Changes in HbA1c Between 2011 and 2017 in Germany/Austria,Sweden,and the United States: A Lifespan Perspective

Introduction

As the incidence of type 1 diabetes (T1D) continues to increase globally, it is essential that people living with the condition have access to high-quality care, sufficient resources, and appropriate education and support to meet glycemic targets and enjoy improved quality of life. ^{1 –3} Initiated in 1983, the landmark Diabetes Control and Complications Trial (DCCT) demonstrated that people who use intensive insulin therapy can achieve glucose levels close to target and significantly reduce their risk for chronic complications, including neuropathy, nephropathy, and retinopathy. ⁴ In 1989, European leaders, under the auspices of the World Health Organization and International Diabetes Federation, met in St. Vincent, Italy, and established universally accepted standards in diabetes care in Europe (St. Vincent Declaration) and subsequently evaluated progress toward improved care and outcomes. ^5,6 Current clinical guidelines in both the United States and Europe recommend a hemoglobin A1c (HbA1c) of less than 7% (53 mmol/mol) for most children and adults with T1D, while the UK recommends ≤6.5% (48 mmol/mol), as does Sweden, but only in pediatrics, beginning in 2017. ^{7 –10}

Following the DCCT and the St. Vincent Declaration, and in parallel with the advent of electronic health records, clinicians and epidemiologists established national and international data registries to track clinical outcomes, including HbA1c. Over the past two decades, these registry data have been used for benchmarking and to inform and evaluate the effect of quality improvement initiatives in T1D care, particularly in European countries such as Germany, Austria, and Sweden. ^{11 –14} Children and adults with T1D who live in these countries enjoy universal access to health care coverage and low out-of-pocket cost for diabetes care.

Iterative analyses have revealed that adolescents and young adults with T1D, in particular, face unique developmental and metabolic barriers to achieving glycemic targets, and past studies utilizing registry data suggest that the rise in HbA1c from early childhood into young adulthood appears to be consistent across countries where data are available. ^{15 –19} Further, within country and between country disparities in glycemic outcomes have been noted among high-income nations. ^{17 –20} These findings have informed targeted quality improvement initiatives in Europe.

In the United States, access to health care is variable, often linked to employment, and the state where a person lives determines access to public insurance benefits. ²¹ The number of uninsured Americans has been increasing since 2016, with an estimated 9.2% or 29.6 million having no health coverage in 2019. In states that did not expand Medicaid under the Affordable Care Act, rates of uninsured individuals were higher, including in populous states such as Florida (13.2% uninsured) and Texas (18.4% uninsured). The percentage of uninsured children under the age of 19 in the United States was 5.7% in 2019. ²² Even when individuals in the United States carry public or private insurance, eligibility requirements and coverage for diabetes technology is variable based on state of residence or policy type. ²³

Clinical data in the United States are often held as proprietary by clinical practices, ²⁴ with the possibility to benchmark T1D data first becoming available only in 2010 with the creation of the T1D Exchange Clinic Network and Registry (T1DX). ^25,26 This registry included a sample of people with T1D who received care at predominantly academic medical centers, potentially producing a curated sample not generalizable to the wider U.S. population of people with T1D. Analyses of T1D Exchange data in 2012 and 2018 showed that most children and adults with T1D were not meeting recommended targets, and despite increased uptake of diabetes technology and therapeutics, glycemic outcomes worsened between the two time points. ¹⁶

Comparing international data provides the opportunity to learn from countries that are achieving better clinical outcomes and adopting effective systems and processes. The purpose of this study was to describe trends in HbA1c in people with T1D across the lifespan between 2011 and 2017 with data from the three registries: Registry data from the Prospective Diabetes Follow-Up Registry/Diabetes-Patienten-Verlaufsdokumentation (DPV) for Germany and Austria; the Swedish Pediatric Diabetes Quality Registry (SWEDIABKIDS) and the Swedish National Diabetes Register (NDR) for data from Sweden; and the T1D Exchange Clinic Registry data from the United States were included in the study.

