Supplementary Online Content

Transcription

1 Supplementary Online Content Phung OJ, Scholle JM, Talwar M, Coleman CI. Effect of Noninsulin Antidiabetic Drugs Added to Therapy on Glycemic Control, Weight Gain, and Hypoglycemia in Type 2 Diabetes. JAMA. 2010;303(14): etable. e Characteristics of Randomized Controlled Trials Evaluating Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 1. Overall Network Diagram of Randomized Controlled Trials Evaluating Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 2. Network Diagram of Randomized Controlled Trials Evaluating Change From e in A1c (%) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 3. Network Diagram of Randomized Controlled Trials Evaluating A1c Goal Achieved (RR) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 4. Network Diagram of Randomized Controlled Trials Evaluating Change From e in Body Weight (kg) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 5. Network Diagram of Randomized Controlled Trials Evaluating Overall Hypoglycemia (RR) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus efigure 6. Results of Mixed Treatment Comparison Meta-analysis Presented as Forest Plots This supplementary material has been provided by the authors to give readers additional information about their work.

2 etable. e Characteristics of Randomized Controlled Trials Evaluating Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus Study Name, Year N DeFronzo, N=370 Ferrannini, N=2789 Goodman, N=247 Nauck, 2009a 13 N=314 Nauck, 2009b 14 N = 608 Bolli, ,16 N=576 Jadad Score* Followup () Inclusion Criteria 2,2,1 24 A1c: 7-10% Dose: Duration: 2 BMI: 40 1,2,1 52 A1c: % Dose: 1500 Duration: >4 BMI: ,1,1 24 A1c: % Dose: 1500 Duration: 12 BMI: ,1,1 26 A1c: 7-10% Dose: 1500 Duration: 12 BMI: ,2,1 26 A1c: 7-11% (prestudy OAD monotherapy 3 ) or 7-10% (prestudy combination OAD 3 ) Dose: 3 week forced titration, final dose 2000 BID Duration: 3 following forced titration (6 total) BMI: 40 1,0,1 52 A1c: % Dose: 1500 Intervention s Evaluated Saxagliptin 5 Population Sample Characteristics size (Age, Males) 191 Age (yrs): 54.7 ± 9.6 Males (%): 53.9 Placebo 179 Age (yrs): 54.8 ± 10.2 Males (%): 53.6 Vildagliptin 100 Glimepiride up to 6 Vildagliptin 100 (in the morning) 1396 Age (yrs): 57.5 ± 9.06 Males (%): Age (yrs): 57.5±9.28 Males (%): Age (yrs): 54.7 ± 10.3 Males (%): 52.8 Placebo 122 Age (yrs): 54.5 ± 9.7 Males (%): 67.2 Alogliptin Age (yrs): 54 ± 11 Males (%): 54.3 Placebo 104 Age (yrs): 56 ± 11 Males (%): 48 Liraglutide 1.8 Glimepiride Age (yrs): 57 ± 9 Males (%): Age (yrs): 57 ±9 Males (%): 57 Placebo 122 Age (yrs): 56 ± 9 Males (%): 60 Vildagliptin Age (yrs): 56.3 ± 9.3 Males (%): 61.7 e A1c (%) 8.1 ± ± 7.31 ± ± 0.65 e Fasting Plasma Glucose (l) e Weight (kg) 180 ± ± ± ± ± ± ± ± ± 7.9 ± ± 50.4 e BMI ( ) 31.2 ± ± ± ± 5.25 NR 31.7 ± 4.6 NR 31.7 ± 4.3 Duration of DM (years) 6.4 ± 4.7 NR 6.7 ± 5.6 NR 5.71 ± ± 5.03 Metformi n Dose (mg) ± ± 408 NR 1889 ± NR ± ± 45 NR 32 ± 5 6 ± ± ± 180 ± 50.4 NR 32 ± 6 6 ± ± ± 41.4 NR 3 ± ± 46.8 NR 31.2 ± ± 41.4 NR 31.6 ± ± ± ± ± 5 NR 8 ± 5 NR 8 ± 6 NR 6.4 ± ± 454

