Ralph B D'Agostino

Wake Forest School of Medicine, Winston-Salem, North Carolina, United States

Are you Ralph B D'Agostino?

Claim your profile

Publications (805)7251.01 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: The SEARCH for Diabetes in Youth (SEARCH) study was initiated in 2000, with funding from the Centers for Disease Control and Prevention and support from the National Institute of Diabetes and Digestive and Kidney Diseases, to address major knowledge gaps in the understanding of childhood diabetes. SEARCH is being conducted at five sites across the U.S. and represents the largest, most diverse study of diabetes among U.S. youth. An active registry of youth diagnosed with diabetes at age <20 years allows the assessment of prevalence (in 2001 and 2009), annual incidence (since 2002), and trends by age, race/ethnicity, sex, and diabetes type. Prevalence increased significantly from 2001 to 2009 for both type 1 and type 2 diabetes in most age, sex, and race/ethnic groups. SEARCH has also established a longitudinal cohort to assess the natural history and risk factors for acute and chronic diabetes-related complications as well as the quality of care and quality of life of persons with diabetes from diagnosis into young adulthood. Many youth with diabetes, particularly those from low-resourced racial/ethnic minority populations, are not meeting recommended guidelines for diabetes care. Markers of micro- and macrovascular complications are evident in youth with either diabetes type, highlighting the seriousness of diabetes in this contemporary cohort. This review summarizes the study methods, describes key registry and cohort findings and their clinical and public health implications, and discusses future directions. © 2014 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.
    Diabetes Care 12/2014; 37(12):3336-3344. · 8.57 Impact Factor
  • Source
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Neoadjuvant chemotherapy for breast cancer leads to considerable variability in clinical responses, with only 10 to 20% of cases achieving complete pathologic responses (pCR). Biological and clinical factors that determine the extent of pCR are incompletely understood. Mounting evidence indicates that the patient's immune system contributes to tumor regression and can be modulated by therapies. The cell types most frequently observed with this association are effector tumor infiltrating lymphocytes (TILs), such as cytotoxic T cells, natural killer cells and B cells. We and others have shown that the relative abundance of TILs in breast cancer can be quantified by intratumoral transcript levels of coordinately expressed, immune cell-specific genes. Through expression microarray analysis, we recently discovered three immune gene signatures, or metagenes, that appear to reflect the relative abundance of distinct tumor-infiltrating leukocyte populations. The B/P (B cell/plasma cell), T/NK (T cell/natural killer cell) and M/D (monocyte/dendritic cell) immune metagenes were significantly associated with distant metastasis-free survival of patients with highly proliferative cancer of the basal-like, HER2-enriched and luminal B intrinsic subtypes. Given the histopathological evidence that TIL abundance is predictive of neoadjuvant treatment efficacy, we evaluated the therapy-predictive potential of the prognostic immune metagenes. We hypothesized that pre-chemotherapy immune gene signatures would be significantly predictive of tumor response. In a multi-institutional, meta-cohort analysis of 701 breast cancer patients receiving neoadjuvant chemotherapy, gene expression profiles of tumor biopsies were investigated by logistic regression to determine the existence of therapy-predictive interactions between the immune metagenes, tumor proliferative capacity, and intrinsic subtypes. By univariate analysis, the B/P, T/NK and M/D metagenes were all significantly and positively associated with favorable pathologic responses. In multivariate analyses, proliferative capacity and intrinsic subtype altered the significance of the immune metagenes in different ways, with the M/D and B/P metagenes achieving the greatest overall significance after adjustment for other variables. Gene expression signatures of infiltrating immune cells carry both prognostic and therapy-predictive value that is impacted by tumor proliferative capacity and intrinsic subtype. Anti-tumor functions of plasma B cells and myeloid-derived antigen-presenting cells may explain more variability in pathologic response to neoadjuvant chemotherapy than previously recognized.
