Appendix A. Literature review and meta-analysis procedures

The objective of the literature review and meta-analysis was to extract clinical efficacy data for both dapagliflozin and glimepiride. Studies carried out in Asian populations were identified that compared the clinical efficacy between dapagliflozin and glimepiride. No publication directly comparing dapagliflozin versus glimepiride was found through the initial search, thus an alternative indirect comparison approach was carried out. Literature regarding efficacy evaluation on dapagliflozin versus placebo and glimepiride versus placebo were extracted, and efficacy differences between dapagliflozin versus glimepiride were calculated using placebo as the common comparator (Figure A.1).

The detailed selection criteria was as follows:

Inclusion criteria:

1) Only randomized controlled trials (RCT) were eligible; 2) The interventions should include comparisons between any combination of dapagliflozin, placebo, and glimepiride; 3) The target population should be Asians; 4) Target population should be T2DM patients.

Exclusion criteria:

1) The target population included subjects less than 18 years old; 2) The study duration was less than 12 weeks; 3) The study was not monotherapy design; 4) The literature didn’t contain the clinical efficacy data including HbA1c, weights (BMI), BP, lipids, hypoglycemia, and other adverse effects; 5) The target population were renal impaired patients (eGFR>60 mL/min/1.73m2; 6) Trials not using dapagliflozin 10mg as daily dosage.

Literature review was conducted solely on English databases including PubMed, MEDLINE and EMBase (January 1, 1990-December 31, 2015). The detailed literature screening process was summarized in Figure A.2. The searching strategy was designed directly following the searching algorithm originally developed by Robinson KA and Dickersin K 68 and later modified by Biondi-Zoccai GG 69. To limit heterogeneity in study dose across different studies, 10 mg daily and 4 mg daily were recommended by the project clinical investigators for data extraction because they were the most common daily dosage for dapagliflozin and glimepiride in Chinese T2DM patients, respectively.

Two reviewers independently extracted data from each study retrieved.Any discrepancy between the two reviewers’ records were addressed by a mutual check and resolved by consensus with a third reviewer.

Two papers were found that were conducted in Asian population comparing the efficacy of dapagliflozin and placebo 24, 44. However, no papers comparing glimepiride and placebo in the Asian population were found. Thus, for glimepiride versus placebo studies the restriction on population location was removed identifyingfour studies comparing glimepiride to placebo 45-48. Data was then extracted by recording the clinical efficacies on HbA1c, SBP, weight, lipids, and adverse effects. Extracted parameters included the means and standard errors (S.E.). Two reviewers independently extracted data from each study retrieved. Any discrepancy between the two reviewers’ records were addressed by a mutual check and resolved by consensus with a third reviewer.

After data extraction, we synthesized the efficacy of dapagliflozin versus placebo and glimepiride versus placebo, and used placebo as the common comparator to calculate the “between treatment difference”. Details regarding the heterogeneities of clinical efficacies among identified cohorts and synthesized clinical efficacies are provided in Fig A.3~A.7. The first panel of each figure shows studies used to calculate the clinical efficacy of dapagliflozin, with the synthesized clinical efficacies on the bottom; the second panel shows studies used to calculate the clinical efficacy of glimepiride, also with the synthesized clinical efficacies on the bottom; the third panel shows the “between treatment difference”. We further summarized the synthesized data into Table A.1. The data was synthesized by calculating the weighted means of the efficacy data from each study, weighting on the inverse standard error respectively. Details regarding equations for synthesizing data can be found in Sterne’s paper70. Stata 12.0 was used to apply the indirect comparison.

The Cardiff model requires the user to input the clinical efficacy of each treatment arm separately. To calculate the clinical efficacy of each treatment arm, we first synthesize the clinical efficacy of Dapagliflozin from identified papers (column one in table A.2), and then added the “between treatment difference (column two in table A.2)” onto it to get the clinical efficacy of Glimepiride (column three in table A.2).

