3.1 Participant Characteristics

Of the 421 eligible participants, 417 participants were enrolled in this study. Among them, four participants were excluded because of certification of long-term care needs (n = 2), withdrawal of consent (n = 1), or incomplete data (n = 1) (Fig. 1). The mean participant ages of the participants, diabetes duration, BMI, HbA1c level, serum albumin level (Alb), serum creatinine level (CRE), and estimated glomerular filtration rate (eGFR) were as follows: 70.1 ± 5.4 years, 15.0 ± 10.9 years, 24.5 ± 3.8 kg/m2, 7.2 ± 1.0%, 0.85 ± 0.30 mg/dL, and 65.6 ± 18.0 mL/min/1.73 m2, respectively (Table 1). The diabetes medications used were as follows: SUs (33.8%); glinides (8.2%); αGI (9.1%); BG (32.6%); TZD (5.3%); DPP4i (65.9%); SGLT2i (10.8%); GLP-1 RA (7.4%); and insulin (29.0%) (Table 1). Regarding diabetic complications, nephropathy, retinopathy, peripheral neuropathy, and CAD were 15.6%, 23.0%, 16.8%, and 15.1%, respectively. Meanwhile, hypoglycemia and serious hypoglycemia were prevalent in 23.7% and 2.9%, respectively (Table 1).

Fig. 1Table 1 Participant characteristics3.2 Clinical Characteristics of People with Type 2 Diabetes by Diabetic Drug

All drugs were used more frequently in the prefrailty condition; the average KCL score was >4 points (Table 2). Patients taking SUs had a longer duration of diabetes (17.2 ± 10.4 years) and slightly higher HbA1c levels (7.5 ± 0.9%) (Table 2). Patients taking glinides had higher medication control scores (better medication adherence; 6.9 ± 0.4) on the Summary of Diabetes Self-Care Activities Measure (Table 2), but they also had a higher frequency of diabetic retinopathy (38.2%), lower grip strength (24.9 ± 8.1 kg), and higher KCL scores (6.5 ± 4.3), indicating a tendency to be frail. Patients taking ɑGI had a longer duration of diabetes (20.7 ± 9.5 years), lower Alb levels (4.09 ± 0.35 mg/dL), and lower renal function (CRE: 0.97 ± 0.39 mg/dL, eGFR: 57.1 ± 15.3) (Table 2). In addition, they had a higher frequency of hypoglycemia (36.8%) and lower physical activity level (1.7 ± 1.5), but nutritional status (0.2 ± 0.4) and cognitive function (0.4 ± 0.5) were better (Table 2).

Table 2 Characteristics of the participants with type 2 diabetes by diabetes treatment

Patients taking were younger (68.2 ± 5.0 years) and had higher body weight (64.6 ± 11.6 kg) and BMI (25.3 ± 4.3 kg/m2) (Table 2). They also had lower dietary scores (4.5 ± 1.6) and higher HbA1c levels (7.4 ± 1.0%) but also had better renal function (CRE: 0.77 ± 0.21 mg/dL, eGFR: 70.0 ± 17.1), albeit higher Alb levels (4.29 ± 0.33 mg/dL) (Table 2). Patients taking TZD were more likely to be smokers (36.4%) and had lower dietary scores (4.2 ± 1.7) and higher Alb levels (4.41 ± 0.44 mg/dL) (Table 2). None of the patients on TZD had diabetic nephropathy (0.0%), but they had lower instrumental ADL (0.9 ± 1.1) (Table 2). Patients takingDPP4is had drinking habits (38.5%) and higher exercise scores (3.7 ± 2.4) (Table 2). Patients taking SGLT2is were younger (67.9 ± 6.2 years) and had higher body weight (67.6 ± 12.7 kg), higher BMI (25.8 ± 4.1), lower dietary scores (3.9 ± 1.8), and higher HbA1c levels (7.8 ± 1.0%) (Table 2). Diabetic retinopathy was also more common (40.0%), and KCL scores were higher (6.0 ± 3.4) (Table 2).