    Genome Medicine 10/2014; 6(10):80. · 4.94 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: -In a murine anthracycline-related cardiotoxicity model, increases in cardiovascular magnetic resonance (CMR) myocardial contrast-enhanced T1-weighted signal intensity are associated with myocellular injury and decreases in left ventricular ejection fraction (LVEF). We sought to determine if T1- and T2-weighted measures of signal intensity associate with decreases in LVEF in human subjects receiving potentially cardiotoxic chemotherapy.
    Circulation Cardiovascular Imaging 10/2014; · 5.80 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Biomarkers of cardiovascular stress have been associated with incident cardiovascular outcomes. Their relations with measures of subclinical atherosclerosis, as assessed by carotid intima-media thickness, have not been well described.
    Clinical Chemistry 09/2014; · 7.77 Impact Factor
  • Statistics in Medicine 08/2014; 33(19). · 2.04 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Microvascular dysfunction is a key event in the development of atherosclerosis, which predates the clinical manifestations of vascular disease including stroke and myocardial infarction. Dysfunction of the microvasculature can be measured as a decreased microperfusion in response to heat.
    Pediatric Diabetes 08/2014; · 2.13 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective Dyslipidemia contributes to the increased risk of cardiovascular disease in persons with type 1 diabetes (T1D). Weight control is commonly recommended as a treatment for dyslipidemia. However, the extent to which decreases in weight affect the lipid profile in youth with T1D is not known. Therefore, we tested the hypothesis that decreases in body mass index z-score (BMIz) were associated with concomitant changes in the lipid profile in youth with T1D.Study DesignWe studied 1142 youth with incident T1D, who had at least two fasting lipid measurements over 2 yr (initial visit mean: age = 10.8 ± 3.9 yr, BMIz = 0.55 ± 0.97, T1D duration = 10.7 ± 7.6 months; 47.5% female, 77.9% non-Hispanic white) in the SEARCH for Diabetes in Youth Study. Longitudinal mixed models were used to examine the relationships between changes in BMIz and changes in total, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), non-HDL cholesterol, and log triglycerides (TG) adjusted for initial age, sex, race/ethnicity, clinical site, season of study visit, T1D duration, and glycated hemoglobin A1c (HbA1c).ResultsWe found that over 2 yr all lipid levels, except LDL-C, increased significantly (p < 0.05). Decreases in BMIz were associated with favorable changes in HDL-C and TG only and the magnitude of these changes depended on the initial BMIz value (interaction p < 0.05), so that greater improvements were seen in those with higher BMIz.Conclusions Our data suggest that weight loss may be an effective, but limited, therapeutic approach for dyslipidemia in youth with T1D.
    Pediatric Diabetes 08/2014; · 2.13 Impact Factor
  • Karol M. Pencina, Michael J. Pencina, Ralph B. D'Agostino
    [Show abstract] [Hide abstract]
    ABSTRACT: The net reclassification improvement (NRI) has become a popular measure of incremental usefulness of markers added to risk prediction models. However, the expected magnitude of the three-category NRI is not well understood, leading researchers to rely on statistical significance. In this paper, we describe a slight modification to the original definition of the NRI, which weighs each reclassification by the number of categories by which a given individual is reclassified. This modification resolves some recent criticisms of the three-category NRI and at the same time has a minimal impact on its magnitude. Then we show that using this modified definition, the event and nonevent NRIs have simple interpretations as sums of changes in sensitivities and specificities calculated at the risk thresholds. We exploit this relationship to arrive at closed-form solutions for the NRI under normality within the event and nonevent subgroups. We observe that the size of the intermediate risk category and the event rate have limited impact on the magnitude of the NRI. As expected, the NRI increases with the strength of the added marker, and this relationship appears fairly proportional for markers with non-weak net effect size (above 0.25). Furthermore, we conclude that using the NRI as a metric, it is harder to improve models that already perform well. Copyright © 2014 John Wiley & Sons, Ltd.
    Statistics in Medicine 08/2014; · 2.04 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This study highlights a simple bedside evaluation of itch and pain for suspicious skin lesions.
