The International Diabetes Federation Global Diabetes Map tenth edition, 2021 estimated that there were 537 million diabetes patients between 20 and 79 years of age worldwide, with about 6.7 million deaths of adults from diabetes or its complications [16]. The results of studies conducted in the last 20 years have confirmed that high glycemic variability increases oxidative stress, induces inflammatory reactions, and promotes endothelial-cell damage ion diabetes patients [2]. It is also an independently associated with increased risk of macrovascular and microvascular complications and neurological dysfunction [17–19]. If glycemic variability is not well monitored and controlled, the occurrence and progression of diabetes complications cannot be delayed even if HbA1c is well controlled [4]. Effective monitoring of glycemic variability cannot be ignored, but the limitations of CGM and SMBG result in low compliance and inadequate monitoring. The GDI is a simple, convenient screening index for high glycemic variability. To explore the screening effectiveness of GDI, the study found that the GDI value was higher in patients with high glycemic variability than it was in patients with normal glycemic variability. ROC curve analysis showed that AUC of the GDI as a screening tool was significantly higher than the AUCs of |2hPG-FPG|, HbA1c%, and GA/HbA1c. Although the specificity (0.973) and positive likelihood ratio (21.59) of |2hPG།FPG| were higher than those of GDI, its sensitivity (0.594) was lower. As only the daily glycemic variability was measured, which are affected by diet, mood, drugs and other factors, it is not suitable for screening. The relationship of GA/HbA1c and glycemic variability was studies in 143 diabetes patients in this study, 62 with normal glycemic variability and 81 with high glycemic variability. The sensitivity of GA/HbA1c to exclude high glycemic variability was high (0.903), but its specificity (0.383) and positive likelihood ratio (1.56) were too low to be suitable from screening. The variability of HbA1c is significantly associated with microvascular and macrovascular complications of diabetes [20, 21]. However, the long cycle of HbA1c variability monitoring is not conducive to timely monitoring of glycemic variability, The screening efficacy of HbA1c% for high glycemic variability, the sensitivity (0.75) and specificity (0.757) of HbA1c% were significantly lower than the GDI in the current study.
GDI had high specificity (0.905) and sensitivity (0781) and was positively correlated with the SD of blood glucose values (r = 0.813, p < 0.01). Its good screening efficacy is attributable to the evaluation factors included in the calculation of the model. It not only includes the low (FPG) and high (2hPG) blood glucose concentrations but also considers that because of the daily differences in blood glucose values, the glycemic variability in a one day may not accurately reflect glycemic variability over a longer period of time, Therefore, the GDI formula includes eAG, which reflects the mean blood glucose concentration in the previous 2–3 months, which buffers daily differences, accidental hyperglycemia, and missed diagnosis of hyperglycemia and hypoglycemia. Three patients included in this study serve as examples.
The FPG of patient No. 48 was 5.92 mmol/L and 2hPG was 11.08 mmol/L the difference between FPG and 2hPG was 5.16 mmol/L, which exceeded the maximal range of < 4.4 mmol/L specified by the Consensus of Chinese Experts in 2017.5 However, the actual glycemic variability of the patient was normal. The SD of blood glucose values was 1.77 mmol/L, the GDI was 3.74, which was less than the cutoff value of 4.21 mmol/L, eAG = 7.90, indicating that the patient's increase of 2hPG increase may have been incidental and little impact on the overall glycemic variability. Under such circumstances, the patient did not need CGM or SMBG monitoring, only continued to observation to see whether there is a significant increase in postprandial blood glucose.
The FPG of the patient No. 136 was 4.83 mmol/L and the 2hPG was 7.41 mmol/L. Blood glucose and glycemic variability seemed to be well controlled, but the SD of blood glucose values was 3.18, the GDI was 9.45, which was greater than the cutoff value of 4.21, the eAG was 12.67, which indicated high glycemic variability, with a large increase in glucose concentration. CGM or SMBG monitoring should be performed to determine the time of the increase in glucose concentration, observe the variability, and investigate the reasons, with timely intervention to adjust living habits and drug intervention.
