Eur J Pediatr DOI 10.1007/s00431-016-2807-6

ORIGINAL ARTICLE

Diabetes screening in overweight and obese children and adolescents: choosing the right test Stefan Ehehalt 1,4 & Susanna Wiegand 2 & Antje Körner 3 & Roland Schweizer 4 & Klaus-Peter Liesenkötter 5 & Carl-Joachim Partsch 6 & Gunnar Blumenstock 7 & Ulrike Spielau 3 & Christian Denzer 8 & Michael B. Ranke 4 & Andreas Neu 4 & Gerhard Binder 4 & Martin Wabitsch 8 & Wieland Kiess 3 & Thomas Reinehr 9

Received: 11 June 2016 / Revised: 2 November 2016 / Accepted: 4 November 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract Type 2 diabetes can occur without any symptoms, and health problems associated with the disease are serious. Screening tests allowing an early diagnosis are desirable. However, optimal screening tests for diabetes in obese youth are discussed controversially. We performed an observational multicenter analysis including 4848 (2668 female) overweight and obese children aged 7 to 17 years without previously

known diabetes. Using HbA1c and OGTT as diagnostic criteria, 2.4% (n = 115, 55 female) could be classified as having diabetes. Within this group, 68.7% had HbA1c levels ≥48 mmol/mol (≥6.5%). FPG ≥126 mg/dl (≥7.0 mmol/l) and/ or 2-h glucose levels ≥200 mg/dl (≥11.1 mmol/l) were found in 46.1%. Out of the 115 cases fulfilling the OGTT and/or HbA1c criteria for diabetes, diabetes was confirmed in

S.E., S.W., and A.K. contributed equally to this work. Communicated by Beat Steinmann * Susanna Wiegand [email protected]

Gerhard Binder [email protected]

Stefan Ehehalt [email protected]

Martin Wabitsch [email protected]

Antje Körner [email protected]

Wieland Kiess [email protected]

Roland Schweizer [email protected]

Thomas Reinehr [email protected]

Klaus-Peter Liesenkötter [email protected] Carl-Joachim Partsch [email protected] Gunnar Blumenstock [email protected]

1

2

3

Ulrike Spielau [email protected] Christian Denzer [email protected]

4

Michael B. Ranke [email protected]

5

Andreas Neu [email protected]

6

Public Health Department of Stuttgart, Department of Pediatrics, Dental Health Care, Health Promotion and Social Services, Schloßstraße 91, 70176 Stuttgart, Germany Department of Pediatric Endocrinology and Diabetes, Charité Children’s Hospital, Universitätsmedizin Berlin, Augustenburger Platz 1, D-13353 Berlin, Germany Hospital for Children and Adolescents, Department of Women and Child Health, University Hospitals, University of Leipzig, Liebigstraße 20a, D-04103 Leipzig, Germany Pediatric Endocrinology and Diabetes, University Children’s Hospital, University of Tübingen, Hoppe-Seyler-Str. 1, 72076 Tübingen, Germany Endokrinologikum Berlin am Gendarmenmarkt, Friedrichstraße 76, Jägerstraße 61, 10117 Berlin, Germany Endokrinologikum Hamburg, Lornsenstraße 4-6, 22767 Hamburg, Germany

Eur J Pediatr

43.5%. For FPG, the ROC analysis revealed an optimal threshold of 98 mg/dl (5.4 mmol/l) (sensitivity 70%, specificity 88%). For HbA1c, the best cut-off value was 42 mmol/mol (6.0%) (sensitivity 94%, specificity 93%). Conclusions: HbA1c seems to be more reliable than OGTT for diabetes screening in overweight and obese children and adolescents. The optimal HbA1c threshold for identifying patients with diabetes was found to be 42 mmol/mol (6.0%). What is Known: • The prevalence of obesity is increasing and health problems related to type 2 DM can be serious. However, an optimal screening test for diabetes in obese youth seems to be controversial in the literature. What is New: • In our study, the ROC analysis revealed for FPG an optimal threshold of 98 mg/dl (5.4 mmol/l, sensitivity 70%, specificity 88%) and for HbA1c a best cut-off value of 42 mmol/mol (6.0%, sensitivity 94%, specificity 93%) to detect diabetes. Thus, in overweight and obese children and adolescents, HbA1c seems to be a more reliable screening tool than OGTT.