    JAMA Dermatology 07/2014; · 4.30 Impact Factor
  • Siyan Xu, Kerry Barker, Sandeep Menon, Ralph B D'Agostino
    [Show abstract] [Hide abstract]
    ABSTRACT: Non-inferiority (NI) clinical trials are getting a lot of attention of late due to its direct application in biosimilar studies. Because of missing placebo arm, NI is an indirect approach to demonstrate efficacy of a test treatment. One of the key assumptions on the NI test is constancy assumption, that is, the effect of reference treatment is the same in current NI trials as in historical superiority trials. However, if a covariate interacts with the treatment arms, then changes in distribution of this covariate will likely result in violation of constancy assumption. In this paper, we propose four new NI methods and compare them with two existing methods to evaluate the change of background constancy assumption on the performance of these six methods. To achieve this goal, we study the impact of three elements: 1) Strength of covariate; 2) Degree of interaction between covariate and treatment and; 3) Differences in distribution of the covariate between historical and current trials have on both the type I error rate and power using three different measures of association: difference, log relative risk and log odds ratio. Based on this research, we recommend using a modified covariate-adjustment fixed margin method.
    Journal of Biopharmaceutical Statistics 07/2014; · 0.72 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Abstract Equivalence trials aim to demonstrate that new and standard treatments are equivalent within pre-defined clinically relevant limits. We focus when inference of equivalence is made in terms of the ratio of two normal means. In the presence of unspecified variances, methods such as likelihood-ratio test use sample estimates for those variances; Bayesian models integrate them out in the posterior distribution. These methods limit the knowledge on the extent that equivalence is affected by variability of the parameter of interest. In this paper, we propose a likelihood approach that retains the unspecified variances in the model and partitions the likelihood function into two components: F-statistic function for variances and t-statistic function for the ratio of two means. By incorporating unspecified variances, the proposed method can help identify numeric range of variances where equivalence is more likely to be achieved, which cannot be accomplished by current analysis methods. By partitioning the likelihood function into two components, the proposed method provides more inference information than a method that relies solely on one component. Using a published real example data, we show that the proposed method produces the same results as the likelihood-ratio test and comparable to Bayesian analysis in general case. In a special case where the ratio of two variances is directly proportional to the ratio of two means, the proposed method yields better results in inference about equivalence than either likelihood-ratio test or Bayesian method. Using a published real example data, the proposed likelihood method is shown to be a better alternative than current analysis methods for equivalence inference.
    Journal of Biopharmaceutical Statistics 07/2014; · 0.72 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Abstract Background: Multiple abnormal metabolic traits are found together or "cluster" within individuals more often than is predicted by chance. The individual and combined role of adiposity and insulin resistance (IR) on metabolic trait clustering is uncertain. We tested the hypothesis that change in trait clustering is a function of both baseline level and change in these measures. Methods: In 2616 nondiabetic Framingham Offspring Study participants, body mass index (BMI) and fasting insulin were related to a within-person 7-year change in a trait score of 0-4 Adult Treatment Panel III metabolic syndrome traits (hypertension, high triglycerides, low high-density lipoprotein cholesterol, hyperglycemia). Results: At baseline assessment, mean trait score was 1.4 traits, and 7-year mean (SEM) change in trait score was +0.25 (0.02) traits, P<0.0001. In models with BMI predictors only, for every quintile difference in baseline BMI, the 7-year trait score increase was 0.14 traits, and for every quintile increase in BMI during 7-year follow-up, the trait score increased by 0.3 traits. Baseline level and change in fasting insulin were similarly related to trait score change. In models adjusted for age-sex-baseline cluster score, 7-year change in trait score was significantly related to both a 1-quintile difference in baseline BMI (0.07 traits) and fasting insulin (0.18 traits), and to both a 1-quintile 7-year increase in BMI (0.21 traits) and fasting insulin (0.18 traits). Conclusions: Change in metabolic trait clustering was significantly associated with baseline levels and changes in both BMI and fasting insulin, highlighting the importance of both obesity and IR in the clustering of metabolic traits.
