Model 1 was identical for all outcomes and included age, sex, duration of diabetes, marital status, income, educational level, and country of birth. Model 2 was further adjusted for smoking, hemoglobin A1c level, estimated glomerular filtration rate, diabetes treatment, and body mass index. Model 3 was further adjusted for albuminuria, heart failure, myocardial infarction, stroke, stage 5 chronic kidney disease, and cancer at baseline.
Model 1 was identical for all outcomes and included age, sex, duration of diabetes, marital status, income, educational level, and country of birth. Model 2 was further adjusted for smoking, hemoglobin A1c level, estimated glomerular filtration rate, diabetes treatment, and body mass index. Model 3 was further adjusted for albuminuria, heart failure, myocardial infarction, stroke, and stage 5 chronic kidney disease.
eTable 1. Baseline Characteristics According to Income Quintile
eTable 2. Baseline Characteristics According to Education
eTable 3. Baseline Characteristics According to Country of Birth
eTable 4. Baseline Characteristics According to Marital Status
eTable 5. Mean Number of Annual Registrations in the National Diabetes Register, According to Socioeconomic Factors
eFigure 1. Hazard Ratios for Diabetes-Related Death in the Fully Adjusted Model
eFigure 2. Hazard Ratios for Death From Prostate Cancer
eFigure 3. Hazard Ratios for Death From Breast Cancer
eFigure 4. Hazard Ratios for Death From Lung and Tracheal Cancer
eFigure 5. Hazard Ratios for Death From Gastrointestinal Cancer, Excluding Cancers of the Liver, Gallbladder, Gall Tracts, and Pancreas
eFigure 6. Hazard Ratios for Death From Kidney and Urinary Tract Cancers
eFigure 7. Hazard Ratios for Death From Cancer of the Liver, Pancreas, Gall Tract, and Gallbladder
Customize your JAMA Network experience by selecting one or more topics from the list below.
Rawshani A, Svensson A, Zethelius B, Eliasson B, Rosengren A, Gudbjörnsdottir S. Association Between Socioeconomic Status and Mortality, Cardiovascular Disease, and Cancer in Patients With Type 2 Diabetes. JAMA Intern Med. 2016;176(8):1146–1154. doi:10.1001/jamainternmed.2016.2940
Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
The association between socioeconomic status and survival based on all-cause, cardiovascular (CV), diabetes-related, and cancer mortality in type 2 diabetes has not been examined in a setting of persons with equitable access to health care with adjustment for important confounders.
To determine whether income, educational level, marital status, and country of birth are independently associated with all-cause, CV, diabetes-related, and cancer mortality in persons with type 2 diabetes.
Design, Setting, and Participants
A study including all 217 364 individuals younger than 70 years with type 2 diabetes in the Sweden National Diabetes Register (January 1, 2003, to December 31, 2010) who were monitored through December 31, 2012, was conducted. A Cox proportional hazards regression model with up to 17 covariates was used for analysis.
Main Outcomes and Measures
All-cause, CV, diabetes-related, and cancer mortality.
Of the 217 364 persons included in the study, mean (SD) age was 58.3 (9.3) years and 130 839 of the population (60.2%) was male. There were a total of 19 105 all-cause deaths with 11 423 (59.8%), 6984 (36.6%), and 6438 (33.7%) CV, diabetes-related, or cancer deaths, respectively. Compared with being single, hazard ratios (HRs) for married individuals, determined using fully adjusted models, for all-cause, CV, and diabetes-related mortality were 0.73 (95% CI, 0.70-0.77), 0.67 (95% CI, 0.63-0.71), and 0.62 (95% CI, 0.57-0.67), respectively. Marital status was not associated with overall cancer mortality, but married men had a 33% lower risk of prostate cancer mortality compared with single men, with an HR of 0.67 (95% CI, 0.50-0.90). Comparison of HRs for the lowest vs highest income quintiles for all-cause, CV, diabetes-related, and cancer mortality were 1.71 (95% CI, 1.60-1.83), 1.87 (95% CI, 1.72-2.05), 1.80 (95% CI, 1.61-2.01), and 1.28 (95% CI, 1.14-1.44), respectively. Compared with native Swedes, HRs for all-cause, CV, diabetes-related, and cancer mortality for non-Western immigrants were 0.55 (95% CI, 0.48-0.63), 0.46 (95% CI, 0.38-0.56), 0.38 (95% CI, 0.29-0.49), and 0.72 (95% CI, 0.58-0.88), respectively, and these HRs were virtually unaffected by covariate adjustment. Hazard ratios for those with a college/university degree compared with 9 years or less of education were 0.85 (95% CI, 0.80-0.90), 0.84 (95% CI, 0.78-0.91), and 0.84 (95% CI, 0.76-0.93) for all-cause, CV, and cancer mortality, respectively.