Keywords OGTT . HbA1c . Children . Overweight . Diabetes . Screening

Abbreviations CI Confidence interval FPG Fasting plasma glucose IFG Impaired fasting glucose IGT Impaired glucose tolerance OGTT Oral glucose tolerance test SDS Standard deviation score

associated with the disease are serious, screening tests that allow an early diagnosis are desirable. However, since the prevalence of type 2 DM in childhood is still assumed to be rather low in many European countries, screening tests are not without controversy [10, 27, 29, 36]. Furthermore, screening with HbA1c or oral glucose tolerance test (oGTT) has limitations [1, 36]. With HbA1c, we find concerns involving such matters as a different diagnostic performance in some groups or a low sensitivity or specificity for prediabetes and type 2 DM in obese children and adolescents identified by OGTT [25, 28]. However, Kapadia and Zeitler state that these studies do not recognize that diabetes diagnostic criteria are based upon long-term health outcomes and that OGTT and FPG have themselves never been validated in the pediatric population [1, 14]. This view is supported by studies showing a poor reproducibility of the OGTT in obese youth or a high conversion rate to normal glucose tolerance in a group of untreated obese white children and adolescents with impaired glucose tolerance [16, 24]. Due to this controversy concerning screening for diabetes in children and adolescents and as the prevalence of obesity is increasing and health problems related to type 2 DM can be serious, we aimed 1. to investigate the test properties of fasting plasma glucose, 2-h glucose, and HbA1c levels for screening of type 2 DM in asymptomatic or oligosymptomatic overweight and obese children and adolescents living in Germany, and 2. to find appropriate cut-off values for the detection of manifest diabetes in children.

Introduction

Materials and methods

Diagnosing diabetes in children used to be a simple task. Diabetes-specific symptoms in combination with random blood glucose values ≥200 mg/dl (≥11.1 mmol/l) were sufficient criteria for diagnosing childhood diabetes, which as a rule was considered type 1 DM [9]. Today, the situation is more complex: With the increase in childhood obesity, we find more and more children and adolescents facing the risk of developing type 2 DM. As type 2 DM can occur without any symptoms, and as health problems

Subjects

7

8

9

Department of Clinical Epidemiology and Applied Biometry, University of Tübingen, Silcherstraße 5, 72076 Tübingen, Germany Division of Pediatric Endocrinology and Diabetes, Interdisciplinary Obesity Unit, Department of Pediatrics and Adolescent Medicine, Ulm University, Eythstr. 24, D-89073 Ulm, Germany Department of Pediatric Endocrinology, Diabetes and Nutrition Medicine, Vestische Children’s Hospital, University Witten/ Herdecke, Dr. F. Steiner Str. 5, D-45711 Datteln, Germany

A total of 4848 children and adolescents were included in the study [Vestische Hospital for Children and Adolescents Datteln, University of Witten/Herdecke, Germany (n = 2934); University Hospital for Children and Adolescents, University of Leipzig, Germany (n = 889); Endokrinologikum Berlin, Germany (n = 415); University Hospital for Children and Adolescents, Charité University Medicine, Berlin, Germany (n = 212); University Hospital for Children and Adolescents, University of Tübingen, Germany (n = 208); Endokrinologikum, Hamburg, Germany (n = 190)]. There were no statistically significant differences in the proportions of overweight, obese, and extremely obese children and adolescents between the participating centers. All these children presented for their overweight at these institutions.

Eur J Pediatr

The inclusion criteria were as follows: (1) overweight, obese, and extremely obese children and adolescents aged 7 to 17 years; and (2) oral glucose tolerance test and HbA1c measurement on the same day. Overweight was defined by a BMI >90th percentile for population-specific percentiles, obesity by a BMI >97th percentile for population-specific percentiles, and extreme obesity by a BMI >99.5th percentile for population-specific percentiles according to German guidelines [2, 19]. Generally, the BMI percentiles of KromeyerHauschild et al. (Germany, 2001), of Cole et al. (international BMI cut-offs, 2000), and of Kuczmarski et al. (CDC growth charts for the U.S.A., 2002) are in relatively good accordance. Within the reference population of Kromeyer-Hauschild et al., the cut-off levels of Cole et al. identify 10.8% boys and 11.2% girls with overweight and 2.7% boys and 2.6% girls with obesity [7, 19–21]. The exclusion criteria were as follows: (1) previously known diabetes; (2) tumor or severe systemic disease; (3) blood transfusion and significant blood loss within the last year; (4) hematologic diseases, hemoglobinopathies, renal insufficiency, chronic lead poisoning, and galactosemia; (5) syndromes associated with obesity; and (6) drugs affecting glucose and HbA1c levels, respectively. The study has been approved by the appropriate research ethics committee and has been performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments. Testing procedures and biochemical analyses An oral glucose tolerance test was performed according to WHO criteria following a standardized diagnostic procedure based on the guidelines of the German Pediatric Obesity Association [2]. Prior to the test, a weight-maintaining diet rich in carbohydrates is recommended for at least 3 days. Testing is performed after an overnight fasting period. During the test, the patient should rest. At time point zero, a glucose solution is given to drink (1.75 g glucose/kg body weight, maximum 75 g). Blood samples are taken for fasting glucose and glucose at 120 min (depending on the clinical or scientific question together with insulin at 0, 30, 60, 90, 120 min) [2, 9]. HbA1c levels were measured on the same day of the OGTT by using immunoassay and highperformance liquid chromatography (HPLC) methods, respectively, which are certified and standardized to the DCCT assay. Each type of glucose tolerance status (NGT, at risk for diabetes/prediabetes, diabetes) was defined according to the criteria of the American Diabetes Association. Diagnoses of type 2 DM without symptoms were confirmed on another day by oGTT or HbA1c, or both [1]. HbA1c levels from 39 mmol/mol (5.7%) to 46 mmol/mol (6.4%) are assumed to be associated with an increased risk for