    Metabolic Syndrome and Related Disorders 07/2014; · 1.92 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background The study provides evidence of the longitudinal association between screen time with hemoglobin A1c (HbA1c) and cardiovascular risk markers among youth with type 1 diabetes (T1D) and type 2 diabetes (T2D).Objective To examine the longitudinal relationship of screen time with HbA1c and serum lipids among youth with diabetes.SubjectsYouth with T1D and T2D.Methods We followed up 1049 youth (≥10 yr old) with recently diagnosed T1D and T2D participating in the SEARCH for Diabetes in Youth Study.ResultsIncreased television watching on weekdays and during the week over time was associated with larger increases in HbA1c among youth with T1D and T2D (p-value <0.05). Among youth with T1D, significant longitudinal associations were observed between television watching and TG (p-value <0.05) (week days and whole week), and low-density lipoprotein cholesterol (LDL-c, p-value <0.05) (whole week). For example, for youth who watched 1 h of television per weekday at the outset and 3 h per weekday 5 yr later, the longitudinal model predicted greater absolute increases in HbA1c (2.19% for T1D and 2.16% for T2D); whereas for youth who watched television 3 h per weekday at the outset and 1 h per weekday 5 yr later, the model predicted lesser absolute increases in HbA1c (2.08% for T1D and 1.06% for T2D).Conclusions Youth with T2D who increased their television watching over time vs. those who decreased it had larger increases in HbA1c over 5 yr. Youth with T1D who increased their television watching over time had increases in LDL-c, TG, and to a lesser extent HbA1c.
    Pediatric Diabetes 07/2014; · 2.13 Impact Factor
  • Steven Y. Hua, Siyan Xu, Ralph B. D'Agostino
    [Show abstract] [Hide abstract]
    ABSTRACT: Bioequivalence of two drugs is usually demonstrated by rejecting two one-sided null hypotheses using the two one-sided tests for pharmacokinetic parameters: area under the concentration-time curve (AUC) and maximum concentration (Cmax). By virtue of the intersection–union test, there is no need for multiplicity adjustment in testing the two one-sided null hypotheses within each parameter. However, the decision rule for bioequivalence often requires equivalence to be achieved simultaneously on both parameters that contain four one-sided null hypotheses together; without adjusting for multiplicity, the family wise error rate (FWER) could fail to be controlled at the nominal type-I error rate α. The multiplicity issue for bioequivalence in this regard is scarcely discussed in the literature. To address this issue, we propose two approaches including a closed test procedure that controls FWER for the simultaneous AUC and Cmax bioequivalence and requires no adjustment of the type-I error, and an alpha-adaptive sequential testing (AAST) that controls FWER by pre-specifying the significance level on AUC (α1) and obtaining it for Cmax (α2) adaptively after testing of AUC. While both methods control FWER, the closed test requires testing of eight intersection null hypotheses each at α, and AAST is at times accomplished through a slight deduction in α1 and no deduction in α2 relative to α. The latter considers equivalence reached in AUC a higher importance than that in Cmax. Illustrated with published data, the two approaches, although operate differently, can lead to the same substantive conclusion and are better than a traditional method like Bonferroni adjustment. Copyright © 2014 John Wiley & Sons, Ltd.
    Statistics in Medicine 07/2014; · 2.04 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: New markers may improve prediction of diagnostic and prognostic outcomes. We aimed to review options for graphical display and summary measures to assess the predictive value of markers over standard, readily available predictors. We illustrated various approaches using previously published data on 3264 participants from the Framingham Heart Study, where 183 developed coronary heart disease (10-year risk 5.6%). We considered performance measures for the incremental value of adding HDL cholesterol to a prediction model. An initial assessment may consider statistical significance (HR = 0.65, 95% confidence interval 0.53 to 0.80; likelihood ratio p < 0.001), and distributions of predicted risks (densities or box plots) with various summary measures. A range of decision thresholds is considered in predictiveness and receiver operating characteristic curves, where the area under the curve (AUC) increased from 0.762 to 0.774 by adding HDL. We can furthermore focus on reclassification of participants with and without an event in a reclassification graph, with the continuous net reclassification improvement (NRI) as a summary measure. When we focus on one particular decision threshold, the changes in sensitivity and specificity are central. We propose a net reclassification risk graph, which allows us to focus on the number of reclassified persons and their event rates. Summary measures include the binary AUC, the two-category NRI, and decision analytic variants such as the net benefit (NB). Various graphs and summary measures can be used to assess the incremental predictive value of a marker. Important insights for impact on decision making are provided by a simple graph for the net reclassification risk.