3 Study Name, Year N Hamann, N=595 Khanolkar, N=50 Raz, N=190 Scott, N=273 Ahrén, ,22 N=71 Bosi, N=273 Jadad Score* Followup () Inclusion Criteria Duration: at least 4 week run-in BMI: ,1,1 52 A1c: 7-10% Dose: >850 Duration: 8 BMI: 25 1,1,1 24 A1c: >6.5% Dose: 2000 Duration: 4 BMI: NR 2,1,1 30 A1c: 8-11% Dose: >1500 Duration: 6 BMI: ,1,1 18 A1c: 7-11% Dose: 1500 Duration: 10 1,1,1 52 A1c: 7-9.5% Dose: Duration: 3 BMI: ,1,1 24 A1c: % Dose: 1500 Duration: 3 BMI: Intervention s Evaluated Pioglitazone 30 Rosiglitazone 4 Glibenclimide 5 or Gliclazide 80 Rosiglitazone 4 Gliclazide 80 Sitagliptin 100 Population Sample Characteristics size (Age, Males) 281 Age (yrs): 57 ± 9.7 Males (%): Age (yrs): 58.5 ± 9.6 Males (%): Age (yrs): 59.3 ± 9.2 Males (%): Age (yrs): 59 ± Males (%): Age (yrs): 56 ± Males (%): Age (yrs): 53.6 ± 9.5 Males (%): 51 Placebo 94 Age (yrs): 56.1 ± 9.5 Males (%): 41.5 Sitagliptin 100 Rosiglitazone 8 94 Age (yrs): 55.2 ± 9.8 Males (%): Age (yrs): 54.8 ± 10.5 Males (%): 63 BMI: NR Placebo 92 Age (yrs): 55.3 ± 9.3 Males (%): 59 Vildagliptin Age (yrs): 58.4 ± 9.2 Males (%): 61.9 Placebo 29 Age (yrs): 54.3 ± 12.2 Males (%): 75.9 Vildagliptin Age (yrs): 53.9 ± 9.5 Males (%): 61.5 Placebo 130 Age (yrs): 54.5 ± 10.3 Males (%): 53.1 e A1c (%) e Fasting Plasma Glucose (l) e Weight (kg) 198 ± ± ± 189 ± ± ± ± ± ± ± 9.1 ± ± 16.6 e BMI ( ) 32.1 ± 5.1 Duration of DM (years) Metformi n Dose (mg) 6.4 ± ± ± ± 5.4 NR 32.2 ± 4.9 NR NR ± 8.83 NR NR ± ± ± ± ± ± 7.6 ± ± ± 81.5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 5.5 NR 29.6 ± 3.7 NR 29.9 ± 3.6 NR 32.9 ± 5.0 NR 33.2 ± ± 5.6 NR NR NR NR NR 6.5 NR 7.3 ± 5.3 NR 4.9 ± 3.5 NR 4.6 ± 4.0 NR 30 ± ± 3.7 NR 5.8 ± 4.2 NR 4.6 ± 3.6 NR 5.8 ± ± ± ± 320