diabetes (prediabetes) [1]. Therefore, we used HbA1c ≥39 mmol/mol (≥5.7%) as cut-off level, too. All children fulfilling the criteria of diabetes were re-tested on another day. Type 1 DM was assumed in the presence of bcell autoantibodies, low C-peptide levels, and initiation of insulin treatment. Genetic screening of MODY type 2 and type 3 was performed if parents and grandparents suffered from diabetes. Statistical analysis Statistical analysis was performed using the JMP® 11.2 software (SAS Institute, Cary, NC). Categorical data are presented as numbers and percentages, and continuous data are reported with the mean and standard deviation. The chi-squared test for independence was used to compare proportions across groups. The level of statistical significance was set at P <0.05. The characteristics of different testing strategies for diabetes screening are reported with the diagnostic sensitivity and specificity and with the positive and negative predictive values of the test. Estimates of screening test performance are given with the associated 95% Clopper-Pearson binomial confidence intervals. Receiver operating characteristics (ROC) analysis was done for FPG and for HbA1c, and the accuracy of both tests is illustrated with an ROC curve. The area under the curve with 95% confidence limits was calculated to compare the diagnostic accuracy of the tests. The trade-off between sensitivity and specificity was examined, and the best cut-off point maximizing the sum of both measures was determined.

Results The study sample encompassed 4848 children and adolescents (2668 girls) with a mean BMI of 30.6 ± 5.4 kg/m2 (BMI-SDS 2.8 ± 0.6). Mean age was 13.1 ± 2.4 years. Within the study group, 15.7% (n = 759) were overweight, 46.4% (n = 2251) were obese, and 37.9% (n = 1838) were extremely obese. Baseline distribution of glucose tolerance status according to OGTT and HbA1c categories (n = 4848) The proportion of each type of glucose tolerance status (NGT, IFG and/or IGT (prediabetes), and diabetes derived from the OGTT) within categories of HbA1c (NGT, at risk for diabetes (prediabetes), diabetes) is summarized in Table 1 [28]. OGTT identified 21.5% of the patients as having diabetes by using the HbA1c criteria. HbA1c identified 32% of the patients as having diabetes by using the OGTT criteria. The comparison of the classification of glucose tolerance status between OGTT and HbA1c showed significant differences (P < 0.001; see Table 2).

Eur J Pediatr Table 1

OGTT versus HbA1c (at baseline) Diabetes HbA1c ≥ 48 mmol/mol (≥6.5%)

Total

844

39

3892

603

277

23

903

14

22

17

53

3626

1143

79

4848

oGTT

NGT HbA1c < 39 mmol/mol (<5.7%)

NGT (FPG < 100 mg/dl (<5.6 mmol/l), 2-h glucose < 140 mg/dl (<7.8 mmol/l) IFG and/or IGT (FPG 100–125 mg/dl (5.6–6.9 mmol/l) and/or 2-h glucose 140–199 mg/dl (7.8–11.0 mmol/l)) Diabetes (FPG ≥ 126 mg/dl (≥7.0 mmol/l) and/or 2-h glucose ≥200 mg/dl (≥11.1 mmol/l)) Total

3009

At risk for diabetes HbA1c 39–46 mmol/mol (5.7–6.4%)

NGT normal glucose tolerance, IFG impaired fasting glucose, IGT impaired glucose tolerance

Baseline distribution of patients classified with diabetes according to OGTT and HbA1c categories (n = 115) Using both HbA1c (≥48 mmol/mol, ≥6.5%) and OGTT (FPG ≥ 126 mg/dl, ≥7.0 mmol/l and/or 2-h glucose ≥200 mg/dl, ≥11.1 mmol/l) as diagnostic criteria, 2.4% of our patients (n = 115, 55 female, mean age 14.0 ± 2.3, age range 8.3–17.9 years) could be classified as having diabetes. Within this group of 115 patients, 22.6% (n = 26) had FPG levels ≥126 mg/dl (≥7.0 mmol/l) and 68.7% (n = 79) had HbA1c ≥48 mmol/mol (≥6.5%). FPG ≥126 mg/dl (≥7.0 mmol/l) and 2-h glucose levels ≥200 mg/dl (≥11.1 mmol/l) were found in 46.1% (n = 53) of the patients, while the combination of FPG ≥126 mg/dl (≥7.0 mmol/l) and HbA1c ≥48 mmol/mol (≥6.5%) was found in 81.7% (n = 94) of the patients. The baseline distribution of glucose tolerance status according to OGTT and HbA1c categories within the entire study group (n = 4848) and in patients who were overweight (n = 759), obese (n = 2251), and who were suffering from extreme obesity (n = 1838) is displayed in Fig. 1. The proportion of abnormal values in the entire study group as well as in the weight categories was highest for HbA1c, which almost doubled the numbers detected using 2 h values, particularly in less severe obesity. Based on these observations, HbA1c measurement seems to be a more promising screening method than FPG/OGTT. In our patient group, however, the sensitivity for HbA1c ≥48 mmol/mol (≥6.5%) was also rather low (68.7%). Out of the 115 patients, 101 patients had HbA1c values ≥39 mmol/mol (≥5.7%) corresponding to a sensitivity of 87.8% (95% CI 80.4–93.2, n = 101). Specificity was found to be 76.3% (95% CI 75.1–77.5, n = 3612). Further analysis revealed a positive predictive value of 8.3% (6.8–10.0%, Table 2 OGTT versus HbA1c (at baseline) according to glucose tolerance status (n = 4848, P < 0.001)

