mitigate the genetic risk. We studied whether greater adherence to the American Heart Association’s cardiovascular health metric, Life’s Simple 7 (LS7), is associated with lower incidence of VTE in individuals across categories of a genetic risk score (GRS) for VTE. Approach AND RESULTS: We followed 9026 White participants from the ARIC (Atherosclerosis Risk in Communities) Study, a prospective cohort enrolled in 1987 to 1989 until 2015. We tested the joint associations with VTE of a validated VTE GRS comprising 5 well-known gene variants and baseline LS7 categories. There were 466 incident VTE events over 22.8 years. Participants with an optimal LS7 score had a lower incidence of VTE (3.9%) than those with inadequate LS7 (5.7%). Compared with the high GRS and inadequate LS7 group (hazard ratio=1), those with high GRS and optimal LS7 indeed had a reduced hazard ratio of VTE: 0.65 (95% CI, 0.48–0.89). The group with low GRS and optimal LS7 had the lowest hazard ratio of VTE (0.39 [95% CI, 0.25–0.61]). Of the LS7 components, in all GRS groups, the factor most strongly protective for VTE was normal weight. Conclusions: Among people at low or high genetic risk for VTE, healthier lifestyle factors, particularly normal weight, were associated with a lower incidence of VTE. Further studies should determine the impact of lifestyle changes among patients at high genetic risk of VTE, such as in thrombophilic families.

Background: Patients with heart failure (HF) have multiple coexisting comorbidities. The temporal trends in the burden of comorbidities and associated risk of mortality among patients with HF with preserved ejection fraction (HFpEF) and HF with reduced ejection fraction (HFrEF) are not well established. Methods: HF-related hospitalizations were sampled by stratified design from 4 US areas in 2005 to 2014 by the community surveillance component of the ARIC study (Atherosclerosis Risk in Communities). Acute decompensated HF was classified by standardized physician review and a previously validated algorithm. An ejection fraction Results: A total of 5460 hospitalizations (24 937 weighted hospitalizations) classified as acute decompensated HF had available ejection fraction data (53% female, 68% white, 53% HFrEF, 47% HFpEF). The average number of comorbidities was higher for patients with HFpEF versus HFrEF, both for women (5.53 versus 4.94; P P P -trendP for interaction by HF type=0.02). The associated mortality risk per 1 higher comorbidity also increased significantly over time for patients with HFpEF and HFrEF, as well ( P for interaction with time=0.002 and 0.02, respectively) Conclusions: The burden of comorbidities among hospitalized patients with acute decompensated HFpEF and HFrEF has increased over time, as has its associated mortality risk. Higher burden of comorbidities is associated with higher risk of mortality, with a stronger association noted among patients with HFpEF versus HFrEF.

Background and aims Elevated serum calcium and phosphorus have been associated with increased risk of cardiovascular disorders. We evaluated whether abnormal calcium and high serum phosphorus are associated cross-sectionally with echocardiographic measures of left ventricular (LV) structure and function, as doing so may provide insight into the etiology of cardiac disorders. Methods and results Included in the analysis were 5213 Atherosclerosis Risk in Communities Study (ARIC) participants who in 2011–2013 had echocardiography and serum calcium and phosphorus measurements. We evaluated the association of serum calcium (corrected for albumin) and phosphorus quintiles with measures of LV structure and function, after adjusting for other cardiovascular risk factors. Participants were on average 75.3 years old; 59.1% were female and 19.8% were African American. Mean (±SD) concentrations of calcium and phosphorus were 9.33 ± 0.38 and 3.46 ± 0.45 mg/dL, respectively. Higher calcium was associated with lower LV end-diastolic diameter (LVEDD) but greater prevalence of concentric remodeling (p-trend: 0.005 and 0.004 respectively). We observed association between high phosphorus and high septal E/e’ (p-trend: 0.02). Likewise, higher serum phosphorus was associated with higher left atrial volume index (p-trend: 0.001) and LV hypertrophy prevalence (p-trend: 0.04). Conclusions In conclusion, higher calcium was associated with more concentric remodeling but lower LVEDD, suggesting complex associations between calcium and cardiac function. Serum phosphorus was related to worse indices of LV diastolic function and LV hypertrophy, but not to LV systolic function. However, the magnitudes of association were modest, so clinical implications of these findings may be limited.

Background Ankle‐brachial index (ABI) is used to identify lower‐extremity peripheral artery disease (PAD). However, its association with severe ischemic leg outcomes (eg, amputation) has not been investigated in the general population. Methods and Results Among 13 735 ARIC (Atherosclerosis Risk in Communities) study participants without clinical manifestations of PAD (mean age, 54 [SD, 5.8] years; 44.4% men; and 73.6% White) at baseline (1987–1989), we quantified the prospective association between ABI and subsequent severe ischemic leg outcomes, critical limb ischemia (PAD with rest pain or tissue loss) and ischemic leg amputation (PAD requiring amputation) according to discharge diagnosis. Over a median follow‐up of ≈28 years, there were 221 and 129 events of critical limb ischemia and ischemic leg amputation, respectively. After adjusting for demographics, ABI ≤0.90 versus 1.11 to 1.20 had a ≈4‐fold higher risk of critical limb ischemia and ischemic leg amputation (hazard ratios, 3.85 [95% CI, 2.09–7.11] and 4.39 [95% CI, 2.08–9.27]). The magnitude of the association was modestly attenuated after multivariable adjustment (hazard ratios, 2.44 [95% CI, 1.29–4.61] and 2.72 [95% CI, 1.25–5.91], respectively). ABI 0.91 to 1.00 and 1.01 to 1.10 were also associated with these severe leg outcomes, with hazard ratios ranging from 1.7 to 2.0 after accounting for potential clinical and demographic confounders. The associations were largely consistent across various subgroups. Conclusions In a middle‐aged community‐based cohort, lower ABI was independently and robustly associated with increased risk of severe ischemic leg outcomes. Our results further support ABI ≤0.90 as a threshold diagnosing PAD and also suggest the importance of recognizing the prognostic value of ABI 0.91 to 1.10 for limb prognosis.