Conclusions and Relevance
Independent of risk factors, access to health care, and use of health care, socioeconomic status is a powerful predictor of all-cause and CV mortality but was not as strong as a predictor of death from cancer.
Socioeconomic status is a fundamental element of public and individual health,1 indicating an individual’s resources and ability to thrive and survive. Socioeconomic measures commonly include race/ethnicity, income, educational level, and occupation. Other featured components are sex and marital status.
Socioeconomic status is generally inversely associated with the risk of cardiovascular (CV) disease and death2-5 as well as the risk of developing type 2 diabetes (referred to as diabetes for the study outcomes).6-8 Studies have demonstrated that mortality rates among individuals with diabetes vary according to neighborhood9,10 as well as individual income and educational level.11-13
However, access to medical care, varying risk factors, and comorbidities may contribute to socioeconomic differences in outcome. Sweden provides a setting in which socioeconomic status has a minimal influence on access to and use of health care services and comprehensive individual-level data on risk factors, as well as clinical and socioeconomic variables, in individuals with diabetes. Given these prerequisites, the present nationwide study aimed at examining how socioeconomic status relates to the adjusted risk of all-cause, CV, diabetes-related, and cancer mortality in diabetes.
Question How important is socioeconomic status in type 2 diabetes?
Findings In this nationwide study in Sweden, low socioeconomic status was associated with almost a 2-fold risk of all-cause, cardiovascular, and diabetes-related mortality but was not as strong as a predictor of death from cancer. Controlling for well-recognized risk factors did not eliminate the disparities; however, non-Western immigrants displayed a substantially lower risk of most outcomes.
Meaning Socioeconomic disparities in survival in type 2 diabetes are pronounced and may not be eliminated by conventional risk factor control.
The National Diabetes Register (NDR) in Sweden was launched in 1996 as a tool for quality control of care for persons with diabetes. Nurses and physicians in primary health care centers and hospital outpatient clinics report information on these patients. Reporting, which is carried out online or by direct data transmission from electronic medical records to the NDR, is done at least once a year and includes information on clinical characteristics. The NDR currently includes the vast majority of all Swedish individuals with diabetes.14,15
The present study was approved by the ethics committee of the University of Gothenburg. All individuals provided verbal informed consent before being included in the NDR, and data were deidentified.
We included 217 364 individuals younger than 70 years with diabetes who had at least 1 entry in the NDR between January 1, 2003, and December 31, 2010. We excluded individuals 70 years or older because income drops steeply after retirement, whereas educational level is unchanged. Without this age limit, a large proportion of the events would occur among the elderly, which would influence estimated hazard ratios (HRs). We intended to study a homogeneous cohort in terms of the association between socioeconomic status and mortality and therefore restricted the cohort to individuals of working age; the youngest age in our cohort was 18 years.
Type 2 diabetes was defined on the basis of epidemiologic data: participants were receiving treatment with diet with or without the use of oral hypoglycemic agents or treatment with insulin with or without the use of oral hypoglycemic agents. The latter category applied only to participants 40 years or older at diagnosis of diabetes.
Data on individual income in Swedish kronor (1 US dollar equals 8 Swedish kronor), highest educational level, country of birth, marital status, and occupation were obtained from the Longitudinal Integration Database for Health Insurance and Labour Market Studies (http://www.scb.se/en_/Services/Guidance-for-researchers-and-universities/SCB-Data/Longitudinal-integration-database-for-health-insurance-and-labour-market-studies-LISA-by-Swedish-acronym/#). Education was stratified into lower (≤9 years, with 9 years being the length of compulsory education in Sweden), intermediate (10-12 years, upper secondary), and higher (college/university). Income was stratified into quintiles. Country of birth was categorized into the following groups: Sweden, high-income Europe, low-income Europe, non-Western, and Nordic. Marital categories were single (never married or had a registered partner), married or had a registered partner, divorced, or widowed.