NGT Prediabetes Diabetes

OGTT

HbA1c

80.3% 18.6% 1.1%

74.8% 23.6% 1.6%

n = 101) and a negative predictive value of 99.6% (95% CI 99.4–99.8%, n = 3612). False-positive test results were found in 1121 patients, and false-negative test results were found in 14 patients. In the subgroup of IFG and IGT patients, an HbA1c cut-off level ≥39 mmol/mol (≥5.7%) detected 39.0% (95% CI 34.4– 43.7, n = 170) of IFG levels and 33.1% (95% CI 29.3–37.0, n = 198) of all IGT cases. Lowering the limit of HbA1c from 39 mmol/mol (5.7%) to 31 mmol/mol (5.0%), 95% of all FPG ≥100 mg/dl (≥5.6 mmol/l) and of all 2-h glucose levels ≥140 mg/dl (≥7.8 mmol/l) would have been identified. Predictive value for confirmed diabetes at follow-up Out of the 115 patients fulfilling the OGTT and/or HbA1c criteria for diabetes, the diagnosis of diabetes was confirmed in only 43.5% of these patients (n = 50). A total of 38 patients suffered from type 2 DM (mean age 14.6 ± 2.0 years, range 11.3–17.9 years). Eight patients were diagnosed with type 1 DM (mean age 11.7 ± 2.2 years, range 9.9–14.9 years), two patients with MODY II, and two patients with MODY III. Unfortunately, seven cases were lost to follow-up. In another six cases, diagnostic follow-up was incomplete. The test properties for different cut-off values are given in Table 3. FPG alone or in combination with 2-h glucose revealed poor sensitivity in diagnosing diabetes, but at the same time good specificity and negative predictive values. Hence, if the FPG was abnormal, there was obviously manifest hyperglycemia. The HbA1c threshold of 48 mmol/mol (6.5%) revealed markedly better, though not ideal, sensitivity with a smaller number of false-negative results. Lowering the threshold of HbA1c to 39 mmol/mol (5.7%) resulted in an appreciably better sensitivity at the expense of very poor positive predictive values. ROC analyses Based on the 50 patients with confirmed diabetes, the ROC analysis revealed for FPG an optimal threshold of 98 mg/dl (5.4 mmol/l) and for HbA1c a best cut-off value of 42 mmol/

Eur J Pediatr Fig. 1 Baseline distribution of glucose tolerance status according to OGTT and HbA1c categories within the entire study group (n = 4848) and in patients who are overweight (n = 759), obese (n = 2251), and who suffer from extreme obesity (n = 1838)

mol (6.0%; see Fig. 2). Further details on diagnostic performance are given in Table 4.

Discussion The aim of the present study was to investigate the diagnostic properties of fasting plasma glucose, 2-h plasma glucose, and HbA1c levels for screening of type 2 DM in asymptomatic or oligosymptomatic overweight and obese children and adolescents living in Germany, and to find appropriate cut-off values for the detection of manifest diabetes in children. This is important because an optimal screening test for diabetes in obese youth seems to be controversial in the literature [1, 6, 15, 28]. Reasons for type 2 DM screening in childhood and adolescence are, among others, the increasing prevalence of obesity and the fact that obesity- and diabetes-related metabolic abnormalities are common in children and adolescents even if type 2 DM is asymptomatic [26, 34]. This is the first study Table 3

investigating this issue in a large number of asymptomatic or oligosymptomatic overweight and obese children in Germany. Country-specific investigations are necessary because HbA1c and glucose levels may vary according to the race and ethnic background of patients and because there are country-specific definitions of overweight and obesity [11, 22, 34, 37]. In our study population of overweight and obese children and adolescents, we found inconsistencies between OGTT and HbA1c in classifying patients with normal glucose tolerance, with prediabetes or diabetes. Thus, OGTT and HbA1c did not necessarily detect diabetes in the same individuals. These findings correspond with those of other studies [6, 28]. According to the American Diabetes Association (ADA) thresholds of hyperglycemia, the sensitivity for detecting diabetes was better for HbA1c than that for OGTT. Figure 1 indicates that this could be due to a lower sensitivity of OGTT in obese children. Further weight-specific analyses have not been performed due to the limited number of overweight children in our study sample.

