Cardiovascular and Renal Disease Burden in Type 1 Compared With Type 2 Diabetes: A Two-Country Nationwide Observational Study

In 2019, an estimated 463 million people worldwide were living with diabetes, and the global prevalence is projected to increase by ∼50% by the year 2045 (1). The majority have type 2 diabetes (T2D), while 5–10% of patients with diabetes suffer from type 1 (T1D). The highest incidence rates of T1D can be found in the Scandinavian countries, with onset often at a young age and, hence, a long lifetime with the disease (2). Globally, cardiovascular (CV) diseases remain the leading cause of death (3), and it has been shown that both T1D and T2D are associated with markedly increased CV risk compared with populations without diabetes (4–7). In addition to disturbed glucose metabolism, T2D is linked to the metabolic syndrome, including obesity, insulin resistance, hypertension, and dyslipidemia. These factors are considered to contribute to the accelerated atherosclerotic CV and renal disease (CVRD) development and increased rates that are seen in T2D (8), especially represented by cardiorenal disease defined by heart failure (HF) and/or chronic kidney disease (CKD) (9). In contrast, these risk factors are not as common in patients with T1D, and their increased CVRD risk has traditionally been explained by the degree and duration of hyperglycemia per se and nephropathy (10). However, there is a paucity of studies directly comparing the frequency of CVRD between the two major forms of diabetes, so it is not known whether the risk differs significantly between these diabetes forms (11,12).

We aimed to directly compare the prevalence and incidence of CVRD in patients with T1D and T2D in large, full-population registries in Sweden and Norway. Furthermore, the presence of CKD, besides being an important microvascular diabetes complication, also in itself is an important risk factor for CV morbidity and mortality.

The present work is part of a large-scale initiative to acquire an understanding of diabetes and its treatment, and it is based on the Diabetes Swedish Full Population Study (DAISY) and Diabetes Norwegian Full Population Study (DAPHNE) studies (13–16). Both Sweden and Norway have comprehensive, nationwide public health care systems. All citizens have a unique personal identification number, which is mandatory for administrative purposes, including any contact with the health care system as well as prescribed drug dispensaries, thus providing comprehensive medical information about the whole population. Individual patient-level data from the prescribed drug, cause of death, and national patient registries covering all hospitalizations and all outpatient hospital visits, were linked using the personal identification number but in a pseudonymized manner. Database management and study protocols are described in the Supplementary Material.

All patients with diabetes in Sweden and all patients >18 years of age in Norway were identified by using the prescribed drug and national patient registries. Subjects between the ages of 18 and 84 years at the index date were included in the study population. T1D was defined as having received at least one prescription of short-acting insulin from treatment start or within 6 months after start of basal insulin treatment and having a hospital diagnosis of T1D. To validate our definition of T1D, we compared our Swedish T1D cohort with data from the Swedish National Diabetes Registry (NDR) (17) and from a previous Swedish registry-based study (18) regarding prevalence, incidence, age distribution, and age at onset for subjects with T1D. We found strong agreements between our cohort and previous data sets in this regard, thus lending support for a valid classification tool (see Supplementary Fig. 1A–D for details).

T2D was defined as having any prescribed blood glucose–lowering drug dispensed by a pharmacy, except for subjects defined as having T1D and those diagnosed with gestational diabetes mellitus or polycystic ovary syndrome according to their respective diagnosis codes in the national patient registries. Subjects were followed in the national patient and cause of death registries until the end of the follow-up period or death.

The index date was 31 December 2013 for both countries, and the follow-up period was 3 years for Sweden (until 31 December 2016) and 2 years for Norway (until 31 December 31 2015). This amounted to a mean follow-up of 2.6 years and 1.0 million person-years. Baseline comorbidities were searched in the national patient registries before the index date, and baseline drug treatment was searched in the prescribed drug registries 1 year before the index date (presented in detail in Supplementary Tables 2 and 3).

Age-stratified CV and CKD event rates were calculated during the follow-up period in each country separately, and results were then pooled. Because the effect of disease duration in T1D is known to have a substantial impact on the risk of complications (6), we calculated disease duration for Swedish patients with T1D and analyzed the impact of duration on the various outcomes. Because of registry structure and limited follow-back time, complete data on duration of T2D could not be obtained.

