Abstract
Aims
To assess hospital‐based care, work absence, associated costs, and mortality in patients with type 2 diabetes with and without established cardiovascular disease (eCVD) compared to matched controls.
Materials and methods
In a population‐based cohort study, we analysed individual‐level data from national health, social insurance and socio‐economic registers for people diagnosed with type 2 diabetes before age 70 years and controls (5:1) in Sweden. Regression analysis was used to attribute costs and days absent due to eCVD. Mortality was analysed using Cox proportional hazard regression, stratified by birth year and adjusted for sex and education.
Results
Thirty percent (n = 136 135 of 454 983) of people with type 2 diabetes had ≥1 person‐year with eCVD (women 24%; men 34%). The mean annual costs of hospital‐based care for diabetes complications were EUR 2629 (95% confidence interval [CI] 2601‐2657) of which EUR 2337 (95% CI 2309‐2365) were attributed to eCVD (89%). The most costly person‐years (10th percentile) were observed in a broad subgroup, 42% of people with type 2 diabetes and eCVD. People with type 2 diabetes had on average 146 days absent (95% CI 145‐147) per year, of which 68 days (47%; 95% CI 67‐70) were attributed to eCVD. The mortality hazard ratio for type 2 diabetes with eCVD was 4.63 (95%CI 4.58‐4.68) and without eCVD was 1.86 (95% CI 1.84‐1.88) compared to controls without eCVD.
Conclusion
The sizable burden of eCVD on both the individual with type 2 diabetes and society calls for efficient management in order to reduce the risks for those living with eCVD and to postpone its onset.
1. INTRODUCTION
Risk management with the aim of postponing and preventing cardiovascular disease is a cornerstone in diabetes management. 1 Recent decades have seen downward trends in all‐cause and cardiovascular mortality among people with type 2 diabetes. 2 , 3 A nationwide Danish study showed that individuals with newly diagnosed type 2 diabetes and no prior atherosclerotic cardiovascular disease had 7‐year risks of myocardial infarction and cardiac death close to those of matched controls from the general population. 4 However, there was a persistent gap in all‐cause mortality, 4 and a recent study indicated rising cumulative costs of cardiovascular and renal disease in type 2 diabetes in six countries with data follow‐up of 2 to 10 years. 5
There is also a trend towards people living longer with type 2 diabetes, therefore, the concern regarding development of cardiovascular disease, albeit later, remains. Global costs of diabetes were estimated to be 1.8% of gross domestic product (GDP) in 2015 6 and 35% of these were attributed to production loss due to premature mortality and the impact of morbidity on work capacity. The projections for 2030 would in an optimistic scenario contain costs at 1.8% GDP if age‐related prevalence of diabetes could revert to the levels in 2010, but would increase to 2.2% of GDP if past trends in prevalence and mortality continue. 7 Data from the United States have highlighted increasing prevalence and increasing cost per person over time. 8 , 9 Preventing long‐term complications of diabetes is an important part of diabetes management, not only because of the costs to the health sector but also because diabetes complications are associated with a negative health impact and a reduced quality of life. 10
Many studies report on the prevalence, incidence and time trends of cardiovascular disease in type 2 diabetes, 2 , 4 , 11 but few have analysed the distribution of the costs of diabetes complications over longer periods. 11 , 12 From the societal perspective, estimates of the costs of diabetes complications are important for understanding the key factors contributing to the burden of the disease. Studies based on individual‐level data have reported costs by subgroups, for example, people with type 2 diabetes with and without established cardiovascular disease (eCVD), 13 cost estimates by complication event, 14 or as annual costs by complication for people living with chronic conditions. 15 Two nationwide studies have also shown estimates of days absent from work and associated costs related to single diabetes complications. 15 , 16 , 17
The aim of this study was to analyse the use of hospital‐based care, days absent from work and their associated costs, and mortality in 10 years of observational data on people with type 2 diabetes and eCVD. The data include costs of diabetes and a set of 19 cardiovascular and other diabetes‐related complications. Results are put into perspective using corresponding data for matched controls from the general population and for people with type 2 diabetes but no eCVD. An additional aim was to describe comorbidity patterns in person‐years with highest costs for people with type 2 diabetes and eCVD.
2. MATERIALS AND METHODS
2.1. Study design and data sources
In a population‐based cohort study, people with type 2 diabetes and eCVD were identified in a Swedish retrospective database, cross‐linking 20 years of longitudinal individual‐level data from national population‐based health, social insurance and socio‐economic registers for people with diabetes and their matched controls from the general population. Data were obtained after ethical approval from nationwide registers at the National Board of Health and Welfare (NBHW), Försäkringskassan (the Swedish Social Insurance Agency) and Statistics Sweden. The NBHW identified people with diabetes and provided data both from the National Patient Register on outpatient and inpatient hospital healthcare 18 and from the Swedish Prescribed Drugs Register on filled prescriptions of selected drugs. 19 Demographic and socioeconomic background characteristics were obtained from the Register of the Total Population and the Longitudinal Integrated database for health insurance and labour market Studies (LISA) at Statistics Sweden. 20 Information on work absence was derived from the Micro Data for the Analysis of Social Insurance database (MiDAS), registering sickness and rehabilitation benefits, and sickness and activity compensation at Försäkringskassan. 21 , 22 Since our merged database contains information for individuals aged 16 years and older, during the period 1997 to 2016, it can provide information on years before first observation of diabetes and covers history of comorbidities since 1997.
