Clinical evolution of beta cell function in youth with diabetes: the SEARCH for Diabetes in Youth study
Few studies have explored the epidemiology of beta cell loss in youth with diabetes. This report describes the evolution and major determinants of beta cell function, assessed by fasting C-peptide (FCP), in the SEARCH for Diabetes in Youth study.
Participants were 1,277 youth with diabetes (948 positive for diabetes autoantibodies [DAs] and 329 negative for DAs), diagnosed when aged <20 years, who were followed from a median of 8 months post diagnosis, for approximately 30 months. We modelled the relationship between rate of change in log FCP and determinants of interest using repeated measures general linear models.
Among DA-positive youth, there was a progressive decline in beta cell function of 4% per month, independent of demographics (age, sex, race/ethnicity), genetic susceptibility to autoimmunity (HLA risk), HbA1c and BMI z score, or presence of insulin resistance. Among DA-negative youth, there was marked heterogeneity in beta cell loss, reflecting an aetiologically mixed group. This group likely includes youths with undetected autoimmunity (whose decline is similar to that of DA-positive youth) and youth with non-autoimmune, insulin-resistant diabetes, with limited decline (~0.7% per month).
SEARCH provides unique estimates of beta cell function decline in a large sample of youth with diabetes, indicating that autoimmunity is the major contributor. These data contribute to a better understanding of clinical evolution of beta cell function in youth with diabetes, provide strong support for the aetiological classification of diabetes type and may inform tertiary prevention efforts targeted at high-risk groups.
KeywordsBeta cell function Decline Determinants Epidemiology Type 1 diabetes Type 2 diabetes Youth
Diabetes Prevention Trial-Type 1
Fasting plasma glucose
Glutamic acid decarboxylase-65 autoantibody
- HLA DR/DQ
Human leucocyte antigen DR/DQ genotypes
National Institute of Diabetes and Digestive and Kidney Diseases
Diabetes mellitus is a spectrum of metabolic diseases with distinct aetiologies, characterised by a progressive decline in beta cell function and mass [1, 2]. An increasing body of evidence demonstrates the importance of preserving endogenous beta cell function in patients with diabetes, regardless of aetiology or presumed diabetes type [3, 4]. However, there is a paucity of data regarding the epidemiology of beta cell function, including the rate and major determinants of beta cell decline, especially among contemporary youth with diabetes . Limited information has been provided by interventional studies in youth with autoimmune diabetes [6, 7, 8, 9] and individuals with adult-onset diabetes . Recent data suggest that there is a progressive deterioration of beta cell function both before and after the diagnosis of autoimmune diabetes [11, 12]; however, limited information is available regarding the factors that predict the rate of decline. Even less is known about the clinical evolution of beta cell function in youth without autoimmune diabetes. Moreover, most previous studies were designed to explore the effects of specific interventions, were usually much smaller and often included selected participants, making it difficult to draw conclusions applicable to the general population of youth with diabetes. Such data may be relevant for the aetiological classification of diabetes type.
SEARCH for Diabetes in Youth is a population-based, multicentre and multi-ethnic study of paediatric diabetes in the USA. With the aim of rigorously characterising the epidemiology of diabetes in youth, this study has carefully collected detailed clinical and pathophysiological data on a large number of youth with all types of diabetes. Moreover, SEARCH has recently developed operational definitions of diabetes types, as recommended by the ADA expert committee, using an aetiological approach based on assessment of autoimmunity and insulin resistance (IR), in which four aetiological subgroups were defined . The purpose of this manuscript is to describe the presentation, short-term clinical evolution and major determinants of decline in beta cell function, as measured by fasting C-peptide (FCP), in a general population of youth with diabetes, using data from the prospective cohort of the SEARCH for Diabetes in Youth study.
Data for these analyses derive from the SEARCH for Diabetes in Youth prospective cohort study. A detailed description of SEARCH study methods has been published elsewhere . Briefly, SEARCH has been conducting population-based ascertainment of youth and young adults with newly diagnosed (incident) diabetes diagnosed when <20 years of age, starting in 2002 and continuing through to the present. SEARCH recruited participants from four geographically defined populations in the USA (Ohio, Colorado, South Carolina and Washington) and also Indian Health Service beneficiaries from several American Indian populations and enrollees in several managed healthcare plans in California and Hawaii. Participants with newly diagnosed diabetes in 2002–2005 were invited to participate in a baseline research visit (median duration of diabetes 8 months, interquartile range [IQR] 4.5–14.5 months) and two follow-up exams centred around 12 and 24 months after their baseline visit. At each research visit fasting blood samples were obtained from metabolically stable participants (defined as no episode of diabetic ketoacidosis during the previous month), physical measurements were conducted and questionnaires were administered. The study was reviewed and approved by the local Institutional Review Board(s) that had jurisdiction over the local study population and all participants provided informed consent and/or assent.