Outcomes were retrieved from in- and outpatient hospital records. In cases of inpatient care, diagnoses at the end of the hospital visit were obtained after completion of clinical and laboratory investigations. Codes from the ICD-10 were used, and only the first position as main diagnosis was considered (for ICD codes of outcomes, see Supplementary Table 4). As a sensitivity analysis, all registered diagnosis codes (i.e., also beyond the first position) were searched. The following outcomes were assessed: all-cause death, myocardial infarction (MI), HF, stroke, CKD, any CVRD, and CV and all-cause death. MI was identified according to the diagnosis code of MI. Stroke included both ischemic and hemorrhagic stroke. HF included both hypertensive and other HF. CKD was defined by a first-position diagnosis of diabetic nephropathy, CKD, or renal failure (hypertensive, acute, chronic, or unspecified). CV death was defined as death as a result of stroke or MI, and any CVRD was defined as any of HF, MI, stroke, CV death, and CKD. Cardiorenal disease was defined as at least one of HF or CKD because these often appear together or sequentially. For subgroup analysis of patients with and without previous CVRD at baseline (see below), CKD was defined as described above, and CV disease was defined as MI, HF, or stroke.

Baseline characteristics were calculated separately for each country and then pooled. Event rates of the outcomes were calculated by dividing the number of incident cases by the duration of follow-up in person-years and expressed as incident cases/1,000 person-years.

Duration of T1D was calculated by retrieving the age at first insulin prescription or first diagnosis registration (whichever came first) available in the national patient and prescribed drug registries and calculating the time difference to age at the index date. For T2D, this was not possible since in most cases, there is not a hospital diagnosis but mainly in primary care where no national registry data are available.

Cox regression models were applied to determine the hazard ratio (HR) and 95% CI for each outcome according to type of diabetes. The effect of age was assessed by interaction analysis and was found to have a significant interaction effect on most outcomes (data not shown). Thus, age-adjusted HRs were calculated within each age category separately. Additionally, we performed analyses in patients who at baseline were free from previous CVRD and in patients who had a history of previous CVRD. This was done to explore whether previous occurrence of a CV or renal event would affect the risk in T1D versus T2D for new or recurring events, respectively.

A total of 59,331 patients with T1D and 484,241 patients with T2D were included (Fig. 1). Data from both countries were pooled because results for the separate country cohorts were comparable. Baseline data are presented in Table 1. At baseline, 87.5% of patients with T2D were on any glucose-lowering drug, with the remaining patients having previously filled at least one prescription but stopped the treatment before the baseline date. Patients with T1D were younger than those with T2D (mean age 45.8 vs. 64.1 years) and had a lower overall prevalence of CV disease (13.3% vs. 26.2%); a similar proportion had CKD (4.3% vs. 3.3%) but more frequent microvascular complications (71.6% vs. 27.8%), mainly driven by an increased occurrence of retinopathy (58.6% vs. 13.9%) (Supplementary Table 2). A higher use of CV preventive drugs was observed in T2D versus T1D, possibly because of the higher average age. Following age stratification at baseline, the prevalence of CV disease was similar in both cohorts, with a clear increase from age ≥40 years in both T1D and T2D. For MI, HF, and stroke, there were no apparent differences in any age-groups, as shown in Fig. 2. In contrast, the age-dependent increase in prevalence of CKD appeared earlier in T1D than in T2D, and it remained consistently higher in T1D across all age-groups. The prevalence of all CV diseases combined was slightly higher in T1D compared with T2D (Supplementary Fig. 2).