    Biometrical Journal 07/2014; · 1.15 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The present study aimed to examine perceptions of shared decision-making (SDM) in caregivers of youth with type 1 diabetes (T1D). Interview, survey data, and HbA1c assays were gathered from caregivers of 439 youth with T1D aged 3-18 years. Caregiver-report indicated high perceived SDM during medical visits. Multivariable linear regression indicated that greater SDM is associated with lower HbA1c, older child age, and having a pediatric endocrinologist provider. Multiple logistic regression found that caregivers who did not perceive having made any healthcare decisions in the past year were more likely to identify a non-pediatric endocrinologist provider and to report less optimal diabetes self-care. Findings suggest that youth whose caregivers report greater SDM may show benefits in terms of self-care and glycemic control. Future research should examine the role of youth in SDM and how best to identify youth and families with low SDM in order to improve care.
    Journal of Clinical Psychology in Medical Settings 06/2014; · 1.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Sensitive general cardiometabolic risk assessment tools of modifiable risk factors would be helpful and practical in a range of primary prevention interventions or for preventive health maintenance.
    American Journal of Preventive Medicine 06/2014; · 4.28 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: People with atherosclerotic renal artery stenosis may benefit from renin-angiotensin inhibitors, angiotensin-converting enzyme inhibitors, and angiotensin-receptor blockers, but little is known about the factors associated with their use.DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS: The Cardiovascular Outcomes in Renal Atherosclerotic Lesions study (ClinicalTrials.gov identified: NCT00081731) is a prospective, international, multicenter clinical trial that randomly assigned participants with atherosclerotic renal artery stenosis who received optimal medical therapy to stenting versus no stenting from May 2005 through January 2010. At baseline, medication information was available from 853 of 931 randomly assigned participants. Kidney function was measured by serum creatinine-based eGFR at a core laboratory.RESULTS: Before randomization, renin-angiotensin inhibitors were used in 419 (49%) of the 853 participants. Renin-angiotensin inhibitor use was lower in those with CKD (eGFR<60 ml/min per 1.73 m(2)) (58% versus 68%; P=0.004) and higher in individuals with diabetes (41% versus 27%; P<0.001). Presence of bilateral renal artery stenosis or congestive heart failure was not associated with renin-angiotensin inhibitor use. Although therapy with renin-angiotensin inhibitors varied by study site, differences in rates of use were not related to the characteristics of the site participants. Participants receiving a renin-angiotensin inhibitor had lower systolic BP (mean±SD, 148±23 versus 152±23 mmHg; P=0.003) and more often had BP at goal (30% versus 22%; P=0.01).CONCLUSIONS: Kidney function and diabetes were associated with renin-angiotensin inhibitor use. However, these or other clinical characteristics did not explain variability among study sites. Patients with renal artery stenosis who received renin-angiotensin inhibitor treatment had lower BP and were more likely to be at treatment goal.
    Clinical Journal of the American Society of Nephrology 06/2014; · 5.25 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The SEARCH for Diabetes in Youth Study prospectively identified youth less than 20 years with physician-diagnosed diabetes. Annual type 1 diabetes (T1D) incidence rate and 95 percent CI, overall, by age group and by sex, were calculated per 100,000 person-years at risk for 2002 through 2009 for non-Hispanic white (NHW) youth. Joinpoint and Poisson regression models were used to test for temporal trends. The age- and sex-adjusted incidence of T1D increased from 24.4/100,000 (95% CI 23.9-24.8) in 2002 to 27.4/100,000 (95% CI 26.9-27.9) in 2009 (p for trend=0.0008). The relative annual increase in T1D incidence was 2.72% (1.18-4.28%) per year; 2.84% (1.12-4.58%) for males and 2.57% (0.68-4.51%) for females. After adjustment for sex, there were significant increases for those 5-9 years (p=0.0023), 10-14 years (p= 0.0008), and 15-19 years (p=0.004), but not among 0-4 year olds (p=0.1862). Mean age at diagnosis did not change. The SEARCH study demonstrated a significant increase in the incidence of T1D among NHW youth from 2002 through 2009 overall and in all but the youngest age group. Continued surveillance of T1D in youth in the United States to identify future trends in T1D incidence and to plan for health care delivery is warranted.