Study visits occurred after an 8 h overnight fast. Participants did not take diabetes medications on the morning of the visit and long-acting insulin was administered on the evening before the visit and then discontinued. Fasting blood samples were used to analyse diabetes autoantibodies (DAs), HbA1c, lipids and FCP and fasting plasma glucose (FPG). Assays were performed at the Northwest Lipid Metabolism and Diabetes Research Laboratories, University of Washington, the Central Laboratory for SEARCH. Glutamic acid decarboxylase-65 (GADA) and insulinoma-associated-2 (IA-2A) autoantibodies were analysed using a standardised protocol and a common serum calibrator developed by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)-sponsored harmonisation group . The cut-off values for positivity were 33 NIDDKU/ml for GADA and 5 NIDDKU/ml for IA-2A . HbA1c was measured in whole blood with an automated nonporous ion-exchange HPLC system (model G-7; Tosoh Bioscience, San Francisco, CA, USA). Measurements of plasma cholesterol, triacylglycerols and HDL-cholesterol were performed on a Roche Modular-P autoanalyzer (Roche Diagnostics, Indianapolis, IN, USA). Levels of FCP were determined by a two-site immunoenzymetric assay (Tosoh 1800; Tosoh Bioscience) with a sensitivity of 0.05 ng/ml. Human leucocyte antigen class II genotyping (HLA DR/DQ) was performed with a PCR-based sequence-specific oligonucleotide probe system in the laboratories of S. Gaur (University of Washington, Seattle, WA, USA) and H. Erlich (Roche Molecular Systems, Pleasanton, CA, USA) on all consenting participants .
Weight and height were measured using standardised procedures and used to calculate BMI. Waist circumference was measured using the National Health and Nutrition Examination Survey (NHANES) III protocol .
Insulin sensitivity (IS) was estimated using the following equation: loge IS = 4.64725 − 0.02032 (waist, cm) − 0.09779 (HbA1c, %) − 0.00235 (triacylglycerol, mg/dl; to convert triacylglycerol values from mg/dl to mmol/l, multiply by 0.0113). This equation was developed and validated in a previous study  using direct measurements of glucose disposal rate from euglycaemic–hyperinsulinaemic clamps. IR was defined as an IS value below the 25th percentile in the general US youth population (<8.15), as previously described .
Data on self-reported race and ethnicity were collected using 2000 US Census questions  and classified as Hispanic, non-Hispanic white (NHW), non-Hispanic black, American Indian and Asian/Pacific islander.
Determinants of interest
Autoimmunity was defined by positive titres for either GADA or IA-2A at baseline. Because many participants were treated with insulin, positive insulin autoantibodies were not used. Genetic susceptibility to autoimmunity was defined based on HLA DR/DQ. HLA genotypes were categorised as susceptible (DR3/4, DR4/4, DR4/8, DR4/1, DR 4/13, DR 3/3, DR 3/9, DR 4/9 and DR 9/9 genotypes) and not susceptible, including neutral and protective genotypes , as recommended by the Type 1 Diabetes Genetic Consortium, with modifications for the multi-ethnic population . For these analyses race/ethnicity was categorised as NHW and other than non-Hispanic white, including all other racial/ethnic groups. BMI z scores from measured height and weight at baseline visit were derived for age and sex using the US Centers for Disease Control and Prevention National Center for Health Statistics growth charts . Finally, SEARCH participants were also classified according to baseline aetiological category of diabetes type in four mutually exclusive groups as previously described : autoimmune and insulin sensitive, autoimmune and insulin resistant, non-autoimmune and insulin sensitive, non-autoimmune and insulin resistant.
The eligible study population consisted of all SEARCH participants who were diagnosed in 2002–2005, who had a baseline visit with non-missing FCP, DA and demographic data, plus at least another follow-up visit with non-missing FCP. There were 2,763 individuals that had a baseline SEARCH visit and 1,895 participants had non-missing FCP, DA and demographic data at baseline. An additional sample of 629 were dropped because they did not have at least one follow-up visit with non-missing FCP, resulting in 1,277 SEARCH individuals contributing to this analysis. Of the 1,277 participants included in these analyses, 634 contributed two visits (baseline plus one follow-up) and 643 contributed three visits (baseline plus two follow-up visits). There were 948 DA-positive participants (487 with double and 461 with single antibody positivity) and 329 DA-negative participants.