The FPG of the patient No. 151 was 11.41 mmol/L and the 2hPG was 15.03 mmol/L. The overall daily blood glucose concentration appeared to increase significantly, and although the difference was relatively small, glycemic variability was high, with an SD of 3.64. The GDI was 7.73, which was higher than the cutoff value of 4.21., indicating high glycemic variability and a hypoglycemic time point. In addition to hypoglycemic treatment, such patients need blood glucose monitoring to determine when hypoglycemia occurs, such as before bed, or early morning, to avoid serious consequences caused by covert hypoglycemia, which is particularly important for elderly patients with diabetes.
The SD of the blood glucose value reflects mainly depends on the difference between each blood glucose value and the mean level of blood glucose, the greater the difference, the greater the dispersion between each blood glucose value and the mean level of blood glucose, which means the greater the variability of glycemic. Therefore, determining the mean level of blood glucose levels is a key step in evaluating glycemic variability. Without frequent blood glucose monitoring, HbA1c is the best indicator to reflect the mean level of blood glucose of patients, and it is not affected by lifestyle changes such as short-term diet and exercise [6]. The formula for converting HbA1c% to eAG in GDI has been recommended by the 2020 Chinese Guidelines for the Prevention and Treatment of Type 2 diabetes [11]. When FPG and 2hPG accurately reflect the low and high blood glucose concentrations, the GDI formula calculates the sum of the differences between FPG, 2hPG and eAG, to measure the maximum dispersion of blood glucose value to evaluate glycemic variability.
The inclusion of HbA1c% increased the sensitivity of GDI to screen high glycemic variability. HbA1c production increases with the increase in blood glucose concentration and of the duration of the increase [6]. If FPG and 2hPG do not accurately reflect the patient's low and high blood glucose concentration, there is hidden hypoglycemia or hyperglycemia. The HbA1c% decreased or increased with glucose concentration and duration of sub-clinical hypoglycemia or hyperglycemia. If difference between FPG or 2hPG and eAG increases, GDI increases. GDI evaluates not only glycemic variability but also shows whether patients have hidden hypoglycemia or hidden hyperglycemia.
Secondly, GDI was effective for screening high glycemic variability and its calculation was simpler than that of traditional indices such as SD and coefficient of variation. More important, it is only necessary to collect whole blood on an empty stomach and 2 h after a meal to monitor GDI, GDI monitoring only requires twice of invasive blood collection, Compared with SMBG monitoring, the number of blood collection is reduced by five times. The operation is simpler, which improves the compliance of patients. Compared with CGM, it greatly reduces the economic burden and pressure of patients to comply.
In summary, GDI was effective for screening glycemic variability, is mor convenient than CGM and SMBG, it is simple to operate, low in cost, and has high clinical applicability, Patients with normal glycemic variability screened by GDI can avoid unnecessary blood glucose monitoring methods such as CGM or SMBG, reduce the financial burden of patients and alleviate anxiety. Patients with `high glycemic variability screened by GDI can perform SMBG or CGM monitoring to obtain more detailed information on blood glucose fluctuations. Therefore, monitoring GDI can expand the population being monitored for glycemic variability, provide early assessment and warning of the risk of occurrence, development of diabetes complications. It can be a target of primary and secondary prevention, guide clinical diagnosis and treatment, provide personalized hypoglycemic treatment programs, and facilitate the development of precision medicine for diabetes.
This study has limitations. First, it excluded patients with hypoproteinemia, anemia, acute infection, severe organic disease, acute diabetes complications, and recent use of glucocorticoids, thyroid hormone, β-adrenergic agonists, and other drugs that affect blood glucose levels. Consequently, GDI may not be useful in such patients. Additionally, to ensure that the study results accurately describe the daily glycemic variability, we monitored blood glucose concentration of patients within 24 h of admission and before making any changes in treatment. As few patients had complete blood glucose data including all seven required assays on the first day, leading to the small sample size included in the study. A large prospective study is required to verify the study results.