The analysis of outcomes of hospital‐based healthcare, work absence, and mortality from 2007 to 2016 accounted for diabetes, diabetes complications, eCVD, and demographic and socioeconomic information. It also included retrospective information from years 1997 to 2006 for the study cohort.
2.2. Study population
People with type 2 diabetes were retrieved from a database including people with first observation of diabetes in the labour‐market‐active age range (16‐70 years). The database inclusion criteria are:
One or more healthcare visits or inpatient admissions with diabetes as the main or secondary diagnosis according to the International Classification of Diseases version 10 (ICD‐10) codes for type 1 diabetes (E10), type 2 diabetes (E11) or unspecified diabetes (E14) in any year in the National Patient Register from 1997 to 2016.
Two or more dispensed prescriptions of glucose‐lowering medication (identified using the Anatomic Therapeutic Chemical [ATC] classification code A10) with dispense dates not more than 6 months apart in the Swedish Prescription Drug register from July 1, 2005 to December 31, 2017.
Women treated with glucose‐lowering medication with a coinciding registration of pregnancy were not included in the study population but were eligible again after 2 years. An index date was defined as the first documented occurrence of diabetes according to criterion A or criterion B, whichever occurred first.
For each person with diabetes, Statistics Sweden matched controls from the general population (5:1) using year of birth, sex and region of residence in the diabetes index year. This enabled incremental analysis of costs of hospital care and absence from work for people with diabetes compared to controls. Further information on the data sources, retrieval process, control selection, and categorization of type 1 and type 2 diabetes can be found in the Supporting Information.
For this study, all individuals with at least 1 year of observation in the period 2007 to 2016 were included. Study panel attrition was attributable to death and migration. The analysis compared outcomes for people with type 2 diabetes to those of controls, with a special focus on additional effects of eCVD. Analyses of work absence were conducted in the subsample of working ages (<66 years). Controls meeting type 2 diabetes criteria received a diabetes index date and were switched to the diabetes group in that year.
2.3. Definition and classification of eCVD
An eCVD status was defined as any registration of the following diagnoses or procedures in the National Patient Register: coronary artery disease; stroke; amputation; peripheral vascular disease; nonfatal cardiac arrest; or related interventions in line with a recently published study 13 , 23 , 24 (Supplementary Table S2). eCVD was a binary variable, taking either the value 0 (no eCVD) or 1 (eCVD). A person with at least one criterion for eCVD was assumed to have eCVD from that year and onwards. The conditions and interventions characterizing switch to eCVD could occur in people with type 2 diabetes and in controls. Observed eCVD in 1997 to 2006 was included at study baseline, January 1, 2007, and eCVD status was updated at switch 2007 to 2016. The design allowed that a person could have eCVD before observed type 2 diabetes status.
2.4. Study variables
Use of hospital‐based care related to cardiovascular disease and a set of selected diabetes‐related complications was identified by the main diagnosis to avoid double counting. Supplementary Table S3 gives a full specification of diagnosis and procedure codes included from the National Patient Register. 15 , 16 These data underlying cost calculations are presented as number of visits and number of inpatient days per diagnosis per 100 000 individuals for the study period in the Supporting information.
Hospital‐based care was costed using diagnosis‐related group (DRG) codes and weights from the Nordic Diagnosis Related Group nomenclature, NordDRG, 25 and the average national price for a DRG weight in 2016. A limited number of healthcare contacts had no assigned DRG code and were costed based on the main diagnosis and its average costs per episode in 2016 in the national Cost Per Patient database. 26 The annual costs of hospital‐based care attributed to eCVD were estimated using the longitudinal panel for the full sample and separately for men and women.
A network diagram of the 10% most costly person‐years shows the prevalence of each comorbidity and the proportion of person‐years having each pairwise combination of included comorbidities for people with type 2 diabetes and eCVD. The diagram illustrates the role of cardiovascular conditions alongside other diabetes complications behind high costs. 15 , 27
Försäkringskassan registers days absent from work. Days absent from work were measured by calendar days (maximum 365 per year) and weighted by the degree of sick leave allowing for part‐time absence (ie, net days). Försäkringskassan grants compensations with reference to the degree of work capacity impairment and not primarily related to a diagnosis, and the method for attributing work absence to eCVD is described in Section 2.5. Costs of days absent from work were valued by multiplying net days by age‐ and sex‐specific earnings for the year 2016 including payroll taxes from labour market data at Statistics Sweden following the human capital approach.
2.5. Statistical methods
The distribution of demographic and socioeconomic characteristics for people with type 2 diabetes and eCVD was compared to that of matched controls in the first year with eCVD (2007 or later for those who switched to eCVD status during the observation period) using standardized differences. Use of hospital‐based care is reported as rates per 100 000 individuals.
Regression analysis accounting for individual‐level clustering and adjusting for education was used to attribute costs of hospital‐based care and days absent from work to eCVD. Further presentation of the empirical strategy is found in the Supporting information. Mortality attributed to eCVD and type 2 diabetes was illustrated in a Kaplan‐Meier survival graph and by Cox proportional hazards regressions, with baseline hazards stratified by year of birth and covariates for sex and level of education including interaction effects.