Statistical analyses were performed using SAS software (version 9.2; SAS Institute, Cary, NC, USA), with p < 0.05 indicating significance. Comparisons of baseline participant characteristics by baseline DA status were examined using χ2 tests, t tests and Wilcoxon tests. Beta cell function was estimated using FCP concentration .
Using all baseline and follow-up data, we used mixed effects regression analysis to derive predicted FCP levels over time, based on fitted models that included up to three observed measures of FCP for each participant. These models adjust for the correlation between repeated observations in the same individual and have the advantage of handling a varying number of observations within individuals, thereby allowing for inclusion of the maximum number of data points . Further, this method allows for a random intercept and slope in disease duration to account for within-individual dependence. These models incorporate duration of diabetes (actual time in months) as a continuous time-varying covariate and explore significant associations (statistical interactions) between determinants of interest (baseline DA status, age at diagnosis, sex, race/ethnicity, HLA DR/DQ susceptibility status, baseline BMI z score and HbA1c) and time (duration of diabetes) on FCP levels. All models are also adjusted for site- and time-varying FPG levels measured at each FCP measurement. Since baseline DA status emerged as the main determinant of FCP decline all analyses were stratified by baseline DA status. For DA-positive individuals, the effect of double vs single DA positivity at baseline on FCP decline over time was also explored, adjusting for all the covariates mentioned above. Age at diagnosis was modelled both continuously and categorically (<10 and ≥10 years, for illustrative purposes) with very similar results. To account for a skewed distribution of FCP levels, values were log-transformed for these analyses. Predicted FCP rate of change was expressed as per cent change in FCP level per month with 95% CIs.
Finally, we were also interested in modelling the rate of change in FCP levels as a function of baseline aetiological category of diabetes type. Therefore, a separate model was built to explore the association between the four-level aetiological diabetes type variable (autoimmune and insulin sensitive, autoimmune and insulin resistant, non-autoimmune and insulin sensitive, non-autoimmune and insulin resistant) and log-transformed FCP over time. This analysis excluded 66 participants due to missing data on variables that are part of the IS equation.
Characteristics of study participants, according to baseline DA status
Age at diagnosis (years)
Age at diagnosis <10 years
HLA DR-DQ group
FCP (nmol/l), median (IQR)
BMI z score
FPG (mmol/l), mean (SD)
Number of participants at each visit, their average diabetes duration and measured FCP and FPG levels, by DA status
Duration (months), median (IQR)
FCP (nmol/l), median (IQR)
Glucose (mmol/l), mean (SD)
Determinants of baseline FCP levels from multivariate analyses, stratified by baseline DA status
Effect, % (95% CI)a
Effect, % (95% CI)a
Age at diagnosis (<10 vs ≥10 years)
44 (38, 49)
57 (47, 65)
Sex (male vs female)
11 (2, 19)
Race (NHW vs others)
19 (9, 28)
30 (15, 42)
HLA (susceptible vs not susceptible)
Baseline HbA1c (per 1% increase)
19 (15, 22)
7 (1, 13)
Baseline BMI z score (per 1 unit decrease)
17 (13, 21)
37 (32, 42)
Baseline FCP levels and estimated monthly decline in FCP, according to baseline aetiological category of diabetesa
Baseline FCP (nmol/l), median (IQR)
Estimated monthly decline, % (95% CI)
Autoimmune + insulin sensitive
4.0 (3.8, 4.3)
Autoimmune + insulin resistant
4.2 (3.8, 4.7)
Non-autoimmune + insulin sensitive
2.4 (1.9, 3.0)
Non-autoimmune + insulin resistant
0.7 (0.3, 1.2)
Our study provides unique quantitative estimates of decline in beta cell function over the first 30 months following diagnosis of diabetes in a large, diverse sample of youth in the USA. SEARCH provides strong evidence that autoimmunity is the main driver of beta cell loss even after diabetes diagnosis. Among DA-positive youth, there is a progressive decline in beta cell function, independent of demographic factors, genetic susceptibility to autoimmunity (HLA risk), HbA1c levels and BMI z score. Among DA-negative youth, there is marked heterogeneity in beta cell loss, reflecting an aetiologically mixed group. Our data contribute to a better understanding of the clinical progression of diabetes in youth in the first years following diagnosis.