Background Black Americans have more atrial fibrillation risk factors but lower atrial fibrillation risk than White Americans. Left atrial (LA) enlargement and/or dysfunction, frequent atrial tachycardia (AT), and premature atrial contractions (PAC) are associated with increased atrial fibrillation risk. Racial differences in these factors may exist that could explain the difference in atrial fibrillation risk. Methods and Results We included 2133 ARIC (Atherosclerosis Risk in Communities) study participants (aged 74±4.5 years[mean±SD], 59% women, 27% Black participants) who had echocardiograms in 2011 to 2013 and wore the Zio XT Patch (a 2‐week continuous heart monitor) in 2016 to 2017. Linear regression was used to analyze (1) differences in AT/day or PAC/hour between Black and White participants, (2) differences in LA measures between Black and White participants, and (3) racial differences in the association of LA measures with AT or PAC frequency. Compared with White participants, Black participants had a higher prevalence of cardiovascular risk factors and disease, lower AT frequency, greater LA size, and lower LA function. After multivariable adjustments, Black participants had 37% (95% CI, 24%–47%) fewer AT runs/day than White participants. No difference in PAC between races was noted. Greater LA size and reduced LA function are associated with more AT and PAC runs; however, no race interaction was present. Conclusions Differences in LA measures are unlikely to explain the difference in atrial fibrillation risk between Black and White individuals. Despite more cardiovascular risk factors and greater atrial remodeling, Black participants have lower AT frequency than White participants. Future research is needed to elucidate the protective mechanisms that confer resilience to atrial arrhythmias in Black individuals.

We thank Dr. Fisher et al.1 for bringing to our attention their previous research directly measuring the effects of premature ventricular contractions on intracranial hemodynamics. We agree that these findings offer some mechanistic perspectives on the epidemiologic observations of the link between cardiac arrhythmias and dementia.

Aims: Both lifestyle factors and genetic background contribute to the development of type 2 diabetes. Estimation of the lifetime risk of diabetes based on genetic information has not been presented, and the extent to which a normal body weight can offset a high lifetime genetic risk is unknown. Methods: We used data from 15,671 diabetes-free participants of European ancestry aged 45 years and older from the prospective population-based ARIC study and Rotterdam Study (RS). We quantified the remaining lifetime risk of diabetes stratified by genetic risk and quantified the effect of normal weight in terms of relative and lifetime risks in low, intermediate and high genetic risk. Results: At age 45 years, the lifetime risk of type 2 diabetes in ARIC in the low, intermediate and high genetic risk category was 33.2%, 41.3% and 47.2%, and in RS 22.8%, 30.6% and 35.5% respectively. The absolute lifetime risk for individuals with normal weight compared to individuals with obesity was 24% lower in ARIC and 8.6% lower in RS in the low genetic risk group, 36.3% lower in ARIC and 31.3% lower in RS in the intermediate genetic risk group, and 25.0% lower in ARIC and 29.4% lower in RS in the high genetic risk group. Conclusions: Genetic variants for type 2 diabetes have value in estimating the lifetime risk of type 2 diabetes. Normal weight mitigates partly the deleterious effect of high genetic risk.

INTRODUCTION A data-driven index of dementia risk based on magnetic resonance imaging (MRI), the Alzheimer's Disease Pattern Similarity (AD-PS) score, was estimated for participants in the Atherosclerosis Risk in Communities (ARIC) study. METHODS AD-PS scores were generated for 839 cognitively non-impaired individuals with a mean follow-up of 4.86 years. The scores and a hypothesis-driven volumetric measure based on several brain regions susceptible to AD were compared as predictors of incident cognitive impairment in different settings. RESULTS Logistic regression analyses suggest the data-driven AD-PS scores to be more predictive of incident cognitive impairment than its counterpart. Both biomarkers were more predictive of incident cognitive impairment in participants who were White, female, and apolipoprotein E gene (APOE) e4 carriers. Random forest analyses including predictors from different domains ranked the AD-PS scores as the most relevant MRI predictor of cognitive impairment. CONCLUSIONS Overall, the AD-PS scores were the stronger MRI-derived predictors of incident cognitive impairment in cognitively non-impaired individuals.

Introduction: Individuals with diabetes often have low heart rate variability (HRV)—a marker of autonomic dysfunction. How diabetes duration, glucose excursions, and poor glycemic control relate to HRV is not well characterized. Objectives: To evaluate the associations of diabetes status, duration of diabetes, and biomarkers of glucose excursions (1,5-anhydroglucitol [1,5-AG]—a measure of glycosuria) and glycemic control with HRV. Methods: We conducted a cross-sectional analysis of ARIC Study participants who underwent a 2-minute electrocardiogram at visit 4 (1996-98). We considered root mean square of successive differences in normal-to-normal R-R intervals (rMSSD, measure of parasympathetic activity), standard deviation of normal-to-normal R-R intervals (SDNN, total variability), low frequency (LF, primarily sympathetic activity), high frequency (HF, reflects parasympathetic activity), and LF/HF ratio (sympathetic-parasympathetic balance) as HRV indices. In persons with diabetes, we compared geometric means of HRV indices by diabetes duration ( Results: Among 10169 participants (mean age 63, 57% female, 22% black), participants with diabetes had significantly lower rMSSD, SDNN, LF, HF, and LF/HF ratio compared to those without diabetes ( Table ). Among the 1667 participants with diabetes, longer duration of diabetes, low 1,5-AG, and poor glycemic control were associated with lower rMSSD, SDNN, LF, and HF. No statistically significant associations were found for diabetes duration, 1,5-AG, glycemic control with LF/HF ratio. Discussion: Our results demonstrate that longer duration of diabetes, glucose excursions, and poor glycemic control are associated with some measures of low HRV. This study underscores that cardiac autonomic neuropathy is a common but overlooked complication of long-standing and/or poorly managed diabetes.