4 Study Name, Year N Nauck, N=1172 Charbonnel, N=701 Garber, N=318 Ristic, ,28 N=213 DeFronzo, N=226 Feinglos, N=122 Jadad Score* Followup () Inclusion Criteria 1,1,1 52 A1c: % Dose: 1500 Duration: 2 BMI: NR 1,1,1 24 A1c: 7-10% Dose:1500 Duration: up to 19 BMI: NR 1,2,1 24 A1c: 7-12% Dose: 1500 Duration: 2 BMI: ,2,1 52 A1c: 6.8-9% Dose:1000 Duration: >2 BMI: ,2,1 30 A1c: % Dose: 1500 Duration: 3 BMI: ,1,1 16 A1c: % Dose: 1000 Duration: 3 BMI: Intervention s Evaluated Sitagliptin 100 Glipizide 5 (titrated up to a maximum of 20) Sitagliptin 100 Population Sample Characteristics size (Age, Males) 588 Age (yrs): 56.8 ± 9.3 Males (%): Age (yrs): 56.6 ± 61.3 Males (%): Age (yrs): 54.4 ± 10.4 Males (%): 55.8 Placebo 237 Age (yrs): 54.7 ± 9.7 Males (%): 59.5 Glibenclamide up to 10 Rosiglitazone up to 8 Nateglinide 180,360 or 540 Gliclazide 80,160 or 240 Exenatide 20 mcg/d 160 Age (yrs): 56 Males (%): Age (yrs): 56 Males (%): Age (yrs): 61.9 ± 1 Males (%): Age (yrs): 61.5 ± 10.2 Males (%): Age (yrs): 52 ± 11 Males (%): 60.2 Placebo 113 Age (yrs): 54 ± 9 Males (%): 59.3 Glipizide GITS 2.5 ay 61 Age (yrs): 57.7 ± 10.7 Males (%): 46 Placebo 61 Age (yrs): 58.8 ± 10 Males (%): 41 e A1c (%) 7.7 ± 7.6 ± 8.0 ± ± ± ± ± ± 8.2 ± 7.45 ± ± 0.78 e Fasting Plasma Glucose (l) ± ± ± ± 41.4 e Weight (kg) 89.5 ± ± ± ± 17.5 e BMI ( ) 31.2 ± ± ± ± 4.9 Duration of DM (years) 6.5 ± 6.1 NR 6.2 ± 5.4 NR 6.0 ± 5.0 NR 6.6 ± 5.5 NR 191 ± ± ± 5 5 ± 4 NR 188 ± ± ± 5 6 ± 5 NR ± ± ± ± ± ± ± ± ± ± 18.4 NR 28.6 ± 3.5 NR 30.0 ± ± ± 5.40 Metformi n Dose (mg) ± ± 4.7 NR 34 ± ± 6.1 NR 31.7 ± ±

5 Study Name, Year N Matthews, N=630 Jadad Score* Followup () Inclusion Criteria 1,2,1 52 A1c: % Dose: 50% of the maximum recommended dose or at the maximum tolerated dose Duration: 3 BMI: NR 1,1,1 26 A1c: NR Dose: 2500 Duration: initial 3-6 week titration phase prior to randomization BMI: NR 2,2,1 24 A1c: % Dose: 1500 Duration: 3 BMI: ,2,1 20 A1c: NR Dose: 2550 Duration: 1 month BMI: for females; for males 2,2,0 32 A1c: % Dose: Duration: 3 BMI: ,1,1 26 A1c: NR Dose: 2500 Duration: NR BMI: Gómez- Perez, N=70 Marre, N=312 Charpentier, N = 222 Van Gaal, N=152 Fonseca, N=223 Intervention s Evaluated Pioglitazone up to 45 Gliclazide up to 320 Rosiglitazone 8 Population Sample Characteristics size (Age, Males) 317 Age (yrs): 56 ± 9.2 Males (%): Age (yrs): 57 ± 9.0 Males (%): Age (yrs): 54.2 ± 9.3 Males (%): 19.4 Placebo 34 Age (yrs): 53.4 ± 7.5 Males (%): 29.4 Nateglinide Age (yrs): 57.3 ± 10.5 Males (%): 61.3 Placebo 152 Age (yrs): 56.4 ± 10.3 Males (%): 55.3 Glimepiride 1 (titrated up to 2, 4 or 6 ) 147 Age (yrs): 56.8 Males (%): 59 Placebo 75 Age (yrs): 56.7 Males (%): 60 Miglitol up to Age (yrs): 57.9 ± 10 Males (%): 42 Placebo 75 Age (yrs): 57.9 ± 8.5 Males (%): 49 Rosiglitazone 8 ay 110 Age (yrs): 58.3 ± 8.8 Males (%): 68.2 Placebo 113 Age (yrs): 58.8 ± 9.2 Males (%): 74.3 e A1c (%) 8.71 ± 8.53 ± 0.89 e Fasting Plasma Glucose (l) ± ± 46.8 e Weight (kg) 91.8 ± ± 17.4 e BMI ( ) 32.6 ± ± 5.8 NR NR NR 27.6 ± 3.2 Duration of DM (years) 5.8 ± ± ±7.0 NR NR NR NR 28.5± ±5.6 NR ± ± ± 6.8 ± ± 8.9 ± ± ± ± ± ± ± ± ± ± ± ± ± 5.5 NR 6.5 ± 6.5 NR NR NR Metformi n Dose (mg) NR 30 ± 4.0 NR 1807 ± 371 NR 29.7 ± 3.9 NR 29.8 ± 3.9 NR 30.3 ± 4.4 NR 1812 ± ± 6.3 NR 7.3 ± 5.7 NR