Results are presented in Fig. 3A–F. Event rates were low in age-groups <40 years, where comparisons between T1D and T2D could not be performed. For HF, event rates were significantly higher in the patients with T1D aged 65–79 years (HR 1.31–1.39), and for MI, there were similar findings in ages 55–79 years (HR 1.30–1.79). Rates of stroke, on the other hand, were significantly higher in T1D only among younger patients aged 40–54 years (HR 1.35–1.72). The incidence of CKD was greater for T1D versus T2D at all ages, similar to the baseline prevalence, although intergroup differences generally decreased with increasing age (HR 2.95 for 40–44 years to 1.53 for 80–84 years). The risk for any CVRD event was significantly higher among patients with T1D across all age-groups (HR 1.20–1.60). CV mortality rates were higher in T1D than T2D at ages 55–64 years and 70–79 years (HR 1.37–1.71) (Fig. 3) and so were total CVD events at ages 55–79 years (HR 1.14–1.42) (Supplementary Fig. 3). Except for the youngest ages of 40–49 years, all-cause mortality rates were also generally higher in the T1D group. As a further sensitivity analysis, we also included all registered diagnosis codes regardless of position (i.e., main as well as additional diagnoses). The comparisons between T1D and T2D were nearly identical to our main analyses (Supplementary Fig. 4), albeit with the expected higher overall event rates.

We performed subgroup analyses among patients who were free of CVRD at baseline and those with previous CVRD (Supplementary Fig. 5A–H). In the CVRD-free group, the only significant differences between T1D and T2D were increased rates of MI for T1D in the age span 70–79 years (HR 1.41–1.70) and CV mortality at 75–79 years (HR 1.78). There were no increased risks of CKD and all-cause mortality among patients with T1D in the CVRD-free cohort compared with T2D. Among patients with previous CVRD at baseline, results showed similar patterns as the main results in the total cohort (Supplementary Fig. 6A–H). We also examined the relative impact of disease duration in the Swedish T1D cohort and found that it was an important risk factor for CV events and CKD, independent of age per se (Supplementary Fig. 7). The effect of disease duration was most marked for CKD but remained substantial for all types of CVRD events. Notably, the effect of duration on all-cause mortality was, however, much smaller.

In this observational, longitudinal, full-population study of all patients with T1D and T2D in Sweden and Norway in 2013, we present several important findings on individuals aged 18–84 years. First, the age-adjusted incidence of CV diseases were, in general, higher for T1D than T2D, with a slightly greater stroke risk at younger ages and a greater risk of MI and HF at older ages. Also, CV mortality was greater in T1D at ages >55 years, except the age intervals 65–69 and 80–84 years. Second, CKD was more common in T1D than T2D at all ages. Third, the risk for future CVRD is modified by the presence or absence of preexisting CVRD, with an almost equal risk in T1D and T2D if CVRD free at baseline, whereas risks for T1D were increased for those with previous CVRD.

To our knowledge, this study is the first to address CV disease in full-population cohorts of both T1D and T2D in modern health care systems (from 2013 to 2016). Some studies have reported risks of CV disease (19) and death (11), but these have focused on only one outcome or performed indirect comparisons through populations without diabetes.

Previously, Juutilainen et al. (11) showed in a small long-term cohort study of 134 patients with T1D and 834 with T2D that CV mortality is similar in both groups. Subjects in this study were of similar age (45–64 years), and the age of onset was >30 years. This would suggest that CV mortality risk is similar in the two diabetes types when taking age into account. Our study was much larger and included nationwide, complete T1D and T2D cohorts, which may explain the finding of higher CV mortality in T1D over the age of 50 years.

Recently, Fang et al. (12) examined data from the U.S. National Health Interview Survey conducted in 2016 and 2017 to compare the burden of complications in young-onset T1D and T2D, defined as being diagnosed before age 40 years. They found that there were similar rates of CV disease between the two groups, which is in agreement with our results, but they also observed similar rates of renal disease, which is different from our findings. This study was purely interview based, including ∼230 subjects with T1D and 1,200 subjects with T2D, and the statistical power may not have been sufficient to detect true differences between the groups. It is, however, also possible that young-onset T2D represents an aggressive diabetes phenotype with greater cardiorenal risk, as previous studies have suggested (20,21), and that this contributed to the increased risk of kidney disease among subjects with T2D in the study.

In South Korea, Lee et al. (22) followed a population of 20.4 million individuals from 2009 to 2016, of whom ∼1.9 million had T2D and 9,400 had T1D. In agreement with our results, they found a significantly higher risk of MI, HF hospitalization, atrial fibrillation, and mortality among the subjects with T1D compared with those with T2D during a mean follow-up of 4.6 years. Furthermore, we show that individuals with T1D who already have established CVRD are at a higher risk of further CVRD events, which suggests that patients with T1D will benefit from a more intensive strategy for both primary and, in particular, secondary prevention of CVRD.