Background: High-sensitivity cardiac troponin T (TnT) is a potent predictor of cardiovascular disease (CVD) in the general population. Recently, the US FDA has approved Roche fifth generation (Gen 5) TnT assay (more sensitive than the fourth generation [Gen 4] TnT assay). Since many previous epidemiological studies used Gen 4 TnT, it is important to characterize the association of Gen 5 TnT with major CVD events. Methods: We first assessed correlation of Gen 5 vs. Gen 4 TnT in a subsample of 91 participants. Then, as the main analysis, we examined the association of Gen 5 TnT at visit 3 (1993-1995) with major CVD events (coronary heart disease [CHD], stroke, heart failure [HF], and peripheral artery disease [PAD]). Gen 5 TnT was categorized as Results: Gen 5 TnT and Gen 4 TnT had a correlation coefficient of 0.98 (0.88 after excluding outliers >3SD). Of 11,979 participants (mean age 60 [SD 6] years, 1,840 [15%] with prevalent CVD), 5,856 (49%) participants had quantifiable levels of TnT. During a median follow-up of 22.1 years, there were 1,850 CHD events, 1,075 stroke events, 2,908 HF cases, and 571 PAD cases. Gen 5 TnT showed a robust dose response association with each CVD type, with adjusted hazard ratio 2-4 for TnT ≥19 (vs.Table ). Even the category 6- Conclusions: Gen 5 TnT was highly correlated with Gen 4 TnT and gave quantifiable values in half middle-aged adults. Gen 5 TnT was robustly associated with major CVD events (especially HF and PAD) in the general population, supporting its usefulness in epidemiological research and clinical practice.

Introduction: Prior research indicates that higher LTPA is associated with better left ventricular (LV) diastolic function, but not with left atrial volume index (LAVI). Left atrial (LA) strain is a more sensitive marker of LA pathology. Hence, we evaluated association of LTPA with LA strain in the ARIC study, a community-based cohort study. Hypothesis: Higher LTPA is associated with better LA reservoir, conduit and contractile function. Methods: We included 4005 participants (mean age 75 years, 63% female, 21% black) with 2D echocardiographic (2DE) LA % strain data, LTPA data, and without prevalent coronary heart disease, valvular heart disease, heart failure, atrial fibrillation, and stroke at Visit 5 (V5, 2011-13). We estimated LTPA as metabolic equivalent-minutes per week from validated Baecke questionnaire and categorized it as poor, intermediate, or ideal as per AHA recommendation. We categorized change in LTPA from V3 (1993-95) to V5 as persistently poor (referent), decreasing, increasing, or persistently active. LA reservoir, conduit, and contractile % strain were measured from speckle-tracking 2DE. Linear regression was used to evaluate association of LTPA with LA function. We modeled LTPA (1) cross-sectionally at V5 as a continuous variable and (2) change from V3 to V5, as continuous and categorical variables (referent- persistently poor). Results: Higher LTPA at V5 (Table) and persistently high (vs poor) LTPA from V3 to 5 (β estimate -0.89, 95% CI -1.40, -0.38) were significantly associated with better LA conduit function after adjusting for CV risk factors, LV function, and LAVI. The same associations with LA reservoir function were no longer significant after adjusting for CV risk factors. LTPA was not associated with LA contractile function (Table). Conclusions: Consistent with LTPA’s salutary effect on LV diastolic function, higher LTPA is associated with better LA conduit function. Maintaining high LTPA from midlife to late-life may be beneficial in mitigating age-related changes in cardiac function.

Introduction: Diabetes and metabolic syndrome (MetS) confer an increased risk of heart failure (HF) through poorly understood mechanisms. Galectin-3 (Gal-3) is a marker of fibrosis linked to greater HF risk. The inter-relationships among diabetes, MetS and Gal-3, and related implications for HF risk, are not well understood. Hypothesis: Diabetes and MetS are associated with elevated Gal-3, and high Gal-3 indicates greater HF risk among those with diabetes or MetS. Methods: We included 8,445 participants (mean age 63, 59% male, 21% Black) at ARIC Visit 4 (1996-1999) without baseline HF. We categorized participants by metabolic risk (no diabetes/no MetS; MetS only; diabetes with or without MetS), and Gal-3 levels (gender-specific quartiles). We assessed the associations of metabolic risk categories with high Gal-3 (≥75 th percentile) using logistic regression. We used Cox regression to evaluate associations of cross-categories of metabolic risk group and Gal-3 quartiles with incident HF. Results: In cross-sectional analyses, those with MetS were more likely to have elevated Gal-3 levels than those with no diabetes or MetS (OR 1.29, 95%CI 1.13-1.47). The additional presence of diabetes did not change the likelihood of elevated Gal-3 (OR 1.29, 95%CI 1.08-1.55). Over 20.5 years of follow-up, there were 1,611 HF events. Higher Gal-3 was associated with higher HF risk in each metabolic risk group, and higher metabolic risk group was associated with greater HF risk in each Gal-3 quartile. There was no interaction between Gal-3 and metabolic risk group on HF risk ( P =0.15). The combination of diabetes &MetS and high Gal-3 was associated with an almost 5-fold higher risk of incident HF (HR 4.93; 95% CI: 3.77 - 6.44) than the combination of no diabetes/MetS and low Gal-3 (first quartile) ( Table ). Conclusions: MetS was associated with higher levels of Gal-3. Metabolic risk group and Gal-3 provided powerful complementary prognostic information regarding HF risk. Fibrosis likely plays a role in the development of HF linked to metabolic risk.

