PEDIATRICS Vol. 118 No. 5 November 2006, pp. 1888-1895 (doi:10.1542/peds.2006-0121)
ARTICLE |
Diabetes Mellitus Screening in Pediatric Primary Care
a Division of General Pediatrics
b Department of Emergency Medicine, Boston University School of Medicine, Boston, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
OBJECTIVE. The goal was to determine the rates of diabetes screening and the prevalence of screening abnormalities in overweight and nonoverweight individuals in an urban primary care clinic.
METHODS. This study was a retrospective chart review conducted in a hospital-based urban primary care setting. Deidentified data for patients who were 10 to 19 years of age and had 
1 BMI measurement between September 1, 2002, and September 1, 2004, were extracted from the hospital electronic health record.
RESULTS. A total of 7710 patients met the study criteria. Patients were 73.0% black or Hispanic and 47.0% female; 42.0% of children exceeded normal weight, with 18.2% at risk for overweight and 23.8% overweight. On the basis of BMI, family history, and race, 8.7% of patients met American Diabetes Association criteria for type 2 diabetes mellitus screening, and 2452 screening tests were performed for 1642 patients. Female gender, older age group, and family history of diabetes were associated with screening. Increasing BMI percentile was associated with screening, exhibiting a dose-response relationship. Screening rates were significantly higher (45.4% vs 19.0%) for patients who met the American Diabetes Association criteria; however, less than one half of adolescents who should have been screened were screened. Abnormal glucose metabolism was seen for 9.2% of patients screened.
CONCLUSIONS. This study shows that, although pediatricians are screening for diabetes mellitus, screening is not being conducted according to the American Diabetes Association consensus statement. Point-of-care delivery of consensus recommendations could increase provider awareness of current recommendations, possibly improving rates of systematic screening and subsequent identification of children with laboratory evidence of abnormal glucose metabolism.
Key Words: type 2 diabetes mellitus • screening • childhood obesity
Abbreviations: T2DM—type 2 diabetes mellitus • ADA—American Diabetes Association • FPG—fasting plasma glucose • OGTT—oral glucose tolerance test • RPG—random plasma glucose • PPCC—pediatric primary care center • EHR—electronic health record • HbA1c—hemoglobin A1c • CI—confidence interval
Although previously considered an adult disease, type 2 diabetes mellitus (T2DM) is increasing in US children1–3 and children around the world.4,5 T2DM now accounts for up to 45% of all newly diagnosed cases of diabetes in US children.6 Black and Latino children show the greatest rates of increase.7 The increasing prevalences of overweight8 and T2DM9 in minority youths make the need for comprehensive T2DM screening an essential part of urban pediatric primary care.
The early detection of pediatric T2DM has become a national priority.6,10 In 2000, the American Diabetes Association (ADA) issued a consensus statement on T2DM screening in children.6 In that statement, the ADA recommends that children who are at risk for overweight or overweight (BMI 85th percentile for age and gender) and have 2 of 3 risk factors should be screened (Table 1). Biannual testing beginning at 10 years of age, or at the onset of puberty if that occurs sooner, is recommended. Discretionary testing is recommended for other high-risk patients, as assessed by their medical providers. Fasting plasma glucose (FPG) is recommended as the primary screening test for T2DM because of decreased cost and greater convenience, compared with the 2-hour oral glucose tolerance test (OGTT), which is recommended as a secondary test. No recommendation has been made regarding the use of random (nonfasting) plasma glucose (RPG) or hemoglobin A1c (HbA1c) measurements for T2DM screening in children.6
|
The purpose of our study was to describe how pediatricians screen for diabetes in an urban pediatric primary care center (PPCC) and to describe the results of screening for those patients. For this report, all possible screening tests related to diabetes, including FPG, OGTT, RPG, and HbA1c, were assessed. Specific outcome measures included the type of screen, factors associated with screening, and results of screening.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study Design and Population
The study was a retrospective chart review of electronic health record (EHR) data from all encounters that occurred during the period between September 1, 2002, and September 1, 2004, for children 10 through 19 years of age in the PPCC at Boston Medical Center. The PPCC serves a population of predominantly low-income black, Caribbean American, and Hispanic children and adolescents. Providers include pediatric residents, nurse practitioners, and general pediatric and adolescent attending physicians.