6 Study Name, Year N Halimi, N=129 Moses, ,39 N = 54 Rosenstock, N = 148 Jadad Score* Followup () Inclusion Criteria 1,1,1 24 A1c: 7-11% Dose: 1700 or 2550 Duration: 2 BMI: ,1,1 12 A1c: > 7.1% Dose: Duration: > 6 BMI: 21 1,1,0 24 A1c: 7-10% Dose: 2000 or 2500 Duration:NR BMI: NR Intervention s Evaluated Acarbose up to 300 Population Sample Characteristics size (Age, Males) 59 Age (yrs): 56 ± 9.2 Males (%): 47 Placebo 70 Age (yrs): 55 ± 10 Males (%): 63 Repaglinide up to Age (yrs): 57.2 ± 8.3 Males (%): 67 Placebo 27 Age (yrs): 57.8 ± 9.5 Males (%): 63 Acarbose up to Age (yrs): 57.2 Males (%): 61 Placebo 74 Age (yrs): 55.9 Males (%): 49 e A1c (%) 8.6 ± 8.5 ± 8.3 ± 8.6 ± e Fasting Plasma Glucose (l) e Weight (kg) e BMI ( ) 189 ± 55.8 NR 30.1 ± ± 48.6 NR 29.7 ± ± 4 NR 33.2 ± ± NR 31.8 ± 6.0 Duration of DM (years) 9.5 ± 7.4 NR 9 ± 7.5 NR Metformi n Dose (mg) 5.9 ± ± ± ± NR NR Values reported as mean±sd when available. * Jadad score presented as subscores for randomization (up to 2 points), double-blinding (up to 2 points), description of withdrawals (up to 1 point). These individual components can be summed to achieve the total Jadad score. Abbreviations: A1c=glycosylated hemoglobin; BMI=body mass index; NR=not reported; yrs=years

7 Placebo 3 studies Reference 12,29,32 1 study Reference 25,26 2 studies Reference 31,36,37 4 studies Reference 13,16,24,28 3 studies Reference 18,30,34 8 studies Reference 8,10,11,17-21,23 3 studies Reference 33,35,38 1 study Reference 12 2 studies Reference 12,27 2 studies Reference 9,22 2 studies Reference 14,15,18 efigure 1. Overall Network Diagram of Randomized Controlled Trials Evaluating Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus All agents are in combination with metformin. Lines represent the presence of direct comparison trial(s). Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones

