[Skip to Navigation]
Sign In
September 2003

Self-reported Social Class, Self-management Behaviors, and the Effect of Diabetes Mellitus in Urban, Minority Young People and Their Families

Author Affiliations

From the Departments of Pediatrics (Drs Lipton, Burnet, Mencarini, and Rich and Mr Cooper), Medicine (Dr Burnet), and Health Studies (Drs Lipton and Drum), University of Chicago, Chicago, Ill.

Arch Pediatr Adolesc Med. 2003;157(9):919-925. doi:10.1001/archpedi.157.9.919

Objective  To document social factors, diabetes mellitus (DM) self-management, and the effect of the disease among children with DM residing in Chicago, Ill.

Methods  Patients were ascertained for a population-based study of childhood-onset DM (ages, 0-17 years). Families and patients (N = 288) were interviewed about health services use, sociodemographics, self-management behaviors, and other characteristics, a mean of 6.1 years after diagnosis. Univariate and multivariate analyses were conducted.

Results  Most patients were non-Hispanic black (68.4%) or Hispanic (28.5%). Overall, 59.8% of the respondents received Medicaid or had no health insurance at the time of disease onset, including 61.7% of non-Hispanic black patients and 59.2% of Hispanic patients. Of the entire group, 61.5% reported that having DM had affected their family life in some way, with 14.9% reporting difficulty in obtaining child care and 18.4% reporting an effect on the parent and/or guardian's job. Having DM negatively affected the personal lives of 33.7% of the young people, with 13.9% having been excluded from sports activities, and 16.8% having had difficulty obtaining insurance. Among the 64 patients who were aged 20 years or older when interviewed, 21.0% had been denied employment because of their DM. The level of blood glucose was measured less than once daily by 8.8% of the respondents, while 12.4% had never been given a dilated ophthalmologic examination or had not been examined within the past year. Fully 30.6% of those aged 20 years or older smoked cigarettes. There were few ethnic differences in these characteristics.

Conclusion  We conclude that DM has had a significant effect in this group of primarily African American and Hispanic young people who have a chronic illness.

DURING THE PAST 4 decades, therapy for diabetes mellitus (DM) has evolved so that young people are raised with the expectation of full participation in work and family life. However, barriers to adequate management of DM persist for health insurance, employment, and education, and they may be even more important among underserved minority young people. Although African Americans and Latinos in the United States are at almost the same risk of developing DM in childhood as non-Hispanic white persons, published data focus largely on middle-class non-Hispanic white persons. The enormous financial and social costs of childhood DM1,2 are likely to affect minority families disproportionately, since they are overrepresented among the poor in the United States. Glasgow et al3 studied readmissions for metabolic stabilization in a racially mixed pediatric clinic population: adolescents aged 14 through 17 years, as well as African Americans, were at increased risk for readmission to the hospital. Glasgow et al concluded that the cause of readmission was largely a result of poor adherence to the treatment regimen. A previous analysis of our Chicago, Ill, data, which controlled for several potential confounding variables, found that having one's DM diagnosed in a community-based setting (compared with a tertiary care facility) was associated with rehospitalization after the initial diagnosis.4

African Americans are at greater risk for both acute (eg, diabetic ketoacidosis) and chronic complications of type 1 DM (eg, kidney disease) than are white persons5; data are still accumulating for Hispanics in the United States. Presumably these race disparities are due to differences in socioeconomic conditions and related factors.6-8 The literature demonstrates that, although there is a clear rationale for strict glycemic control, there are large groups of patients with unacceptably poor management and/or adherence to treatment. While there has been much speculation, few population-based data have been published on access to health services for minority children; a few studies relate ethnicity and socioeconomic factors to morbidity and mortality resulting from DM.7,9 Our research group demonstrated a 9-fold increased risk of death for young African Americans with DM in Chicago compared with non-Hispanic white patients with DM over an 8-year period.10

The goal of the Chicago Childhood Diabetes Registry (CCDR) is to describe the epidemiology of childhood DM with particular attention to young people of African and Hispanic descent. These groups represent most children in Chicago, and they are diverse in terms of economic advancement and the length of time since immigration. The 2000 US Census enumerated a total of 759 840 Chicagoans aged between 0 and 17 years. These included 332 296 non-Hispanic blacks, 265 857 Hispanics, 122 672 non-Hispanic whites, and 39 015 from other ethnic groups,11 providing the opportunity to measure epidemiological parameters with precision in a large at-risk population. This descriptive report presents sociodemographic characteristics, self-reported health care use, DM self-management behaviors, and the effect of DM on personal and family life in non-Hispanic white, non-Hispanic black, and Hispanic children and teenagers whose condition was diagnosed when they were between the ages of 0 and 17 years and during the interval January 1, 1985, through December 31, 2001.