The PPCC has used an EHR (Centricity; GE Medical Systems, Waukesha, WI) since September 1, 2001. To allow providers to adapt to the EHR, the review period was initiated 1 year after EHR implementation. Children were included if they had 1 visit to the PPCC during the study period and height and weight measured at a PPCC visit on the same day. Because of the effects of pregnancy on glucose metabolism and our inability to determine the precise period of pregnancies, patients who were pregnant or ever pregnant were excluded.
Although electronic laboratory data from the PPCC were available from June 1, 2001, onward, only laboratory tests that occurred within 90 days before the study period were included in the analyses. This ensured that the laboratory tests included were temporally associated with available BMI measurements. Although it is possible that some patients were excluded from analyses because they had been screened before the beginning of the analytic observation period, the ADA recommends repeat screening biannually and most eligible patients should have been screened at least once during the 2-year study period.
Data Extraction
The following data were extracted from the EHR database: age, height, weight, family history, and HbA1c, RPG, FPG, and OGTT results. No medical record was extracted in its entirety. BMI was calculated automatically from each height/weight pair obtained, with the formula BMI = weight/height2 (with weight in kilograms and height in meters). BMI percentile was determined by using Centers for Disease Control and Prevention reference data.11
To ensure that we captured all possible screening results, any laboratory test that was performed within 90 days before a BMI measurement or any time after a BMI measurement and within the 2-year study period was included in our analyses. If a patient had >1 BMI measurement associated with a screening test, then only the BMI measurement closest in time to the most abnormal laboratory result was used in the analyses. For patients who had no abnormal laboratory test results but >1 BMI measurement, we used the highest BMI, to ensure recording of all patients with elevated BMI measurements within the study period.
The ADA considers a family history of T2DM in a first- or second-degree relative to be a risk factor for T2DM in a child and uses family history as a screening criterion (Table 1). At the PPCC, family histories are recorded by clinicians as unstructured free text in the EHR. Free-text fields were transferred into a spreadsheet, where the data were searched by using the string search terms "dm," "di," and "dai." Family history fields containing any of those strings were reviewed by a physician (S.G.A.) for the presence of any family history of diabetes. Records with a family history of T2DM or an unspecified history of diabetes were coded as positive for a family history of T2DM. Patients with a family history of diabetes in distant relatives (not first or second degree) were coded as negative for a family history of T2DM.
Patients were considered to meet the ADA criteria for T2DM screening if they met the following criteria: family history of T2DM, minority race, and BMI of 85th percentile for age and gender. Because of our inability to extract physical examination data, physical examination findings and the presence of comorbidities of diabetes (Table 1) were not included in our definition of patients meeting ADA screening criteria.
Laboratory Measures
Patients were classified as having undergone screening if they had a FPG measurement, RPG measurement, 2-hour OGTT, or HbA1c measurement during the study period. Patients with laboratory evidence of abnormal glucose metabolism (diabetes, impaired FPG, impaired glucose tolerance, or unspecified) were classified according to test type and level of impairment. Classifications were defined as follows: diabetes: FPG level of 126 mg/dL or OGTT result of 200 mg/dL6; impaired FPG: FPG level of 100 to 125 mg/dL12; impaired glucose tolerance: OGTT result of 140 to 199 mg/dL12; unspecified: RPG level of 200 mg/dL or HbA1c level of 6.0%.