Screening test properties (at confirmed follow-up diagnosis) FPG ≥ 126 mg/ FPG ≥ 126 mg/dl (≥7.0 mmol/l) and/ HbA1c ≥ 48 mmol/ HbA1c ≥ 39 mmol/ HbA1c ≥ 39 mmol/mol (≥5.7%) dl (≥7.0 mmol/ or 2-h glucose ≥ 200 mg/dl mol (≥6.5%) mol (≥5.7%) and/or FPG ≥ 100 mg/dl l) (≥11.1 mmol/l) (≥5.6 mmol/l)

Sensitivity (95% CI) Specificity (95% CI) Positive predictive value (95% CI) Negative predictive value (95% CI) False positive False negative

18.0% (8.6–31.4) 99.8% (99.7–99.9) 52.9% (27.8–77.0)

44.0% (30.0–58.7)

84.0% (70.9–92.8)

96.0% (86.3–99.5)

98.0% (89.4–99.9)

99.6% (99.3–99.7)

99.3% (99.0–99.5)

75.6% (74.3–76.8)

70.0% (68.7–71.3)

51.2% (35.5–66.7)

56.0% (44.1–67.5)

4.0% (2.9–5.2)

3.3% (2.5–4.3)

99.1% (98.8–99.4)

99.4% (99.2–99.6)

99.8% (99.7–99.9)

99.9% (99.8–100)

99.97% (99.8–100)

n=8 n = 41

n = 21 n = 28

n = 33 n=8

n = 1168 n=2

n = 1435 n=1

Eur J Pediatr

Fig. 2 ROC Curves (at follow-up) for fasting plasma glucose (a) and HbA1c (b) for type 2 DM

In the study group, only one third of all patients diagnosed as having diabetes according to OGTT criteria would have been identified by means of HbA1c. Based on similar observations, some authors conclude that HbA1c is a poor diagnostic tool for prediabetes and at least asymptomatic diabetes in obese children and adolescents [28]. As prediabetes is a common metabolic complication of childhood obesity, we also aimed at determining the threshold for HbA1c at which 95% of all patients with IFG and IGT could have been detected [18, 30, 35]. The analysis revealed an HbA1c cut-off level of 31 mmol/mol (5.0%) leading at the same time to an unacceptably high number of false-positive results. Therefore, we do not recommend using this cut-off point. This is in line with the fact that the prognostic and therapeutic value of identifying IFG and IGT in youth is controversial [16, 34]. To understand which test might better predict the later development of diabetes, we examined how many patients developed diabetes during follow-up. The diagnosis of diabetes was confirmed in 50 patients. This means about one in a hundred of our whole study population and in 43.5% out of the 115 patients fulfilling the OGTT and/or HbA1c criteria for

Table 4 ROC analysis (at confirmed follow-up diagnosis) Optimal threshold AUC (95% CI) Sensitivity (95% CI) Specificity (95% CI) Positive predictive value (95% CI) Negative predictive value (95% CI) False positive False negative

diabetes. This is less frequent than reported in a Korean study conducted by Lee and more frequent than in a Swedish study conducted by Berhan [4, 23]. The overweight and obese children with diabetes developed mainly type 2 DM but also type 1 DM and MODY. With respect to diabetes screening, type 2 DM overall seems to be a rare condition, but is increasing in prevalence and with detrimental long-term consequences regarding morbidity and mortality. For daily clinical practice, it is important to notice that our youngest patient with type 2 DM was 11.3 years, which is consistent with the ADA recommendation of starting testing for type 2 DM in asymptomatic children at the age of 10 or at onset of puberty if puberty occurs at a younger age [1]. As OGTT and HbA1c do not necessarily detect diabetes in the same individuals, and as pathological test results can normalize with time, we decided to derive screening test properties and thresholds based on cases with a confirmed diagnosis of diabetes (see Tables 3 and 4). Again, HbA1c ≥48 mmol/ mol (≥6.5%) identified more children and adolescents with diabetes than FPG ≥126 mg/dl (≥7.0 mmol/l) alone or in combination with 2-h glucose ≥200 mg/dl (≥11.1 mmol/l). This stands in contrast to a French adult study [8]. A systematic

FPG

HbA1c

≥98 mg/dl (≥5.4 mmol/l) 83.4% (76.2–90.6) 70.0% (55.4–82.1) 88.0% (87.1–89.0) 5.8% (4.0–7.9) 99.6% (99.4–99.8) n = 572 n = 15

≥42 mmol/mol (≥6.0%) 97.0% (93.4–100) 94.0% (83.5–98.7) 92.9% (92.2–93.6) 12.2% (9.1–15.9) 99.9% (99.8–100) n = 338 n=3