Several factors could explain the high risk of CV disease in individuals with T1D. Previous studies have established the importance of albuminuria and CKD in increasing the risk of mortality and CV disease in T1D (23,24). We did not have access to laboratory assessments and, therefore, could not adjust for the degree of renal impairment, but our results indicate that both prevalent and incident CKD are more common in T1D than in T2D, and this could contribute to the higher rates of CV disease in T1D found in several age-groups. We did not have complete data on T2D duration, but microvascular complications, including diabetic nephropathy, often progressing to manifest CKD, are known to be associated with disease duration (25), and it is likely that the earlier age of onset of T1D and, hence, longer average duration, would lead to a greater risk of microvascular complications in later life. Hyperglycemia is another risk factor, and higher mean HbA_1c levels and long-term glycemic exposure are associated with increased mortality (5) and greater risk of macro- and microvascular complications (26,27). A study that was based on Swedish NDR data explored the prognostic significance of various risk factors among subjects with T1D and found HbA_1c and albuminuria to be the most important predictors for mortality and CV disease (28). However, other conventional risk factors such as LDL cholesterol and systolic blood pressure were also strongly associated with adverse CV outcomes.

Interestingly, there was an excess risk in the CVRD-free T1D versus T2D cohorts only of MI and CV death in some age strata but not of other CV events and CKD. The reasons for this are not readily apparent. It is possible that these cohorts were more homogenous, with relatively low risks overall. However, it is worth noting that previous studies have shown that excess mortality in T1D is largely accounted for by nephropathy (24,29), which may explain the finding of a relatively lower risk elevation in T1D versus T2D in the CVRD-free cohorts. In T1D, aggressive nephropathy leading to advanced CKD typically has an early onset, and there is seldom development of nephropathy leading to end-stage renal disease after 15 years of T1D (24,30,31). Notably, among patients with T2D, those with onset at younger ages have a relatively higher risk of serious nephropathy (21,30). Taken together, these findings are compatible with an increased CKD risk in T1D across age strata. However, a similar risk for CKD events in previously CVRD-free patients with T1D versus T2D is somewhat surprising and warrants further detailed investigations.

In our cohort, established drugs to prevent CV disease, such as statins and antihypertensive and antiplatelet drugs, were prescribed far less among patients with T1D compared with those with T2D, albeit this is largely explained by their overall younger age. This may have contributed to the increased cardiorenal risk in T1D. More aggressive intervention with CV disease preventive drugs should therefore be considered at an earlier disease stage, and thus at younger ages in patients with T1D, than what is currently done in routine care. Furthermore, several CV outcome trials have shown beneficial effects of sodium–glucose cotransporter 2 (SGLT2) inhibitors to prevent progression of HF and CKD in both medium- and high-risk T2D populations (32–35). SGLT2 inhibitors might be further explored regarding their risk/benefit profile in future interventional outcome studies in patients with T1D, considering their high cardiorenal risk. In patients with T2D with low-medium CVRD risk, interventional studies should also address initiation of SGLT2 inhibitors at early disease stages. Such studies are currently under way; for example, in Sweden, the SGLT2 inhibitor dapagliflozin and metformin are being compared as first-line therapy in a large randomized prospective trial (SGLT2 Inhibitor or Metformin as Standard Treatment of Early Stage Type 2 Diabetes [SMARTEST] trial, ClinicalTrials.gov identifier NCT03982381).

Strengths of the current study include the large population-based cohorts from modern health care covering essentially all patients with T1D and T2D in two countries. The representative populations constitute a real-world data set and include patients of a wide age range. To our knowledge, this is the first study in European populations to directly compare CV and renal complications in T1D and T2D in nationwide full populations. Importantly, we were also able to validate our definition of T1D and T2D by comparing our T1D cohort (concerning prevalence and patient characteristics) with that of the Swedish NDR, showing an essentially complete overlap.