Microalbuminuria was defined as 2 positive tests from 3 samples taken within 1 year, with an albumin to creatinine ratio of 3 to 30 mg/mmol or urinary albumin level of 20 to 200 µg/min or 20 to 300 mg/L. Macroalbuminuria was defined as an albumin to creatinine ratio greater than 30 mg/mmol or urinary albumin level greater than 200 µg/min or greater than 300 mg/L. Estimated glomerular filtration rate was estimated with the Modified Diet in Renal Disease equation.16,17 Systolic blood pressure was the mean value of 2 supine readings after at least 5 minutes of rest with a cuff of appropriate size. Diabetes treatment categories were diet and lifestyle modification, oral hypoglycemic agents, insulin only, or insulin with oral hypoglycemic agents. Use of statins, antihypertensives, and aspirin was dichotomized. Smoking was coded as positive if the individual was a current smoker.
Data on comorbidities and events were collected at baseline and during follow-up by linking the NDR to the Inpatient Register and the Cause of Death Register, both of which are maintained by the National Board of Health and Welfare. The Inpatient Register includes mandatory information on all principal and secondary hospital discharge diagnoses and has had near-complete coverage since 1987. Codes from the International Classification of Diseases are used to classify diagnoses in the Inpatient Register, which has been validated.18
The following International Classification of Diseases, Ninth Revision (ICD-9) and International Statistical Classification of Diseases, Tenth Revision (ICD-10) codes were collected: coronary heart disease, 410-414 (ICD-9) and I20-I25 (ICD-10); acute myocardial infarction, 410 (ICD-9) and I21 (ICD-10); stroke, 431-434, 436 (ICD-9), and I61-I64 (ICD-10); hospitalization for heart failure, 428 (ICD-9) and I50 (ICD-10); and atrial fibrillation, 427D (ICD-9) and I48 (ICD-10). For renal dialysis and transplant, the following codes were used: V42A, V45B, V56A, and V56W (ICD-9) and Z94.0, Z49, and Z99.2 (ICD-10). Stage 5 chronic kidney disease was defined as the need for renal dialysis or renal transplant or estimated glomerular filtration rate less than 15 mL/min/1.73 kg.
The outcomes assessed were all-cause, CV (ICD-10 codes I00-I99), and cancer mortality (ICD-10 codes C00-C97). Post hoc assessment included diabetes-related death (ICD-10 codes E10-E14) and the following cancers: prostate (ICD-10 code C61); breast (ICD-10 code C50); lung and tracheal (ICD-10 codes C33-C34); gastrointestinal (ICD-10 codes C15-C21); liver, pancreas, gall bladder, and biliary tract (ICD-10 codes C22-C26); and kidney and urinary system (ICD-10 codes C64-C68).
All individuals were monitored from baseline until an event or death through December 31, 2012. Median follow-up was 5.6 years.
Crude mortality rates are described as events per 1000 person-years. A multivariable Cox proportional hazards regression model was used to perform survival analyses. We computed 3 sequentially adjusted models for each outcome. Model 1 was identical for all outcomes and included age, sex, duration of diabetes, marital status, income, educational level, and country of birth. Covariate adjustment in the remaining models differed slightly to suit the outcome.
For all-cause mortality, model 2 was additionally adjusted for smoking, hemoglobin A1c, estimated glomerular filtration rate, diabetes treatment, and body mass index. Model 3 was also adjusted for albuminuria, heart failure, myocardial infarction, stroke, stage 5 chronic kidney disease, and cancer at baseline. Adjustments to models 2 and 3 for CV mortality were identical to all-cause mortality except that previous cancer was not included. Adjustments to models 2 and 3 for cancer mortality were identical to all-cause mortality.
For diabetes-related death, we computed one model that included the same covariates as model 3 for CV mortality. Similarly, for specific cancers, only one model was used, and it was adjusted for the same covariates as model 3 for overall cancer mortality. Competing risks were not taken into consideration since we were interested in estimating only cause-specific hazards.