Background: Statins are prescribed to reduce LDL-cholesterol (LDL-c) and the risk of cardiovascular disease. Statins may influence other outcomes such as dementia through pleiotropic effects unrelated to lipid metabolism. We compared the proteome of statin users and nonusers. We hypothesized that statin use would be associated with proteins in biological pathways unrelated to lipid metabolism. Methods: Among 11,471 participants enrolled in the ARIC study at visit 3 (1993-95), plasma concentrations of 4,638 proteins were determined using the SOMAlogic DNA aptamer-based capture array. We identified 402 participants who initiated statin use within the 2 years prior. Propensity scores (PS) were calculated based on visit 2 (1990-92) LDL-c levels and visit 3 demographic characteristics, CVD risk factors and medications, kidney function, and prevalent CVD. Statin users were matched to nonusers (controls) based on PS similarity, yielding 382 matched pairs. Log2-transformed and standardized protein levels were compared between statin users and controls using t-tests. The false discovery rate (FDR) adjusted for multiple comparisons. Ingenuity Pathway Analysis (IPA) was used to identify potentially altered biological pathways. Results: After PS matching, LDL-c levels at visit 3 were lower in statin users vs controls (125.4 vs 147.1 mg/dL; p Conclusion: Several proteins unrelated to lipid metabolism differed by statin use. Pending external validation, exploring the biological functions of these proteins could elucidate pleiotropic effects of statins.

Introduction: Age-related homeostatic dysregulation (Dm), characterized by multi-biomarker composite measures, has shown promise as a rate-of-aging metric. We tested the association of midlife Dm and its change over ~20 years of follow-up with later life physical function. Methods: We studied 4617 ARIC cohort participants (mean age 54.8 years, 58% female, 20% Black) who completed in-person examinations from 1990-92 (baseline) through 2011-13. Dm quantifies the multivariate statistical deviation of 15 physiology-motivated biomarkers from the distribution in a healthy sample at baseline. Midlife Dm was grouped by quartiles (higher quartiles indicating greater homeostatic dysregulation) and temporal change in Dm was quantified continuously. Physical function was assessed in 2011-13 and included measures from the Short Physical Performance Battery (SPPB) including repeated chair stands, balance, 4-meter walk and grip strength. Associations were quantified using linear regression and ordinal logistic regression as applicable, adjusting for age, sex, race, and education. Results: A graded association was observed between midlife Dm and physical functional performance in late life (Table). Compared to the 1 st quartile of midlife Dm, the odds ratio of a lower SPPB score in late life was 1.19 (95%CI 1.04, 1.35) for the 2 nd quartile, 1.63 (1.41, 1.90) for the 3 rd quartile, and 3.14 (2.60, 3.79) for the 4 th quartile. Similar associations were observed for each of SPPB components and grip strength, either modeled ordinally or continuously. Each standard deviation increase in 10-year Dm change was associated with ~2-3 times the odds of having a poorer physical functional performance. Conclusions: Greater homeostatic dysregulation at midlife and increases in dysregulation during follow-up were associated with poorer physical function in late life. Insights into the factors that lead to progression of multisystem deterioration during midlife may highlight opportunities to preserve functional abilities in late life.

Background: It is established that a family history of coronary heart disease (FHCHD) is associated with coronary atherosclerosis in healthy first-degree relatives, but the extent to which FHCHD is associated with silent cerebrovascular disease (cSVD) is unknown. We hypothesized a higher prevalence of cSVD in healthy persons with FHCHD, independent of traditional risk factors, compared to those without FHCHD. Methods: ARIC is a community-based cohort study with self-reported family history data and brain magnetic resonance imaging (visit 5; 2011-13). The association between binary markers of cSVD (lacunar infarcts and/or cerebral microbleeds), or log-transformed white matter hyperintensity volume (WMH), and FHCHD (parent and/or sibling), or number of relatives was examined using separate adjusted multivariable logistic or linear regression models respectively. Sensitivity analysis (N=183) excluded prevalent CHD. Race interaction terms were included. Results: Of 1828 participants (76±5yo, 60% female, 28% black), 787 had FHCHD (699 parental, 209 sibling FHCHD). There were increased adjusted odds of lacunar infarct among those with parental FHCHD (Table). An increased odds of cerebral microbleeds were seen among those with sibling history but not parental. Effect estimates were similar when excluding those with prevalent CHD (Table). Greater number of siblings affected was associated with higher odds of lacunar infarct (OR 1.35, CI 1.04-1.74), lobar (OR 1.53, CI 1.12-2.09) and subcortical microbleeds (OR 1.30, CI 1.01-1.66). Odds of a lacunar infarct being present were higher among blacks (p-interaction 0.04) with paternal FHCHD (OR 2.20, CI 1.35-3.58) compared to whites (OR 1.17, CI 0.87-1.56). Neither FHCHD nor number of affected relatives was associated with WMH. Conclusions: Our results suggest that some cSVD manifestations are associated with FHCHD, potentially representing shared mechanisms in different vascular beds, and perhaps a genetic propensity for vascular disease.