Eur J Pediatr

review of primary cross-sectional studies (n = 9, mainly adult studies) using OGTT as the reference standard and FPG as a comparison reported three different cut-off points for HbA1c [41 mmol/mol (5.9%), 43 mmol/mol (6.1%), and 44 mmol/ mol (6.2%)] which were advised as optimum cut-offs for detecting diabetes in at least two different studies [3]. At a cut-off point of 41 mmol/mol (5.9%), the sensitivity ranged from 76 to 95% and specificity from 67 to 86% [31, 32]. At a cut-off point of 43 mmol/mol (6.1%), the sensitivity ranged from 78 to 81% and specificity from 79 to 84% [17, 32]. Finally, at a cut-off point of 44 mmol/mol (6.2%), the sensitivity ranged from 43 to 81% and specificity 88 to 99% [3, 12, 13, 32]. The authors of the review stated that the HbA1c cut-off point of >43 mmol/mol (>6.1%) was the recommended optimum cutoff point for HbA1c in most reviewed studies [3]. The proposed cut-off value is close to our result (HbA1c ≥ 42 mmol/ mol, ≥6.0%). However, there are also studies reporting distinctly lower or higher HbA1c cut-off levels and different sensitivities and specificities compared to our study [5, 23, 28, 33]. These differences can mainly be explained by varying methodological approaches (e.g., children and adolescents vs. adults, the use of OGTT as reference method vs. the use of both OGTT and HbA1c, cross-sectional analysis vs. follow-up investigations, implementation of HbA1c standards at different times in different countries). Another explanation could be that the identified cut-off point also depends on the diabetes risk within the investigated study population [28]. In our study group, the ROC analysis revealed for FPG an optimal threshold of 98 mg/dl (5.4 mmol/l) and for HbA1c a best cut-off value of 42 mmol/mol (6.0%) to detect diabetes. An HbA1c cut-off value of 42 mmol/mol (6.0%) seems to be a suitable screening tool in overweight and obese children and adolescents. However, even if the cut-off value is lowered from 48 mmol/mol (6.5%) to 42 mmol/mol (6.0%), there are still false-negative cases together with an increase in false positives. For HbA1c values between 42 and 48 mmol/l (between 6.0 and 6.5%), further testing is mandatory to make the diagnosis of diabetes. For HbA1c ≥48 mmol/mol (≥6.5%), diagnosis should also be confirmed with a repeat HbA1c test unless clinical symptoms and glucose levels >200 mg/dl (>11.1 mmol/l) are present [1]. The strength of our multicenter study lies in the large number of patients. However, there are also some limitations: There were no patient data available concerning nationality, ethnicity, pubertal status, diabetes-specific antibodies, birth weight, and family history for diabetes. Furthermore, a selection bias cannot be excluded due to the following: the number of cases reported by the individual centers was different, the proportion of overweight patients was relatively low, and retesting was performed only in patients with pathological test results at first presentation. Therefore, further prospective population-based studies in different ethnicities are necessary to confirm our findings.

In conclusion, we found that in overweight and obese children and adolescents, HbA1c seems to be a more reliable screening tool than OGTT. In our patient group, the best cutoff value for HbA1c was found to be 42 mmol/mol (6.0%). However, even if the cut-off value is lowered from 48 mmol/ mol (6.5%) to 42 mmol/mol (6.0%), it should be noted that there are still false-negative cases together with an increase in false positives. For HbA1c values between 42 and 48 mmol/l (between 6.0 and 6.5%), further testing is mandatory to make the diagnosis of diabetes. For HbA1c ≥48 mmol/mol (≥6.5%), diagnosis should also be confirmed with a repeat HbA1c test unless clinical symptoms and glucose levels >200 mg/dl (>11.1 mmol/l) are present. We suggest making the decision for or against type 2 DM screening based both on representative country and target group data and on the general guidelines in justifying a screening test applied to type 2 DM. Further prospective population-based studies in different ethnicities should be performed to confirm our findings. Authors’ contributions S.E. and T.R. researched data, wrote the manuscript, and contributed to the discussion. A.K. and S.W. researched data, reviewed the manuscript, and contributed to the discussion. K.P.L., C.J.P., and R.S. researched data and contributed to the discussion. The data analysis was carried out by G. Bl. G. Bi., G. Bl., C.D., W.K., A.N., M.B.R., and M.W. reviewed the manuscript and contributed to the discussion. Compliance with ethical standards All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.^ Conflict of interest The authors declare that they have no competing interests. Funding W.K., T.R., M.W., and S.W. were supported by grants of the German Federal Ministry of Education and Research (BMBF) (project funding reference number 01GI1120A). This part of the study is integrated in the German Competence Network on Obesity (CNO). S.W. was supported by German Research Foundation (DFG), project BHormonal regulation of body weight maintenance^ (KFO 218). G. Bi., S.E., A.N., and M.B.R. were supported by a research grant from the Tübingen University Hospital (AKF-Program, nr. 221-2-0). A.K. was supported by German Research Foundation (DFG) for the Clinical Research Center BObesity Mechanisms^ CRC1052/1 C05 and the Integrated Research and Treatment Centre (IFB) Adiposity Diseases FKZ: 01EO1001.