3. RESULTS
The study included 454 983 people with type 2 diabetes (60% men, n = 271 222; 40% women, n = 183 861) with at least one observation during the years 2007 to 2016 and 2 126 405 matched controls. Thirty percent (n = 136 135) of people withtype 2 diabetes had eCVD at least one year during the study period. Table 1 describes the demographic and socioeconomic characteristics of people with type 2 diabetes when first observed with eCVD (n = 61 671 prevalent in year 2007 and incident 6000‐7000/year in 2008‐2016, n = 61 436) and their controls. The corresponding data for all individuals with type 2 diabetes and controls are shown in Supplemental Table S6. The dynamics of the study panel, including new eCVD, switch from control to type 2 diabetes, and deaths, are shown in Supplemental Table S4. Men with type 2 diabetes were more likely to have eCVD (women 24%, n = 43 847; men 34%, n = 92 288; P < 0.001).
TABLE 1.
Distribution of demographic and socioeconomic characteristics for people with type 2 diabetes and established cardiovascular disease, and their matched controls
| Variable | Type 2 diabetes and eCVD | Controls | Std. diff. |
|---|---|---|---|
| Sample a | n = 123 107 | n = 580 166 | |
| Year of birth, n (%) | |||
| <1940 | 37 032 (30.1) | 145 118 (25.0) | 0.117 |
| 1940‐1949 | 52 399 (42.6) | 259 223 (44.7) | |
| 1950‐1959 | 24 382 (19.8) | 129 260 (22.3) | |
| 1960‐1969 | 7751 (6.3) | 39 090 (6.7) | |
| 1970‐1979 | 1297 (1.1) | 6305 (1.1) | |
| 1980 or later | 246 (0.2) | 117 (0.2) | |
| Age, mean (SD) years | 64.5 (8.8) | 63.4 (8.63) | −0.122 |
| Sex, n (%) | |||
| Men | 81 665 (66.3) | 385 037 (66.4) | 0.001 |
| Women | 41 442 (33.7) | 195 129 (33.6) | |
| Education, n (%) | |||
| Compulsory | 48 969 (39.8) | 176 308 (30.4) | 0.299 |
| Upper secondary | 52 449 (42.6) | 245 223 (42.3) | |
| University | 19 585 (15.9) | 154 533 (26.6) | |
| Missing | 2104 (1.7) | 4102 (0.7) | |
| Type 2 diabetes | Control | ||
| Sample in labour‐market‐active age (age <66 years) | n = 61 227 | n = 326 000 | |
| Employed b , n (%) | 28 532 (48.5) | 236 120 (72.8) | 0.515 |
| Work absence at least once during year, n (%) | 42 032 (68.6) | 84 997 (26.1) | 0.943 |
| Number of days absent from work | |||
| All <66 years, mean (SD) | 153.9 (150.3) | 54.0 (114.0) | −0.749 |
| Days absent for people with ≥1 day absence | |||
| Mean (SD) | 224.2 (130.9) | 207.1 (134.6) | |
| Median (25th percentile, 75th percentile) | 249.5 (92; 365) | 208 (74; 361) | <0.001 c |
Note: Measurement in the first year of observed eCVD (2007, or later in the study period for individuals who switched status to eCVD 2008‐2016). a Employment and work absence only reported for sample in labour‐market‐active ages (<66 years). Supplemental Table S6 gives corresponding information for type 2 diabetes without limitation to eCVD.
Abbreviations: eCVD, established cardiovascular disease; SD, standard deviation; Std. diff., standardized difference.
Table data based on 123 107 individuals with type 2 diabetes and eCVD of in total 135 136 individuals (91%). Remaining 12 029 (9%) already had eCVD at first indication of diabetes. Supplemental Table S3 describes the sample dynamics 2007‐2016 with switches to type 2 diabetes and to eCVD, and the observed number of deaths.
Information on employment status was missing for type 2 diabetes n = 2367 and controls n = 1794.
Wilcoxon rank‐sum test.
3.1. Costs of hospital‐based care
Our data show high use of hospital‐based care for individuals with type 2 diabetes and concomitant eCVD, which was also reflected in associated costs (Table 2). The regression analysis controlling for educational status estimated that individuals with type 2 diabetes and eCVD on average incurred approximately EUR 1000 higher total annual costs of hospital‐based care compared to matched controls (EUR 2629 [95% confidence interval {CI} EUR 2601‐2657] vs. EUR 1578 [95% CI EUR 1566‐1589]). The regression analyses attributed 89% of these costs for people with type 2 diabetes to eCVD status (EUR 2337 [95% CI 2309‐2365]). Men with type 2 diabetes had slightly higher annual costs attributed to eCVD (EUR 2425 [95% CI EUR 2390‐2460]) compared to women (EUR 2141 [95% CI EUR 2096‐2187]), but the proportion of costs attributed to eCVD did not differ between the sexes.
TABLE 2.