Characteristics of the population are presented in Table 1 and eTables 1 to 4 in the Supplement. Of the 217 364 persons included in the study, the mean (SD) age was 58.3 (9.3) years and 130 839 of the population (60.2%) was male. Individuals who were married were older, had higher educational levels, were more frequently born outside of Sweden, and had more coexisting conditions compared with those who were single. Individuals in quintile 5 (highest income) had more favorable characteristics than those in quintile 1 (lowest income). Individuals from non-Western countries (1699 [9.7%] originated from Latin America and the Caribbean; 2902 [16.5%] from East or South Asia; 10 506 [59.7%] from the Middle East or North Africa; and 2482 [14.1%] from Sub-Saharan Africa) were younger than native Swedes (mean [SD], 52.1 [9.8] vs 58.8 [9.2] years). Non-Western immigrants were approximately 6 years younger at diagnosis of diabetes and had lower income, higher educational levels, and a higher prevalence of albuminuria. As presented in eTable 5 in the Supplement, there were no noteworthy differences regarding the number of entries in the register per year regarding socioeconomic categories.
There were a total of 19 105 all-cause deaths with 11 423 (59.8%), 6984 (36.6%), and 6438 (33.7%) CV, diabetes-related, or cancer deaths, respectively. Unadjusted mortality rates are presented in Table 2. Mortality increased gradually with declining income. In quintile 5, there were 8.92 deaths per 1000 person-years (highest income) compared with 18.33 deaths per 1000 person-years in quintile 1 (lowest income) with comparable age distribution. The age of individuals with 10 to 12 years of education was comparable to the age of those with a college/university degree. The latter category had lower rates for all outcomes. Persons from high-income Europe, low-income Europe, Sweden, and Nordic countries were of similar ages. Individuals from Nordic countries had higher incidence rates for all outcomes. The mean ages of married and divorced individuals were nearly the same (59.3 vs 59.2 years). The mortality rate was 13.00 deaths per 1000 person-years among married individuals compared with 18.82 deaths per 1000 person-years among those who were divorced.
Figure 1 presents the HRs (95% CIs) for all-cause mortality. Being single was the reference group for marital status. The HR for individuals who were married was 0.73 (95% CI, 0.70-0.77) in model 3, which was similar to the estimates in models 1 (HR, 0.70 [95% CI, 0.67-0.73]) and 2 (HR, 0.71 [95% CI, 0.67-0.74]). Being widowed was associated with a slightly elevated risk in model 1 (HR, 1.08 [95% CI, 1.02-1.16]), but the statistical significance was eliminated in models 2 (HR, 1.02 [95% CI, 0.95-1.10]) and 3 (HR, 1.05 [95% CI, 0.97-1.14]). The HR in model 3 for individuals who were divorced was 0.94 (95% CI, 0.88-0.99). Those with the highest income (quintile 5) served as the reference group for income. The HR for death increased gradually with declining income. In the fully adjusted model (model 3), the HRs for individuals in quintiles 1 and 2 were 1.71 (95% CI, 1.60-1.83) and 1.68 (95% CI, 1.58-1.80), respectively.
The effect of education was robust against covariate adjustment. Compared with having 9 years or less of education, HRs for individuals with 10 to 12 years of education and those with a college/university degree were 0.90 (95% CI, 0.86-0.93) and 0.85 (95% CI, 0.80-0.90), respectively (model 3). Similarly, the association between country of birth and death withstood covariate adjustment. In model 3, HRs for individuals born in low-income European or non-Western countries were 0.72 (95% CI, 0.62-0.83) and 0.55 (95% CI, 0.48-0.63), respectively.
For CV death, we noted the same associations as for all-cause mortality, but the differences were slightly more pronounced. The HR for married individuals compared with those who were single, determined using the fully adjusted model, was 0.67 (95% CI, 0.63-0.71). Individuals in the 2 lowest income quintiles had an almost 90% elevated risk of CV death compared with those in the highest income quintile. The HR for the lowest vs highest income quintile was 1.87 (95% CI, 1.72-2.05) and for the second lowest vs highest income quintile was 1.87 (95% CI, 1.72-2.03). Hazard ratios for persons with a college/university degree compared with those having 9 years or less education was 0.84 (95% CI, 0.78-0.91). The HR for non-Western immigrants compared with native Swedes was 0.46 (95% CI, 0.38-0.56) (Figure 2).
