This is the first nationwide study to estimate the prevalence and socioeconomic burden of diabetes using a nationally representative claims database in South Korea. This study showed that the prevalence of diabetes in Korean adults was 10.7%. The diabetes-related economic burden was USD 18,293 million, with an average per capita cost of USD 4,090 in 2019. Our results also showed that type 2 diabetes, presence of diabetic complications or related comorbidities, diabetes medication use, and hospitalization were associated with a large economic burden of diabetes.
The estimated prevalence of diabetes in Korean adults was slightly higher than the global prevalence of diabetes in the adult population (8.5%) (3). According to the Korean National Health and Nutrition Examination Survey (2007–2009), 11.0% of males and 8.9% of females among the adult population had diabetes (11). These results are similar to ours (11.4% in males and 10.1% in females) considering the increasing trend in the prevalence of diabetes.
The economic burden of diabetes was USD 18,293 million in South Korea in 2019, which is equivalent to approximately 1.14% of Korea’s gross domestic product (GDP). A study published in the U.S. showed that the estimated national cost of diabetes in 2017 (USD 327 billion) accounted for 1.69% of the GDP (1). It could be said that the relatively high percentage of GDP in the U.S. is caused by the high per capita cost for diabetes (USD 16,752), especially considering that the prevalence of diabetes in the U.S. is slightly lower (9.7% of the adult population) than it is in Korea. Additionally, the total annual cost of major diseases in Korea, such as cancer (21), liver disease (22), and cardio-cerebrovascular disease (23, 24), has been reported in the range of USD 1 billion to 3 billion, which was much lower than the total cost of diabetes. Moreover, the economic burden of diabetes was higher than the economic burden for overall cancers, at USD 15 billion (21).
Type 2 diabetes is the most common type of diabetes, accounting for approximately 90% of all cases of diabetes (7). Our study showed that 88.0% cases of all types of diabetes were type 2, which also accounted for most of the economic burden associated with diabetes (86.9%). In Korea, the direct medical costs of type 2 diabetes corresponded to 10.6% of all healthcare expenditure (USD 85.5 billion, calculated as only direct medical costs excluding out-of-pocket costs (25)). This was higher than the cost of type 2 diabetes in France, where it corresponded to ~ 5% of all healthcare expenditures (26). The economic burden makes type 2 diabetes a major clinical and public health problem in Korea (2). The type 2 diabetes are associated with overweight and obesity. Therefore, efforts to reduce the global health and economic burden of diabetes should emphasize the prevention of type 2 diabetes, or delaying its onset, by promoting healthy behavior and diet at the population level (4).
As with most other diseases, elderly patients require more healthcare resources to treat diabetes than younger patients: approximately half of all health care expenditure related to diabetes account for health resources used by those over the age of 65. Adult patients over the age of 65 in this study spent approximately 1.2 times more in annual per capita cost than those under the age of 65. Because productivity loss was assumed to be zero for people over the age of 65, indirect costs included only caregivers’ costs related to hospitalization. High medical expenditures among the elderly, along with high caregivers’ costs, can be partly attributed to the increased risk of hospitalization that comes with aging. The hospitalization rate of those over the age of 65 among our study subjects was approximately 1.5 times that of those under the age of 65 (28.2% vs. 18.9%).
We confirmed that hospitalization was a cost driver, which is associated with high costs for diabetes (2). Inpatient services accounted for 42.7% of the total economic costs for all diabetes patients. Diabetes incurred higher spending for inpatient services than other diseases that are common among the elderly, such as hypertension (18.3%), rheumatic arthritis (7.9%), heart failure (29.2%), and asthma (11.7%) (27–29). Furthermore, the total costs for diabetes patients with an experience of hospitalization represented 76.4% of the total economic costs for all diabetes patients, and the per capita cost in inpatient settings was much higher than that in outpatient settings, by 10.8 times. Our results show that inpatients represented a higher percentage of the elderly (55% vs. 42%) and those with complications (71.1% vs. 53.4%) than did outpatients. Patients requiring hospitalization generally had severe conditions, and because of the high cost of premature deaths from such conditions, indirect costs for inpatients accounted for a higher proportion of total costs (34% vs. 8%) than for outpatients. These findings suggest that an effective intervention to prevent hospitalization should be a critical component of a disease management strategy to minimize the economic and clinical burden of diabetes.