Established cardiovascular disease and its association with annual total costs of hospital‐based care for all individuals in the study cohort and, in the sample aged <66 years, days absent from work and costs of work absence related to diabetes complications per individual in the period 2007 to 2016
| Annual total costs of hospital‐based care of diabetes complications: Mean per individual with eCVD | |||
|---|---|---|---|
| Type 2 diabetes | Control | ||
| (n = 454 983; person‐years 3 707 449) | (n = 2 126 405; person‐years 19 440 908) | ||
| Total annual costs of hospital‐based care including eCVD (EUR) | Of these, hospital‐based care costs attributed to eCVD (EUR) | Total annual costs of hospital‐based care including eCVD (EUR) | |
| Mean (95% CI) | Mean (95% CI) | Mean (95% CI) | |
| All | 2629 (2601‐2657) | 2337 (2309‐2365) | 1578 (1566‐1589) |
| Men | 2733 (2698‐2768) | 2425 (2390‐2460) | 1662 (1648‐1676) |
| Women | 2413 (2368‐2459) | 2141 (2096‐2187) | 1364 (1345‐1383) |
| Total number of days absent from work in sample aged <66 years: Mean per individual with eCVD | |||
|---|---|---|---|
| Type 2 diabetes | Control | ||
| (n = 337 090; person‐years 2 320 072) | (n = 1 738 831; person‐years 12 951 021) | ||
| Total number of annual days including eCVD | Of these, attributed to eCVD | Total number of annual days including eCVD | |
| Mean (95% CI) | Mean (95% CI) | Mean (95% CI) | |
| All | 145.9 (144.6‐147.2) | 68.2 (66.9‐69.6) | 105.7 (104.7‐106.6) |
| Men | 130.8 (129.3‐132.4) | 69.1 (67.5‐70.7) | 94.6 (93.6‐95.7) |
| Women | 179.5 (177.2‐181.8) | 82.6 (80.2‐84.9) | 139.7 (137.8‐141.6) |
| Costs of total number of days absent from work in sample <66 years old mean per individual with eCVD | |||
|---|---|---|---|
| Type 2 diabetes | Control | ||
| (n = 337 090; person‐years 2 320 072) | (n = 1 738 831; person‐years 12 951 021) | ||
| Total costs of days absent (EUR) | Of these, costs attributed to eCVD (EUR) | Total costs of days absent (EUR) | |
| Mean (95% CI) | Mean (95% CI) | Mean (95% CI) | |
| All | 19 782 (19 603‐19 961) | 9337 (9150‐9523) | 14 256 (14 131‐14 381) |
| Men | 18 888 (18 663‐19 112) | 9697 (9461‐9934) | 13 479 (13 331‐13 627) |
| Women | 21 783 (21 501‐22 065) | 9823 (9531‐10 115) | 16 707 (16 479‐16 935) |
Note: Calculated from coefficients in regression analysis controlling for educational level.
Abbreviations: eCVD, established cardiovascular disease; EUR, Euro.
The calculated annual costs are based on the hospital visits and admissions due to diabetes complications included in the study database. Tables S5 and S7 in the Supporting information present the data for resource use and cost outcomes for each of the complications in greater detail. The unadjusted differences in costs were particularly large for end‐stage renal disease (nearly ninefold), peripheral vascular disease (eightfold), osteoarthritis (sevenfold), kidney disease (nearly sevenfold) and heart failure (fivefold). In addition, angina pectoris, acute myocardial infarction and ischaemic heart disease were all associated with more than threefold higher annual costs per person‐year for type 2 diabetes compared to controls. Among individuals with type 2 diabetes, those with concomitant eCVD had overall higher costs for hospital‐based care compared to those without eCVD. Nevertheless, costs of diagnoses defining eCVD (and therefore not present for persons with no eCVD) were key cost drivers in the eCVD group: the estimated cost per 1000 person‐years was more than EUR 350 000 for stroke, angina pectoris and acute myocardial infarction.
Figure 1 shows the comorbidity pattern in person‐years with high resource use: more than four in 10 individuals with type 2 diabetes and eCVD (42%; n = 56 531 of 136 135) contributed with at least one costly year to the upper 10% of person‐years ranked by annual costs of hospital‐based care (90th percentile value EUR 8064). The diagram shows the pairwise observed presence of the 15 most common long‐term diabetes complications. Eye disease, angina pectoris and acute myocardial infarction were observed in approximately half of all the most costly person‐years (40%‐56%).
FIGURE 1.
Network diagram of comorbidity patterns for persons with type 2 diabetes and established cardiovascular disease. Upper 10% of person‐years ranked by total annual costs of hospital‐based care. ESRD, end‐stage renal disease; IHD, ischaemic heart disease; PVD, peripheral vascular disease
3.2. Days absent from work
Regression analysis showed that, for a person with type 2 diabetes and eCVD, more than 2 months (68.2 calendar days, 95% CI 66.9‐69.6) of the total 5 months of annual work absence (145.9 days, 95% CI 144.6‐147.2) were attributed to eCVD. Table 2 shows results for days absent and the associated costs for the type 2 diabetes and control groups, for both the full sample and stratified by sex. The total annual costs of absence from work were EUR 19 782 (95% CI EUR 19 603‐19 961) per year, of which EUR 9 337 (95% CI EUR9 150‐9523) were attributed to eCVD. The corresponding total costs of work absence for controls were EUR 14 256 (95% CI EUR 14 131‐14 381). Forty‐seven percent of all days absent were attributed to presence of eCVD. The proportion of costs of days absent attributed to eCVD compared to total costs of work absence was higher in men than in women (51% vs. 45%).
3.3. Mortality
Figure 2 shows Kaplan‐Meier survival curves for four subgroups of type 2 diabetes and eCVD status, respectively. The presence of eCVD increased the risk of death in young and early middle‐aged adults. While the number of individuals at risk was lower in the control group than in the type 2 diabetes group, mortality was high for people with eCVD in both groups at this stage in life. The solid lines in Figure 2 show a flatter survival curve after 40 years of age for controls with eCVD, indicating fewer deaths despite increasing numbers of people at risk. The corresponding curve for type 2 diabetes with eCVD has approximately the same slope from young adulthood past normal retirement ages. The dashed line in Figure 2 shows persistently higher mortality among people with type 2 diabetes without eCVD compared to controls without eCVD.
