National trends in the prevalence of glycemic control among patients with type 2 diabetes receiving continuous care in Thailand from 2011 to 2018

Diabetes is one of the largest global health problems and exhibits a constantly increasing trend. A series of nationwide hospital-based cross-sectional surveys of clinical outcomes was performed annually from 2011 to 2015 and 2018 among patients with type 2 diabetes aged ≥ 20 years receiving medical care for at least 12 months. A two-stage stratified cluster that was proportional to the size sampling technique was used to select a nationally and provincially representative sample of patients with type 2 diabetes in Thailand. A total of 186,010 patients with type 2 diabetes were enrolled in the study from 2011 to 2018. The prevalence of adequate glycemic control (hemoglobinA1c level < 7.0%) among patients with type 2 diabetes were estimated to be 34.5% (95%CI 33.8–35.2%) in 2011, 33.0% (95%CI 32.4–33.6%) in 2012, 34.7% (95%CI 34.1–35.4%) in 2013, 35.5 (95%CI 34.9–36.1%) in 2014, 35.6 (95%CI 35.0–36.2%) in 2015, and 35.6% (95%CI 35.0–36.2%) in 2018, respectively (p for trend < 0.001). Independent factors related to poor glycemic control (hemoglobinA1c ≥ 7%) were being female, younger aged, living in the northeastern region, received care form hospitals lower than regional level, under universal health coverage scheme, greater duration of diabetes, higher body mass index level and absence of hypertension comorbidity.

www.nature.com/scientificreports/ Asian countries 2 . Additionally, Asians may be more susceptible to the development of type 2 diabetes and may have a higher risk of developing diabetes complications than other ethnicities 3 .
In Thailand, the 5th National Health Examination Survey (NHES V) conducted in 2014 found that the prevalence of diabetes among Thais aged over 20 years was 9.9, 8.9 and 10.8% among total adults, males and females, respectively 4 . Nevertheless, the percentage of undiagnosed type 2 diabetes was 51.2 and 41.3% among male and female patients with diabetes, respectively. Additionally, 4.2% of males and 1.7% of females received a diagnosis type 2 diabetes but were untreated. In terms of residence, the prevalence of type 2 diabetes, undiagnosed and controlled among Thai adults in urban and rural areas were comparable 4 . Currently, diabetes is a major health problem in Thailand. The disability-adjusted life years from diabetes in Thailand constituted the 7th leading cause of morbidity among men and the 2nd leading cause among women in 2013 5 . Proper glycemic control will reduce the risk of diabetes complications, particularly microvascular complications 6,7 . Adult diabetes cases benefit from proper glycemic control, indicated by glycosylated hemoglobin (HbA1c) levels < 7% (53 mmol/mol) 8 . The HbA1c level was adopted as the assessment method for glycemic control by the Diabetes Clinical Practice Guidelines from the National Health Security Office (NHSO), Thailand, in 2011 9 . The indicated target for the attainment of glycemic control is an HbA1c value < 7% (53 mmol/mol) in nonpregnant adult diabetes cases.
In 2015, many countries worldwide began to move toward providing universal health coverage based on the Sustainable Development Goals. In Thailand, universal health coverage was implemented in 2002 10 and included free services for all basic health problems, including type 2 diabetes. The present study comprised an annual national study among patients with type 2 diabetes aiming to determine national trends in the prevalence of glycemic control among patients with type 2 diabetes receiving continuous care in Thailand from 2011 to 2018 which was a decade after implementing the universal health coverage scheme in Thailand. Furthermore, we identified factors associated with poor achievement of glycemic control.

Methods
Study design and participants. A series of annual cross-sectional studies was performed from 2011 to 2015, and 2018 to evaluate the status of diabetes care among patients with type 2 diabetes attending public hospitals of the Ministry of Public Health (MoPH) nationwide in Thailand. The participating hospitals also included public hospitals and private clinics in the Bangkok Metropolitan Region supported by Thailand's NHSO program. The inclusion criteria comprised patients with type 2 diabetes aged 20 years in 2011, 2012, 2014, 2015 and 2018 studies and aged 35 years in the 2013 study and older who visited clinics and received medical care in the targeted hospitals for at least 12 months (Fig. 1). Patients participating in any other clinical trials were excluded.