Chicago childhood diabetes registry

The CCDR is a citywide registry of patients with DM aged between 0 and 17 years at the initial diagnosis. The primary source of ascertainment was hospital records, augmented by outpatient sources (American Diabetes Association camping lists, medical records from 16 Chicago Department of Public Health clinics, a survey of freestanding community clinics, and the pharmacy payment database of the Illinois Department of Public Aid). Ascertainment was estimated to be about 85% complete through 1994, using the capture-mark-recapture method.12,13 Retrospective ascertainment of non-Hispanic black and Hispanic patients began in 1991, and those of other ethnic backgrounds began to be ascertained in 1999. Eligible cases were identified by review of medical records at 37 of 40 area hospitals that had a pediatrics ward and at least 200 total inpatient beds; patients were included if they were diagnosed as having DM after December 31, 1984, and were a resident of Chicago at the time of the diagnosis. If the condition was diagnosed before January 1, 1992, only those children discharged from the hospital receiving insulin treatment and those who were classified as non-Hispanic black or Hispanic were registered. Later, all young Chicago residents diagnosed as having DM were included, irrespective of the initial treatment or ethnic group. Ethnicity was defined as being listed as non-Hispanic white, non-Hispanic black, or Hispanic on the medical record, and/or having a surname classified as Hispanic by the US Bureau of the Census.14 Those who had a Hispanic surname but were classified as non-Hispanic white or black on the medical record were considered to be Hispanic.

A subset of 288 patients was contacted and interviewed a mean of 6.1 years after diagnosis, beginning in 1996. The attending physician was required to give permission before patients could be approached. Locating and contacting physicians, followed by the tracing and contacting of patients, was a time-consuming process. Therefore, all of those who were traced were invited to participate in the interviews, of whom 89% did. These individuals represented about 20% to 25% of the entire CCDR database at the time they were interviewed. Human subjects research committees at the University of Chicago and the other participating institutions citywide approved the study protocol. Written informed consent was obtained from participants prior to the interview.

Exposure and outcome measures were abstracted from medical records and derived from patient and/or family interviews. We distinguished a group of patients with probable type 2 DM from the body of registered cases based on documentation in the medical record of one or more of the following: obesity, a body mass index exceeding 27 kg/m2 at onset; a physician note of "possible type 2 DM" or "atypical" DM; markers of insulin resistance such as acanthosis nigricans or polycystic ovary syndrome; or the patient's receiving oral medications for the treatment of DM. In addition, those who responded positively to certain questions during the interview were considered likely to have type 2 DM. These questions elicited information on cessation of insulin use for more than 2 weeks' duration and more than 6 months after the initial diagnosis, that is, after the "honeymoon" period. We also classified those who reported current treatment with any oral medication for DM as having type 2 DM. Those who did not meet any of these criteria were considered to have type 1 DM.

Participants were asked about the type of health insurance that they had over the past year. Private insurance was defined as using any health maintenance organization, preferred provider organization, or fee-for-service plan that was not affiliated with Medicaid or the Illinois Department of Public Aid. To assess the effects of the child's DM, we asked about its effect on family activities such as child care, travel, and others as well as on patient-specific concerns such as school activities, sports, and employment. Families were considered to be affected by the child's DM if they reported at least one affected domain, and patients were considered to be affected themselves if they responded to any of the individual-level items. Health care resources were considered to be tertiary care facilities if they supported a pediatric DM specialty clinic (this category included 5 that were affiliated with a university, the large public hospital serving the medically indigent, and 3 hospitals with pediatric endocrinology clinics). All other sources of medical care were considered to be community-based health care providers or clinics.