FIGURE 2.
Kaplan‐Meier survival estimates, stratified by status of established cardiovascular disease (eCVD) and by study group: type 2 diabetes (T2D; red) and control (blue), and presence of eCVD (solid) or not (dash)
Cox regression analysis showed an eCVD hazard ratio (HR) of mortality of 4.63 (95% CI 4.58‐4.68) for individuals with type 2 diabetes compared to an HR 2.78 (95% CI 2.75‐2.81) for controls (reference control without eCVD; Table 3). However, the data did not support a stronger eCVD effect for individuals with type 2 diabetes than for controls: HR 0.892 (95% CI 0.876‐0.909).
TABLE 3.
Factors associated with mortality in Cox proportional hazard regression analysis, stratified by year of birth
| 95% confidence interval | |||
|---|---|---|---|
| Variable | Hazard ratio | Lower | Upper |
| Woman | 0.781 | 0.774 | 0.788 |
| Type 2 diabetes | 1.865 | 1.841 | 1.890 |
| eCVD | 2.782 | 2.752 | 2.812 |
| Interaction type 2 diabetes and eCVD | 0.892 | 0.876 | 0.909 |
| Education | |||
| Upper secondary | 0.851 | 0.844 | 0.859 |
| University | 0.625 | 0.618 | 0.633 |
| Missing | 1.157 | 1.119 | 1.196 |
Note: Baseline hazards by grouped variable for year of birth: before 1940, 1940‐1949, … 1970‐79, 1980 and younger. Regression reference category: Man in control group without eCVD and with education at compulsory level only.
Abbreviation: eCVD, established cardiovascular disease.
4. DISCUSSION
We investigated hospital‐based care, work absence and associated costs, and mortality in people with type 2 diabetes and eCVD. We found that, compared to matched controls, type 2 diabetes was associated with higher hospital‐based resource use and associated costs for nearly all investigated diagnoses. Data confirmed that eCVD was a main driver of costs, especially for hospital‐based care. However, the average annual costs of work absence were greater: more than EUR9 000 for days absent compared to less than EUR2500 for hospital‐based care. This resulted from a persistent loss of work capacity as measured by full or parttime early retirement. Moreover, eCVD and type 2 diabetes, independently increased risk of death. Presence of eCVD in young and early middle‐aged adults was as detrimental for individuals with type 2 diabetes as for controls, but eCVD was observed in a larger proportion of the type 2 diabetes group.
Our findings are in line both with previous findings on increased resource use and costs, 5 , 13 , 14 , 15 , 28 work absence 16 , 17 , 29 and mortality 2 , 3 , 4 in people with type 2 diabetes and with epidemiological data on prevalence of long‐term diabetes complications 30 in type 2 diabetes. Empirical cost data distributions are typically right‐skewed, with a high proportion of the patient group with low or no costs and a limited number of individuals with very high costs. Interestingly, our 10‐year data, with attrition only due to death or migration, indicate that nearly half of the population with type 2 diabetes and eCVD (42%) had occasional high costs and contributed to the 10% most costly person‐years in hospital‐based care. Hence, interventions aimed at reducing the eCVD burden need to be broadly implemented.
The study is based on a “young” type 2 diabetes sample but has national coverage. Inclusion in the database was conditioned on meeting diabetes inclusion criteria in ages 16 to 70 years. Our estimates of the impact of diabetes and eCVD on work absence exceeds reported average estimates of the burden of diabetes in general in a systematic review. 31 However, the review did not include studies with a focus on eCVD or other specific complications of type 2 diabetes and reported early retirement separately. In contrast, our data contain short‐ and longer‐term/permanent work absence, and we focus on absence attributed to eCVD. A recent meta‐analysis showed high risks of vascular disease and mortality associated with diagnosis of type 2 diabetes at a young age and our findings add that there is a considerable loss of work capacity among those with type 2 diabetes before normal retirement age. 32 This study adds new knowledge through its comparison with matched controls from the general population and provides estimates of excess costs related to eCVD in type 2 diabetes. Using controls, we safeguarded against exaggerating the impact of type 2 diabetes on costs. The regression‐based costs of eCVD are the incremental costs on top of what could be expected for someone without type 2 diabetes. These incremental costs stem from a higher incidence of these conditions, a greater need for healthcare interventions, high rates of comorbidity and greater impact on work capacity. These findings also underline the value of study designs with long observation periods for better understanding of resource use, labour market activity, and cost implications from a person‐centred perspective. Our data clearly show that all cost outcomes are higher for type 2 diabetes compared to matched controls and that eCVD is a main contributor to those higher costs.
Our data had a higher proportion of men (60%) and more men also had at least 1 year with eCVD (34% men vs. 24% for women). Nevertheless, data confirmed limited differences in average costs attributed to eCVD between men and women once they had the condition. However, this study confirms the need for adequate levels of resources targeted to eCVD in type 2 diabetes for efforts to minimize the risk of progression and new events. It is also expected that implementation of guideline‐supported treatment 33 , 34 will have the greatest potential for health and quality‐of‐life gains in groups with eCVD, or at high risk of eCVD, where the burden of disease is currently great.