Hazard ratios for diabetes-related death in the fully adjusted model are presented in eFigure 1 in the Supplement. We noted associations similar to those of all-cause mortality. Using the same reference groups for all-cause mortality, we found that individuals who were married displayed an HR of 0.62 (95% CI, 0.57-0.67). Being in the lowest income quintile yielded an HR of 1.80 (95% CI, 1.61-2.01). Non-Western immigrants displayed an HR of 0.38 (95% CI, 0.29-0.49), which was the lowest observed in the present study.
There was no statistically significant association between marital status and death from any cancer (Figure 3). Individuals in income quintiles 1, 2, and 3 had approximately the same elevated risk (30% higher) of death from cancer compared with those in the highest income quintile. The HR for persons with a college/university degree was 0.84 (95% CI, 0.76-0.93) compared with having 9 years or less of education. A smaller but significant reduction in risk was also noted for individuals with 10 to 12 years of education. Compared with native Swedes, persons from non-Western countries had an HR of 0.72 (95% CI, 0.58-0.88).
Hazard ratios are presented in eFigure 2 to eFigure 7 in the Supplement. Compared with men who were single, those who were married displayed an HR of 0.67 (95% CI, 0.50-0.90) for death from prostate cancer; the HR in men with 10 to 12 years of education was 0.77 (95% CI, 0.60-0.99). Remaining socioeconomic status variables were not associated with a risk of prostate cancer (eFigure 2 in the Supplement).
None of the socioeconomic status variables were associated with the risk of breast cancer. However, there was a tendency for increasing risk with declining income (eFigure 3 in the Supplement) in the judgment of point estimates.
Being divorced, having a low income, or being born in nearby Nordic countries were each associated with higher risks for lung and tracheal cancer. The HR for individuals with a college/university degree compared with those who had 9 years or less of education was 0.58 (95% CI, 0.44-0.76) (eFigure 4 in the Supplement).
Compared with the highest income quintile, all other income groups displayed a 30% to 50% elevated risk of death from gastrointestinal cancers. Being well educated appeared to be associated with a lower risk of the outcome, although statistical significance was absent. The HR for non-Western immigrants compared with Swedish natives was 0.54 (95% CI, 0.31-0.93) (eFigure 5 in the Supplement).
Socioeconomic status was not associated with death from kidney and urinary tract cancer except for individuals in the second highest income quintile, who displayed a lower risk than persons in the highest income quintile (eFigure 6 in the Supplement). Low income was associated with an elevated risk of death from cancer in the liver, pancreas, or biliary tract (eFigure 7 in the Supplement).
In this nationwide study of 217 364 individuals with diabetes, we demonstrated that components of socioeconomic status are independent indicators of all-cause, CV, diabetes-related, and cancer mortality. The effect on cancer mortality was much less pronounced compared with the other factors, indicating that socioeconomic status probably has a stronger influence on CV outcomes. In general, low compared with high income was associated with almost twice the risk of all-cause, CV, and diabetes-related mortality and a 30% elevated risk of overall cancer mortality. Being married was associated with a 30% to 40% reduced risk of all-cause, CV, and diabetes-related mortality but had no effect on death from cancer with the exception of prostate cancer mortality, for which married men displayed a 33% lower risk than single men. Immigrants displayed a 30% to 60% lower risk of all-cause, CV, and diabetes-related death; non-Western immigrants also had a reduced risk (approximately 30%) of death from cancer.
The implications of the present study proceed from its prerequisites. The Swedish health care system is arguably one of the most equitable worldwide. All citizens have equal access to all aspects of the health care system, and individual costs represent a fraction of actual costs; being hospitalized costs approximately $10 per day regardless of level of care or the type and number of interventions and examinations carried out. In addition, a previous study by our group19 showed that immigrants received evidence-based treatments earlier than did native Swedes. Thus, health care access bias should be a minor issue.19 Furthermore, we made rigorous adjustments for potential confounders. The results are likely to present the influence of socioeconomic variables independent of their effect on health care use and access.