Data collection and statistical methods

Data were abstracted from medical records by trained research assistants onto paper forms or, more recently, handheld electronic devices (Palm IIIx; Palm Inc, Milpitas, Calif). Trained interviewers conducted telephone interviews using a standardized protocol. Univariate associations of demographics, clinical variables, and survey responses with ethnicity, insurance type, and age group, were tested using χ2 tests for homogeneity, t tests, or analysis of variance as appropriate. Where relevant, P values were adjusted for disease duration (rehospitalization and/or personal effect) or age at onset (family effect) using logistic regression; for models involving rehospitalization, the duration effects were quadratic. In the analysis examining the relationship of current age with health care use and behaviors, P values were not adjusted for disease duration, since duration was correlated with current age. None of the P values was adjusted for multiple comparisons.


Characteristics of interviewed sample

The interviewed patients did not differ from the remaining patients on relevant variables (sex, family history, type 1 DM vs presumed type 2 DM). However, those not interviewed were, on average, 2 years older at the time of their diagnosis, 11.5 (4.3) years (mean [SD]), compared with those interviewed (9.5 [4.5] years; P<.001). Fewer non-Hispanic white patients and those with onset of disease after 1999 have been interviewed because they were more recently ascertained.

Demographic comparisons

In these young people with DM, there were few demographic or clinical differences between African American and Hispanic patients (Table 1); the number of non-Hispanic white patients who were interviewed was small. About 40% of the African American patients and 27% of the Hispanic patients were living in poverty as defined by a self-reported annual household income of less than $15 000. This level closely corresponds to the US poverty threshold guidelines for the years 1996-2003 for families of 3 to 4. Slightly more than half of the sample used Medicaid or had no health insurance. Educational attainment of the head of household was particularly low for Hispanic patients, reflecting the large numbers of newly arrived immigrants in the Chicago Hispanic community. The overwhelming majority of these young patients were being treated with insulin, although about one fourth were classified as having early-onset type 2 DM according to our criteria.

Table 1. 
Demographic and Clinical Characteristics of Interviewed Patients With Diabetes Mellitus (DM)*
Demographic and Clinical Characteristics of Interviewed Patients With Diabetes Mellitus (DM)*

Comparisons by ethnic group

There were few ethnic differences with respect to the effect of DM (Table 2). More than 60% of the patients reported that DM had affected their family life in some way, with 14.9% reporting specific problems in obtaining child care and 18.4% reporting an effect on the parents' and/or guardian's job. Among the young people themselves, 33.7% reported an adverse effect of DM: 13.9% had been excluded from sport activities, and 16.8% reported difficulty obtaining insurance. Non-Hispanic black patients were significantly more likely to report being excluded from insurance than were Hispanic patients (P = .04). Among those who were at least 20 years old when interviewed, 21.0% had been denied employment because of their DM.

Table 2. 
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Interviewed Patients by Race/Ethnicity*
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Interviewed Patients by Race/Ethnicity*

More than half of the participants, 57.6%, had been rehospitalized for DM-related causes at least once, including 61.9% of African American patients and 47.6% of Hispanic patients. Somewhat more patients were being followed up by a practitioner affiliated with a tertiary care facility (39.9%) than a community-based provider; many (27.2%) of the patients saw providers at both types of facilities during the past year. Almost all of the respondents had seen a physician at least once during the past year, although about 10% reported not seeking medical care when they should have. Self-monitoring of the blood glucose level was performed infrequently by 9.2% of those taking insulin, while fully 20.0% of the patients rarely or never checked their feet for cuts and sores. Twelve percent of the patients had not had a dilated ophthalmologic examination within the past year or had never been examined.

Comparisons by type of insurance

In comparing groups by the type of health insurance reported at the time of interview, some differences were noted, although there were no major differences in health services use (Table 3). Not surprisingly, those with private insurance were much less likely to report an annual household income less than $15 000. Twice as many families receiving Medicaid or no insurance reported an adverse effect of DM on their ability to get child care or on the parent's employment, but there was no overall difference between insurance categories in reported effects on family life. More of those receiving Medicaid or those who had no insurance had been readmitted to the hospital for DM as well as for other reasons. The type of facility where medical care was obtained during the past year did not correlate either with the rate of readmission to the hospital or with self-management behaviors.