In youth with autoimmune diabetes there was an estimated decline in beta cell function of ~4% per month, or approximately 40% per year. Similarly, in a series of 204 consecutive patients with type 1 diabetes, Snorgaard et al  reported that FCP levels (on average 0.5 ng/ml at diagnosis) declined in a homogeneous manner at a rate of 0.24 ng/ml annually. Other studies have estimated the decline in beta cell function during the first year after diagnosis of type 1 diabetes, with variable results. Most of the recent studies are based on data provided by the control arm of intervention studies, include small numbers of highly selected participants and use stimulated C-peptide as the measure of beta cell function. While some report modest reduction in beta cell function at 1 year [24, 25], others report a 50% decline in the first year post-diagnosis [1, 26, 27, 28], consistent with our data in an unselected population.
There are limited data on factors influencing beta cell function in youth with autoimmune diabetes [12, 29, 30, 31, 32, 33, 34, 35]. We found that younger age, male sex, NHW race/ethnicity, higher HbA1c and lower BMI z score were each associated with lower FCP levels at baseline (Table 3). The Diabetes Prevention Trial-Type 1 (DPT-1) study has demonstrated that C-peptide levels at diagnosis are much higher in individuals diagnosed through surveillance of those at high risk  compared with those diagnosed in the community, with signs and symptoms of hyperglycaemia. Our findings that higher HbA1c and lower BMI z score are associated with lower baseline FCP levels are consistent with DPT-1 findings, suggesting a more severe or longer preclinical course in those with less residual beta cell function at presentation. Consistent with our data, several reports have shown that younger age is associated with lower C-peptide levels at diagnosis with type 1 diabetes [8, 9, 36]. This may suggest a more aggressive preclinical course or an overall lower beta cell mass in younger children , while in older patients greater IR may lead to a clinical presentation at an earlier preclinical stage of disease [21, 37].
Although age, sex, race/ethnicity, HbA1c and BMI z score were each associated with lower FCP levels at baseline (Table 3), none of them were significantly associated with the subsequent rate of decline (Fig. 1). Based on our data, on average, residual beta cell function at any point in time is a function of baseline C-peptide levels and disease duration. Others have also reported that baseline C-peptide level is the strongest predictor of decline in beta cell function in youth  and young adults  with autoimmune diabetes.
One of our most novel findings is that markers of obesity and IR do not significantly influence the decline in FCP levels after diagnosis of autoimmune diabetes. Investigators and clinicians have previously described a group of youth with autoimmunity and obesity as having ‘hybrid’ or ‘double’ diabetes , suggesting a mixed aetiology and, likely, a more aggressive clinical evolution. The aetiological classification of diabetes type proposed by SEARCH  and findings from other recent studies  provide evidence that rather than being a distinct aetiological category this group represents the upper tail of the distribution of IR and obesity among youth with autoimmune diabetes. We now provide novel evidence that the rate of FCP decline in youth with autoimmune, insulin-resistant diabetes is not significantly different from that observed in youth with autoimmune, insulin-sensitive diabetes (4.2% [95% CI 3.8, 4.7] per month vs 4.0% [95% CI 3.8, 4.3] per month, p = 0.5, Table 4). This was also suggested by baseline BMI z score not being significantly associated with decline in FCP levels in DA-positive youth. These findings lend additional support to the notion that type 1 diabetes is a distinct aetiological category and that the main driver of beta cell loss is autoimmunity, regardless of prevailing IR that may occur simultaneously. These findings also have important clinical implications indicating the importance of a correct diagnosis of autoimmune type 1 diabetes in obese youth, and thus adequate insulin treatment.
Less is known about the clinical history of beta cell function in youth with non-autoimmune diabetes. Although small studies suggested that youth with type 2 diabetes have some degree of beta cell dysfunction at diagnosis [41, 42], there are no large prospective studies of the natural evolution and determinants of beta cell function in youth with type 2 diabetes. The estimated rate of decline among SEARCH youth who had non-autoimmune, insulin-resistant diabetes was 0.7% per month, or approximately 8% per year, in the first 30 months following diagnosis (Table 4). This is consistent with studies in adults showing a 7% decline annually [3, 43]. However, a recent study has suggested no signs of progressive beta cell damage during 20 years of prospective follow-up of autoantibody-negative, adult-onset diabetes , which contrasts with the generally held view of progressive beta cell loss in type 2 diabetes. There is an obvious need for long-term follow-up studies of individuals with youth-onset, non-autoimmune, type 2 diabetes.