Materials and Methods

Data collection and extraction were approved by the Institutional Review Boards at participating clinics in the DPV, SWEDIAB/NDR, and T1DX registries. Informed consent was obtained as directed by the relevant Institutional Review Board. The DPV was established in 1995 and is currently composed of 502 diabetes care centers in Germany and Austria, with each center responsible for obtaining approval from its Institutional Review Board. Few institutions participate from Switzerland and Luxembourg; therefore, both countries were excluded from this analysis. Data are collected prospectively by each center using a standardized electronic health record, the DPV software. Every 6 months, data are shared in an anonymized format with the University of Ulm for quality assurance (benchmarking) and patient-centered research. ¹¹

For Sweden, the NDR was established in 1996 and SWEDIABKIDS started in 2000. Registry results are available publicly online for real-time benchmarking, and participating centers can access their own data and utilize it for quality improvement. The registries include information on risk factors and complications. ^9,13 The NDR includes 90% of all adults with T1D, and more than 95% of Swedish children and adolescents with T1D are registered in the SWEDIABKIDS database. ⁹ Participation by patients/parents is voluntary, and all provide consent before their data are included in the registry.

The T1DX Clinic Registry was established in 2010 and comprised 81 U.S. pediatric and adult endocrine practices in 35 states. Written consent was obtained from adult participants, and parents/guardians provided consent for minors, and children aged ≥7 provide written assent. Data are obtained from medical charts and questionnaires completed by participants/parents. Anonymized data are shared with the JAEB Center for Health Research for storage and quality assurance. ²⁵

Participant data were included in the analysis if the patient belonged to one of the three registries and had a diagnosis of T1D. Mean HbA1c was calculated per year of age. HbA1c values that were obtained within 3 months of the diagnosis of T1D as well as those that were improbable, for example, less than 4% (20 mmol/mol) or greater than 20% (195 mmol/mol), were excluded from the analysis. Statistical analyses were performed using SAS 9.4 software (SAS Institute, Cary, NC).

Results

For 2011, data for 86,323 registry participants were included in the analysis; in 2017, 100,219 participants were included, with the DPV contributing 25,651 in 2011 and 29,442 in 2017; SWEDIABKIDS/NDR contributing 44,474 in 2011 and 53,690 in 2017; and T1DX contributing 16,198 in 2011 and 17,087 in 2017. Participant characteristics by registry are shown in Table 1.

Overall mean HbA1c by registry adjusted for age, sex, and diabetes duration are shown in Table 1. In 2011, mean overall HbA1c in participants from Germany/Austria was 7.8% (62 mmol/mol) and was 7.8% (62 mmol/mol) in 2017, demonstrating no change between the two time points (P = 0.48). In Sweden, mean HbA1c was 8.0% (64 mmol/mol) in 2011 and 7.6% (60 mmol/mol) in 2017, representing an absolute decrease in HbA1c of 0.4% (4 mmol/mol) (P < 0.001). In the United States, mean HbA1c was 8.1% (65 mmol/mol) in 2011 and 8.3% (67 mmol/mol) in 2017, representing an absolute increase of 0.2% (2 mmol/mol) in HbA1c (P < 0.001). The age group analyses revealed a significant decrease in HbA1c among adolescents aged 13 to <18 years in Germany/Austria, an increase among German/Austrian adults aged 25 and older, and no change for other age groups. In Sweden, there was a statistically significant decrease in HbA1c for people of all ages. In the United States, there was no change in HbA1c in children <6 years and adults aged 45 to <65 years, and a significant increase in mean HbA1c among all other age groups.

