This study was designed as a cross-sectional study. We enrolled 669 patients with T2DM who were admitted to the Hospital of the University of Occupational and Environmental Health or its affiliated hospitals for glycemic control and education on diabetes management. All patients underwent CGM between April 2010 and March 2019 (excluding those receiving steroids). We analyzed the CGM data of 247 patients except those treated with insulin and those treated with no drug who started CGM within 5 days of hospitalization and did not change the doses or preparations after wearing a CGM device. Of these 247 patients, the data of 123 patients with TIR > 70% were analyzed (Fig. 1).
The meal during CGM was adjusted to 25–30 kcal/kg of ideal body weight (60% carbohydrates, 15–20% fats, and 20–25% protein) for patients with nephropathy at stage II or earlier, and 30–35 kcal/kg of ideal body weight (60–70% carbohydrates, 15–20% fats, and 15–20% protein) for patients with nephropathy at stage III or later. The meal contents were not changed during the CGM, and the patients consumed all inpatient meals. Moreover, the amount of exercise was kept constant.
Diabetic neuropathy was defined as neuropathy meeting two of the following three criteria: (1) the presence of subjective symptoms presumably caused by diabetic polyneuropathy, (2) weak or absent Achilles tendon reflex bilaterally, and (3) decreased vibration perception at the medial malleolus bilaterally. Diabetic retinopathy was defined as the presence of any of the following fundoscopic findings: simple retinopathy, preproliferative retinopathy, and proliferative retinopathy. Diabetic nephropathy was defined as a urinary albumin to creatinine ratio ≥ 30 mg/g, the presence of overt proteinuria, or an estimated glomerular filtration rate (eGFR) < 30 mL/min/1.73 m2. Micro-angiopathy was defined as the presence of any of the following conditions: neuropathy, retinopathy, or nephropathy. Macroangiopathy was defined as the presence of history of any of the following conditions: ischemic heart disease, stroke, or arteriosclerosis obliterans.
This study was conducted in accordance with the Declaration of Helsinki and the current applicable ethical codes. The study protocol was approved by the ethics committees of the Hospital of the University of Occupational and Environmental Health and its affiliated hospitals (Trial registration: H27-186, Registered 25 December 2015). This research is registered in University Medical Information Network [UMIN] ID: UMIN000025433. Before the start of the study intervention, all patients received information on the study and provided informed consent.
The CGM devices used were Gold (CGMS System Gold, Medtronic Inc.; Fridley, MN) and iPro2 (Medtronic MiniMed Inc.; Northridge, CA). These CGM devices continuously measure interstitial glucose levels within the range of 40–400 mg/dL. The sensor placed in the subcutaneous tissue converts the interstitial glucose levels into electrical signals and records measurements every 5 min, up to a maximum of 288 measurements per day. Interstitial glucose levels measured by the CGM devices have been reported to correlate well with venous blood glucose levels10. The CGM data used in the present study were calibrated before each of the three meals and at bedtime with data from a self-monitored blood glucose device (MEDISAFE MINI; Terumo, Inc.).
We analyzed 288 measurements taken between 00:00 and 24:00 after placement of the CGM device for 24 h. The CGM parameters included average glucose (AG); standard deviation (SD); percentage coefficient of variation (%CV); maximum glucose level; minimum glucose level; large amplitude of glycemic excursions (LAGE); mean postprandial glucose excursion (MPPGE) following breakfast, lunch, and supper; low blood glucose index (LBGI); high blood glucose index (HBGI); time above range (TAR, defined as the percent time with glucose level above 180 mg/dL); TIR (defined as the percent time with glucose level between 70 and 180 mg/dL); time below range < 70 mg/dL (TBR < 70, defined as the percent of time with glucose level less than 70 mg/dL); time below range < 54 mg/dL (TBR < 54, defined as the percent time with glucose level less than 54 mg/dL). %CV was calculated by the following formula: (SD/AG) × 100. LAGE was calculated as the difference between the maximum and minimum glucose levels. To calculate MPPGE, preprandial glucose levels were measured at 07:00 for breakfast, 12:00 for lunch, and 18:00 for supper, and the corresponding postprandial levels were measured from 07:00 to 12:00, from 12:00 to 18:00, and from 18:00 to 24:00, respectively.
Then, the difference between the maximum postprandial glucose level and the preprandial glucose level for each meal was calculated as MPPGE. In addition, hypoglycemia was defined as glucose level < 70 mg/dL as measured by CGM, and severe hypoglycemia was defined as glucose level < 54 mg/dL.
Measurements of biochemical variables
HbA1c levels (%) were measured on admission and converted to National Glycohemoglobin Standardization Program (NGSP) values for assessment. The HbA1c levels measured as Japan Diabetes Society (JDS) values were converted using the following formula: HbA1c (NGSP) (%) = HbA1c (Japan relationship of HbA1c, JDS) × 1.02 + 0.25 (%)11. The eGFR was calculated as follows: 194 × serum creatinine level (mg/dL) − 1.094 × age − 0.287 for men, and 194 × serum creatinine level (mg/dL) − 1.094 × age − 0.287 × 0.739 for women.
Data are expressed as mean ± SD. The Shapiro–Wilk test was used to test for normal distribution of data. For comparisons between two groups, we used the Student’s t-test for parameters with normal distribution, and the Mann–Whitney U-test for parameters with skewed data distribution. For nominal scale, the Fisher’s exact test was performed when some cells had an expected value within 5, whereas the χ2 test was performed otherwise. Univariate and multivariate linear regression analyses were performed to estimate the regression coefficients for %CV and TBR < 54. Since multicollinearity was observed between age and the duration of diabetes; between HbA1c and fasting blood glucose (FBG) levels; and between neuropathy, retinopathy, and nephropathy; we excluded diabetes duration, FBG, neuropathy, retinopathy, and nephropathy from the model of multivariate analysis, and included age, HbA1c and microangiopathy. In the model, age, sex, body mass index (BMI), microangiopathy, macroangiopathy, HbA1c level, eGFR, SU use, thiazolidinedione (TZD) use, biguanide (BG) use, alpha-glucosidase inhibitor (α-GI) use, glinide use, dipeptidyl peptidase-4 inhibitor (DPP4i) use, glucagon-like peptide 1 receptor agonist (GLP1RA) use, and sodium-glucose cotransporter 2 inhibitor(SGLT2i) use were entered as independent variables. In addition, we used the Fisher’s exact test to assess the association between TBR < 70 and TBR < 54 based on the use of SUs (after dividing the patients into the high-dose, recommended-dose, and non-SU users). The Statistical Program for Social Sciences version 25.0 (IBM-SPSS Statistics) was used for all statistical analyses. We determined the level of significance p < 0.05.
The primary endpoint was the difference in glycemic variability, based on CGM parameters (%CV), between the SU use and non-SU use groups in patients with TIR > 70%. The secondary endpoint was the comparison of proportions of patients with severe hypoglycemia (as defined above).