    Diabetes 06/2014; · 7.90 Impact Factor

Publication Stats

67k Citations
7,251.01 Total Impact Points


  • 1997–2014
    • Wake Forest School of Medicine
      • • Section on Cardiology
      • • Department of Cancer Biology
      • • Department of Biostatistical Sciences
      • • Division of Public Health Sciences
      • • Section on Infectious Diseases
      Winston-Salem, North Carolina, United States
  • 1987–2014
    • Boston University
      • • Department of Mathematics and Statistics
      • • Division of Mathematics
      • • Endocrinology, Diabetes, and Nutrition
      • • Department of Neurology
      • • Section of Preventive Medicine and Epidemiology
      Boston, Massachusetts, United States
  • 2012–2013
    • Colorado Department of Public Health and Environment
      Denver, Colorado, United States
    • McGill University
      Montréal, Quebec, Canada
    • Dana-Farber Cancer Institute
      • Department of Biostatistics and Computational Biology
      Boston, MA, United States
    • Children's Hospital Colorado
      Aurora, Colorado, United States
  • 2007–2013
    • Brigham and Women's Hospital
      • Department of Medicine
      Boston, MA, United States
    • University of Toronto
      • Institute for Clinical Evaluative Sciences
      Toronto, Ontario, Canada
  • 2005–2013
    • Karl Jaspers Society of North America
      United States
    • University of Texas Health Science Center at Houston
      • Department of Pediatrics
      Houston, TX, United States
    • Durham University
      Durham, England, United Kingdom
  • 1998–2013
    • National Institutes of Health
      • Branch of Liver Diseases Branch (LDB)
      Maryland, United States
    • University at Buffalo, The State University of New York
      Buffalo, New York, United States
  • 1997–2013
    • National Heart, Lung, and Blood Institute
      • Division of Cardiovascular Sciences (DCVS)
      Maryland, United States
  • 1982–2013
    • Boston Medical Center
      Boston, Massachusetts, United States
  • 2008–2012
    • University of Colorado
      • • Department of Epidemiology
      • • Barbara Davis Center for Childhood Diabetes
      Denver, CO, United States
    • University Medical Center Utrecht
      • Julius Center for Health Sciences and Primary Care
      Utrecht, Provincie Utrecht, Netherlands
    • The University of Manchester
      • Institute of Cardiovascular Sciences
      Manchester, ENG, United Kingdom
  • 2003–2012
    • University of South Carolina
      • Department of Epidemiology & Biostatistics
      Columbia, SC, United States
    • U.S. Department of Veterans Affairs
      Washington, Washington, D.C., United States
  • 1992–2012
    • University of Massachusetts Boston
      • Clinical Epidemiology Research and Training Unit
      Boston, Massachusetts, United States
    • Mass College of Liberal Arts
      Boston, Massachusetts, United States
    • Tufts Medical Center
      • Department of Medicine
      Boston, Massachusetts, United States
  • 2011
    • Boston Biomedical Research Institute
      Boston, Massachusetts, United States
  • 2010–2011
    • Duke University Medical Center
      • • Duke Cancer Institute
      • • Department of Community and Family Medicine
      Durham, NC, United States
    • University of Chicago
      • Department of Medicine
      Chicago, IL, United States
  • 2007–2011
    • University of Illinois at Chicago
      • • Department of Medicine (Chicago)
      • • Section of Endocrinology, Diabetes and Metabolism
      Chicago, IL, United States
  • 2005–2011
    • Rhode Island Hospital
      Providence, Rhode Island, United States
  • 1993–2011
    • University of Washington Seattle
      • • Cardiovascular Health Research Unit (CHRU)
      • • Department of Otolaryngology/Head and Neck Surgery
