Type 2 Diabetes mellitus (T2DM) affects people at all ages. Nearly, 200 million people (or over 5% of the global adult population) have diabetes worldwide, and this number will increase up to 333 million (or 6.3% of the global adult population) by 2025 [1]. In the developing world, the prevalence was highest in Europe and Central Asia and lowest in Sub-Saharan Africa. In China, the total number of diabetes patients is currently 92.4 million [2, 3] and expected to double by 2030 [4].

T2DM is also a heavy economic burden on national economies and healthcare systems, especially in developing countries[4]. The direct healthcare costs of diabetes range from 2.5 to 15.0 percent of annual healthcare budgets, depending on local prevalence and sophistication of the treatments available[5]. In China, it is expected that the economic burden associated with T2DM will increase dramatically in the next few decades with the sharp increase of the number of diabetic patients. Previous studies suggested that healthcare expenses in people with diabetes were more than two times greater than that in those without [6, 7]. Major costs for the disease include medical expenditures for blood glucose level control and additional medicine and service for diabetes related chronic complications, which may lead to excess costs and additional consumption of healthcare services [8].

Several factors were reported to have impact on overall economic burden in diabetic patients. Studies showed that level of Hemoglobin A1c (HbA1c) and insulin treatment, were highly associated with total medical costs [9–12]. Some other studies suggested sharing of health expenditure, an index of equity in healthcare financing, was associated with medical costs, and affected by clinical characteristics and demographic factors as well [13, 14]. Unreasonable healthcare utilization, i.e., occurrence of hospitalization and number of outpatient visits, which was associated with the increased healthcare costs of T2DM [13, 15], was affected by insurance schemes and income levels [16, 17].

The burden of diabetes has been a big challenge on clinical and public health. So far, only limited studies have been reported to identify the impact factors of medical costs and healthcare utilization in T2DM population. Moreover, those studies rarely addressed the issues in the setting of developing countries. An analysis of the demographic, socioeconomic and clinical influencing factors that affect medical cost, sharing of expenditure and health utilizations will help better understand the increasing medical cost of T2DM, especially in developing countries. The purpose of our study is to investigate the determinants of direct medical cost (DMC), out-of-pocket proportion (OOP) of the cost, and healthcare utilization associated with T2DM in a population of T2DM outpatients by a cross-sectional survey conducted in four Chinese cities in order to help healthcare policy makers to control the high medical costs of T2DM and reduce the inequity across different socioeconomic populations.

This cross-sectional study provided quantitative analysis of the changes in healthcare expenditure and healthcare service utilization caused by the incremental changes of T2DM related complications, medical insurance schemes, individual income and the levels of admission hospitals. To our knowledge, there is no similar data are available for the comparison in China as well as in other developing countries. Although a number of cost-of-illness studies have indicated that T2DM related complications significantly increased the costs for patients [16, 19–22], there have been few epidemiological or clinical studies to systematically address the influncing factors of health expenditure of T2DM, especially in developing countries like China.

As expected, there was an association between complications and annual DMC. By using the GEE model, we found that for each increase in number of complications, there was a 33% increase in annual DMC, which quantitatively proved that the high prevalence of diabetic patients who currently live with complications (52% of the outpatients) has significantly increased the economic burden of the type 2 diabetes. Moreover, each increase in the number of complications resulted in extra 6% hospital visits, which may explain the 33% increase of DMC by complications. Therefore, these results strongly indicate that the prevention and treatment of diabetes-related complications could therefore not only improve the life quality and survival of patients but also substantially reduce costs.

Our study show that different insurance schemes were significantly associated with increased annual DMC of T2DM. Although not significant, those covered by urban employee insurance or resident insurance plan had a 24% increase in annual DMC than those without insurance plan; commercial insurance covered patients were found having a significantly higher annual DMC (28% increase). The result implies that the insured patients may overuse the service, as indicated by the widely accepted conventional theory of health insurance based on the concept of moral hazard [23], which is characterized by incremental average cost.

The latest Chinese national health survey in 2003 also revealed that about 73% of people in rural areas who should have sought medical treatment chose not to do so due to cost [24]. However, the results did not correlate the outpatient visit with insurance schemes, which had been reported in other settings [25, 26]. It implies that the types of insurance plans might affect the average charges of medical cost per visit, other than frequency of seeking healthcare services for T2DM under the current Chinese Health Insurance System. In fact, the patients covered by health insurance scheme(s) had significantly reduced the proportions of medical cost from pocket, with 19% to 44% decrease in comparison to those uncovered.

