This study was a population-based retrospective observational study in which historical data on healthcare consumption and mortality among T2D patients with and without established CVD were collected and analyzed for a 5-year period between 2012 and 2016. As possible, we applied the inclusion and exclusion criteria of the EMPA REG Outcomes Study , in which study patients with T2D and established CVD were administered empagliflozin, to define patients with established CVD.
The study was approved by the Regional Ethics Research Committee in Östergötland, Sweden (Dnr: 2017/535-31) and was performed in accordance with the Helsinki Declaration of 1964, and its later amendments.
The primary data sources were patient data extracted from national health registers, including the Pharmaceutical Registry and the Cause of Death Registry, the National Diabetes Registry (NDR), and the Östergötland regional administrative registry (Healthcare Data Warehouse [VDL]).
The Pharmaceutical Registry contains all prescription medicines collected in pharmacies . This register also contains data on extracted items other than medicinal products—if they are part of the national benefit system. There are just over 100 million entries per year in this registry. In this study, the Anatomical Therapeutic Chemical Classification System (ATC) code, date, and withdrawal costs between 1 January 2012 and 31 December 2016 were extracted for the study population.
The Cause of Death Registry covers all deaths that occur in Sweden . It also includes the deaths of persons not registered in Sweden at the time of death. In this study, the register was used solely to extract the date of death of patients in the study cohort.
The NDR  was founded in 1996 as a tool for local quality control and to better align with national treatment targets. The NDR contains data on the type of diabetes (type 1 diabetes [T1D] or T2D), demographic factors, diabetes duration, treatments, risk factors, and diabetes complications. The register's coverage rate is estimated to be 94% . Registration immediately before 1 January 2012 was used in this study to extract data on patient status on 1 January 2012 with respect to weight, height, body mass index (BMI), diabetes duration, age at onset, year of first glycated hemoglobin measurement, low-density lipoprotein, blood pressure, glomerular filtration rate (GFR), albuminuria, and type of diabetes.
The Östergötland VDL contains information on patients' characteristics, visits to healthcare providers, and hospital stays (including number of inpatient days). For each care episode registered in the VDL, there is information on diagnoses, inpatient care, outpatient care, and primary care. The VDL system is linked to a cost-per-patient database (KPP) in which one cost entity is linked to each contact. The VDL registry has almost complete coverage (close to 100%) as all contacts provided by the Swedish publicly financed healthcare system are registered. The volume of healthcare visits to private providers is available, but not their prices. For this study, these prices were taken to be the average of publicly provided visits.
All inhabitants in Sweden have unique ID-numbers that enable data from the different registers to be linked. In this study, the actual linking of the data was performed by the National Board of Health and Welfare, which also held the identification key. All diagnoses were coded according to the International Classification of Diseases, tenth revision (ICD-10).
The study population was defined by first identifying all patients in the region of Östergötland with at least one diabetes ICD-10 code between 1 January 2009 and 31 December 2011 in the VDL. Thereafter, the Östergötland VDL and the NDR were searched to exclude patients with T1D, leaving only T2D patients in the study population. These T2D patients were then screened and the exclusion criteria used in the EMPA REG Outcome study  applied: age < 18 years; BMI > 45; reduced kidney function (GFR < 30 ml/min); and recent (2 months prior to study start) stroke, transient ischemic attack (TIA) or acute coronary syndrome (ACS).
Those T2D patients who met the inclusion criteria were subsequently divided into two cohorts, those with established CVD (T2D-CVD cohort) and those without established CVD (T2D-others cohort). Established CVD was defined based on the inclusion criteria in the EMPA REG Outcome study and translated into diagnoses and measures registered in inpatient care during a 10-year period prior to 2012, as shown in Table 1. The most common reasons for inclusion in the T2D-CVD cohort were myocardial infarction (MI) and stroke. A flow chart of study enrollment and allocation to the two patient cohorts is shown in Fig. 1.
For the analysis of costs during the 5-year study period (2012–2016), healthcare consumption was retrieved from the VDL and drug consumption from the national Pharmaceutical Registry. Date of death was retrieved from the Cause of Death Registry.
This was a descriptive analysis of two patient cohorts: T2D-CVD and T2D-others. The use of healthcare resources and associated costs, as well as mortality, were described for the two cohorts. Proportions were reported as percentages, and Chi-square tests were used to test the significance of differences between groups. Means were used as measures of central tendencies, and the standard deviation was used as a measure of distribution. Group mean differences were tested with t tests. A significance level of p = 0.05 was used throughout. Survival in the different groups was shown using Kaplan–Meier curves. Group differences in survival were also tested with Cox regression due to differences in age and sex distribution.