For data by age group by registry, see Table 2. These data are depicted visually by registry and by year (Fig. 1), with mean HbA1c data from across the lifespan for participants from all three registries overlaid for comparison in 2011 and 2017 (Fig. 2). The data from 2011 reveal that HbA1c values for children, adolescents, and young adults in Sweden and Germany/Austria were similar, with U.S. HbA1c values ∼0.5% higher. By 2017, Swedish HbA1c values in these age groups were approximately 0.5% lower than in Germany/Austria and were between 0.8% and 1.5% lower than in the United States, demonstrating a widening gap, particularly among older adolescents and young adults.

OPEN IN VIEWER

FIG. 1.

HbA1c by Age Group by Registry, 2011 and 2017. (A) DPV Data for Germany/Austria (B) SWEDIABKIDS/NDR Data for Sweded (C) T1DX Data for the United States. 2011 data are shown in blue and 2017 data are shown in red. HbA1c, hemoglobin A1c.

OPEN IN VIEWER

FIG. 2.

Changes in HbA1c Across the Lifespan, All Registries. (A) 2011 Data (B) 2107 Data. German/Austrian data are represented in burgundy (stars), Swedish data are represented in yellow (circles), U.S. data are represented in green (triangles). Circles/triangles/stars express HbA1c for each year of age by registry. The line and shaded area represent the LOESS fit +95% confidence interval.

The rise in HbA1c from early childhood through adolescence was similar across all three registries, with a lower baseline in young children in Sweden and a significantly higher baseline in the United States. Among adults, HbA1c values in 2011 were lowest in Germany/Austria and in the United States and highest in Sweden. In 2017, HbA1c values for middle-aged adults were lowest in Sweden and in the United States and essentially the same in all four countries for older adults.

Discussion

This study revealed that mean HbA1c increased in the United States between 2011 and 2017, decreased in Sweden, and did not change in Germany/Austria. Mean HbA1c was lowest at baseline in Germany/Austria and highest in the United States. The age group analyses revealed that HbA1c declined among adolescents aged 13 to <18 years between 2011 and 2017 in Germany/Austria, which is notable given the metabolic and developmental challenges in diabetes self-management faced by teenagers. The DPV data also revealed a slight increase in HbA1c among German/Austrian adults aged 25 and older.

In Sweden, there was a statistically significant decrease in HbA1c for people of all ages and, strikingly, the largest decrease in HbA1c was realized among children, adolescents, and young adults. In the United States, there was a trend toward lower mean HbA1c in children <6 years that was not statistically significant, with no change for adults aged 45 to <65 years and an increase in mean HbA1c among all other age groups, which was most pronounced among children, adolescents, and young adults. As noted above, while a widening gap among HbA1c was noted for children, adolescents, and young adults, HbA1c values begin to converge in middle age and were essentially the same across registries among older adults.

This study has important implications for clinicians, policymakers, and people living with T1D. The data reveal significant differences in HbA1c trajectories in these three registries, with static or improved outcomes in high-income countries in Europe and worsening outcomes in the United States. Previous analyses have examined HbA1c across the lifespan in the DPV and T1DX, ¹⁵ differences in clinical care and outcomes in young children in the T1DX and DPV registries, ²⁷ disparities in diabetes technology adoption and access, ^{28 –30} risk of severe hypoglycemia, ³¹ differences in BMI trajectories, ³² and coverage for emerging diabetes technology ³³ ; however, this study is unique because it is the first to include data from Sweden, a country with an established and successful history of quality improvement in diabetes care and improving T1D outcomes for people of all ages. ¹³ While a causal line cannot be drawn to explain the results of this study, several potential explanations should serve as the basis of future inquiry.

