OBJECTIVE: To determine whether patients with prediabetes can be accurately and easily identified in clinical settings using a predictive clinical and laboratory model. METHODS: This retrospective study examined demographic and laboratory data from patients who had undergone 2-hour glucose testing for suspected prediabetes or diabetes between 2000 and 2004. Patients who met the diagnostic criteria for diabetes mellitus were excluded. Prediabetes was defined as a fasting glucose concentration > or = 100 mg/dL and < or = 125 mg/dL or a 2-hour postprandial glucose concentration > or = 140 mg/dL and < 200 mg/dL. Multivariate logistic regression was conducted to identify calculated or measured clinical and laboratory attributes that predict the presence of prediabetes, including fasting insulin quartiles, homeostasis model assessment of insulin resistance (HOMA-IR), and quantitative insulin sensitivity check index. RESULTS: Of 965 patients, 287 (29.7%) had prediabetes. The study population primarily consisted of white, obese, female patients. A multivariate model revealed that compared with the referent lowest quartile of fasting insulin (mu = 4.9 [+/-SD] +/-1.2 microIU/mL), subsequent insulin quartiles increased the likelihood of identifying prediabetes (quartile 2: mu = 8.0 +/-0.8 microIU/mL, odds ratio [OR] = 2.076, confidence interval [CI] = 1.241-3.273; quartile 3: mu = 12.2 +/-1.7 microIU/mL, OR = 3.151, CI = 1.981-5.015; quartile 4: mu = 25.9 +/-12.4 microIU/mL, OR = 5.035, CI = 3.122-8.122). Older age and increased diastolic blood pressure also contributed modestly to this model. Further analysis using the area under the curve revealed that at a fasting insulin level > 9.0 microIU/mL, prediabetes would be correctly identified in 80% of affected patients. A second model revealed that increased HOMA-IR index (OR = 1.303, CI = 1.205-1.410) and older age (OR = 1.037, CI = 1.024-1.05) predicted prediabetes. CONCLUSIONS: The most robust model, which used fasting insulin levels, may provide the most utility as a clinical tool because the highest quartiles suggest significantly greater likelihood of identifying prediabetes.
OBJECTIVE: To determine whether patients with prediabetes can be accurately and easily identified in clinical settings using a predictive clinical and laboratory model. METHODS: This retrospective study examined demographic and laboratory data from patients who had undergone 2-hour glucose testing for suspected prediabetes or diabetes between 2000 and 2004. Patients who met the diagnostic criteria for diabetes mellitus were excluded. Prediabetes was defined as a fasting glucose concentration > or = 100 mg/dL and < or = 125 mg/dL or a 2-hour postprandial glucose concentration > or = 140 mg/dL and < 200 mg/dL. Multivariate logistic regression was conducted to identify calculated or measured clinical and laboratory attributes that predict the presence of prediabetes, including fasting insulin quartiles, homeostasis model assessment of insulin resistance (HOMA-IR), and quantitative insulin sensitivity check index. RESULTS: Of 965 patients, 287 (29.7%) had prediabetes. The study population primarily consisted of white, obese, female patients. A multivariate model revealed that compared with the referent lowest quartile of fasting insulin (mu = 4.9 [+/-SD] +/-1.2 microIU/mL), subsequent insulin quartiles increased the likelihood of identifying prediabetes (quartile 2: mu = 8.0 +/-0.8 microIU/mL, odds ratio [OR] = 2.076, confidence interval [CI] = 1.241-3.273; quartile 3: mu = 12.2 +/-1.7 microIU/mL, OR = 3.151, CI = 1.981-5.015; quartile 4: mu = 25.9 +/-12.4 microIU/mL, OR = 5.035, CI = 3.122-8.122). Older age and increased diastolic blood pressure also contributed modestly to this model. Further analysis using the area under the curve revealed that at a fasting insulin level > 9.0 microIU/mL, prediabetes would be correctly identified in 80% of affected patients. A second model revealed that increased HOMA-IR index (OR = 1.303, CI = 1.205-1.410) and older age (OR = 1.037, CI = 1.024-1.05) predicted prediabetes. CONCLUSIONS: The most robust model, which used fasting insulin levels, may provide the most utility as a clinical tool because the highest quartiles suggest significantly greater likelihood of identifying prediabetes.
Authors: Kevin O Murray; Hunter L Paris; Alyce D Fly; Robert F Chapman; Timothy D Mickleborough Journal: J Sports Sci Med Date: 2018-03-01 Impact factor: 2.988
Authors: Kristen M Ward; Larisa Yeoman; Cora McHugh; A Zarina Kraal; Stephanie A Flowers; Amy E Rothberg; Alla Karnovsky; Arun K Das; Vicki L Ellingrod; Kathleen A Stringer Journal: Pharmacotherapy Date: 2018-06 Impact factor: 4.705
Authors: Fiona Yuen; Arthi Thirumalai; Frances A Fernando; Ronald S Swerdloff; Peter Y Liu; Youngju Pak; Laura Hull; Rachelle Bross; Diana L Blithe; Jill E Long; Stephanie T Page; Christina Wang Journal: Andrology Date: 2021-05-22 Impact factor: 4.456
Authors: N R den Braver; E de Vet; G Duijzer; J Ter Beek; S C Jansen; G J Hiddink; E J M Feskens; A Haveman-Nies Journal: Int J Behav Nutr Phys Act Date: 2017-06-12 Impact factor: 6.457
Authors: Dennis Schrijnders; Steven H Hendriks; Nanne Kleefstra; Pauline A J Vissers; Jeffrey A Johnson; Geertruida H de Bock; Henk J G Bilo; Gijs W D Landman Journal: PLoS One Date: 2018-01-25 Impact factor: 3.240