Table 3. 
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Interviewed Patients by Current Health Insurance Status*
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Interviewed Patients by Current Health Insurance Status*

Comparisons by age

As children pass through adolescence into young adulthood, their health care use and self-care behaviors are likely to deteriorate, and this is of particular concern for young patients with DM.7 Among patients 15 years of age or older at the time of interview, about one third reported that DM had affected their personal lives (Table 4). Among those 20 years of age and older, 21% reported having been excluded from a job because of DM. Of those 20 years of age and older, 28.6% had graduated from high school, while an additional 49.2% of the group reported having at least some college or technical training. Although about 40% of teenagers and young adults fit our criteria for type 2 DM, 89.7% of those older than 14 years were being treated with insulin exclusively. The rate of smoking was 7.8% in those 15 through 19 years of age and 30.6% in those 20 years of age and older. Compared with those 15 through 19 years of age, those 20 years of age and older were more likely to rely on a community-based health care provider and to have been rehospitalized for DM (70.3% vs 56.2%) as well as for other reasons (51.6% vs 25.7%). The older patients were less likely to be up-to-date with their ophthalmologic examinations and, if using insulin exclusively, to test their blood glucose level regularly. However, they were more likely to check their feet regularly.

Table 4. 
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Adolescents and Young Adults*
Self-reported Health Care Behaviors, Health Care Use, and Effect of Diabetes Mellitus (DM) in Adolescents and Young Adults*


These self-reported data confirm published reports from elsewhere that indicate a significant effect of DM in the lives of young people and their families.1,3,4,7,8 To our knowledge, however, the current report is the first to emerge from a population sample including significant numbers of underserved minority and impoverished children. The patients we surveyed were generally similar to the US population in terms of socioeconomic indicators: about 40% of the African American children and 27% of the Hispanic children in our study were living in poverty compared with 36% of African American children and 34% of Hispanic children nationwide during 1999.15 Among those participants who were 20 years of age and older, 78% were at least high school graduates compared with 85% of adults aged between 18 and 25 years in the United States in 1999.16 In the current sample, including slightly more than average numbers of impoverished African American patients and slightly fewer impoverished Hispanic patients, we found that having a severe chronic disease carries an important burden overall for pediatric patients and the families with whom they live. When more non-Hispanic white subjects are interviewed, we may be able to address these issues across a broader range of socioeconomic strata.

Fully two thirds of those with a type 2 DM phenotype were using insulin exclusively at the time they were interviewed. This likely reflects the US Food and Drug Administration guidelines for treatment of pediatric DM, which until very recently endorsed only insulin, as well as the inescapable fact that many years are likely to elapse between changes in "official" treatment guidelines and their adoption by the general medical community. During the period when these patients were interviewed, a change in diagnostic practice was occurring as the problem of type 2 DM in youth became more generally recognized.17 The medical records–based algorithm we used to identify subjects with type 2 DM in the CCDR database is admittedly crude; however, it is more likely to misclassify patients with type 2 DM as having type 1 DM than vice versa. While the extent of oral treatment for youth-onset type 2 DM is unknown, anecdotal evidence that young patients with type 2 DM "take themselves off insulin" is widespread and has been cited by clinicians as a major obstacle in establishing a trusting therapeutic relationship.

Irrespective of their phenotype, many patients reported adverse self-management and health behaviors. The rate of smoking among these young patients with DM is considerably higher than national data showing that 7.4% of African American 12th-grade students, and 13.6% of Hispanics, acknowledged smoking cigarettes in 1998.18 Nine percent of patients who were using insulin exclusively tested their blood glucose levels less frequently than once a day. Fully 20% of the young patients we talked with did not inspect their feet, despite the universal recommendation that this be encouraged as a daily activity for patients with DM. A dilated ophthalmologic examination is recommended annually for those who have had DM for longer than 5 years or those attaining the age of 13 years.19 Almost all of the respondents reported seeing a physician at least once during the past year, although 18% had some difficulty obtaining health care. These findings are likely to reflect financial as well as emotional obstacles in dealing with the very real risks associated with having DM.