Introduction: Greater late-life physical function decline is associated with incident adverse outcomes including disability and death. Hypertension is the strongest risk factor for stroke, the major cause of physical disability. Hypertension in mid-life has previously been associated with poor physical functioning in late-life; however, more evidence is needed to evaluate whether higher blood pressure in mid-life is associated with the rate of physical function decline during late-late in the absence of stroke. We hypothesized that elevated blood pressure in mid-life would be associated with greater physical function declines in late life. Methods: We studied 5,559 older adults in the ARIC Study (Visit 5; mean age: 75.8 years; range: 66.7-90.9 years; 58% women; 21% Black/79% White) without prior stroke or Parkinson disease who completed the Short Physical Performance Battery (SPPB, scored 0-12). Repeated SPPB assessments occurred at Visits 6 and 7 (median follow-up: 4.2 years). The exposure was a history of elevated blood pressure (BP) (Visit 1; mean age: 52.0 years; mean gap between mid- and late-life exams: 23.7 years). BP was modeled both categorically (hypertensive: SBP 140+ mmHg, DBP 90+ mmHg, or antihypertensive medication use; pre-hypertensive: SBP 120-139 mmHg or DBP 80-89 mmHg; else normotensive) and continuously. Random-slope, random-intercept mixed models with an independent covariance structure tested the association between BP and SPPB score change, adjusted for age, sex, race-site, BMI, education, heart disease and heart failure. Continuous analysis also adjusted for antihypertensive medication use. Results: SPPB scores declined an average of 1.60 points per 10 years (95% CI: -1.75, -1.46; p Conclusions: Elevated BP in mid-life provides insight into the rate of physical function decline decades later, with higher mid-life systolic blood pressure corresponding with steeper declines in late-life physical function even in the absence of stroke. Future research should investigate whether elevated blood pressure at multiple points in mid-life further informs the association.

Background: The association between pre-stroke PA and long-term stroke outcomes is still unclear. We examined the association of pre-stroke PA with adverse health outcomes after incident stroke in the ARIC study. Methods: We included 881 participants with incident ischemic or hemorrhagic stroke occurring between 1993-1995 and December 31 st , 2016. Follow-up for all-cause and cardiovascular disease (CVD) mortality, CVD or recurrent stroke occurred through December 31 st , 2017, allowing for at least 1-year of follow-up after incident stroke. Pre-stroke total and domain-specific (work, sports and leisure) PA was assessed with the modified Baecke questionnaire at the baseline visit (1987-1989) and in 1993-1995. The total PA score was calculated by summing scores at both visits. Distribution-based tertiles of the summed score were derived. Cox proportional hazards models estimated the hazard ratios of adverse outcomes by tertiles of pre-stroke PA, adjusted for demographic and clinical characteristics. Results: During a median follow-up of 3.1 years after incident stroke, 77% participants developed adverse outcome. Compared to low pre-stroke PA, high pre-stroke PA was associated with a lower risk of all-cause mortality (HR=0.78; 95% CI: 0.63-0.97, Table). In the analysis by domain-specific PA, the risk of all-cause mortality was lowest among participants in the highest tertile of work PA (HR; 95% CI=0.77; 0.62-0.96) and leisure PA (HR=0.72; 95% CI: 0.58 to 0.89) compared to the lowest tertiles. The risk of CVD mortality was also lower in the highest tertile of work PA (HR=0.45; 95% CI: 0.29-0.70) compared to the lowest tertile. Conclusions: Higher levels of total pre-stroke PA, as well as work and leisure PA, were associated with a lower risk of mortality after incident stroke. Strategies to increase lifetime PA should be encouraged to decrease long-term mortality after stroke.

Background: Reduced physical function, an exemplary phenotype of aging, has been associated with cardiovascular disease (CVD). However, few studies have comprehensively investigated the association of physical function with risk of incident and recurrent CVD in community-dwelling older adults. Methods: Physical function, evaluated with the Short Performance Physical Battery (SPPB), was assessed in the ARIC study at visit 5 (2011-2013) among 5548 participants (mean age 75±5 years, female 58%, black 22%). The SPPB score was categorized into low (0-6), intermediate (7-9), and high (10-12) physical function. We assessed the associations of SPPB categories with composite and individual outcomes of coronary heart disease (CHD), stroke, or heart failure (HF) using Cox proportional hazards models adjusting for potential confounders. We also evaluated improvement in c-statistics by adding SPPB to covariates. Results: During a median follow-up of 6.2 years, there were 821 composite events (329 CHD, 226 stroke, and 467 HF cases). Compared to high SPPB score, low and intermediate SPPB score were robustly associated with higher risk of the composite CVD outcome (hazard ratio [HR] 1.59 [95% CI 1.29-1.95] and 1.32 [1.12-1.56], respectively) (Table). These associations were largely consistent between participants with and without a history of CVD at baseline. Among individual outcomes, low SPPB score showed the highest HR for stroke, whereas intermediate SPPB score was only significantly associated with HF. The addition of SPPB significantly improved c-statistic for composite outcome (Δc-statistic 0.006 [95% CI 0.002-0.009]), and the improvement was especially evident in participants without history of CVD (Δc-statistic 0.013 [0.003-0.023]). Conclusions: Lower physical function was robustly associated with the risk of CVD and improved its prediction beyond established predictors in older adults. Clinicians should pay attention to physical function when managing CVD risk in older adults.

Background: Carotid to femoral pulse wave velocity (cfPWV), the gold-standard measure of aortic stiffness, is associated with the incidence of cardiovascular disease (CVD) events. Most cfPWV measurements depend on technical proficiency, which limit broader implementation in clinical settings. Recently, an estimated pulse wave velocity (ePWV) measure derived from age and mean arterial pressure (MAP) has been shown to be associated with the risk of CVD events and all-cause mortality and may be a simple alternative to cfPWV. However, ePWV is poorly predictive in non-White individuals, possibly because it was originally developed from European Cohorts. Consequently, our purpose was to: estimate the strength of association between cfPWV and ePWV, and determine whether agreement is consistent across race. Hypothesis: cfPWV and ePWV are in good agreement and association would be consistent across race. Methods: We included Black and White older adults (n= 4,478; 75.2 [5.0] years) from visit 5 (2011-13) of the community-based Atherosclerosis Risk in Communities Study (ARIC). cfPWV was measured using an automated cardiovascular screening device. ePWV was calculated as presented in Fig 1. Association between cfPWV and ePWV was determined using Pearson’s correlation coefficient (r) and Bland-Altman plots. Results: As reported in Fig 1, there was a weak (r=0.35) correlation between cfPWV and ePWV for the total population, with similar correlations when stratified by Blacks (r=0.31) and Whites (r=0.36). Bland-Altman plots indicated significant systematic differences between cfPWV and ePWV (-0.17 m/s, 95% confidence interval: -0.25 to -0.09 m/s, P Conclusion: In older adults, there was weak correlation between cfPWV and ePWV and systematic differences in agreement. Our results do not support ePWV as a surrogate measure of cfPWV in a sample of older White and Black adults.

