Young adult patients with type 1 diabetes have a higher risk of mortality than those of similar age with type 2 diabetes: A nationwide analysis in Hungary

There are few papers comparing complications of type 1 diabetes with those of a similarly young age with type 2 diabetes. The aim of our nationwide study was to compare the risks of mortality and morbidities between the two types of diabetes (age ≤ 40).


| INTRODUCTION
The two most common forms of diabetes mellitus are type 1 and type 2. Each has an effect on diabetic complications, but the degree of the effect may differ from the standpoint of cardiovascular and renal complications and with regard to cancer, hypoglycemia, and diabetic ketoacidosis.

| Type 1 diabetes
Several observational studies showed substantially increased mortality and diabetes-related comorbidity in patients with type 1 diabetes. 1 Data collected from the Swedish registry showed a 3.52 adjusted hazard ratio in patient with type 1 diabetes for death from any causes compared with that in the control population. 2 Despite substantial improvement in diabetes care, a Danish analysis still demonstrated higher rates of mortality among patients with type 1 diabetes with lowest age at onset, and in men. 3 Lung et al conducted a meta-analysis studying the risk of all-cause mortality in patients with type 1 diabetes and reported a 3.82 relative risk compared with the general population. However, they showed a significant improvement, because relative risk of mortality decreased from 5.8 in 1971 to 3.11 after 1990. 4 Another analysis of the Swedish National Register presented a roughly 40% decrease in the mortality of patients with type 1 diabetes and in the incidence of cardiovascular morbidities. 5 In Finland, the survival of patients with type 1 diabetes also improved, but in Estonia and Lithuania a significantly higher diabetic ketoacidosis-related mortality risk was found compared with that of Finland in the same study. 6

| Type 2 diabetes
While the prevalence of type 2 diabetes is higher among the elderly population, there is an increasing incidence and prevalence among younger people resulting in higher long-term cardiovascular morbidity as a consequence of the long exposure to hypoglycemia. 7 Earlier diagnosis and manifestation of renal complications in type 2 diabetes were found in a Canadian clinical record-based study, but no difference in retinopathy was detected. In this study, macrovascular comorbidities were rare, while microvascular complications, eg, dialysis manifested within 10 years from the diagnosis of diabetes. 8 In an Australian population, long-term morbidity and mortality risk was also the highest in patients with type 2 diabetes starting at younger age (15-31 years) compared with those with usual onset (40-50 years, HR 3.4). 9

| Comparison of type 1 and type 2 diabetes
Several studies compared morbidity and mortality of type 1 and type 2 diabetes in young cohorts. 10,11 For type 1 compared with type 2 diabetes, lower risks of death from any cause and of cardiovascular morbidity were characteristic. 5,10 Despite the presence of more severe comorbidities in young patients with type 2 diabetes, longterm survival was similar in both types of diabetes. 12 Constantino demonstrated a mortality excess in patients with type 2 diabetes compared with type 1 (11% vs 6.8%, P = .03) and an elevated hazard risk for death from any causes (HR 2.0, P = .003). Cardiovascular mortality was also more frequent in the population of type 2 diabetes (HR 3.5, P = .004). 13 Similarly, three-fold higher mortality rate was found in a 15 to 34 age group of patients with type 2 diabetes in a large, prospective, cohort-based, Swedish study. 14 Dart found a significantly higher 10-year survival rate in patients with type 1 diabetes (aged 1-18 years) compared with patients with type 2 (99.5% and 91.4%), which further decreased to 97.6% and to 77.5% after 20 years. 15 Dart also found a higher risk of end-stage renal failure in young patients with type 2 diabetes when compared with the risk of patients with type 1 diabetes (aged 1-18 years). 15 This finding has been confirmed between the two types of diabetes in Chinese and Japanese studies. 16,17 The prevalence of diabetic ketoacidosis was high in young patients  18 Another report across three registries and five nations also presented a high (5.0% to 7.1%) frequency of diabetic ketoacidosis in young patients with type 1 diabetes. 19 Barski evaluated the mortality risk of diabetes ketoacidosis in both types of diabetes and found that in type 2 diabetes it is more severe, with worse, outcomes compared with type 1 diabetes. 20 The incidence of cancer in diabetes is increased among type 2 diabetic patients compared with that of the nondiabetic population and is well reported in case of type 2 diabetes. 21 On the other hand, only a limited amount of data is available on cancer-related morbidity and the mortality of patients with type 1 diabetes. 22,23 The aim of our nationwide analysis was to compare the mortality and morbidity risks of young patients with type 1 and type 2 diabetes, using data of the Hungarian National Health Insurance Fund.