References 1.

American Diabetes Association (2015) (2) Classification and diagnosis of diabetes. Diabetes Care 38(Suppl. 1):S8–S16

Eur J Pediatr 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Arbeitsgemeinschaft Adipositas im Kindes- und Jugendalter (2014) Konsensbasierte (S2) Leitlinie zur Diagnostik, Therapie und Prävention von Übergewicht und Adipositas im Kindes- und Jugendalter. Available at http://www.aga.adipositas-gesellschaft. de/index.php?id=9. Accessed 05 Nov 2015 Bennett CM, Guo M, Dharmage SC (2007) HbA(1c) as a screening tool for detection of Type 2 diabetes: a systematic review. Diabet Med 24:333–343 Berhan YT, Mollsten A, Carlsson A, Hogberg L, Ivarsson A, Dahlquist G (2014) Five-region study finds no evidence of undiagnosed type 2 diabetes in Swedish 11- to 13-year-olds. Acta Paediatr 103:1078–1082 Brar PC, Mengwall L, Franklin BH, Fierman AH (2014) Screening obese children and adolescents for prediabetes and/or type 2 diabetes in pediatric practices: a validation study. Clin Pediatr (Phila) 53:771–776 Buse JB, Kaufman FR, Linder B, Hirst K, El Ghormli L, Willi S, HEALTHY Study Group (2013) Diabetes screening with hemoglobin A(1c) versus fasting plasma glucose in a multiethnic middleschool cohort. Diabetes Care 36:429–435 Cole TJ, Bellizzi MC, Flegal KM, Dietz WH (2000) Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 320:1240–1243 Droumaguet C, Balkau B, Simon D, Caces E, Tichet J, Charles MA, Eschwege E, DESIR Study Group (2006) Use of HbA1c in predicting progression to diabetes in French men and women: data from an Epidemiological Study on the Insulin Resistance Syndrome (DESIR). Diabetes Care 29:1619–1625 Ehehalt S, Neu A (2011) Insulin resistance in childhood and adolescence. In: Ranke MB, Mullis PE (eds) Diagnostics of endocrine function in children and adolescents, 4th edn. Karger, Basel, pp. 294–309 Ehtisham S, Hattersley AT, Dunger DB, Barrett TG, British Society for Paediatric Endocrinology and Diabetes Clinical Trials Group (2004) First UK survey of paediatric type 2 diabetes and MODY. Arch Dis Child 89:526–529 Hagman E, Reinehr T, Kowalski J, Ekbom A, Marcus C, Holl RW (2014) Impaired fasting glucose prevalence in two nationwide cohorts of obese children and adolescents. Int J Obes 38:40–45 Herdzik E, Safranow K, Ciechanowski K (2002) Diagnostic value of fasting capillary glucose, fructosamine and glycosylated haemoglobin in detecting diabetes and other glucose tolerance abnormalities compared to oral glucose tolerance test. Acta Diabetol 39:15–22 Jesudason DR, Dunstan K, Leong D, Wittert GA (2003) Macrovascular risk and diagnostic criteria for type 2 diabetes: implications for the use of FPG and HbA(1c) for cost-effective screening. Diabetes Care 26:485–490 Kapadia C, Zeitler P, Drugs and Therapeutics Committee of the Pediatric Endocrine Society (2012) Hemoglobin A1c measurement for the diagnosis of Type 2 diabetes in children. Int J Pediatr Endocrinol 2012:31 Kester LM, Hey H, Hannon TS (2012) Using hemoglobin A1c for prediabetes and diabetes diagnosis in adolescents: can adult recommendations be upheld for pediatric use? J Adolesc Health 50:321–323 Kleber M, Lass N, Papcke S, Wabitsch M, Reinehr T (2010) Oneyear follow-up of untreated obese white children and adolescents with impaired glucose tolerance: high conversion rate to normal glucose tolerance. Diabet Med 27:516–521 Ko GT, Chan JC, Yeung VT, Chow CC, Tsang LW, Li JK, So WY, Wai HP, Cockram CS (1998) Combined use of a fasting plasma glucose concentration and HbA1c or fructosamine predicts the likelihood of having diabetes in high-risk subjects. Diabetes Care 21: 1221–1225 Körner A, Wiegand S, Hungele A, Tuschy S, Otto KP, l'AllemandJander D, Widhalm K, Kiess W, Holl RW, APV initiative, German Competence Net Obesity (2013) Longitudinal multicenter analysis