However, our study also has a number of important limitations. First, because of the registry structures and limited follow-back time, it was not possible to accurately determine diabetes duration for all patients with T2D. Longer duration of T2D has been linked with a greater risk of CV disease and death (36,37), and since we did not have complete information about T2D duration, we could not adjust for this. In 2013 (i.e., the time of our baseline survey), the median duration in Swedish patients with T2D was 8 years according to the 2013 NDR annual report (38) and as reported in a previous publication (4). It is clear that diabetes duration overall is longer in our T1D cohort, with a median of 14 years (range 0 to >27 years) in the Swedish patients. This undoubtedly contributes to the high CVRD risk in T1D, which for several event types, overrides that of T2D, despite its metabolic components. In our cohort of Swedish patients with T1D for whom duration data were available, we found a higher risk of adverse outcomes in those with a longer age-adjusted T1D duration, which is in agreement with previous findings (39,40). Assuming similar glycemic levels in T1D and T2D and an ∼10% increase in risk of CVRD for every year of diabetes duration (Supplementary Fig. 6), the longer average duration of T1D versus T2D at any given age could account for the higher risks we found. Our study did not include a control population. However, previous results comparing T2D with the general Swedish population (16) indicated an ∼1.7-fold increase in the risk for MI and HF, and by extrapolation from our T2D cohort, we estimate that patients with T1D had an ∼2.5-fold greater risk of MI and HF compared with the background population.

Second, no data were available regarding several important risk factors associated with increased rates of CVRD in diabetes. These include HbA_1c, blood lipids, blood pressure, and obesity (24). Furthermore, no information was available regarding socioeconomic and lifestyle risk factors, which are important for outcomes in both T1D (41) and T2D (42). Such confounding factors should be addressed in future studies.

Third, hospital ICD codes were used, and we had no access to primary care data. This may have led to an underestimation of events in patients with T2D because the majority of them are seen in primary care. However, the main scope of the study was to address risk of major clinical CV and renal events, which almost invariably will lead to in- or outpatient hospital care. For example, early stages of CKD in T2D may be underestimated because these patients are typically found in primary care, but the overall occurrence of end-stage renal disease was clearly higher among patients with T1D versus T2D (Supplementary Table 2).

In conclusion, the CVRD burden among patients with T1D is at least as heavy as in patients with T2D following age stratification. For several CVRD events like HF and MI and clearly for CKD, T1D displays the highest risk. There is a remaining need to reinforce and evaluate additional preventive approaches, beyond glycemic control, also in patients with T1D.

Acknowledgments. The authors are grateful to Susanna Jerström and Helena Goike at AstraZeneca for logistic support and valuable comments on the manuscript. Urban Olsson, Statisticon AB, is acknowledged for database management. Data from the Norwegian Patient Register, Norwegian Cause of Death Registry, and Norwegian Prescription Database have been used in this publication. The interpretation and reporting of these data are the sole responsibility of the authors, and no endorsement by the Norwegian Patient Register is intended or should be inferred.

Duality of Interest. This work was sponsored and funded by AstraZeneca. J.B. holds a full-time position at AstraZeneca as an epidemiologist. A.N. has honoraria from MSD, AstraZeneca, Eli Lilly, Boehringer Ingelheim, and Novo Nordisk. M.T. is employed by an independent statistical consultant company, Statisticon AB, of which AstraZeneca Nordic-Baltic is a client. D.N. has received consultancy fees from Novo Nordisk, AstraZeneca, and Eli Lilly. T.N. has received unrestricted grants from AstraZeneca and Novo Nordisk and is on the national board of Novo Nordisk, Sanofi, Eli Lilly, and Boehringer Ingelheim. K.I.B. has received grants to his institution from AstraZeneca for this study and for lectures and consulting from Novo Nordisk, Sanofi, Eli Lilly, Boehringer Ingelheim, Roche, and Merck Sharp & Dohme. J.W.E. has received research support to his institution from AstraZeneca and for lectures and consulting from Novo Nordisk, Sanofi, AstraZeneca, and Merck Sharp & Dohme. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. R.K., J.D., A.N., M.T., D.N., T.N., K.I.B., and J.W.E. participated in the research design. M.T. performed the data management and statistical analyses for all countries after discussion with all authors. All authors participated in data interpretation and in writing the manuscript. All authors took final responsibility in the decision to submit for publication. J.B. is the guarantor of this work and, as such, had full access to all the data in the study takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 79th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 9–11 June 2019.