A stratified two-stage cluster sampling proportional to the size method was used to select a nationally and provincially representative sample of patients with type 2 diabetes in Thailand. For each province outside the Bangkok Metropolitan Region, the hospital universe is defined by all hospitals that are public hospitals under the MoPH. For the Bangkok Metropolitan Region, the hospital universe is defined as all hospitals and clinics that participate in Thailand's NHSO program. The study sample was a stratified sample drawn from a subset of all MoPH hospitals in Thailand, including all public and private clinics in the Bangkok Metropolitan Region under the NHSO program. Hospitals were stratified in two levels. The first level was the province, which constituted 77 strata (from the 77 provinces in Thailand), and the second level was the hospitals in each province. The hospitals in each province were stratified in 5 strata based on their sizes as follows: regional center hospital (> 500 beds), provincial general hospital (200-500 beds), large community hospital (90-120 beds), medium-sized   11 . Population weighted numbers, weighted prevalence rates, and odds ratio (OR) were calculated. The prevalence of outcome indicators was described as a percentage with a 95% confidence interval (95% CI). P for trend was calculated using chi-square statistics. ANOVA was used for statistical analysis to determine differences between the means of the continuous outcomes of interest. Binary logistic regression analysis was used to determine the risk factors for poor control of HbA1c (HbA1c ≥ 7.0%), and the magnitude of association was presented as crude odds ratio (OR) with 95%CI. Multivariate analysis was performed using logistic regression analysis. Adjusted odds ratio (AOR) from multivariate analysis was presented with corresponding 95% CI, and a p-value less than 0.05 was considered statistically significant.

Results
Demographic characteristics. From 2011 to 2015, and 2018, a total of 186,010 patients with type 2 diabetes attending the participated clinics were enrolled in the study, including 26,860, 28,941, 27,512, 33,288, 32,616 and 36,793 patients in 2011, 2012, 2013. 2014, 2015 and 2018, respectively. Characteristics of the enrolled patients with type 2 diabetes by survey years are presented in Table1. The mean age of the enrolled patients was 61.1 ± 10.9 years. Approximately one half of the participants were agriculturists. Almost one third (31.5%) of the enrolled patients were male. The majority (74.6%) of the study participants were under the universal health coverage scheme, providing free health care for all Thais. Approximately three fourths (74.4%) of patients with type 2 diabetes had comorbid hypertension. The mean duration of the type 2 diabetes diagnosis was 7.3 ± 4.8 years. One half (50.4%) of the enrolled patients had a BMI ≥ 25 kg/m 2 . Additionally, we observed an increasing trend among the patients with a BMI ≥ 30 kg/m 2 during the study period (p for trend < 0.001) ( Table 2 and Fig. 2). The majority (66.2%) of patients were enrolled from community hospitals located at the district level in Thailand. A total of 20.1, 33.5, 32.6, and 13.8% of the patients were located in the north, central, northeast, and south regions, respectively.