Background: In the 2018 AHA/ACC Cholesterol guideline, risk stratification is an essential element. The use of a Pooled Cohort Equation (PCE) is recommended for individuals without atherosclerotic cardiovascular disease (ASCVD), and the new dichotomous classification of very high-risk vs. high-risk has been introduced for patients with ASCVD. These distinct risk stratification systems mainly rely on traditional risk factors, raising the possibility that a single model can predict major adverse cardiovascular events (MACEs) in persons with and without ASCVD. Methods: We studied 11,335 ARIC participants with (n=885) and without (n=10,450) a history of ASCVD (myocardial infarction, ischemic stroke, and symptomatic peripheral artery disease) at baseline (1996-98). We modeled factors in the PCE and the new classification for ASCVD patients (Figure legend) in a single CVD prediction model. We examined their associations with MACEs (myocardial infarction, stroke, and heart failure) using Cox models and evaluated the discrimination and calibration for a single model including those factors. Results: During a median follow-up of 18.4 years, there were 3,658 MACEs (3,105 in participants without ASCVD). In general, the factors in the PCE and the risk classification system for ASCVD patients were associated similarly with MACEs regardless of baseline ASCVD status, although age and systolic blood pressure showed significant interactions. A single model with these predictors and the relevant interaction terms showed good calibration and discrimination for those with and without ASCVD (c-statistic=0.729 and 0.704, respectively) (Figure). Conclusion: A single CVD prediction model performed well in persons with and without ASCVD. This approach will provide a specific predicted risk to ASCVD patients (instead of dichotomy of very high vs. high risk) and eliminate a practice gap between primary vs. secondary prevention due to different risk prediction tools.

Background High blood pressure (BP) is a well‐known risk factor for atrial fibrillation (AF), but a single BP measurement may provide limited information about AF risk in older adults. Methods and Results This study included 1256 MESA (Multi‐Ethnic Study of Atherosclerosis) and 1948 ARIC (Atherosclerosis Risk in Communities) study participants who underwent extended ambulatory electrocardiographic monitoring and who were free of clinically detected cardiovascular disease, including AF. Using BP measurements from 6 examinations (2000–2018 in MESA and 1987–2017 in ARIC study), we calculated individual long‐term mean, trend, and detrended visit‐to‐visit variability in systolic BP and pulse pressure for each participant. Outcomes, assessed at examination 6, included subclinical AF and supraventricular ectopy. Results from each study were combined with inverse variance‐weighted meta‐analysis. At examination 6, the mean age was 73 years in MESA and 79 years in ARIC study, and 4% had subclinical AF. Higher visit‐to‐visit detrended variability in systolic BP was associated with a greater prevalence of subclinical AF (odds ratio [OR], 1.20; 95% CI, 1.02–1.38) and with more premature atrial contractions/hour (geometric mean ratio, 1.08; 95% CI, 1.01–1.15). For pulse pressure as well, higher visit‐to‐visit detrended variability was associated with a greater prevalence of AF (OR, 1.18; 95% CI, 1.00–1.37). In addition, higher long‐term mean pulse pressure was associated with a greater prevalence of subclinical AF (OR, 1.36; 95% CI, 1.08–1.70). Conclusions Antecedent visit‐to‐visit variability in systolic BP and pulse pressure, but not current BP, is associated with a higher prevalence of subclinical atrial arrhythmias. Prior longitudinal BP assessment, rather than current BP, may be more helpful in identifying older adults who are at higher risk of atrial arrhythmias.

Aims The aim of this study was to evaluate the associations of blood pressure categorization based on the 2017 American College of Cardiology and American Heart Association guideline with the risk of peripheral artery disease (PAD). Methods Among 13,113 middle-aged participants, we investigated the associations of 2017 blood pressure categories (systolic Results During a median follow-up of 25.4 years, 466 incident PAD occurred (271 cases in 9858 participants without anti-hypertensive medications). In participants without anti-hypertensive medications, we observed significant hazard ratios of PAD in elevated blood pressure (1.80 (1.28–2.51)) and stage 2 hypertension (2.40 (1.72–3.34)), but not in stage 1 hypertension. Analyzing systolic and diastolic blood pressure separately, higher systolic blood pressure categories showed significant associations with incident PAD in a graded fashion whereas, for diastolic blood pressure, only ≥90 mmHg did. Generally similar patterns were seen among participants on anti-hypertensive medication, while they had higher risk of PAD than those without at each blood pressure category. Conclusions Systolic blood pressure, including the category of 130–139 mmHg, showed stronger associations with incident PAD than did diastolic blood pressure. Consequently, elevated blood pressure conferred similar or even greater risk of PAD than stage 1 hypertension, with implications on how to interpret new blood pressure categories in terms of leg vascular health.