A limitation of this study is that our broad register‐based inclusion criteria could lead to individuals beging mistakenly identified inclusion as type 2 diabetes but the accuracy of diagnosis codes in the National Patient Register has been confirmed. 35 We required two filled prescriptions within 6 months and these data are based on individuals' actions. Checking data, we found that 93% of individuals included in the period 1997 to 2016 in the database also had at least one hospital visit for type 2 diabetes or at least one filled prescription for glucose‐lowering medications in the period 2007 to 2016. Of those who did not, 14% died. However, absence of registration of diabetes diagnosis or medication does not necessarily mean that the person does not have type 2 diabetes and approximately 20% of people with type 2 diabetes treated in primary care are on diet‐only therapy according to the National Diabetes Register. 36 Furthermore, the study dataset is large, and some false positive identified as type 2 diabetes are unlikely to influence results noticeably.
We conclude that eCVD accounts for a sizable portion of the burden of type 2 diabetes. Our findings support increased use of efficient treatments to reduce the risks for those of living with eCVD and to postpone the onset of eCVD, with the aim of mitigating the burden of the disease on the individual and on society.
AUTHOR CONTRIBUTIONS
All authors developed the study concept and design, interpreted data, and critically revised and completed the manuscript. Sofie Persson, Kristoffer Nilsson and Katarina Steen Carlsson performed the statistical and economic analyses and drafted the manuscript. Sofie Persson, Johan Jendle and Katarina Steen Carlsson are the guarantors of this work.
CONFLICT OF INTEREST
Sofie Persson, Kristoffer Nilsson and Katarina Steen Carlsson are employees of the Swedish Institute for Health Economics. None of these authors has received payment outside of normal salary at Swedish Institute for Health Economics related to the subject matter of this work. Kristina Karlsdotter, Josefin Skogsberg and Staffan Gustavsson are employees of Boehringer Ingelheim. No other conflicts of interest relevant to this article were reported. The authors had full editorial control of the manuscript and had the final responsibility for submitting the manuscript.
PEER REVIEW
The peer review history for this article is available at https://publons.com/publon/10.1111/dom.14919.
Supporting information
Figure S1. Population flow chart describing inclusion of people with diabetes and controls from national health and population data registers. Distribution type 1 diabetes and type 2 diabetes.
Figure S2. Study dataset with study subsamples presented as number of unique individuals and number of person‐years for type 2 diabetes and controls including selection subsample with type 2 diabetes and eCVD and their matched controls.
Figure S3. Data availability, population selection for diabetes and controls, exposure to eCVD, and study period for resource use, costs, and work absence. eCVD, established cardiovascular disease; NPDR, Swedish Prescription Drug Register; NPR, National Patient Register.
Table S1. Summary of data sources.
Table S2. Diagnostic codes and procedure codes used to identify established cardiovascular disease.
Table S3. List ICD‐10 codes or procedural codes (KVÅ) obtained from the National Patient Register, NPR, at the National Board of Health and Welfare, NBHW, 1997‐2016 in the existing study database.
Table S4. Description of the study dynamics for people with type 2 diabetes, with and without eCVD, during the study period: number of people, number of deaths, first observations with eCVD, and previous controls added to the sample because of observed diabetes status each year.
Table S5. Use of hospital care measured as number of outpatient visits and number of inpatient hospital days per 100 000 individuals in 2007‐2016. Comparisons of type 2 diabetes vs. matched controls, and type 2 diabetes with/without eCVD.
Table S6. Distribution of demographic and socio‐economic characteristics for type 2 diabetes and matched controls. Measurement in the first year with observed type 2 diabetes (2007, or 2008‐2016 if switch to type 2 diabetes during the study period).† Employment and work absence only reported for sample in labour‐market active ages (<66 years).
Table S7. Costs of outpatient visits and inpatient admissions per 1000 person‐years 2007 to 2016 for people with type 2 diabetes compared to matched controls, and subgrouped for people with type 2 diabetes on eCVD and no eCVD. Data are costs in Euro per 1000 person‐years per study group based on real‐world use of hospital‐based care 2007 to 2016.
ACKNOWLEDGMENTS
The authors acknowledge the editorial assistance of Karin Wahlberg, the Swedish Institute for Health Economics, Lund Sweden. This research was supported by a grant from Boehringer Ingelheim to the Swedish Institute for Health Economics.
Persson S, Nilsson K, Karlsdotter K, et al. Burden of established cardiovascular disease in people with type 2 diabetes and matched controls: Hospital‐based care, days absent from work, costs and mortality. Diabetes Obes Metab. 2023;25(3):726‐734. doi: 10.1111/dom.14919
Persson and Nilsson share equal first authorship, Jendle and Steen Carlsson share equal last authorship.
Funding information Boehringer Ingelheim, Grant/Award Number: Grant
DATA AVAILABILITY STATEMENT
Individual‐level data hosted by national authorities in Sweden are available for research after formal evaluation of the research protocol by the ethical review authorities and by the respective national authorities providing data. Permission to conduct research are granted on a case‐by‐case basis to a limited number of named persons. The secondary individual‐level data of this study cannot be shared by the authors for this reason. Informed consent is not required for register data research in Sweden.