Patients taking GLP-1RAs had a higher body weight (70.8 ± 13.5 kg), higher BMI (26.5 ± 4.5), and longer duration of diabetes (20.8 ± 11.8 years) (Table 2). They also had lower diet (4.2 ± 1.4) and exercise (2.6 ± 2.2) scores, higher HbA1c level (7.6 ± 1.1%), higher CRE level (1.00 ± 0.41 mg/dL), and more cardiovascular complications (29.0%) (Table 2). Patients taking insulin had a longer duration of diabetes (20.4 ± 11.0 years), higher HbA1c level (7.7 ± 1.0%), poorer renal function (CRE: 0.94 ± 0.39 mg/dL, eGFR: 60.5 ± 18.7), and lower Alb levels (4.09 ± 0.31 mg/dL) (Table 2). Hypoglycemia (44.6%), severe hypoglycemia (6.6%), diabetic nephropathy (27.6%), diabetic retinopathy (38.0%), diabetic peripheral neuropathy (25.6%), and cardiovascular disease (25.6%) were also common. In addition, they had lower grip strength (26.2 ± 7.8 kg), a higher frequency of frailty (28.9%), higher KCL scores (5.8 ± 3.7), and lower social ADL (1.0 ± 1.0) and physical activity (1.6 ± 1.3) (Table 2).

3.3 Relationship Between Each Diabetic Drug and Diabetic Complications

The relationship of each diabetes drug with hypoglycemia, severe hypoglycemia, and microvascular and macrovascular damage is shown in Table 3. The frequency of hypoglycemia was not higher in patients taking SUs. There was also no relationship between ɑGI, BG, TZD, DPP4i, or GLP-1RA and hypoglycemia or diabetic complications. Patients taking glinides experienced severe hypoglycemia (5.715 [1.309–24.947], p = 0.020). In contrast, hypoglycemia showed a significant association with insulin use (4.160 [2.497–6.931], p < 0.001); however, severe hypoglycemia did not show a relationship. The concomitant medications for SU, SGLT2i, and αGI are shown in the Electronic Supplementary Material (ESM). Retinopathy was associated with the use of glinides (2.443 [1.113–5.363], p = 0.026), SGLT2i (2.758 [1.352–5.627], p = 0.005), and insulin (2.138 [1.269–3.601], p = 0.004). Nephropathy, peripheral neuropathy, CAD, and CAD and arteriosclerosis obliterans were significantly associated with insulin use (2.519 [1.409–4.502], p = 0.002; 1.946 [1.103–3.433], p = 0.022; 2.274 [1.265–4.086], p = 0.006; and 2.035 [1.181–3.508], p = 0.011, respectively).

Table 3 Diabetes medicines and their association with diabetes complications3.4 Relationship of Each Diabetes Drug with Body Composition and Frailty

The relationships of each diabetes drug to body composition and frailty were also examined (Table 4). Sulfonylurea, αGI, and DPP4i use was not associated with body composition or frailty. Glinide use was associated with lower grip strength (−2.089 [−4.038 to −0.140], p = 0.036), higher KCL score (1.424 [0.219–2.629], p = 0.021), and lower physical activity (0.442 [0.038–0.846], p = 0.032). Biguanide use was associated with higher BMI (1.235 [0.437–2.033], p = 0.003), higher body fat mass (0.999 [0.019–1.979], p = 0.046), and greater grip strength (1.251 [0.073–2.429], p = 0.037). Thiazolidinedione use was associated with lower instrumental ADL levels (0.489 [0.194–0.784]; p = 0.001). Sodium-glucose cotransporter 2 inhibitor use was associated with a higher BMI (1.579 [0.385–2.772], p = 0.010), higher KCL score (1.319 [0.233–2.404], p = 0.017), poorer oral function (0.310 [0.030–0.590], p = 0.030), and worse depressive mood (0.405 [0.034–0.777], p = 0.033). Glucagon-like polypeptide receptor agonist use was associated with higher BMI (2.206 [0.810–3.602], p = 0.002), higher body fat (2.021 [0.299–3.744], p = 0.022), and poor nutritional status (0.200 [0.021–0.379], p = 0.029). Insulin therapy was associated with lower grip strength (−1.629 [−2.870 to −0.387], p = 0.010), higher KCL score (1.021 [0.253–1.789], p = 0.009), lower social ADL (0.285 [0.080–0.490], p = 0.007), and lower physical activity (0.316 [0.059–0.574], p = 0.016). The results are summarized in the ESM.

Table 4 Diabetes medicines and their association with body composition and frailty