A Novel Risk Classification Paradigm for Patients With Impaired Glucose Tolerance and High Cardiovascular Risk

• Published in: The American Journal of Cardiology
• Main Authors: Bethel, M. Angelyn, Chacra, Antonio R., Deedwania, Prakash, Fulcher, Gregory R., Holman, Rury R., Jenssen, Trond, Kahn, Steven E., Levitt, Naomi S., McMurray, John J. V., Califf, Robert M., Raptis, Sotirios A., Thomas, Laine, Sun, Jie-Lena, Haffner, Steven M.
• Other Authors: Univ Oxford, Duke Univ, Universidade Federal de São Paulo (UNIFESP), Univ Calif San Francisco Program Fresno, Vet Adm Cent Calif Hlth Care Syst, Univ Sydney, Univ Tromso, Vet Affairs Puget Sound Hlth Care Syst, Univ Washington, Univ Cape Town, Univ Glasgow, Attikon Univ Hosp
• Format: Article in Journal/Newspaper
• Language: English
• Published: Elsevier B.V. 2013
• Subjects: Norway; Tromso
• Online Access: http://repositorio.unifesp.br/handle/11600/36542; https://doi.org/10.1016/j.amjcard.2013.03.019

We used baseline data from the NAVIGATOR trial to (1) identify risk factors for diabetes progression in those with impaired glucose tolerance and high cardiovascular risk, (2) create models predicting 5-year incident diabetes, and (3) provide risk classification tools to guide clinical interventions. Multivariate Cox proportional hazards models estimated 5-year incident diabetes risk and simplified models examined the relative importance of measures of glycemia in assessing diabetes risk. the C-statistic was used to compare models; reclassification analyses compare the models' ability to identify risk groups defined by potential therapies (routine or intensive lifestyle advice or pharmacologic therapy). Diabetes developed in 3,254 (35%) participants over 5 years median follow-up. the full prediction model included fasting and 2-hour glucose and hemoglobin A1c (HbA1c) values but demonstrated only moderate discrimination for diabetes (C = 0.70). Simplified models with only fasting glucose (C = 0.67) or oral glucose tolerance test values (C = 0.68) had higher C statistics than models with HbA1c alone (C = 0.63). the models were unlikely to inappropriately reclassify participants to risk groups that might receive pharmacologic therapy. Our results confirm that in a population with dysglycemia and high cardiovascular risk, traditional risk factors are appropriate predictors and glucose values are better predictors than HbA1c, but discrimination is moderate at best, illustrating the challenges of predicting diabetes in a high-risk population. in conclusion, our novel risk classification paradigm based on potential treatment could be used to guide clinical practice based on cost and availability of screening tests. (C) 2013 Elsevier Inc. All rights reserved. Novartis Pharmaceuticals Univ Oxford, Oxford Ctr Diabet Endocrinol & Metab, Diabet Trials Unit, Oxford, England Duke Univ, Med Ctr, Dept Med, Div Metab Endocrinol & Nutr, Durham, NC 27710 USA Universidade Federal de São Paulo, São Paulo, Brazil Univ Calif San Francisco Program Fresno, Fresno, CA USA Vet Adm Cent Calif Hlth Care Syst, Fresno, CA USA Univ Sydney, Royal N Shore Hosp, Sydney, NSW 2006, Australia Univ Tromso, Oslo Univ Hosp, Rigshosp, Inst Clin Med, Oslo, Norway Vet Affairs Puget Sound Hlth Care Syst, Seattle, WA USA Univ Washington, Seattle, WA 98195 USA Univ Cape Town, Dept Med, Groote Schuur Hosp, Endocrine Unit, ZA-7925 Cape Town, South Africa Univ Glasgow, Glasgow Cardiovasc Res Ctr, British Heart Fdn, Glasgow, Lanark, Scotland Duke Univ, Med Ctr, Duke Translat Med Inst, Durham, NC USA Attikon Univ Hosp, Hellen Natl Diabet Ctr, Dept Internal Med Endocrinol & Diabetol 2, Athens, Greece Duke Univ, Med Ctr, Dept Biostat & Bioinformat, Durham, NC USA Duke Univ, Med Ctr, Duke Clin Res Inst, Durham, NC USA Universidade Federal de São Paulo, São Paulo, Brazil Web of Science