REFERENCES
- 1. American Diabetes Association . 10. Cardiovascular disease and risk management: standards of medical care in diabetes—2021. Diabetes Care. 2021;44(Supplement 1):S125‐S150. [DOI] [PubMed] [Google Scholar]
- 2. Gregg EW, Li Y, Wang J, et al. Changes in diabetes‐related complications in the United States, 1990–2010. N Engl J Med. 2014;370(16):1514‐1523. [DOI] [PubMed] [Google Scholar]
- 3. Rawshani A, Rawshani A, Franzén S, et al. Mortality and cardiovascular disease in type 1 and type 2 diabetes. N Engl J Med. 2017;376(15):1407‐1418. [DOI] [PubMed] [Google Scholar]
- 4. Gyldenkerne C, Knudsen JS, Olesen KKW, et al. Nationwide trends in cardiac risk and mortality in patients with incident type 2 diabetes: a Danish cohort study. Diabetes Care. 2021;44(10):2353‐2360. [DOI] [PubMed] [Google Scholar]
- 5. Norhammar A, Bodegard J, Eriksson JW, et al. Cost of healthcare utilization associated with incident cardiovascular and renal disease in individuals with type 2 diabetes: a multinational, observational study across 12 countries. Diabetes Obes Metab. 2022;24(7):1277‐1287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bommer C, Heesemann E, Sagalova V, et al. The global economic burden of diabetes in adults aged 20‐79 years: a cost‐of‐illness study. Lancet Diabetes Endocrinol. 2017;5(6):423‐430. [DOI] [PubMed] [Google Scholar]
- 7. Bommer C, Sagalova V, Heesemann E, et al. Global economic burden of diabetes in adults: projections from 2015 to 2030. Diabetes Care. 2018;41(5):963‐970. [DOI] [PubMed] [Google Scholar]
- 8. American Diabetes Association . Economic costs of diabetes in the U.S. in 2012. Diabetes Care. 2013;36(4):1033‐1046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. American Diabetes Association . Economic costs of diabetes in the U.S. in 2017. Diabetes Care. 2018;41(5):917‐928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Beaudet A, Clegg J, Thuresson PO, Lloyd A, McEwan P. Review of utility values for economic modeling in type 2 diabetes. Value Health. 2014;17(4):462‐470. [DOI] [PubMed] [Google Scholar]
- 11. Einarson TR, Acs A, Ludwig C, Panton UH. Prevalence of cardiovascular disease in type 2 diabetes: a systematic literature review of scientific evidence from across the world in 2007‐2017. Cardiovasc Diabetol. 2018;17(1):83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Einarson TR, Acs A, Ludwig C, Panton UH. Economic burden of cardiovascular disease in type 2 diabetes: a systematic review. Value Health. 2018;21(7):881‐890. [DOI] [PubMed] [Google Scholar]
- 13. Bernfort L, Husberg M, Wirehn AB, et al. Disease burden and healthcare costs for T2D patients with and without established cardiovascular disease in Sweden: a retrospective cohort study. Diabetes Ther. 2020;11(7):1537‐1549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Kähm K, Laxy M, Schneider U, Rogowski WH, Lhachimi SK, Holle R. Health care costs associated with incident complications in patients with type 2 diabetes in Germany. Diabetes Care. 2018;41(5):971‐978. [DOI] [PubMed] [Google Scholar]
- 15. Andersson E, Persson S, Hallén N, et al. Costs of diabetes complications: hospital based care and absence from work for 392,200 people with type 2 diabetes and matched controls in Sweden. Diabetologia. 2020;63(12):2582‐2594. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Persson S, Johansen P, Andersson E, et al. Days absent from work due to complications associated with type 2 diabetes: evidence from 20 years of linked national registry data in Sweden. Diabetes Obes Metab. 2020;22(9):1586‐1597. [DOI] [PubMed] [Google Scholar]
- 17. Sørensen J, Ploug UJ. The cost of diabetes‐related complications: registry‐based analysis of days absent from work. Econ Res Int. 2013;2013:1‐8. [Google Scholar]
- 18. National Board of Health and Welfare (NBHW) . National Patient Register. 2022. https://www.socialstyrelsen.se/en/statistics-and-data/registers/national-patient-register/. Accessed 2022 Jul 04.
- 19. Statistical database National Prescription Drug Register . 2018. http://www.socialstyrelsen.se/statistik/statistikdatabas/lakemedel. Accessed April 10, 2018.
- 20. Statistics Sweden . Longitudinal integrated database for health insurance and labour market studies (LISA) 2022. https://www.scb.se/en/services/ordering-data-and-statistics/ordering-microdata/vilka-mikrodata-finns/longitudinella-register/longitudinal-integrated-database-for-health-insurance-and-labour-market-studies-lisa/. Accessed 2022 Jul 04.
- 21. Försäkringskassan (the Swedish Social Insurance Agency) . Beställning av individdata för forskning [in Swedish, presention of data for research including the MiDAS database]. 2022. https://www.forsakringskassan.se/statistik/bestallning-av-individdata-och-statistik/bestallning-av-individdata-for-forskning. Accessed 2022 Jul 04.