Although RPG and HbA1c measurements are not ADA-recommended screening tests for children, they have been suggested by experts as possible modes of testing and have been used previously for screening in settings other than pediatric primary care.13–17 They were included in our analyses because providers may use them for screening to avoid having patients return in the fasting state. Threshold values were determined by using ADA standards for adults with symptoms of diabetes for RPG and those for known diabetic patients for HbA1c.12 This study was approved by the institutional review boards of Boston University School of Medicine and the Boston Medical Center.
Data Analyses
Frequencies and 
2 analyses of screening rates and results were performed for the following variables: age, race, gender, family history of T2DM, BMI percentile, and presence of ADA criteria for screening. Age and BMI percentile were analyzed as categorical variables. Univariate analyses were performed for each of the aforementioned variables, and 95% confidence intervals (CIs) were calculated. Logistic regression analyses were used to analyze the independent effects of each sociodemographic characteristic on the outcomes of screening and laboratory evidence of abnormal glucose metabolism, by using backward likelihood ratio testing. The 17 Native American subjects in the sample (6 of whom were screened for T2DM) were categorized as other/unknown for logistic regression analyses. Variables that demonstrated significant associations at the level of P < .05 in univariate analyses were included in the multivariate models. ADA risk was not included in the final regression analyses because of its colinearity with other risk factors in the model. Data were analyzed by using Stata SE 8.0 for Windows software (Stata Corp, College Park, TX).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patient Characteristics
The 7710 patients who met the study criteria were distributed nearly equally among all age categories (10–12, 13–15, and 16–19 years of age) (Table 2). The majority of children were black (57.6%) and female (53.0%). A large proportion (42.0%) of children exceeded normal weight, with 18.2% at risk for overweight (BMI of 85th to 94th percentile for age and gender) and 23.8% overweight (BMI of
95th percentile for age and gender). A high percentage (17.7%) had BMI values of >97th percentile. Family history was positive for T2DM for 19.6% of patients. On the basis of BMI, family history, and race, 8.7% of patients met ADA criteria for screening (Table 2).6
|
Screening
Overall, 2452 screening tests were performed for 1642 patients (21.3%) in the cohort (Table 2). In bivariate analyses, screening was significantly more common for the 16- to 19-year-old age group, female patients, patients with a family history of diabetes, and patients who met ADA criteria for screening with our limited criteria. White, black, Hispanic, and Asian patients were all screened at similar rates (20.1%–24.4%) (Table 2). Increasing BMI increased the likelihood of screening, with a dose-response relationship between increasing BMI and screening rates.
Of 1642 patients who were screened, 61.2% (n = 1005) had only 1 type of test, 30.1% (n = 495) had 2 types of tests, and 8.6% (n = 142) had 3 types of tests (Table 3). The most common test used was the RPG (86.7%; n = 1424). The least common test used was the FPG (12.4%; n = 203). Of tested patients, 76.9% (n = 1263) had only nonrecommended tests (RPG or HbA1c). Only 4.6% of screened patients (n = 76) had only recommended screening tests (OGTT or FPG). Among patients who were screened, patients who met ADA criteria were more likely to have recommended screening tests than were those who did not (22.6% vs 10.0%; P < .001).
|
In logistic regression analyses, older children, female patients, patients with higher BMI values, and patients with a family history of T2DM were more likely to be screened (Table 4). For example, 57.8% of female patients who had a family history of T2DM and a BMI of >97th percentile were screened, compared with 18.0% of female patients who had a negative family history and a BMI of <85th percentile.
|
Laboratory Evidence of Abnormal Glucose Metabolism
Among 1642 screened subjects, 9.2% (n = 151) had some laboratory evidence of abnormal glucose metabolism; 62.9% (n = 95) were abnormal RPG results (Table 5). Abnormal glucose metabolism was more common for male patients and patients with a family history of diabetes, BMI between the 90th percentile and 94th percentile or of >97th percentile, and ADA criteria for screening (Tables 5 and 6). Of all patients who showed laboratory evidence of abnormal glucose metabolism (n = 151), 11 (7.3%) met the criteria for T2DM with FPG results, 14 (9.3%) had impaired FPG, and 12 (0.1%) had impaired glucose tolerance. Abnormal glucose metabolism was demonstrated by HbA1c or RPG results for 71.3% of patients (n = 107) with laboratory abnormalities.