According to the provided patient care processes and clinical outcomes, the percentage of patients with type 2 diabetes receiving a fasting plasma glucose test during their last follow-up hospital visit increased from 82.1% in 2011 to 95.1% in 2018 (p for trend < 0.001). We observed a decreasing trend for patients with type 2 diabetes achieving the target fasting plasma glucose level (FPG) level (70-130 mg/dL or 3.89-7.22 mmol/l) at their last follow-up visit from 40.8% in 2011 to 36.8% in 2018 (p for trend < 0.001) ( Factor associated with poor glycemic control among patients with type 2 diabetes. Regarding factors related to uncontrolled (HbA1c ≥ 7% or 53 mmol/mol) type 2 diabetes, univariate analysis revealed that age, sex, region of residence, hospital level, health insurance scheme, BMI, duration of type 2 diabetes diagnosis and comorbidity with hypertension were significantly associated with poor glycemic control ( Table 3). As shown using multivariate analysis (

Discussion
We successfully enrolled 186,010 patients with type 2 diabetes receiving care at public hospitals and clinics from 2011 to 2015 and 2018 nationwide. This study constituted the largest epidemiological study among patients with type 2 diabetes in Thailand to date. We found that one third of patients with type 2 diabetes receiving continuous care in Thailand maintained adequate glycemic control (HbA1c < 7% or 53 mmol/mol). Because our study population was selected from those patients attending the clinic on the enrollment date the outcomes from our study may represent only patients who were relatively engaged in care. Approximately two thirds of the enrolled participants with type 2 diabetes were female. However, data from a National Health Examination Survey in Thailand in 2014 (the NHES V) showed that the prevalence rates of type 2 diabetes among people older than 20 years were relatively comparable, with prevalence rates of 8.9% (95%CI 8.3-9.5) and 10.8% (95%CI 10.2-11.4) among males and females, respectively. Nevertheless, the percentage of undiagnosed type 2 diabetes was 51.2% (95%CI 45.9-56.6) and 41.3% (95%CI 36.6-46.1) among males and female patients with type 2 diabetes in Thailand, respectively 4 . This phenomenon of sex difference in type 2 diabetes care may be in part explained by gender values and health seeking behavior in that male patients exhibited lower levels of health awareness including seeking out and engaging in diabetes self-management behaviors 12,13 . The higher ratio of female patients with type 2 diabetes receiving diabetes care in hospitals compared with the ratio found in the community may reflect the lower accessibility of care among male patients with type 2 diabetes in Thailand. This finding indicates that specific programs aimed at increasing early detection and accessibility to continuous care among male patients with diabetes require further attention in Thailand.
The current universal health care policy in Thailand comprised 3 major health care schemes: the civil servant medical benefit scheme (for all civil servants and their immediate family members, ~ 5 million people), the social security scheme (for private employees, ~ 10 million people) and the universal health coverage scheme for the rest of the Thai population (~ 50 million people). All of these schemes provide free medical care for patients, including patients with type 2 diabetes. Our study found that patients with type 2 diabetes under the civil servant medical benefit scheme had a lower proportion of uncontrolled (HbA1c ≥ 7% or 53 mmol/mol) type 2 diabetes (59.57%) than patients under the universal health coverage scheme (66.65%). However, these proportions did not differ significantly from patients under the social security scheme (66.23%). These findings suggested that an opportunity to improve the quality of type 2 diabetes care exists in the universal health coverage and social security schemes.
We also found a difference in diabetes control among hospital levels. Patients with type 2 diabetes from regional hospitals had a lower proportion of uncontrolled diabetes (HbA1c ≥ 7% or 53 mmol/mol) than that of patients from general and community hospitals. One explanation for this finding is that the more comprehensive care provided by specialists in regional hospitals may produce better clinical outcomes.
The proportion of patients receiving a blood test for fasting plasma glucose, HbA1c, and LDL increased annually, which might indicate an increase in care quality. The mean HbA1C levels among the enrolled patients with type 2 diabetes were 8.03% (64.3 mmol/mol) in 2011 and slightly decreased to 7.97% (63.6 mmol/mol) in 2018. The mean HbA1C level found in this study was comparable to that in a study conducted in Malaysia during the www.nature.com/scientificreports/ same period 14 . A total of 34.47% and 35.59% of the patients in this study achieved the target HbA1c of less than 7% (53 mmol/mol) in 2011 and 2018, respectively, which was also comparable to the data from Asian countries including Malaysia 14 , Saudi Arabia 15 and China 16 . However, the percentages of controlled patients with diabetes from the present study were lower than those in the reported data from the US from 1999 to 2010 (52.2%) 17 . Women in this study were more likely to have poor glycemic control than male patients, which was consistent with other reports [18][19][20][21] . One explanation for this sex difference in glycemic control may be related to biological factors, including the effects of female sex hormones on the action of insulin 22 . Additionally, the observed www.nature.com/scientificreports/ differences may be due to differences in health care behavior, including lower treatment adherence among women 23 . In our study, younger patients with type 2 diabetes exhibited poorer glycemic control than older patients, which was consistent with findings from other settings, including the US 17 . Older patients may be more motivated to take care of their diabetes and more compliant with their medication adherence and healthy eating behavior 24 . Younger patients may be more likely to disregard diabetes as important and be less adherent to medication, lifestyle modifications and diet restrictions 25 . A related study found that the persistence of HbA1c elevation among younger individuals could be due to an inadequate low medication dosage or the infrequent use of combined drug regimens 26 . Additionally, younger patients are also less likely to have an established diagnosis of hypertension which might have mediated the observed association between poorer glycemic control and the absence of a diagnosis of clinical hypertension. Therefore, younger patients with type 2 diabetes should constitute a target population to improve glycemic control and other diabetes care. More specific programs customized for this target working-age population should be implemented in Thailand. An effective program for younger patients would be beneficial and decrease the burden of diabetes complications in the near future.