- 22. Försäkringskassan (the Swedish Social Insurance Agency) . Social insurance system 2022. https://www.forsakringskassan.se/english/moving-to-working-studying-or-newly-arrived-in-sweden/social-insurance-system
- 23. Eliasson B, Ekelund J, Amberntsson R, Miftaraj M, Svensson AM. Cardiovascular disease in patients with type 2 diabetes and in patients starting empagliflozin treatment: nationwide survey. Diabetes Ther. 2019;10(4):1523‐1530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Zinman B, Wanner C, Lachin JM, et al. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med. 2015;373(22):2117‐2128. [DOI] [PubMed] [Google Scholar]
- 25. National Board of Health and Welfare (NBHW) . Sekundär klassificering till diagnosrelaterade grupper (DRG) [in Swedish; Secondary classification for Diagnosis Related Groups (DRG)]. 2022. https://www.socialstyrelsen.se/utveckla-verksamhet/e-halsa/klassificering-och-koder/drg/. Accessed 2022 04 Jul.
- 26. Swedish Association of Local Authorities and Regions (SALAR) . KPP databas [in Swedish; Cost Per Patient database]. 2022. https://skr.se/ekonomijuridikstatistik/statistik/kostnadperpatientkpp/kppdatabas.1079.html.
- 27. Kähm K, Laxy M, Schneider U, Holle R. Exploring different strategies of assessing the economic impact of multiple diabetes‐associated complications and their interactions: a large claims‐based study in Germany. Pharmacoeconomics. 2019;37(1):63‐74. [DOI] [PubMed] [Google Scholar]
- 28. Keng MJ, Leal J, Bowman L, Armitage J, Mihaylova B, Group tASC . Hospital costs associated with adverse events in people with diabetes in the UK. Diabetes, Obes Metab. 2022;24(11):2108‐2117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Magliano DJ, Martin VJ, Owen AJ, Zomer E, Liew D. The productivity burden of diabetes at a population level. Diabetes Care. 2018;41(5):979‐984. [DOI] [PubMed] [Google Scholar]
- 30. International Diabetes Federation (IDF) . IDF diabetes atlas 9th edition. https://diabetesatlas.org/atlas/ninth-edition/2019
- 31. Breton MC, Guenette L, Amiche MA, Kayibanda JF, Gregoire JP, Moisan J. Burden of diabetes on the ability to work: a systematic review. Diabetes Care. 2013;36(3):740‐749. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Nanayakkara N, Curtis AJ, Heritier S, et al. Impact of age at type 2 diabetes mellitus diagnosis on mortality and vascular complications: systematic review and meta‐analyses. Diabetologia. 2021;64(2):275‐287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Buse JB, Wexler DJ, Tsapas A, et al. 2019 update to: management of hyperglycaemia in type 2 diabetes, 2018. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetologia. 2020;63(2):221‐228. [DOI] [PubMed] [Google Scholar]
- 34. National Board of Health and Welfare (NBHW) . National guidelines for the care of diabetes. Support for governance and decisions. In Swedish: Nationella riktlinjer för diabetesvård. Stöd för styrning och ledning. In. The webpage of the National Board of Health and Welfare 2018.
- 35. Ludvigsson JF, Andersson E, Ekbom A, et al. External review and validation of the Swedish national inpatient register. BMC Public Health. 2011;11:450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Swedish National Diabetes Register (NDR) . Nationella Diabetesregistret ‐ Årsrapport 2016 [in Swedish; Swedish National Diabetes Register ‐ Annual Report 2016 ]. 2017; www.ndr.nu.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Population flow chart describing inclusion of people with diabetes and controls from national health and population data registers. Distribution type 1 diabetes and type 2 diabetes.
Figure S2. Study dataset with study subsamples presented as number of unique individuals and number of person‐years for type 2 diabetes and controls including selection subsample with type 2 diabetes and eCVD and their matched controls.
Figure S3. Data availability, population selection for diabetes and controls, exposure to eCVD, and study period for resource use, costs, and work absence. eCVD, established cardiovascular disease; NPDR, Swedish Prescription Drug Register; NPR, National Patient Register.
Table S1. Summary of data sources.
Table S2. Diagnostic codes and procedure codes used to identify established cardiovascular disease.
Table S3. List ICD‐10 codes or procedural codes (KVÅ) obtained from the National Patient Register, NPR, at the National Board of Health and Welfare, NBHW, 1997‐2016 in the existing study database.
Table S4. Description of the study dynamics for people with type 2 diabetes, with and without eCVD, during the study period: number of people, number of deaths, first observations with eCVD, and previous controls added to the sample because of observed diabetes status each year.
Table S5. Use of hospital care measured as number of outpatient visits and number of inpatient hospital days per 100 000 individuals in 2007‐2016. Comparisons of type 2 diabetes vs. matched controls, and type 2 diabetes with/without eCVD.
Table S6. Distribution of demographic and socio‐economic characteristics for type 2 diabetes and matched controls. Measurement in the first year with observed type 2 diabetes (2007, or 2008‐2016 if switch to type 2 diabetes during the study period).† Employment and work absence only reported for sample in labour‐market active ages (<66 years).
Table S7. Costs of outpatient visits and inpatient admissions per 1000 person‐years 2007 to 2016 for people with type 2 diabetes compared to matched controls, and subgrouped for people with type 2 diabetes on eCVD and no eCVD. Data are costs in Euro per 1000 person‐years per study group based on real‐world use of hospital‐based care 2007 to 2016.
Data Availability Statement
Individual‐level data hosted by national authorities in Sweden are available for research after formal evaluation of the research protocol by the ethical review authorities and by the respective national authorities providing data. Permission to conduct research are granted on a case‐by‐case basis to a limited number of named persons. The secondary individual‐level data of this study cannot be shared by the authors for this reason. Informed consent is not required for register data research in Sweden.