|
|
Of 17 patients who met the criteria for T2DM with FPG results, 8 (47.1%) also had abnormal RPG results and 6 (35.3%) also had abnormal HbA1c results. Of 12 patients who had positive OGTT results, 1 (8.3%) also had abnormal RPG results and 1 (8.3%) also had abnormal HbA1c results. Comparisons could not be made among test sensitivities because tests for the same patients might have been performed at different points during the observation period and thus at different stages in a subject's disease trajectory.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In our sample of 7710 patients, 21.3% (n = 1642) were screened for diabetes by using FPG, RPG, OGTT, HbA1c, or a combination of tests. Children were more likely to be screened if they were older, were female, were more overweight, or had a family history of diabetes. These attributes were cited previously as risk factors for diabetes.1,6,10,18
Unfortunately, many children who should have been screened were not. Of 672 children who met ADA criteria for screening in our sample, only 305 (45.4%) were screened. In addition, many of the tests performed were not those recommended by the ADA. Overall, only 4.6% of the patients screened had a FPG test or OGTT. The majority of patients screened (71.3%) had a RPG test. This might be attributable to the ease of performing the RPG test, in comparison with recommended tests. A RPG sample can be drawn at a routine visit, eliminating the difficulty (and the associated increase in nonadherence rates) of having the patient return in the fasting state. The frequency of nonrecommended testing in our clinic suggests that additional investigation into the use of RPG and HbA1c measurements as screening tests is warranted. Results of direct comparisons of test sensitivities in this setting, which have not been published to date, combined with cost and feasibility considerations, could render a role for pediatric T2DM testing that does not require a fasting state.
Despite increased national emphasis on the early detection of T2DM, little is known about the actual screening practices of clinicians in pediatric primary care settings. To our knowledge, only one published study is available, a 2004 evaluation of the implementation of a screening protocol in a Chicago pediatric clinic.19 In comparison with that study, our overall screening rate was much higher (21.3% vs 7%), with similar rates of screening for ADA criteria-positive patients (45.4% vs 38%). Prevalences of T2DM (0.2% vs 0.0%), impaired FPG (0.2% vs 0.0%), and impaired glucose tolerance (3.1% vs 3.0%) were similar in the 2 studies. Although the 2004 study included all of the screening variables in the ADA criteria, including the presence of acanthosis nigricans (a driving factor in provider screening), it was limited by its role in a quality improvement initiative and its failure to account for nonrecommended testing. Our larger study offered a view of current screening practices and also allowed us to determine more-precise estimates of diabetes risk in a high-risk urban population.
Children meeting ADA criteria might not have been screened for a number of reasons. First, familiarity with ADA criteria among pediatric providers is limited, with less than one half of pediatricians screening routinely for diabetes.20 Therefore, providers may not be identifying accurately patients who should be screened. Second, providers may be less apt to recommend diabetes screening for overweight children because providers lack referral resources for nutrition and exercise counseling. Third, it is possible that providers recommend screening for patients who ultimately do not receive screening because of nonadherence. Despite these barriers to appropriate screening in our cohort of providers, we think that the ADA consensus guidelines, if used appropriately, would allow identification of many additional patients at especially high risk for T2DM and thus should be used systematically in the pediatric primary care setting.
Abnormal glucose metabolism was seen for 1.9% (n = 151) of the overall sample and for 9.2% of patients screened; 13 of those patients (0.8% of the overall sample) met the criteria for diabetes. Although good population data are lacking in the literature, our prevalence seems to be higher than that in the Third National Health and Nutrition Examination Survey, in which type 1 diabetes mellitus and T2DM combined to account for 4.1 cases per 1000 children.10 This is likely attributable to the increased prevalences of overweight and minority status in our population, compared with national data.