The long-term commitment to systematically collect, analyze, and share actionable registry data at a national level in Sweden, Germany, and Austria has contributed to improved glycemic outcomes, particularly among children and young adults with T1D. ^{9,11 –14,34} In Sweden, the purpose of the NDR and SWEDIABKIDS registries is to use data benchmarking to inform evidence-based diabetes care and support quality improvement initiatives at participating clinical sites. Quality improvement collaboratives (QICs) have been formed to standardize diabetes care and improve clinical processes. The activities of the SWEDIABKIDS pediatric QICs, which included 70% of children and adolescents with diabetes in Sweden, had a direct and significant impact on lowering mean HbA1c among pediatric patients and in reducing the gap between center mean HbA1c between 2010 and 2014 with a concomitant decrease in severe hypoglycemia. ^35,36

Data transparency is a hallmark of the Swedish registry, and site-specific differences among centers in outcomes in care can be easily ascertained online. It is of note that the introduction of lower HbA1c targets in pediatrics in Sweden in 2017 did not translate to lower HbA1c until 2018. ⁹ Despite the overall improvement seen in glycemic control over time in adults with T1D in Sweden, there are still differences between regions and between individual hospital clinics that need to be addressed. In addition, Sweden's universal health care system is nationally regulated, but locally administered, and this can give rise to unequal availability of diabetes devices (meters and pumps) based on where a person lives. ³⁷

The DPV in Germany and Austria was established in 1995 to collect pediatric data, and the effort was expanded to adult clinics in 2000. The DPV provides software to facilitate data collection, external benchmarking for participating centers, and access to a database that can be used to inform patient-centered research. ¹¹ DPV participation is voluntary, and analyses are reported in an anonymized format on a nationwide basis, but Regional Quality Circles allow for open discussion of results. This approach to benchmarking and quality improvement seeks to engender an atmosphere of trust and openness and reduce the risk that centers will stop sharing data out of fear of losing patients to other centers or incurring reputational harm. ³⁴ DPV data have been widely used to improve quality, standardize documentation and clinical practices, and inform international comparisons in Germany and Austria. ³⁸ Over the past decade, improvements in HbA1c have slowed, but the incidence of severe hypoglycemia has dropped, an indicator that is possibly more relevant for children and their families. ³⁹

A recent review summarized the strengths and challenges in pediatric diabetes care in Germany. ⁴⁰ With respect to strengths, diabetes education is well established and reimbursed in Germany; specialty training for pediatricians and diabetes nurses has been available for more than two decades; and psychologists with experience in pediatric diabetes are embedded into most large centers. Every citizen in Germany has health insurance, and all costs related to diabetes care, including pumps and sensors, are generally reimbursed. Intensive insulin therapy and carb counting are widely accepted standards in pediatric diabetes care. There is, however, room for improvement. In recent years, there has been an increase in the number of small, single-physician practices in pediatrics that lack full-time diabetes education and psychology expertise.

A considerable number of teenagers are seen by adult providers, and there is no standardized process to transition patients from pediatric to adult treatment facilities. Most diabetes education for patients and families is provided during hospitalization, leading on average to a 12-day hospitalization at diagnosis. Care for adult patients with T1D in Germany is mostly provided by specialized practices, where trained diabetes educators play a major role, but usually without a psychologist. There is insufficient experience with T1D in the adult hospital setting, and only a fraction of adult diabetes institutions participates in standardized quality improvement initiatives.

The T1D Exchange Network and Clinic Registry (T1DX), composed of a select group of 81 pediatric and adult diabetes clinics, was in operation from 2010 to 2018, and filled a significant benchmarking void in the United States. The registry did not function as a QIC but did facilitate multisite clinical trials. Data from the registry were first used to describe the state of T1D outcomes for the first time in 2012. ^25,26 The initial analysis revealed that the majority of U.S. participants were not meeting HbA1c targets, and later analyses revealed that clinical outcomes in the United States had deteriorated among adolescents and young adults between 2012 and 2018, despite increased use of diabetes technology. ¹⁶ The T1DX data also magnified the impact of social determinants of health on disparities in clinical outcomes.