Our study has some limitations and unique strengths. Because SEARCH is a large epidemiological study, we used fasting, rather than stimulated, C-peptide levels to assess baseline and change in beta cell function over time. FCP level is easy to obtain and correlates well with stimulated C-peptide [45, 46, 47], although it may be insufficient to detect subtle effects of therapy. In addition, we used a standard protocol to assess beta cell function, with all measurements conducted in the fasting state, in the absence of hypo-or hyperglycaemia, which are known to affect FCP levels . Only a fraction of participants had FCP measures in the first 6 months after diagnosis and, therefore, during this period the rate of decline in FCP may have been estimated with less precision. Only two autoantibodies were measured in SEARCH. Thus, individuals who may have autoimmune markers other than GADA or IA-2A were misclassified as having non-autoimmune diabetes. However, the predicted rates of change in FCP levels based on models that included other characteristics, such as HLA DR/DQ susceptibility genotypes, race/ethnicity and age at diagnosis, identified groups of participants with likely undetected autoimmune diabetes. In addition, SEARCH is planning to assess ZnT8 autoantibodies in stored samples once assays are standardised. We also do not have follow-up data on all SEARCH participants. However, minimal differences existed between those who did (n = 1,277) and did not participate (n = 629) in the follow-up assessment. In particular, the proportion of DA-positive and DA-negative participants, the racial/ethnic distribution, the mean age and FCP levels according to DA status were not substantially different. Finally, our data are not the result of a randomised clinical trial, which is much better suited to address specific questions regarding the role of glycaemic control or that of specific therapies, on beta cell decline in youth with diabetes. Our study also has very important strengths. This is a unique cohort of youth with diabetes ascertained without regard to presumed diabetes type, family history or related factors, such as age, treatment or race/ethnicity, which has permitted us to accurately quantify progression of beta cell damage in a large sample of youth with diabetes.
In summary, SEARCH provides unique quantitative estimates of beta cell function decline over the first 30 months following diagnosis in a large sample of youth with diabetes in the USA. Our data provide strong evidence that autoimmunity is the main driver of beta cell loss even after diabetes diagnosis. We also provide novel evidence that markers of obesity and IR do not influence the decline in beta cell function in youth with autoimmune diabetes. Finally, youth with non-autoimmune, insulin-resistant diabetes have a much slower progression of beta cell dysfunction. These data contribute to a better understanding of clinical evolution of beta cell function in youth with diabetes, provide strong support for the aetiological classification of diabetes type and may inform tertiary prevention efforts targeted at high-risk groups.
The SEARCH for Diabetes in Youth study is indebted to the many young individuals, their families and their healthcare providers, whose participation made this study possible. The writing group for this manuscript wishes to acknowledge the contributions of the following individuals to the SEARCH for Diabetes in Youth study:
California: J. M. Lawrence, K. Reynolds, J.-W. Hsu, M. H. Black, K. Holmquist and H. S. Takhar for the Department of Research and Evaluation and A. K. Kershnar for the Department of Pediatrics, Kaiser Permanente Southern California; and D. J. Pettitt for the Sansum Diabetes Research Institute.
Colorado: D. Dabelea, R. F. Hamman and L. Testaverde for the Department of Epidemiology, Colorado School of Public Health, University of Colorado Denver; G. J. Klingensmith, M. J. Rewers, D. Maahs and P. Wadwa for the Barbara Davis Center for Childhood Diabetes; S. Daniels and K. Nadeau, Department of Pediatrics and Children’s Hospital; C. A. Bloch for the Pediatric Endocrine Associates; C. Sorrelman and A. Beartrack for the Navajo Area Indian Health Prevention Program.
Hawaii: B. L. Rodriguez, B. Waitzfelder, W. Fujimoto, J. D. Curb, F. Kennedy, G. Uramoto, S. Waxman, T. Hillier and R. Chung for Kuakini Medical Center.
Ohio: L. M. Dolan, M. Seid, E. Urbina and D. A. Standiford for the Cincinnati Children’s Hospital Medical Center.
Carolina: E. J. Mayer-Davis and J. Thomas for the University of North Carolina, Chapel Hill; A. Merchant, A. D. Liese, R. R. Moran, G. Gaillard-McBride, D. Lawler and M. Jackson for the University of South Carolina; D. Bowlby for the Medical University of South Carolina; J. Amrhein for Greenville Hospital Systems; P. Clark for McLeod Pediatric Subspecialists; M. Parker for Pediatric Endocrinology & Diabetes Specialists, Charlotte, North Carolina.