This study has some important limitations. Like other retrospective chart reviews, our data were limited to information recorded in the chart. We were unable to extract information such as the presence of acanthosis nigricans from the physical examination results; therefore, the number of children who actually met the ADA criteria was likely greater than indicated by our data. The limited observation period might have led to the exclusion of patients who were screened or diagnosed before or after our study period. Patients might have been screened for diabetes in other settings that were not recorded in our analyses. In addition, although we limited our study to primary care visits, it is possible that tests included in our analyses might not have been related to diabetes or might have been used for monitoring, as opposed to initial screening.
Population-based prevalence studies for T2DM in children were recommended by the International Diabetes Federation Consensus Workshop,10 particularly for high-risk populations. Our data represent results for screened patients in such a population. Because screening was not universal, our estimates of prevalence might not reflect the true prevalence of T2DM in this population. A better understanding of the prevalence will require assessment of screening results for a high-risk population after implementation of a comprehensive screening protocol or a true population-based prevalence study, as recommended by expert committees.
In this study of urban children, screening rates for children at highest risk were substantially lower than recommended and many children who did not meet ADA criteria were screened. The EHR has the potential to improve screening by gathering more-comprehensive risk assessment data and providing decision support and prompting when indicated. Additional research regarding how to design and to implement EHR functionality effectively in these areas is needed.
The results of this study confirm the concerning problem of overweight in urban children. Although few screened children had results consistent with T2DM, a considerable number had evidence of abnormal glucose metabolism. Although longitudinal population-based studies are needed to determine outcomes with certainty, these findings may be an early warning sign for children at increased risk for developing T2DM as young adults. We think that better provider education regarding consensus recommendations, screening strategies, and guidelines for interpretation of screening results are needed to allow for earlier provision of effective lifestyle interventions for children at especially high risk for T2DM.
|
ACKNOWLEDGMENTS |
|---|
Financial support was provided by National Institutes of Health grants 1-K24-HD0424891-A2 and 2-T32-HP10014-10.
We thank Howard Bauchner, MD, for his mentorship and thoughtful review of the manuscript.
|
FOOTNOTES |
|---|
Accepted Jun 13, 2006.
Address correspondence to Shikha G. Anand, MD, MPH, Boston Medical Center, 91 East Concord St, Maternity Building, Boston, MA 02118. E-mail: shikha.anand{at}bmc.org
Presented in part at the annual meeting of the Pediatric Academic Societies; May 15, 2006; Washington, DC.
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
1. Pinhas-Hamiel O, Dolan LM, Daniels SR, Standiford D, Khoury PR, Zeitler P. Increased incidence of non-insulin-dependent diabetes mellitus among adolescents. J Pediatr. 1996;128 :608 –615[CrossRef][Web of Science][Medline]
2. Fagot-Campagna A, Pettitt DJ, Engelgau MM, et al. Type 2 diabetes among North American children and adolescents: an epidemiologic review and a public health perspective. J Pediatr. 2000;136 :664 –672[CrossRef][Web of Science][Medline]
3. Dabelea D, Hanson RL, Bennett PH, Roumain J, Knowler WC, Pettitt DJ. Increasing prevalence of type II diabetes in American Indian children. Diabetologia. 1998;41 :904 –910[CrossRef][Web of Science][Medline]
4. Hotu S, Carter B, Watson PD, Cutfield WS, Cundy T. Increasing prevalence of type 2 diabetes in adolescents. J Paediatr Child Health. 2004;40 :201 –204[CrossRef][Web of Science][Medline]
5. Kitagawa T, Owada M, Urakami T, Yamauchi K. Increased incidence of non-insulin dependent diabetes mellitus among Japanese schoolchildren correlates with an increased intake of animal protein and fat.