As the U.S. health care system began to move from a fee-for-service to value-based care model under the Affordable Care Act, clinical practices began to focus more actively on quality improvement initiatives. In 2016, the T1DX established the T1D Exchange Quality Improvement Collaborative (T1DX QI) comprised of 10 clinical sites at academic medical centers. ⁴¹ Under new leadership since 2019, the T1DX QI has expanded from 10 to 40 clinical sites, and the current foci are on facilitating data extraction to improve benchmarking capacity, and creating opportunities for member sites to participate in collaborative efforts to improve health outcomes and care delivery by sharing best practices. ⁴² The T1DX QI will have the opportunity to partner with and learn from more established QICs, including those in Sweden, Germany, and Austria.

However, these aspirational efforts at quality improvement must be undertaken in the context of the country's decentralized public health infrastructure and fragmented and increasingly dysfunctional health care system. U.S. health care is characterized by high cost for medications and services, variable access, waste, fraud, lack of patient-centeredness, and systemic inequities. ²¹ While electronic medical records (EMR) are widely used and have the potential to lower costs and improve care, most exist in their own silos and are unique to each clinic or hospital system; data remain largely proprietary, and EMR systems are primarily utilized for billing and coding rather than patient care, making it difficult to extract actionable data. ²⁴ Per capita health care spending has skyrocketed in the United States. In 1980, during the early days of the DCCT, U.S. spending on health care was consistent with that of its European peers; it is now nearly double that amount, with projections that it will reach 20% of Gross Domestic Product in 2025. ^22,24

Despite high per-capita spending, quality of care and clinical outcomes are poorer than in Europe, and the personal cost for individuals remains high, with an estimated out-of-pocket cost of $800/month for children and adults with T1D in the United States. ⁴³ The retail price of insulin remains prohibitively high in the United States, despite relatively low production costs, and the list prices of commonly used insulins nearly tripled between 2002 and 2013. ^44,45 The impact of rising insulin prices for people with T1D is significant, with estimates that the gross insulin spending per person increased by $2841 between 2012 and 2016. ⁴⁶

In the current U.S. health care model, diabetes care is poorly reimbursed, and obtaining remuneration for diabetes education, a requisite for effective self-management, is difficult and rarely covers the cost of services, resulting in only a small number of patients receiving basic education, medical nutrition therapy, and training for diabetes device use. Higher family income and private health insurance increase the likelihood of meeting glycemic targets, as demonstrated by an analysis of T1D Exchange participant data that revealed that 86% of children with HbA1c <7% (53 mmol/mol) were privately insured and 72% of them reported family income ≥$75,000; conversely only 57% of children with HbA1c ≥9% (75 mmol/mol) were privately insured and 36% had family income ≥$75,000. ⁴⁷ There is no billing pathway for many aspects of diabetes care in the United States, including outpatient advice for acute complications such as hypoglycemia and ketones or consultations with school nurses.

Finally, excessive administrative costs, including processing prior authorizations and other paperwork, have led to an increase in administrative staff often offset by reductions in clinical support and the number of diabetes educators. In contrast, all children and adolescents in Sweden have had full coverage for health care costs since 2016, including free diabetes care and access to equipment such as insulin pumps and continuous glucose monitors when indicated. Adults pay for care and medications, but the cost ceiling is low, and in 2017 was ∼$12 USD per day maximum for hospital care, ∼$130 annual maximum for clinic visits, and ∼$260 annual maximum for diabetes-related medications. While QIC hold promise to improve outcomes, as evidenced by the realized success in Germany, Austria, and Sweden, their efficacy in the United States may be circumscribed until meaningful health care reform occurs.