Young adults are particularly apt to face barriers to adequate health care. Indeed, the current data provide evidence of a marked change in access to health care as children moved into young adulthood: 16% of those 20 years of age or older had no health insurance compared with 6% of those younger than 20 years of age. Those who were 20 years of age and older were less likely to see a physician at a tertiary care facility. Few comparable data are available; however, in an adult sample in Oklahoma, 16% of those with DM and 10% of a matched sample of nondiabetic adults reported that they had no health insurance in 1990 through 1995.20 The long-term follow-up of patients with youth-onset type 1 DM at the University of Pittsburgh (duration, 29 years) found that 7% had no health insurance, and 6.8% reported problems getting medical care.21 Obtaining care from a nonspecialist was a risk factor for developing long-term complications in that cohort, particularly for microvascular disease.22 It is distressing that young people with DM are largely excluded from health insurance at age 18 years or soon thereafter. The result is that, all too often, they are unable to obtain comprehensive health care until the development of life-threatening chronic complications. This study documents a decline in the quality of DM care as reported by young adult patients, and provides the impetus for researching systems to improve the transition for young adults as they move from pediatric to adult care providers.

There are several limitations to the current study. First, the citywide CCDR, on which this report is based, did not include non-Hispanic white children until recently, and then only those with onset after 1991. Since the primary source of case ascertainment was hospital records, patients who did not reach medical attention would have been missed; these may have been those with less severe onset symptoms. However, patterns of health care for Chicago residents demonstrate that almost all of new-onset children with DM were hospitalized throughout the 1990s. Another difficulty is that consistent clinical data at onset were unavailable on all patients, a result of using medical records from 37 different institutions. Classifying the diabetic phenotype of subjects on the basis of medical records information alone does not permit analysis of the role of pubertal stage in the development of type 2 DM, nor the role of obesity as a potential factor in the "acceleration" of autoimmune beta-cell destruction.23

Self-reported data carry some inherent limitations, including recall bias and the potential misattribution of cause. In this article, the analysis assesses only specific, limited aspects of the effects of DM on a patient's life, namely, perceived exclusion from sports, school activities, insurance, or employment. Individual personality characteristics can make it more or less likely that blame for adverse life events will be credited to the presence of a chronic disease. However, patient and family perceptions are key elements in the long-term adjustment of young people to having a chronic disease. Self-reporting of income presents another limitation: 24% of respondents declined to provide information on household income. These families with missing data, if systematically different, could skew the overall findings. For this reason, we also examined educational attainment and health insurance status as proxies for social class. Finally, those who were traced and interviewed for this report may differ from those who were not interviewed in ways that have not been accounted for.

We anticipate that in future we will be able to more clearly document health disparities between minority and non-Hispanic white young patients as the numbers of whites accrue in the CCDR. Documenting the characteristics and outcomes of diabetes in diverse groups of patients can serve as a basis for the rational targeting of health care services and will serve to justify a new approach to treatment protocols for noninsulin-requiring diabetic young people. Additional priorities for future research include documentation of the perceived discriminatory practices reported here with respect to employment, school-related activities, and child care.


Corresponding author and reprints: Rebecca Lipton, PhD, Department of Pediatrics, University of Chicago Pritzker School of Medicine, 5841 S Maryland Ave, M/C 1027, Chicago, IL 60637 (e-mail: lipton@uchicago.edu).

Accepted for publication March 15, 2003.

This study was supported by grants 1R29DK44752 and 1R01DK44752 from the National Institutes of Health, Bethesda, Md.

We gratefully acknowledge the patients and families who have participated in the CCDR. We also acknowledge the members of the Chicago Childhood Diabetes Study Group for their assistance in our efforts to achieve complete case ascertainment.

What This Study Adds

To our knowledge, no other study in the literature has specifically examined pediatricians' perceptions of their preparation to treat youths with type 2 diabetes mellitus. This study found that most pediatricians questioned were in favor of following the ADA clinical practice recommendations; however, only about 15.3% believed they were well prepared and confident to treat and counsel children and adolescents with type 2 diabetes mellitus.