The fact that controlling for risk factors and covariates did not eliminate the effect of socioeconomic indicators does not imply that risk factor control is ineffective in reducing these disparities. In fact, strict risk factor control might be the most accessible means for reducing these disparities. It may be wise to initiate preventive strategies for well-recognized CV risk factors. Because of some overlap in the causes, it is likely that this intervention would also mitigate disparities in cancer mortality. Regardless, it is unlikely that such measures will eliminate existing inequalities. Health policy and societal reforms will be needed.
Previous studies9,11 have shown that low socioeconomic status is associated with mortality from diabetes, but these studies did not adjust for important covariates, did not have access to detailed individual-level socioeconomic data, or were conducted in a setting in which access to health care is inequitable. The present study overcame these obstacles and assessed cancer mortality to show the greater effect of socioeconomic factors on CV outcomes.
The mechanisms by which socioeconomic status exert an influence have been extensively studied.20-23 Absence of socioeconomic privilege is coupled with psychosocial stress, unemployment, financial difficulties, unhealthy habits, health obstacles, living in an at-risk neighborhood, lack of social support, and insufficient cohesion.
Non-Western immigrants displayed a paradoxical association between socioeconomic status and survival. Compared with Swedish natives, non-Western immigrants had a lower income and higher educational level but a lower risk of virtually all outcomes. These associations persisted despite rigorous adjustments, which indicates that there are as-yet unmeasured (protective) factors that put immigrants at lower risk than their Swedish compatriots. This phenomenon has been termed the healthy migrant effect,24 suggesting that individuals who migrate from non-Western to Western countries represent a select and stronger subgroup of their native country. Our study shows that this healthy migrant effect is substantial (HRs from 0.38 to 0.72 ), applies to a range of outcomes, and, perhaps even more intriguing, is not affected by adjustment for well-recognized risk factors and comorbidities.
The study has limitations. Because of its observational design, we cannot exclude residual confounding. Although the regression methods and number and quality of included variables restrict residual confounding, they do not correct for confounding factors that were not considered. We did not have access to data on alcohol consumption, and information regarding smoking was limited to a dichotomous variable. Furthermore, the epidemiologic definition of type 2 diabetes makes it possible that persons with autoimmune (ie, type 1) diabetes may have been included in the cohort, provided that they had disease onset after age 40 years. However, we compared this definition with the physician’s classification of diabetes type (which is available in the NDR), and the epidemiologic definition was concordant with the physician’s in 95% of the cases.
In this nationwide study in Sweden, where access to and use of health care resources is equitable and affordable and management is well developed, low socioeconomic status was associated with a 2-fold risk of all-cause, CV, and diabetes-related mortality. A significant but less pronounced increased risk of death from cancer was also noted. Controlling for conventional risk factors does not eliminate these disparities, but socioeconomically tailored management and aggressive treatment are likely to reduce them.
Corresponding Author: Araz Rawshani, MD, PhD, National Diabetes Register, Centre of Registers, Västra Götaland Region, SE-41345 Gothenburg, Sweden (firstname.lastname@example.org).
Accepted for Publication: May 2, 2016.
Published Online: June 27, 2016. doi:10.1001/jamainternmed.2016.2940.
Author Contributions: Drs Rawshani and Gudbjörnsdottir had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Rawshani.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Rawshani.
Obtained funding: Rawshani, Svensson, Gudbjörnsdottir.
Administrative, technical, or material support: Rawshani, Svensson, Gudbjörnsdottir.
Study supervision: Zethelius, Eliasson, Rosengren, Gudbjörnsdottir.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was financed by the Swedish National Diabetes Register (NDR), which is funded by the Swedish Association of Local Authorities and Regions. The study was also financed by grants from the Swedish government under the agreement with county councils for financial support of research and education of medical practitioners, the Swedish Heart and Lung Foundation, Diabetes Wellness, the Swedish Research Council (SIMSAM), the Swedish Council for Working Life and Social Research, and Diabetesfonden.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The results and views of the present study represent those of the authors and not necessarily those of the Swedish Medical Products Agency, which employs Dr Zethelius.
Additional Contributions: We thank the various regional NDR coordinators, as well as contributing nurses, physicians, and individuals with diabetes. The Swedish Diabetes Association and the Swedish Society of Diabetology support the NDR.