Because most patients with diabetes (75.8%) have been prescribed diabetes drugs, the total economic burden of patients using diabetes medication was high. However, per capita cost was similar between patients using diabetes medication and those with no use of diabetes medication, and indirect costs accounted for a higher proportion of the total for patients with no use of diabetes medication (37%) because of the higher costs incurred by diabetes complications and premature deaths than for those using diabetes medication (25%). Continuous treatment of diabetes is particularly important for preventing diabetes-related complications (30). In particular, one large cohort study found that improving antidiabetic medication adherence among newly diagnosed type 2 diabetes patients decreased the risks of macrovascular complications (31). Therefore, receiving medication reduces the clinical burden at the individual level and also reduces the socioeconomic burden by reducing indirect costs at the population level.
Our study found that per capita costs of diabetes increased with the number of complications or related comorbidities. As the number of complications increased, hospitalization also increased; in particular, the percentage of hospitalizations through emergency departments increased from 19.8–43.6% (data not shown). This result is in line with that of other studies that showed that medical costs incurred by complications led to a high economic burden in diabetic patients (8, 32). The American Diabetes Association reported how diabetes contributed to the economic costs of major complications in the healthcare system: the proportion of expenditures attributed to diabetes for peripheral vascular, neurological, renal, and cardiovascular diseases over total U.S. health care expenditures (39%, 36%, 29%, and 27%, respectively) was higher than that for other general medical conditions (8%) (1). We confirmed that peripheral vascular disease was the most common complication in Korean adult diabetic patients, followed by neurological, cardiovascular, and renal diseases. Therefore, management of diabetic patients with these major complications is expected to significantly reduce not only diabetes but also the economic burden of these diseases.
This study had several limitations. First, study subjects with diabetes were identified only based on ICD-10 codes, which potentially allowed misclassification or miscoding. Because the HIRA-NPS data do not provide information on laboratory test parameters such as fasting plasma glucose, oral glucose tolerance, and HbA1c levels, we were not able to confirm diabetes cases based on diagnostic test results. However, a previous study indicated that diabetes could be accurately identified in administrative data: The definition of diabetes, 2 physician claims within 1 years or 1 hospitalization with the ICD-10 codes E10.x–E14.x, had high validity (sensitivity 91.6%, specificity 97.2%) (13). Therefore, we consider that the administrative data can be used to establish the population-based prevalence of diabetes as a reasonable alternative to biochemical assay data (33). Second, health insurance claims data did not include information about subjects with undiagnosed or untreated diabetes; hence, the prevalence as well as the cost of diabetes might be underestimated. It was estimated that more than 50% of adults with diabetes in the Western Pacific region were undiagnosed according to the IDF Diabetes Atlas (6). Won et al. (2018) also reported that the estimated prevalence of undiagnosed or diagnosed diabetes was 13.7% during 2013–2014 in Korean adults (≥ 30 years of age) (34). However, our study was conducted among diagnosed patients who had paid for healthcare service using very comprehensive health care claims data that cover the nationwide Korean population. We regard our results to be conservative in terms of estimating COI and proper as being representative for the prevalence of diabetes in South Korea. Third, cross-sectional studies using claims data make it difficult to identify causal relationships between diabetes and its complications or related comorbidities. Thus, healthcare resource use by diabetes patients with related complications can be overestimated. To reduce the potential of overestimation in our definition of cases with complications, we excluded those for whom diabetes and complications were not diagnosed in the same prescription, although complications may occur within a certain period after the initial diagnosis of diabetes. Additionally, we mainly used diabetic complication codes (such as E10.1-E10.5 to E14.1-14.5, Additional File 2) to increase the association with diabetes, and all codes defining complications or related comorbidities were validated (1, 14). The known prevalence of diabetic complications or related comorbidities varies from one country to another; the prevalence of diabetic complications in Korea was similar or slightly lower than the average prevalence globally (7). Despite these limitations, we believe that the administrative data used in this study provide a powerful resource for a population-based evaluation of the economic burden of diabetes (33).