Background: We assessed whether plasma troponin I measured by a high-sensitivity assay (hs-TnI) is associated with incident cardiovascular disease (CVD) and mortality in a community-based sample without prior CVD. Methods: ARIC study (Atherosclerosis Risk in Communities) participants aged 54 to 74 years without baseline CVD were included in this study (n=8121). Cox proportional hazards models were constructed to determine associations between hs-TnI and incident coronary heart disease (CHD; myocardial infarction and fatal CHD), ischemic stroke, atherosclerotic CVD (CHD and stroke), heart failure hospitalization, global CVD (atherosclerotic CVD and heart failure), and all-cause mortality. The comparative association of hs-TnI and high-sensitivity troponin T with incident CVD events was also evaluated. Risk prediction models were constructed to assess prediction improvement when hs-TnI was added to traditional risk factors used in the Pooled Cohort Equation. Results: The median follow-up period was ≈15 years. Detectable hs-TnI levels were observed in 85% of the study population. In adjusted models, in comparison to low hs-TnI (lowest quintile, hs-TnI ≤1.3 ng/L), elevated hs-TnI (highest quintile, hs-TnI ≥3.8 ng/L) was associated with greater incident CHD (hazard ratio [HR], 2.20; 95% CI, 1.64–2.95), ischemic stroke (HR, 2.99; 95% CI, 2.01–4.46), atherosclerotic CVD (HR, 2.36; 95% CI, 1.86–3.00), heart failure hospitalization (HR, 4.20; 95% CI, 3.28–5.37), global CVD (HR, 3.01; 95% CI, 2.50–3.63), and all-cause mortality (HR, 1.83; 95% CI, 1.56–2.14). hs-TnI was observed to have a stronger association with incident global CVD events in white than in black individuals and a stronger association with incident CHD in women than in men. hs-TnI and high-sensitivity troponin T were only modestly correlated ( r =0.47) and were complementary in prediction of incident CVD events, with elevation of both troponins conferring the highest risk in comparison with elevation in either one alone. The addition of hs-TnI to the Pooled Cohort Equation model improved risk prediction for atherosclerotic CVD, heart failure, and global CVD. Conclusions: Elevated hs-TnI is strongly associated with increased global CVD incidence in the general population independent of traditional risk factors. hs-TnI and high-sensitivity troponin T provide complementary rather than redundant information.

Background and aims Peripheral arterial disease (PAD) is a clinical manifestation of extracoronary atherosclerosis. Many risk factors are involved in the process of PAD, but the association between serum magnesium (Mg) and PAD is not clear. Our study aimed to investigate whether serum Mg is associated with PAD incidence. Methods A total of 13,826 participants (aged 40–64 years) in the Atherosclerosis Risk in Communities (ARIC) study (1987–1989) without prior PAD were included in the final analysis. Serum Mg levels were measured at visits 1 and 2. PAD was defined as an ankle brachial index less than 0.9, or hospitalization with a PAD diagnosis. Cox regression was used to calculate hazard ratios (HRs) for incidence of PAD and serum Mg. Results During a median follow-up of 24.4 years, 1364 (48.4% female) PAD events were observed. After multiple adjustment, participants in the lowest (≤1.4 mEq/L) category of serum Mg compared with the highest (≥1.8 mEq/L) ones were at higher PAD risk (HR: 1.3; 95% confidence interval (CI): 1.06–1.58) (p value = 0.004). The HRs for PAD in 1.5, 1.6 and 1.7 mEq/L of serum Mg were 1.29 (95% CI: 1.08–1.54), 1.05 (95% CI: 0.89–1.24), and 1.0 (95% CI: 0.85–1.18), respectively. Conclusions Low serum Mg was independently associated with an increased prevalence of PAD in the large population-based study; further studies are needed to confirm our findings.

Resting heart rate (RHR) is independently associated with cardiovascular disease (CVD) risk. We determined whether RHR, measured in mid-life, is also associated with cognitive decline. We studied 13,720 middle-aged white and black ARIC participants without a history of stroke or atrial fibrillation. RHR was obtained from a 12-lead resting electrocardiogram at the baseline visit (1990 to 1992) and categorized into groups as60 (reference), 60 to 69, 70 to 79 and ≥80 beats/min. Cognitive scores were obtained at baseline and at up to 2 additional visits (1996 to 1998 and 2011 to 2013). The primary outcome was a global composite cognitive score (Z-score) derived from 3 tests: delayed word recall, digit symbol substitution, and word fluency. The associations of RHR with cognitive decline and incident dementia were examined using linear mixed-effects and Cox hazard models, respectively, adjusting for sociodemographics, CVD risk factors, and AV-nodal blockade use. Multiple imputation methods were used to account for attrition over follow-up. Participants had mean ± SD age of 58 ± 6 years; 56% were women, 24% black. Average RHR was 66 ± 10 beats/min. Over a mean follow-up of 20 years, those with RHR ≥80 beats/min had greater global cognitive decline (average adjusted Z-score difference -0.12 [95% confidence interval -0.21, -0.03]) and increased risk for incident dementia (hazard ratio 1.28 (1.04, 1.57), compared with those with RHR60 beats/min. In conclusion, elevated RHR is independently associated with greater cognitive decline and incident dementia over 20 years. Further studies are needed to determine whether the associations are causal or secondary to another underlying process, and whether modification of RHR can affect cognitive decline.

H63D has been identified as a novel locus associated with the development of hypertension. The quantitative risks for hypertension, cardiac remodeling, and adverse events are not well studied. We analyzed white participants from the ARIC study (Atherosclerosis Risk in Communities) with H63D genotyping (N=10 902). We related genotype status to prevalence of hypertension at each of 5 study visits and risk for adverse cardiovascular events. Among visit 5 participants (N=4507), we related genotype status to echocardiographic features. Frequencies of wild type (WT)/WT, H63D/WT, and H63D/H63D were 73%, 24.6%, and 2.4%. The average age at baseline was 54.9±5.7 years and 47% were men. Participants carrying the H63D variant had higher systolic blood pressure ( P =0.004), diastolic blood pressure (0.012), and more frequently had hypertension ( P 65 years. After 25 years of follow-up, there was no relationship between genotype status and any outcome ( P >0.05). H63D/WT and H63D/H63D genotypes were associated with small differences in cardiac remodeling. In conclusion, the HFE H63D variant confers an increased risk for hypertension per allele and, given its frequency, accounts for a significant number of cases of hypertension. However, there was no increased risk for adverse cardiovascular events or substantial left ventricular remodeling.