      Seattle, WA, United States
    • University of Alabama at Birmingham
      • Department of Medicine
      Birmingham, AL, United States
  • 2007–2010
    • Emory University
      • Division of Cardiology
      Atlanta, GA, United States
  • 1996–2010
    • Wake Forest University
      • • Department of Biostatistical Sciences
      • • School of Medicine
      • • Department of Public Health Sciences
      Winston-Salem, North Carolina, United States
  • 1995–2010
    • University of Pittsburgh
      • • Department of Epidemiology
      • • Center for Research on Health Care
      Pittsburgh, PA, United States
  • 1992–2010
    • Massachusetts General Hospital
      • • Department of Medicine
      • • Cardiovascular Disease Prevention Center
      Boston, MA, United States
  • 2009
    • Johannes Gutenberg-Universität Mainz
      Mayence, Rheinland-Pfalz, Germany
    • The University of Arizona
      • Division of Epidemiology and Biostatistics
      Tucson, AZ, United States
    • Kaiser Permanente
      Oakland, California, United States
  • 2000–2009
    • University of Texas Health Science Center at San Antonio
      • • Division of Clinical Epidemiology
      • • Division of Hospital Medicine
      San Antonio, TX, United States
  • 2006–2008
    • Partners HealthCare
      Boston, Massachusetts, United States
    • Harvard University
      • Department of Society, Human Development, and Health
      Boston, MA, United States
  • 2003–2008
    • Tufts University
      • Department of Medicine
      Georgia, United States
  • 1999–2008
    • Beth Israel Deaconess Medical Center
      • • Division of General Medicine and Primary Care
      • • Department of Medicine
      Boston, MA, United States
    • National Eye Institute
      Maryland, United States
  • 1993–2008
    • University of Pennsylvania
      • Division of Cardiovascular Medicine
      Philadelphia, PA, United States
  • 2003–2007
    • IMIM Hospital del Mar Medical Research Institute
      Barcino, Catalonia, Spain
  • 2003–2006
    • Harvard Medical School
      • Department of Medicine
      Boston, Massachusetts, United States
  • 1993–2006
    • Massachusetts Institute of Technology
      • Laboratory for Computer Science
      Cambridge, Massachusetts, United States
  • 2002–2005
    • Royal North Shore Hospital
      Sydney, New South Wales, Australia
  • 1993–2005
    • Medical University of South Carolina
      • • General Clinical Research Center
      • • Division of Neuroradiology
      Charleston, SC, United States
  • 1988–2005
    • Beverly Hospital, Boston MA
      Beverly, Massachusetts, United States
    • University of Miami Miller School of Medicine
      • Department of Neurology
      Miami, FL, United States
  • 2004
    • Hospital De Clínicas De Porto Alegre
      Pôrto de São Francisco dos Casaes, Rio Grande do Sul, Brazil
    • University of Alberta
      • Department of Medicine
      Edmonton, Alberta, Canada
    • Northwestern University
      • Department of Preventive Medicine
      Evanston, IL, United States
    • Countess Of Chester Hospital NHS Foundation Trust
      Chester, England, United Kingdom
  • 1996–2001
    • Texas Tech University Health Sciences Center
      • Department of Medicine
      Lubbock, TX, United States
  • 1995–1997
    • University of Maine
      • Department of Psychology
      Orono, MN, United States
  • 1991–1997
    • New England Baptist Hospital
      Boston, Massachusetts, United States
  • 1994
    • University of Illinois, Urbana-Champaign
      Urbana, Illinois, United States
    • University of Florida
      Gainesville, Florida, United States
  • 1990
    • New England Research Institutes
      Watertown, Massachusetts, United States
    • Erasmus Universiteit Rotterdam
      Rotterdam, South Holland, Netherlands