| Study design
In our retrospective cohort study, patients aged 40 years or under and starting antidiabetic therapy (ATC A10) between 1 January 2001 and 31 October 2014 were extracted from the database of the National Health Insurance Fund as anonymized, aggregated patient data. We used data from this time frame of all listed Hungarian patients with type 1 diabetes. We compared two similarly young adult, but not matched populations, and the ratio of type 1 to type 2 was more than 1:4.
Codes of the version 10th of International Classification of Diseases (ICD) were used to define diabetes and comorbidities. The data source includes information of mortality from any causes, incidence of myocardial infarct (ICD-10 I21-24), ischemic and haemorrhagic stroke (ICD-10 I61-63, G4630, G4640,), dialysis, cancer (ICD-10 C and D class), diabetes ketoacidosis (ICD-10 E1010, E1110, and E1410), and hypoglycemia (ICD-10 E1600, E1610, and E1620). As dialysis reimbursement is quite expensive, it is recorded in a separate database of Hungarian NHIF, where all dialysis is followed as an intervention based on patient ID, without using ICD codes. We investigated all dialysis intervention in this separate database.
We defined type 1 diabetes (T1DM) (a) as being when a patient was recorded as having an E10 ICD code, (b) the age of the patient was less than 40 years at the time of diagnosis, and (c) had no oral medication prescriptions during the first 6 months. Patients having antidiabetic treatment (ATC A10) but not matching previously detailed criteria were taken as having type 2 diabetes (T2DM). A detailed specification of type 1 and type 2 diabetes has been described in an earlier publication. 24 Patients with polycystic ovarium syndrome (ICD 10 E282) were excluded from our analysis.
The onset of diabetes was defined either as the first occurrences of the diabetes-related ICD code in the database for a patient or the first insulin/noninsulin diabetic treatment. We searched for previous stroke and myocardial infarction events as well. We only used events (myocardial infarction, stroke, dialysis, cancer, diabetes ketoacidosis, and hypoglycemia) from the in-hospital records and only those that occurred after the diagnosis of diabetes. Dates of death were also retrieved from the National Health Insurance Fund. However, because this database does not differentiate causes of death, we used only allcause mortality for analysis.
All-cause mortality, as well as morbidities of patients with type 1 and type 2 diabetes, were compared. This study was approved by the Pécs University School of Medicine's Regional Research Ethics Committee of the Medical Center, Hungary (study licence number: 6962/2017) and ran without commercial sponsorship. The study protocol was also reviewed and confirmed by the National Health Insurance Fund (NHIF) (identification number: S04/161/2016).  Based on the examined 2001 to 2014 period, patients with type 1 diabetes had a higher risk for all-cause mortality than those patients with type 2 diabetes (Figure 1, panel A)