AIM: —Our goal was to investigate the effect of a global XYZ median beat construction and the heart vector origin point definition on predictive accuracy of ECG biomarkers of sudden cardiac death (SCD). METHODS: —Atherosclerosis Risk In Community study participants with analyzable digital ECGs were included (n=15,768; 55% female, 73% white, mean age 54.2±5.8 y). We developed an algorithm to automatically detect the heart vector origin point on a median beat. Three different approaches to construct a global XYZ beat and two methods to locate origin point were compared. Global electrical heterogeneity was measured by sum absolute QRST integral (SAI QRST), spatial QRS-T angle, and spatial ventricular gradient (SVG) magnitude, azimuth, and elevation. Adjudicated SCD served as the primary outcome. RESULTS: —There was high intra-observer (kappa 0.972) and inter-observer (kappa 0.984) agreement in a heart vector origin definition between an automated algorithm and a human. QRS was wider in a median beat that was constructed using R-peak alignment than in time-coherent beat (88.1±16.7 vs. 83.7±15.9 ms; P

Background Antiplatelets, anticoagulants, and statins are commonly prescribed for various indications. The associations between these medications and the risk of intracerebral hemorrhage (ICH) and cerebral microbleeds (CMBs) are unclear. Methods and Results We performed a retrospective study of the ARIC (Atherosclerosis Risk in Communities) study cohort, recruited from 4 US communities in 1987 to 1989 with follow‐up. In 2011 to 2013, a subset (N=1942) underwent brain magnetic resonance imaging with CMB evaluation. Time‐varying and any antiplatelet, anticoagulant, or statin use was evaluated at subsequent study visits in participants not on each medication at baseline. To determine the hazard of ICH and odds of CMB by medication use, logistic and Cox proportional hazard models were built, respectively, adjusting for the propensity to take the medication, concomitant use of other medications, and cognitive, genetic, and radiographic data. Of 15 719 individuals during up to 20 years of follow‐up, 130 participants experienced an ICH. The adjusted hazard of ICH was significantly lower among participants taking an antiplatelet at the most recent study visit before ICH versus nonusers (hazard ratio [HR], 0.53; 95% CI, 0.30–0.92). Statin users had a significantly lower hazard of an ICH compared with nonusers (adjusted HR, 0.13; 95% CI, 0.05–0.34). There was no association of CMB and antiplatelet, anticoagulant, or statin use in adjusted models. Conclusions In this US community‐based study, antiplatelet and statin use were associated with lower ICH hazard, whereas no association was noted between CMBs and antiplatelets, anticoagulants, and statins. Further study is needed to understand the differential roles of these medications in cerebral microhemorrhages and macrohemorrhages.

Background: Coronary heart disease, heart failure (HF), and stroke are complex diseases with multiple phenotypes. While many risk factors for these diseases are well known, investigation of as-yet unidentified risk factors may improve risk assessment and patient adherence to prevention guidelines. We investigated the diet domain in FHS (Framingham Heart Study), CHS (Cardiovascular Heart Study), and the ARIC study (Atherosclerosis Risk in Communities) to identify potential lifestyle and behavioral factors associated with coronary heart disease, HF, and stroke. Methods: We used machine learning feature selection based on random forest analysis to identify potential risk factors associated with coronary heart disease, stroke, and HF in FHS. We evaluated the significance of selected variables using univariable and multivariable Cox proportional hazards analysis adjusted for known cardiovascular risks. Findings from FHS were then validated using CHS and ARIC. Results: We identified multiple dietary and behavioral risk factors for cardiovascular disease outcomes including marital status, red meat consumption, whole milk consumption, and coffee consumption. Among these dietary variables, increasing coffee consumption was associated with decreasing long-term risk of HF congruently in FHS, ARIC, and CHS. Conclusions: Higher coffee intake was found to be associated with reduced risk of HF in all three studies. Further study is warranted to better define the role, possible causality, and potential mechanism of coffee consumption as a potential modifiable risk factor for HF.

Background Circulating lipids have been implicated as important modulators of immune response, and altered lipid levels correlate with the severity of infection. However, long-term prognostic implications of lipid levels regarding future infection risk remain unclear. The current project aims to explore whether baseline lipid levels are associated with risk of future serious infection, measured by hospitalization for pneumonia. Methods A retrospective analysis was performed in 13,478 participants selected from the Atherosclerosis Risk in Communities (ARIC) study, a large community-based longitudinal cohort in the United States with a median follow-up time of >20 years. First incident of hospitalization for pneumonia was identified through hospital discharge records. Cox proportional hazard models were used to assess the association of baseline major lipid levels (total cholesterol, low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], triglycerides) with time to first pneumonia hospitalization. Results A total of 1969 (14.61%) participants had a pneumonia hospitalization during a median follow-up time of 21.5 years. The hazard ratio (HR) for pneumonia hospitalization was 0.90 (95% confidence interval, 0.87-0.92) for every 10-mg/dL increase in baseline HDL-C, and 1.02 (95% confidence interval, 1.02-1.03) for every 10-mg/dL increase in baseline triglycerides. HDL-C and triglycerides both remained significant predictors of pneumonia hospitalization after multivariable adjustment. Such associations were not seen with baseline LDL-C or total cholesterol levels. Conclusion Lower baseline HDL-C and higher triglyceride levels were strongly associated with increased risk of long-term pneumonia hospitalization in a large longitudinal US cohort.