Clin Pediatr (Phila). 1998;37
:111
–115
6. American Diabetes Association. Type 2 diabetes in children and adolescents. Diabetes Care. 2000;23 :381 –389[Web of Science][Medline]
7. Lipton R, Keenan H, Onyemere KU, Freels S. Incidence and onset features of diabetes in African-American and Latino children in Chicago, 1985–1994. Diabetes Metab Res Rev. 2002;18 :135 –142[CrossRef][Web of Science][Medline]
8. Crawford PB, Story M, Wang MC, Ritchie LD, Sabry ZI. Ethnic issues in the epidemiology of childhood obesity. Pediatr Clin North Am. 2001;48 :855 –878[CrossRef][Web of Science][Medline]
9. Dabelea D, Pettitt DJ, Jones KL, Arslanian SA. Type 2 diabetes mellitus in minority children and adolescents: an emerging problem. Endocrinol Metab Clin North Am. 1999;28 :709 –729[CrossRef][Web of Science][Medline]
10. Alberti G, Zimmet P, Shaw J, et al. Type 2 diabetes in the young: the evolving epidemic: the International Diabetes Federation Consensus Workshop.
Diabetes Care. 2004;27
:1798
–1811
11. Kuczmarksi RJ, Ogden CL, Grummer-Strawn LM, et al. CDC growth charts: United States. Adv Data. 2000;(314):1–27
12. American Diabetes Association. Standards of medical care in diabetes. Diabetes Care. 2004;27 (suppl 1):S15–S35
13. Wollitzer A, Giammattei J, Jovanovic L. HbA1c as a potential screening test for childhood type 2 diabetes mellitus. Diabetes. 2001;50 (suppl 2):A227
14. Peters AL, Davidson MB, Schriger DL, Hasselblad V. A clinical approach for the diagnosis of diabetes mellitus: an analysis using glycosylated hemoglobin levels: Meta-analysis Research Group on the Diagnosis of Diabetes Using Glycated Hemoglobin Levels.
JAMA. 1996;276
:1246
–1252
15. Pettitt DJ, Giammattei J, Wollitzer AO, Jovanovic L. Glycohemoglobin A1c distribution in school children: results from a school-based screening program. Diabetes Res Clin Pract. 2004;65 :45 –49[CrossRef][Web of Science][Medline]
16. Martin DD, Shephard MD, Freeman H, et al. Point-of-care testing of HbA1c and blood glucose in a remote Aboriginal Australian community. Med J Aust. 2005;182 :524 –527[Web of Science][Medline]
17. Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care. 1997;20 :1183 –1197[Web of Science][Medline]
18. Rosenbloom AL, Joe JR, Young RS, Winter WE. Emerging epidemic of type 2 diabetes in youth.
Diabetes Care. 1999;22
:345
–354
19. Drobac S, Brickman W, Smith T, Binns HJ. Evaluation of a type 2 diabetes screening protocol in an urban pediatric clinic.
Pediatrics. 2004;114
:141
–148
20. Ditmyer MM, Price JH, Telljohann SK, Rogalski F. Pediatricians' perceptions and practices regarding prevention and treatment of type 2 diabetes mellitus in children and adolescents.
Arch Pediatr Adolesc Med. 2003;157
:913
–918
PEDIATRICS (ISSN 1098-4275). ©2006 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  B. B. Dean, J. Lam, J. L. Natoli, Q. Butler, D. Aguilar, and R. J. Nordyke Review: Use of Electronic Medical Records for Health Outcomes Research: A Literature Review Med Care Res Rev, December 1, 2009; 66(6): 611 - 638. [Abstract] [PDF]
|
|
|
|
 |
|
 Screening for Diabetes in Pediatric Primary Care Journal Watch Pediatrics and Adolescent Medicine, December 6, 2006; 2006(1206): 5 - 5. [Full Text]
|
|
|
|
 |
|
 Screening for Diabetes in Pediatric Primary Care Journal Watch (General), November 21, 2006; 2006(1121): 3 - 3. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