Washington: C. Pihoker, M. Afkarian, A. Badaru, L. Gilliam, I. Hirsch, L. L. Liu, J. Neff, and J. Yi-Frazier for the University of Washington; B. Loots, R. O’Connor, S. Kearns, S. Semana and M. Yaptangco for Seattle Children’s Hospital; and C. Greenbaum for Benaroya Research Institute.
Centers for Disease Control and Prevention: G. Imperatore, D. E. Williams, H. S. Kahn, B. Moore G. W. Edward and S. H. Saydah.
National Institute of Diabetes and Digestive and Kidney Diseases, NIH: B. Linder.
Central Laboratory (University of Washington): S. M. Marcovina, V. P. Gaur and K. Gadbois
Coordinating Center (Wake Forest University School of Medicine): R. Bell, R. D’Agostino, Jr, B. Beech, D. Case, T. Morgan, M. J. Naughton, S. Moxley, G. Hargis, D. Kronner, A. Anderson, C. Davis, J. Andrews and J. W. Talton.
SEARCH for Diabetes in Youth is funded by the Centers for Disease Control and Prevention (PA Nos 00097, DP-05-069 and DP-10-001) and supported by the National Institute of Diabetes and Digestive and Kidney Diseases.
Site contract Nos: Kaiser Permanente Southern California (U48/CCU919219, U01 DP000246 and U18DP002714), University of Colorado Denver (U48/CCU819241-3, U01 DP000247 and U18DP000247-06A1), Kuakini Medical Center (U58CCU919256 and U01 DP000245), Children’s Hospital Medical Center (Cincinnati) (U48/CCU519239, U01 DP000248 and 1U18DP002709), University of North Carolina at Chapel Hill (U48/CCU419249, U01 DP000254 and U18DP002708-01), University of Washington School of Medicine (U58/CCU019235-4, U01 DP000244 and U18DP002710-01), Wake Forest University School of Medicine (U48/CCU919219, U01 DP000250 and 200-2010-35171).
The authors wish to acknowledge the involvement of General Clinical Research Centers (GCRC) at the South Carolina Clinical & Translational Research (SCTR) Institute, at the Medical University of South Carolina (NIH/NCRR Grant No. UL1RR029882); Children’s Hospital and Regional Medical Center (Grant No. M01RR00037); Colorado Pediatric General Clinical Research Center (Grant No. M01 RR00069) and the Barbara Davis Center at the University of Colorado at Denver (DERC NIH P30 DK57516); and the Institutional Clinical and Translational Science Award (CTSA), NIH/NCRR at the University of Cincinnati (Grant No. 1UL1RR026314-01).
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
DD designed the study, obtained funding, researched the data, wrote the manuscript and takes full responsibility for the contents of the article. EJM-D, LMD, CP, SM, RD’A Jr, RFH, CG, WF, BL and GI designed the study, obtained funding, contributed to discussion and edited the manuscript. JSA analysed the data and edited the manuscript. The final version of this manuscript has been approved by all authors. The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention and the NIDDK.
- 2.Turner RC, Cull CA, Frighi V, Holman RR (1999) Glycemic control with diet, sulfonylurea, metformin, or insulin in patients with type 2 diabetes mellitus: progressive requirement for multiple therapies (UKPDS 49). UK Prospective Diabetes Study (UKPDS) Group. JAMA 281:2005–2012CrossRefPubMedGoogle Scholar
- 19.Ingram DD, Parker JD, Schenker N et al. (2003) United States Census 2000 population with bridged race categories. Vital Health Stat 21–55Google Scholar
- 20.Kuczmarski RJ, Ogden CL, Guo SS et al (2002) 2000 CDC Growth Charts for the United States: methods and development. Vital Health Stat 11:1–190Google Scholar
- 22.Brown H, Prescott R (2011) Applied mixed models in medicine. Wiley, New YorkGoogle Scholar
- 34.Fernandez CM, Montana E, Camps I et al (1996) Ketoacidosis at diagnosis is predictive of lower residual beta-cell function and poor metabolic control in type 1 diabetes. Diabetes Metab 22:349–355Google Scholar
- 35.Montana E, Fernandez-Castaner M, Rosel P, Gomez J, Soler J (1991) Age, sex and ICA influence on beta-cell secretion during the first year after the diagnosis of type 1 diabetes mellitus. Diabetes Metab 17:460–468Google Scholar