Songer  TJ Costs and insurance coverage associated with insulin-dependent diabetes mellitus (IDDM): a case-control study.  Diabetes. 1986;3533PubMedGoogle ScholarCrossref
Chin  MHCook  SJin  L  et al.  Barriers to providing diabetes care in community health centers.  Diabetes Care. 2001;24268- 274PubMedGoogle ScholarCrossref
Glasgow  AMWeissburg-Benchell  JTynan  WD  et al.  Readmission of children with diabetes to a children's hospital.  Pediatrics. 1991;8898- 104PubMedGoogle Scholar
Lipton  RBDrum  MLZierold  KKlein-Gitelman  MKohrman  A Hospitalization after diagnosis varies by gender and insurance status among minority children with type 1 diabetes mellitus.  Pediatr Diabetes. 2002;316- 22Google ScholarCrossref
Cowie  CCPort  FKWolfe  RASavage  PJMoll  PPHawthorne  VM Disparities in incidence of diabetic end-stage renal disease according to race and type of diabetes.  N Engl J Med. 1989;3211074- 1079PubMedGoogle ScholarCrossref
Gay  ECHamman  RFCarosone-Link  PJ  et al.  Colorado IDDM registry: lower incidence of IDDM in Latinos: comparison of disease characteristics and care patterns in a biethnic community.  Diabetes Care. 1989;12701- 708PubMedGoogle ScholarCrossref
Auslander  WFThompson  SDreitzer  DWhite  NHSantiago  JV Disparity in glycemic control and adherence between African-American and Caucasian youths with diabetes.  Diabetes Care. 1997;201569- 1575PubMedGoogle ScholarCrossref
David  RJCollins  JW Differing birth weights among infants of US-born blacks, African-born blacks, and US-born whites.  N Engl J Med. 1997;3371209- 1214PubMedGoogle ScholarCrossref
Palta  MShe  GAllen  CKlein  RD'Alessio  D Longitudinal patterns of glycemic control and diabetes care from diagnosis in a population-based cohort with type I diabetes.  Am J Epidemiol. 1996;144954- 961PubMedGoogle ScholarCrossref
Lipton  RGood  GMikhailov  TFreels  SDonoghue  E Ethnic differences in mortality from type 1 diabetes mellitus among young people ages 1-24.  Pediatrics. 1999;103952- 956PubMedGoogle ScholarCrossref
US Department of Commerce, US Bureau of the Census, 2000 Population Report.  Washington, DC US Bureau of the Census2001;
Lipton  RKeenan  HOnyemere  KUFreels  S Incidence and onset features of diabetes in African American and Latino children, Chicago, 1985-1994.  Diabetes Metab Res Rev. 2002;18135- 142PubMedGoogle ScholarCrossref
Sekar  CCDeming  WE On a method of estimating birth and death rates and the extent of registration.  J Am Stat Assoc. 1949;1949101- 115Google ScholarCrossref
Word  LPerkins  R Building a Spanish Surname List for the 1990's: A New Approach to an Old Problem.  Washington, DC Population Division, US Bureau of the Census March1996;
US Bureau of the Census, March Current Population Survey, Current Population Reports, Consumer Income.  Washington, DC US Bureau of the Census1999;Series P-60
US Bureau of the Census, October Current Population Survey, Current Population Reports, Education Statistics.  Washington, DC US Bureau of the Census1999;
Jones  KL Non–insulin-dependent diabetes in children and adolescents: the therapeutic challenge.  Clin Pediatr (Phila). 1998;37103- 110PubMedGoogle ScholarCrossref
Johnston  LDO'Malley  PMBachman  JG Table 2-2. National Survey Results on Drug Use From the Monitoring the Future Study, 1975-1998.  Bethesda, Md National Institutes of Health1999;NIH publication 99-4660.
National Institute of Diabetes and Digestive and Kidney Diseases, Your Guide to Diabetes: Type 1 and Type 2.  Washington, DC National Institute of Diabetes and Digestive and Kidney Diseases October2001;NIDDK publication 02-4016.
Valdemanis  VSmith  DWPage  MR Productivity and economic burden associated with diabetes.  Am J Public Health. 2001;91129- 130PubMedGoogle ScholarCrossref
Zgibor  JCSonger  TJKelsey  SF  et al.  The association of diabetes specialist care with health care practices in patients with type 1 diabetes.  Diabetes Care. 2000;23472- 476PubMedGoogle ScholarCrossref
Zgibor  JCSonger  TJKelsey  SFDrash  ALOrchard  TJ Influence of health care providers on the development of diabetes complications: long-term followup from the Pittsburgh Epidemiology of Diabetes Complications Study.  Diabetes Care. 2002;251584- 1590PubMedGoogle ScholarCrossref
Wilkin  TJ For debate—the accelerator hypothesis: weight gain as the missing link between type 1 and type 2 diabetes.  Diabetologia. 2001;44914- 922PubMedGoogle ScholarCrossref