8 Placebo N=4; (-0.32, 0.11) (-0.31, 0.13) N=1; 0.00 (-0.28, 0.28) (-0.77, 0.12) N=3; (-5, -0.43) (-7, -0.62) N=3; -0 (-1.62, -0.38) (-8, -0.66) N=2; -9 (-9, -0.78) -7 (-1.30, -0.65) (-0.80, 0.36) N=1; (-0.42, 0.16) (-0.49, 0.25) (-0.56, 0.14) N=2; (-1.24, -0.18) (-7, -0.36) (-0.76, 0.14) N=2; (-0.13, -0.01) (-0.22, 0.22) N=2; (-1, -0.19) (-3, -0.26) (-0.63, 0.34) N=8; (-4, -0.63) (-3, -0.64) (-0.45, 0.20) (-0.51, 0.48) (-0.65, 0.42) N=2; (-0.16, 0.10) (-0.44, 0.16) (-0.55, 0.26) (-0.84, 0.17) (-0.52, 0.17) efigure 2. Network Diagram of Randomized Controlled Trials Evaluating Change From e in A1c (%) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus All agents are in combination with metformin. Solid lines represent the presence of direct evidence along with indirect evidence. Dotted lines represent the presence of indirect evidence only. Traditional pairwise meta-analysis results are reported as the top line in the format n=number of studies; change in A1c (95%CI). Mixed-treatment comparison meta-analysis results, which combine both direct and indirect evidence are reported in the format change in A1c (95%CrI). Arrows represent the favored drug in the mixed-treatment comparison metaanalysis and results reported are referent to the arrow origin. Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones

9 Placebo N=1; 1.27 (3, 1.58) 0 (0.69, 1.47) N=1; 7 (4, 1.46) 1.29 (0.79, 2.45) N=1; 3.38 (2.02, 5.83) 2.49 (1.95, 3.32) N=1; 1.69 (1.24, 2.33) 2.71 (1.74, 3.80) N=1; 3.96 (2.37, 6.79) 3.20 (2.01, 6.24) N=1; 1.33 (9, 1.82) 0 (0.68, 1.64) 1.21 (0.60, 1.92) 1.42 (0.73, 2.92) N=2; 1 (3, 0) 1 (0.79, 1.29) N=1; 3.20 (1.47, 7.58) 2.25 (1.48, 3.90) N=6; 2.44 (1.78, 3.33) 2.51 (2.04, 3.22) 2 (0.65, 1.72) 7 (0.71, 2.69) N=1; 5 (0, 1.47) 9 (0.67, 1.47) 1.27 (0.77, 2.45) efigure 3. Network Diagram of Randomized Controlled Trials Evaluating A1c Goal Achieved (RR) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus All agents are in combination with metformin. Solid lines represent the presence of direct evidence along with indirect evidence. Dotted lines represent the presence of indirect evidence only. Traditional pairwise meta-analysis results are reported as the top line in the format n=number of studies; relative risk (95%CI). Mixed-treatment comparison meta-analysis results, which combine both direct and indirect evidence are reported in the format relative risk (95%CrI). Arrows represent the favored drug in the mixed-treatment comparison metaanalysis and results reported are referent to the arrow origin. Thus, results may not correspond to numerical results reported elsewhere. Results referent to the second agent in a comparison equal 1/RR. Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones

10 Placebo N=2; 0.20 (-2.35, 2.74) (-5, 4) (-5.41, -2.75) N=2; (-3.12, -0.86) (-2.96, -5) N=1; (-2.90, -1.70) (-3.17, -8) N=2; (-2.90, -0.62) (-3.11, -0.48) (-5.98, -0.48) (-1.75, 1.46) (-2.00, 1.60) N=2; (-2.80, -1.42) (-3.02, -1.32) N=2; -1 (-1.46, -0.35) (-3.28, -0.46) (-5.63, -1.70) N=1; (-2.83, -0.77) (-3.79, 0.21) (-5.89, -1.66) N=4; (-0.47, 0.30) (-4, 0.63) (-3.65, -0.40) (-6.11, -1.20) (-5.41, -9) N=2; (-2.60, -1.63) (-2.91, -0.46) (-3.79, 0.54) (-2.42, 2.46) (-3.18, -0.07) efigure 4. Network Diagram of Randomized Controlled Trials Evaluating Change From e in Body Weight (kg) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus All agents are in combination with metformin. Solid lines represent the presence of direct evidence along with indirect evidence. Dotted lines represent the presence of indirect evidence only. Traditional pairwise meta-analysis results are reported as the top line in the format n=number of studies; change in weight (95%CI). Mixed-treatment comparison meta-analysis results, which combine both direct and indirect evidence are reported in the format change in weight (95%CrI). Arrows represent the favored drug in the mixed-treatment comparison meta-analysis and results reported are referent to the arrow origin. Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones

11 Placebo N=3; 0.11 (0.04, 0.29) 0.12 (0.04, 0.30) N=1; 0.17 (0.08, 0.36) 0.20 (0.04, 7) N=3; 0.38 (0.11, 1.32) 0.22 (0.09, 0.47 N=2; 2.04 (0.50, 8.23) 0.56 (0.19, 1.69) N=2; 4 (0.42, 2.12) 0.89 (0.22, 3.96) 0.73 (0.01, 19.50) N=1; 3 (0.51, 1.69) 0.60 (0.12, 2.22) 0.07 (0.01, 0.33) N=2; 0.13 (0.02, 6) 0.13 (0.02, 0.47) 0.12 (0.01, 0.79) N=2; 0.13 (0.09, 0.19) 0.14 (0.05, 0.36) N=2; 0.60 (0.08, 4.55) 0.42 (0.01, 9.00) 0.09 (0.001, 2.16) N=8; 0.67 (0.30, 1.50) 0.63 (0.26, 1.71) 0.08 (0.01, 0.37) 0.05 (7.69E-4, 1.56) 0.62 (0.10, 3.70) N=2; 0.65 (0.08, 5.26) 0 (0.24, 2.86) 0.65 (0.01, 16.67) 0.45 (0.01, 4.29) 0.69 (0.13, 4.00) efigure 5. Network Diagram of Randomized Controlled Trials Evaluating Overall Hypoglycemia (RR) With Noninsulin Antidiabetic Drugs in Addition to in Type 2 Diabetes Mellitus All agents are in combination with metformin. Solid lines represent the presence of direct evidence along with indirect evidence. Dotted lines represent the presence of indirect evidence only. Traditional pairwise meta-analysis results are reported as the top line in the format n=number of studies; relative risk (95%CI). Mixed-treatment comparison meta-analysis results, which combine both direct and indirect evidence are reported in the format relative risk (95%CrI). Arrows in the mixed-treatment comparison meta-analysis represent the favored drug and results reported are referent to the arrow origin. Thus, results may not correspond to numerical results reported elsewhere. Results referent to the second agent in a comparison equal 1/RR. Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones

12 A. Change in A1c (%) B. A1c Goal Achieved (-7, -0.62) 2.49 (1.95, 3.32) (-7, -0.36) 2.25 (1.48, 3.90) (-8, -0.66) 2.71 (1.74, 3.80) (-3, -0.26) (-3, -0.64) No data 2.51 (2.04, 3.22) -7 (-1.30, -0.65) 3.20 (2.01, 6.24) Placebo (Referent) Placebo (Referent) Weighted Mean Difference (95% Credible Interval) Relative Risk (95% Credible Interval) C. Change in Body Weight (kg) D. Overall Hypoglycemia 2.06 (5, 2.96) 4.57 (2.11, 11.45) 1.77 (0.46, 3.28) 7.50 (2.12, 41.52) 2.08 (8, 3.17) 0.56 (0.19, 1.69) (-3.79, 0.21) 0.42 (0.01, 9.00) (-4, 0.63) 0.63 (0.26, 1.71) (-3.11, -0.48) 0.89 (0.22, 3.96) Placebo (Referent) Placebo (Referent) Weighted Mean Difference (95% Credible Interval) Relative Risk (95% Credible Interval) efigure 6. Results of Mixed Treatment Comparison Meta-analysis Presented as Forest Plots The squares represent the pooled effect size for each class of oral antidiabetic drug. Error bars represent 95% credible intervals (CrIs). The number of trials included in each mixed-treatment comparison analysis is as follows: A=26 trials, B=13 trials, C=15 trials, and D=24 trials. Abbreviations: AGI=alpha-glucosidase inhibitor; DPP-4=dipeptidyl peptidase-4; GLP- 1=glucagon-like peptide-1; SU=sulfonylurea; TZD=thiazolidinediones