This study found that patients with type 2 diabetes from the northeast region had a lower percentage of adequate glycemic control than patients from other regions. Factors related to the health system and personal factors may play roles in the poor glycemic control among patients from the northeast. The northeast region has the lowest physician-to-population ratio relative to the rest of the country 27 . The relatively limited health care resources in this region might have influenced the outcome of diabetes care. In terms of personal factors, glutinous rice is a staple food item in this part of the country. Glutinous rice has a high carbohydrate content, high glycemic index and a high risk of hypertriglyceridemia 28 compared with nonglutinous rice, which is the staple food of the other regions.
We found that patients with type 2 diabetes with greater BMI level tended to be at risk for poor glycemic control. Similarly, one related study in the US reported positive associations between being overweight or obese and having uncontrolled type 2 diabetes 29 . Additionally, strong evidence was found that obesity management www.nature.com/scientificreports/ could be beneficial in treating type 2 diabetes 30,31 . Related studies have reported that the sustained weight loss has been illustrated to improve glycemic control 32,33 . Thus, the dietary therapy and weight management should be encouraged for patients with type 2 diabetes to better glycemic control. One limitation of this study was the representativeness of the study subject. The study included only patients with type 2 diabetes visiting a hospital for diabetes care and did not include patients with type 2 diabetes receiving care at primary care units, accounting for approximately one half of the overall patients with type 2 diabetes in Thailand in 2015. Additionally, the study did not include subjects from university hospitals. Because this study was the first endeavor to examine type 2 diabetes care outcomes systematically, nationwide in Thailand, we did not have an opportunity to compare the outcomes with the period before the implementation of universal health care coverage in 2002. Limitations of the validity related to factors associated with uncontrolled type 2 diabetes may have occurred in this study. Because the HbA1c blood test was used to define uncontrolled patients with type 2 diabetes (HbA1C > 7% or 53 mmol/mol), the analysis was limited to the group of patients tested for HbA1c during the previous 12 months (72.70, 76.23, 78.80, 77.00, 79.69 and 76.39 in 2011, 2012, 2013, 2014, 2015 and 2018, respectively). Patients not receiving HbA1C tests were not randomized and were more likely to be under the universal health coverage scheme, visit community hospitals, have a shorter duration of diabetes diagnosis and reside in the northern region of the country. Moreover, we only collected data from May to August of each year between 2011 and 2018, although glycemic control may vary by season [34][35][36][37] .

Conclusion
In conclusion, the study examined the current standards of diabetes care nationwide in hospitals in Thailand. The results showed room for further improvement in the quality of diabetes care. The study findings are useful for health care providers comparing performance and planning quality improvement initiatives. Variations in the outcome of diabetes care were observed among patients of different ages, sex, durations of diabetes, and hypertension comorbidity status and among different hospital levels. At the policy level, the pay-for-performance and chronic disease management programs discussed earlier may improve the current situation 38 . At the hospital level, regular educational activities, practice guideline development and clinical audits are useful for continuous quality improvement.