| DISCUSSION
In our long-term, retrospective, nationwide study, we assessed the differences in risk for mortality and morbidity, comparing adult patients with young age (below 40 years) and follow-up with type 1 and 2 diabetes.
In our analysis, the following main findings were assessed: (a) a higher risk was found for mortality in patients with diabetes type 1 compared with type 2; (b) no difference in risks were detected for cardiovascular diseases, while increased risks were verified for cancer, diabetic ketoacidosis, hypoglycemia, and dialysis in type 1 diabetes.
A study comparing patients with type 1 and type 2 diabetes reported a better and improving mortality risk in the type 1 group, 11 which is opposite to our findings. In the Constantino report, an increased hazard ratio for death was proved in the case of young patients with type 2 diabetes (HR (95%CI), 2.0 (1.2-3.2)) compared with those patients with type 1, 13 which was confirmed by a Swedish study, 14 and in the analysis by Dart et al. 15 On the contrary, in our study, a higher risk was found for all-cause mortality in patients with type 1 diabetes compared with those patients with type 2. These differences could be explained as a consequence of the divergent populations.
However, an unexpectedly higher mortality in the T1DM group was found despite the younger (12 years) mean age of this population.
These outstanding results could be explained as follows. In the cardiovascular outcomes, we did not find difference between the two diabetic groups, but other main causes of diabetic mortality, eg, ketoacidosis and hypoglycemia, together with renal impairment and cancer may impact the higher mortality of T1DM. Indeed, our findings in size of mortality are similar to Estonian and Lithuanian results, which detected a higher mortality in patients with type 1 diabetes. 6 Survival after 10-year duration was 94.3% in Estonia and 94.0% in Lithuania versus 99.1% in Finland, 6 while we found 93.84% survival during a similar duration of follow-up. There was verified a higher incidence of diabetic ketoacidosis in Estonia and Lithuania compared with Finland as a most common cause of death. 6 In our study, the incidence of diabetic ketoacidosis was 15.46% at the diagnosis of type 1 diabetes, which is higher compared with another report 19 and remained high during the follow-up period.
Scibilia also reported diabetic ketoacidosis as a cause of early mortality in type 1 diabetes using data Children's Hospital of Pittsburgh between 1950 and 1980. Diabetic ketoacidosis was a main cause of death at early phase of type 1 diabetes, but no death was associated with hypoglycemia. 26 Similarly, Edge reported diabetic ketoacidosis as 83% of diabetes related death, while hypoglycemia was associated in 8% only as a main cause of mortality in young population with type 1 diabetes. 27 Agerelated incidence of DKA was also found at analysis of DVP database. 28 Karges at al found the mean admission rate for DKA was 4.81/100 patient-years, highest rates were observed in patients with HbA1c ≥ 9.0%, age 15 to 20 years, and diabetes duration of 2 to 4.9 years.
Besides the high incidence of DKA, we also found high occurrence (HR 12.52) for hypoglycemia in T1DM. These unexpectedly high occurrence of two acute complications of type 1 diabetes may be the underlying cause of death in T1DM cohort and may lead to the higher risk (17%) versus T2DM.
In another, comprehensive, population-based cohort study of Karges,29 where T1DM patients using insulin treatment (n = 9814) were paired with those, using insulin pump therapy (n = 9,814), In spite of the Kaplan-Meier curve demonstration of the survival which suggest that in T2DM could be higher the risk of cancer, the age adjustment in the Cox analysis proved that the risk of cancer was higher in the T1DM. This finding also supports the higher mortality outcome for T1DM and emphasizes the need of further studies of cancer risk in young adult.
Our study contains a variety of strengths and weaknesses. The length of the follow-up period, size of patient groups with type 1 and type 2 diabetes, the analyses of many diabetes-related comorbidities, and the high number of events provided sufficient power for the analyses.
The limitation of our analysis is that it included patients with diabetes using antidiabetic drug therapy, while it did not involve patients with type 2 diabetes solely leading a modified lifestyle, without antidiabetic drugs. We are also emphasizing that the mean age was different in the two groups, in T1DM 21.63 (21.42-21.85) and in T2DM 33.47 (33.41-33.53). Further limitations of our trial are that NHIF data source has no information about some important confounders, like haemoglobin A1c, cholesterol level, BMI, blood pressure, and body weight, which could be the explanation for the differences found but have not been evaluated because data were unavailable.
As our NHIF database has no specific cause of death, we were able to investigate only all-cause mortality leading to relevant weakness of our analysis as significant amount of the first CVD events is fatal.
Based on these findings, in spite of the higher prevalence of cardiovascular and cancer morbidity in patients with type 2 diabetes, the higher prevalence of diabetic ketoacidosis and hypoglycemia found in our study may contribute to the higher mortality risk of patients with type 1 diabetes compared with those with type 2. A higher rate of dialysis, as a precipitation factor for both hypoglycemia and ketoacidosis, may sustain these differences in the long term.

FUNDING
None.