on the course of glucose metabolism in obese children. Int J Obes 37:931–936 19. Kromeyer-Hauschild K, Wabitsch M, Kunze D, Geller F, Geiß HC, Hesse V, von Hippel A, Jaeger U, Johnsen D, Korte W, Menner K, Müller G, Müller JM, Niemann-Pilatus A, Remer T, Schaefer F, Wittchen HU, Zabransky S, Zellner K, Ziegler A, Hebebrand J (2001) Perzentile für den Body-Mass-Index für das Kindes- und Jugendalter unter Heranziehung verschiedener deutscher Stichproben. Monatsschr Kinderheilk 149:807–818 20. Kromeyer-Hauschild K (2005) Definition, Anthropometrie und deutsche Referenzwerte für BMI. In: Wabitsch M, Hebebrand J, Kiess W, Zwiauer K (eds) Adipositas bei Kindern und Jugendlichen. Springer, Berlin, Heidelberg, pp. 9–10 21. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, Mei Z, Wei R, Curtin LR, Roche AF, Johnson CL (2002) CDC Growth Charts for the United States: methods and development. Vital Health Stat 11 246:1–190 22. Kumar PR, Bhansali A, Ravikiran M, Bhansali S, Dutta P, Thakur JS, Sachdeva N, Bhadada SK, Walia R (2010) Utility of glycated hemoglobin in diagnosing type 2 diabetes mellitus: a communitybased study. J Clin Endocrinol Metab 95:2832–2835 23. Lee HS, Park HK, Hwang JS (2012) HbA1c and glucose intolerance in obese children and adolescents. Diabet Med 29(7):e102–e105 24. Libman IM, Barinas-Mitchell E, Bartucci A, Robertson R, Arslanian S (2008) Reproducibility of the oral glucose tolerance test in overweight children. J Clin Endocrinol Metab 93:4231–4237 25. Lippi G, Mattiuzzi C, Targher G (2010) Glycated hemoglobin, diabetes, and cardiovascular risk in nondiabetic adults. N Engl J Med 362:2030 author reply 2031 26. Morrison KM, Xu L, Tarnopolsky M, Yusuf Z, Atkinson SA, Yusuf S (2012) Screening for dysglycemia in overweight youth presenting for weight management. Diabetes Care 35:711–716 27. Neu A, Feldhahn L, Ehehalt S, Hub R, Ranke MB, DIARY group Baden-Württemberg (2009) Type 2 diabetes mellitus in children and adolescents is still a rare disease in Germany: a population-based assessment of the prevalence of type 2 diabetes and MODY in patients aged 0–20 years. Pediatr Diabetes 10:468–473 28. Nowicka P, Santoro N, Liu H, Lartaud D, Shaw MM, Goldberg R, Guandalini C, Savoye M, Rose P, Caprio S (2011) Utility of hemoglobin A(1c) for diagnosing prediabetes and diabetes in obese children and adolescents. Diabetes Care 34:1306–1311 29. Rami B, Schober E, Nachbauer E, Waldhor T, Austrian Diabetes Incidence Study Group (2003) Type 2 diabetes mellitus is rare but not absent in children under 15 years of age in Austria. Eur J Pediatr 162:850–852 30. Sinha R, Fisch G, Teague B, Tamborlane WV, Banyas B, Allen K, Savoye M, Rieger V, Taksali S, Barbetta G, Sherwin RS, Caprio S (2002) Prevalence of impaired glucose tolerance among children and adolescents with marked obesity. N Engl J Med 346:802–810 31. Tanaka Y, Atsumi Y, Matsuoka K, Mokubo A, Asahina T, Hosokawa K, Shimada S, Matsunaga H, Takagi M, Ogawa O, Onuma T, Kawamori R (2001) Usefulness of stable HbA(1c) for supportive marker to diagnose diabetes mellitus in Japanese subjects. Diabetes Res Clin Pract 53:41–45 32. Tavintharan S, Chew LS, Heng DM (2000) A rational alternative for the diagnosis of diabetes mellitus in high risk individuals. Ann Acad Med Singap 29:213–218 33. Tsay J, Pomeranz C, Hassoun A, Zandieh SO, Rutledge J, Vogiatzi MG, Oberfield SE, Motaghedi R (2010) Screening markers of impaired glucose tolerance in the obese pediatric population. Horm Res Paediatr 73:102–107 34. Weiss R, Taksali SE, Tamborlane WV, Burgert TS, Savoye M, Caprio S (2005) Predictors of changes in glucose tolerance status in obese youth. Diabetes Care 28:902–909

Eur J Pediatr 35.

36.

Wiegand S, Maikowski U, Blankenstein O, Biebermann H, Tarnow P, Grüters A (2004) Type 2 diabetes and impaired glucose tolerance in European children and adolescents with obesity—a problem that is no longer restricted to minority groups. Eur J Endocrinol 151: 199–206 Zeitler P, J F, Tandon N, Nadeau K, Urakami T, Barrett T, Maahs D, International Society for Pediatric and Adolescent Diabetes (2014)

ISPAD Clinical Practice Consensus Guidelines 2014. Type 2 diabetes in the child and adolescent. Pediatr Diabetes 15(Suppl. 20):26–46 37. Ziemer DC, Kolm P, Weintraub WS, Vaccarino V, Rhee MK, Twombly JG, Narayan KM, Koch DD, Phillips LS (2010) Glucose-independent, black-white differences in hemoglobin A1c levels: a cross-sectional analysis of 2 studies. Ann Intern Med 152: 770–777