Promisingly, advances in diabetes technology may ultimately bend the curve in terms of glycemic control and patient outcomes. International data suggest that the increase in pediatric continuous glucose monitoring (CGM) use is contributing to lower mean HbA1c. ^16,30 First generation automated hybrid closed loop (HCL) insulin delivery systems have demonstrated improved glycemic control (increased time in range) and reduced hypoglycemia. ⁴⁸ Real-world data from second generation HCL systems show not only improvements in glycemic control but also decreased disease burden and high device-related satisfaction, predisposing them to higher adoption rates. ^49,50 The widespread use of HCL systems will likely play a key role in blunting the decline in glycemic outcomes from early childhood through young adulthood, and likewise improving outcomes for adults with T1D. However, there are already disparities in technology use, with those in the lowest SES quintile having the lowest levels of adoption and a large number of children and adults unable to benefit from emerging technologies due to variability in reimbursement and unequal access. ^23,28,33

There is little doubt that countries with universal health care systems, and established benchmarking and quality improvement networks, will integrate these technological advances into care more quickly and more equitably than countries with weaker or more fragmented systems, and the gap between technology haves and have-nots will likely widen in middle- and lower-resourced countries.

This analysis was not without limitations. There are fundamental differences in the registries, with the Swedish registry representing more than 95% of the pediatric and 90% of the adult population in the country with T1D; the German/Austrian data representing greater than 80% of the pediatric population with T1D, but only a sample of adults; and the T1DX Registry, which was clinic-based, representing a sample of people from diabetes specialty clinics in the United States. The mean age at diagnosis and diabetes duration in Table 1 reflects the different composition of study participants and not an actual difference between age and duration in persons with T1D in the respective countries.

There were a varying number of HbA1c results per year per participant, and HbA1c results were clinic reported and not performed in a standardized research laboratory. As has been noted elsewhere, methods in all four countries are DCCT standardized, and sensitivity analyses have shown that differences in laboratory methods between the DPV and T1DX registries did not introduce bias into prior analyses. ²⁷

Minority status and the effects of social determinants of health were not included in this analysis, nor was the effect of diabetes technology use. Finally, and importantly, HbA1c is an imperfect and incomplete proxy for T1D outcomes, and this indicator alone is not sufficient to describe the quality of diabetes care or success with self-management. Other factors, including frequency of severe hypoglycemia, incidence of diabetic ketoacidosis, psychosocial well-being, quality of life, and reduced risk for micro- and macrovascular complications, are also important. While a comparison of these additional outcomes is not feasible in the context of this analysis, their relevance and worth should be acknowledged.

Where to go from here? In the United States, clinicians and diabetes educators must continue to work within the current system while actively advocating for universal access to health care coverage and a system that incentivizes a quality-based, population health approach to improve outcomes and reduce inequities. In Germany and Austria, there is an opportunity to expand the benchmarking model and quality initiatives developed in pediatrics to include adult clinics. Similarly, in Sweden, the quality improvement processes that have contributed to substantial improvements in glycemic outcomes in pediatrics can be adapted for adult care. Meanwhile, international collaboration, benchmarking, and quality improvement must expand, with a focus on equitably integrating diabetes education, psychosocial support, and emerging diabetes technologies that contribute to improved outcomes and reduced burden into clinical care. On this score, European countries provide a model that diabetes centers in the United States can emulate, as well as a pathway for middle- and low-resourced countries as they expand their national health care systems and continue to build public health infrastructure.

Conclusion

Between 2011 and 2017, overall HbA1c improved in Sweden, stayed the same in Germany and Austria, and worsened in the United States. The comparison of international trends in HbA1c creates the possibility to identify differences, explore underlying causes, and share quality improvement processes where applicable. National quality improvement initiatives that are well accepted in Sweden, Germany, and Austria, but have yet to be implemented systematically in the United States, could be adopted and may lead to improvements in glycemic outcomes. However, disparities created by the lack of universal access to health care coverage, unequal access to diabetes technologies (e.g., CGM) regardless of insurance status, and high out-of-pocket cost for the underinsured, ultimately limit the potential of quality improvement initiatives. Advocacy is required to improve access to health care coverage and address high out-of-pocket cost for people with diabetes in the United States.

Role of Study Sponsor or Funder

The study sponsor/funder was not involved in the design of the study; the collection, analysis, and interpretation of data; writing the article; and did not impose any restrictions regarding the publication of the article.