This study examined patient factors associated with quality indicators for diabetes care in South Korea. Our study is unique in that it included the quality indicators of preventive care processes for diabetic complications and diabetes-related clinical outcomes comprehensively. In this study, we found that those with lower education level and shorter duration of diabetes illness had relatively less experience in receiving preventive care services for diabetes complications. Patients with longer duration of diabetes illness and women showed poor glycemic control. Additionally, obese diabetic patients were less likely to accomplish adequate control of blood pressure and LDL-cholesterol.
Our analyses resulted in 5 key findings. Firstly, our findings showed that people with lower education level were less likely to have received fundus and microalbuminuria examinations in the past year, and education about diabetes as preventive care services for diabetes complications. Educational level, the most widely used measure of socioeconomic position, imparts health-related knowledge capacity, reflects access to resources including preventive health care, and determines health behaviors [19
], which may explain our results. We can assume that those with a lower level of education have inadequate understanding about possible diabetic complications and fewer opportunities to meet physicians, and thus are less likely to receive appropriate preventive care services for diabetes complications. Previous studies have shown that diabetes patients with lower levels of education have lower rates of eye examinations [21
]. A study in 2011 demonstrated that lower education level was associated with poor achievement of services such as dilated eye examination, microalbuminuria test, and diabetes education [23
]. These results can imply that stronger public health efforts are needed to increase rates of receiving preventive care for diabetes complications among those with lower education levels. On the other hand, education level was not associated with diabetes-related clinical outcomes in our study. Additionally, Haffner SM et al. found no association between education level and glycemic control in Mexican Americans [24
]. Our results revealed the discordance in association of education level with preventive care services for diabetes complications and diabetes-related clinical outcomes. After controlling for other covariates, education level could be an independent factor associated with receiving preventive care services for diabetes complication, but not with clinical outcomes.
Secondly, income level did not seem to affect quality indicators in this study. Several studies found that lower income was related to poor quality of diabetic care [23
]. A variety of reasons explaining this association have been suggested, such as medical costs of preventive care and medication as well as psychosocial factors [23
]. In our study, however, there were hardly any differences in the quality indicators based on income level, which could be attributable to the use of a universal health insurance system in South Korea. The National Health Insurance covers the majority of citizens (96%), while the Medical Aid program covers the poor and other specified groups (4%) [27
]. Since the financial burden is low within the healthcare system, individual income level is less likely to play an important role [28
Thirdly, in our study, shorter duration of diabetes illness was associated with lower levels of receiving preventive care services for diabetes complications and longer duration was associated with lower achievement of recommended glycemic goal. In the cross-sectional analysis of Indian Health Service Diabetes Care and Outcomes Audit, < 5 years’ duration of diabetes illness was strongly associated with poor reception of preventive care services for diabetes complications [29
]. Recently-diagnosed diabetes patients may have lower reception of preventive care for diabetes complications because they have not received as much education about diabetes complications and recommendation to get preventive care from physicians than patients with longer diabetes duration. Additionally, patients who have been recently diagnosed might have fewer complications, and thus are not as motivated to receive preventive diabetes care [30
]. preventive care services for diabetes complications According to the UK Prospective Diabetes Study (UKPDS), the proportion of patients who achieved the target HbA1c level below 7% HbA1c was only 50% of the patients and this percentage decreased dramatically as the duration of illness increases even with intensive treatment [6
]. The major reason for poor glycemic control in people with long duration of diabetes may be due to disease progression following the natural course of diabetes.
Fourthly, women were less likely to reach the glycemic target level. Up to date, there is a lack of consistency in the findings about gender differences in glycemic control among patients with diabetes [31
]. A few studies have indicated that female patients are at greater risk of not achieving the recommended HbA1c levels [31
], while other studies found no significant gender-related differences [33
]. There may be several possible explanations for lower achievement rates of glycemic target levels in female diabetes patients, such as fewer opportunities for treatment, less aggressive treatment, sex-based physiology, in which therapeutic interventions such as diet or drug therapy may not be as effective [31
]. Gender differences in our study were detected only for glycemic control. In contrast, in a cross-sectional study of 5082 men and 4293 women in Sweden, female patients had poor control for all clinical outcomes (HbA1c, BP, and LDL-cholesterol) than corresponding male patients in the subgroup aged 60-75 [32
]. Further studies with larger data using age-group analyses are necessary for investigating gender differences in clinical outcomes management in Korea.
Finally, obese patients with diabetes were less likely to meet the target values of BP and LDL-cholesterol, although we failed to prove an association between specific obesity grades and glycemic control. In one study, Harris et al. revealed that BMI was not related to glycemic control, which is consistent with our result [35
]. One possible explanation for this is that BMI could both affect glycemic control and be influenced by the level of glycemic control. For example, a low BMI causes insulin sensitivity and therefore good glycemic control. Improved glycemic control causes weight gain, which is a finding consistent with the UKPDS, where the intensive group gained 2 to 5 kg compared with the conventional group [6
]. As weight increases, glycemic control worsens over time; thus BMI and glycemic control are coupled with bi-directional influence. Furthermore, overall obesity as determined by BMI values in our study could be less strongly associated with insulin resistance and poor glycemic control than abdominal obesity [36
]. Similar to our results, previous studies have also showed a positive association between BMI and BP or LDL-cholesterol [37
]. This provides a meaningful lesson for physicians to pay attention to BP and LDL-cholesterol levels among obese diabetes patients.
There are some limitations in our study. Firstly, since our study was based on a cross-sectional design, there was no information on the temporal relationship and therefore, causal associations were unable to be made. Thus, prospective studies are helpful to determine the causal effect of patient factors on quality indicators. Secondly, we could not consider all possible patient factors which may include confounding factors, and account for other important outcome measures (e.g., foot examination, anti-platelet therapy, and smoking cessation etc.) due to lack of information from the KNHANES III. Thirdly, our study used self-administered questionnaires for a majority of the information, which may be subject to recall bias. For instance, patients might not know that they had been tested for microalbuminuria. There were also relatively many missing values in the HbA1c and LDL-C variables. The excluded and included patients for the analyses may have different associations between patient factors and quality indicators, which can lead to non-response bias. Lastly, since non-biometric categorical variables may have low reliability and may not capture complex characteristics of interest very well, differences that exist are masked. For example, in our study, exercise was roughly divided into two groups according to regular exercise in leisure time. Quantitative classification of exercise using metabolic equivalent of task (MET) may be a better choice to grasp the differences in detail.