China has a urban health insurance system that provides virtually sound coverage for people employed in urban state enterprises and relatively inexpensive coverage for their families. The situation for workers in the rural areas or in urban employment outside the state sector is far more varied. Currently, about 55% of Chinese urban residents are insured through this urban programs[27]. For those who were from rural areas and thus did not hold legal permanent urban residency status might benefit from the Rural Cooperative Medical System. Under this plan, each rural participant pays only 10-20 CNY per year and is then eligible to receive partial coverage of medical expenses, which ranges up to 65 percent of total medical expenses[28]. There were few exceptions for the above two plans, e.g., for unemployed urban residents, who may get insured with a relatively lower reimbursement proportions through a so-called residents insurance plan. When the above plans are absent, patients have to rely on commercial health insurance or individually pay all medical costs.

We found patients with higher income (> 2000 CNY/month) had 23% less payment from their pocket in comparison to those with lower income. Under current China's healthcare system, patients have to pay for treatment out of their own pockets with large differences in quality and access among income groups [29]. Under the current health insurance system in China, the higher income patients are more likely to be covered by insurance or have a better insurance coverage than those with lower income. Our result regarding the OOP is consistent with the known fact. It should be particularly noted that the fact may lead to a potential vicious circle - the poorer the patients are, the relative higher propotion of OOP cost they incur. The result is more worthy of note when it was shown that more than 35% of urban households of China had difficulty to afford healthcare cost, or were poverished by the health costs [30]. The employed outpatients paid less DMC but had to pay more from their pocket (OOP), in comparison to unemployed subjects. It may be caused by a fact that many of the unemployed people were retired residents with high prevalence of chronic diseases but subjected to a lower level of OOP according to the urban insurance plan[31].

In response to growing public concerns over widening inequalities in health, the Chinese Government officially launched another round of the national health system reform plan in early 2009 with a target of 90% health insurance coverage by the end of 2010 and universal coverage of essential healthcare by 2020. [32] Assuming that the reform would be successful, the control of the costs from diabetes may be more likely to depend on the re-distributions of patients in different levels of hospitals or community health centers. In this study, the outpatients admitted at secondary hospitals had 23% more hospital visits, but had 7% less payment of DMC than those seeking medical services at tertiary hospitals. This result is consistent with a study indicating that 6% (997/17148) of tertiary hospitals nationwide consumed 55% of total healthcare expenditure in China [33]. As T2DM can be mostly managed in primary care, our study suggested an urgent need of redistribution patients from tertiary hospitals, charcterized by providing comprehensive and intensive medical services, to facilities where basic healthcare services is easy to be accessed with lower prices.

Furthermore, the association between insulin therapy and higher healthcare costs, that has been reported in other settings(18, 19), was also confirmed in our study. However, we did not find the association between DMC and HbA1c, which had also been reported in other studies [10, 34]. The reason may be that cost savings were only significant among a portion of patients who have the highest baseline HbA1c levels (≥10%) but not in patients who have moderate or low HbA1c levels.

The findings from this study may be subject to several limitations. First, the subjects of this study were patients at secondary and tertiary hospitals, which may not be applicable to other groups of patients. The generalizability of the results was also limited by the geographic and demographic characteristics of the study population, considering that only 4 cities were selected as the sites. Second, the cost data of this study, especially for direct non-medical costs and indirect costs, were partially based on patient recall, which may introduce a recall bias. The DMC data that was based primarily on medical history combined with billing information, however, was more precise and accurate. Finally, it was difficult to differentiate T2DM-related complications from coincidental diseases, particularly among the elderly, and possibly some patients were misdiagnosed and therefore stratified incorrectly.

T2DM related DMC varied with disease characteristics (e.g., the number of complications), whereas the OOP proportion of DMC was largely associated with socioeconomic factors. Our study implied that preventing complications through the use of more effective treatment regimens is important in order to control the healthcare costs of the diseases. With early detection and treatment, it is likely that the serious health consequences of diabetes can be prevented or delayed and thus the costs can be reduced. To reduce the inequity in healthcare expenditures and utilizations of T2DM in different groups, healthcare reform needs to be focused on the reforms of medical insurance system and redistribution of patients in different levels of hospitals.