Figure A.1 Indirect Comparison between Dapagliflozin vs Glimepiride

Figure A.2 Literature Screening Flow Chart

Table A.1 Indirect comparison of efficacy parameters

E.S. / S.E. / 95% CI / Data Source
Lower / Upper
HbA1c (%)
Dapa vs PLA / -0.815 / 0.066 / -0.94436 / -0.68564 / 24, 44
GLI vs PLA / -0.762 / 0.11825 / -0.99377 / -0.53023 / 45-48
Dapa vs GLI / -0.053 / 0.135422 / -0.31843 / 0.212427 / Calculated
SBP (mmHg)
Dapa vs PLA / -4.305 / 1.29175 / -6.83683 / -1.77317 / 24, 44
GLI vs PLA / 4.6 / 1.7345 / 1.20038 / 7.99962 / 48
Dapa vs GLI / -8.905 / 2.162662 / -13.1438 / -4.66618 / Calculated
Weight (Kg)
Dapa vs PLA / -1.908 / 0.20125 / -2.30245 / -1.51355 / 24, 44
GLI vs PLA / 1.692 / 0.3265 / 1.05206 / 2.33194 / 45-48
Dapa vs GLI / -3.6 / 0.383541 / -4.35174 / -2.84826 / Calculated
TC (mmol/l)
Dapa vs PLA / 0.138 / 0.0445 / 0.05078 / 0.22522 / 24, 44
GLI vs PLA / -0.11 / 0.196 / -0.49416 / 0.27416 / 48
Dapa vs GLI / 0.248 / 0.200988 / -0.14594 / 0.641937 / Calculated
HDL (mmol/l)
Dapa vs PLA / 0.12 / 0.0255 / 0.07002 / 0.16998 / 24, 44
GLI vs PLA / 0.02 / 0.1595 / -0.29262 / 0.33262 / 48
Dapa vs GLI / 0.1 / 0.161526 / -0.21659 / 0.41659 / Calculated

Table A.2 Model Input: efficacy data

Dapa* / Dapa vs GlI** / GLI***
HbA1c (%) / -0.792 / -0.053 / -0.739
S.E. / 0.04725 / 0.135422 / 0.143428
SBP (mmHg) / -3.279 / -8.905 / 5.626
S.E. / 0.91475 / 2.162662 / 2.348164
Weight (Kg) / -2.048 / -3.6 / 1.552
S.E. / 0.1435 / 0.383541 / 0.409507
TC (mmol/l) / 0.092 / 0.248 / -0.156
S.E. / 0.03275 / 0.200988 / 0.203639
HDL (mmol/l) / 0.117 / 0.1 / 0.017
S.E. / 0.01825 / 0.161526 / 0.162553

*Data synthesized from Ji et al [44] and kaku et al[24], methods regarding data synthesize can be found here [70].

** From Table A.1

*** Data calculated based on Dapagliflozin + “Dapa vs Gli”

Figure A.3 Forest graph for synthesizing clinical efficacy on HbA1c (%)

Figure A.4Forest graph for synthesizing clinical efficacy on Weight (Kg)

Figure A.5 Forest graph for synthesizing clinical efficacy on SBP (mmHg)

Figure A.6Forest graph for synthesizing clinical efficacy on Total Cholesterol (mmol/l)

Figure A. 7Forest graph for synthesizing clinical efficacy on High Density Lipids (mmol/l)

Appendix B. Sensitivity analysis using global wide efficacy data from both 5 mg/d and 10 mg/d trials of dapagliflozin.

To further test the validity of our study,we expanded our inclusion criteria from Asian population to any population and from 10mg/d to either 5mg/d or 10mg/d. 7 extra cohorts were included17,24,26,28,44. Clinical efficacy data was synthesized using the same method applied in the base case analysis. Details regarding synthesizing the data are provided in table B.1 and B.2. Also, details regarding the heterogeneities of clinical efficacies for dapagliflozin are provided in Fig B.1~B.5. The simulation results are provided in table B.3. We also reported the ICER in table 8 of the main text.

Table B.1 Indirect comparison of efficacy parameters
E.S. / S.E. / 95% CI
Lower / Upper
HbA1c (%)
Dapa vs PLA / -0.730 / 0.019 / -0.810 / -0.660
GLI vs PLA / -0.762 / 0.118 / -0.994 / -0.530
Dapa vs GLI / 0.032 / 0.120 / -0.203 / 0.267
SBP (mmHg)
Dapa vs PLA / -4.150 / 0.327 / -5.430 / -2.870
GLI vs PLA / 4.600 / 1.735 / 1.200 / 8.000
Dapa vs GLI / -8.750 / 1.765 / -12.209 / -5.291
Weight (Kg)
Dapa vs PLA / -1.780 / 0.056 / -2.000 / -1.560
GLI vs PLA / 1.692 / 0.327 / 1.052 / 2.332
Dapa vs GLI / -3.472 / 0.331 / -4.121 / -2.823
TC (mmol/l)
Dapa vs PLA / 0.110 / 0.020 / 0.030 / 0.190
GLI vs PLA / -0.110 / 0.196 / -0.494 / 0.274
Dapa vs GLI / 0.220 / 0.197 / -0.166 / 0.606
HDL (mmol/l)
Dapa vs PLA / 0.180 / 0.009 / 0.150 / 0.220
GLI vs PLA / 0.020 / 0.160 / -0.293 / 0.333
Dapa vs GLI / 0.160 / 0.160 / -0.153 / 0.473
Table B.2 Model Input: efficacy data
Dapa* / Dapa vs GlI** / GLI***
HbA1c (%) / -0.73 / 0.03 / -0.76
S.E. / 0.02 / 0.12 / 0.12
SBP (mmHg) / -4.15 / -8.75 / 4.60
S.E. / 0.33 / 1.76 / 1.79
Weight (Kg) / -1.78 / -3.47 / 1.69
S.E. / 0.06 / 0.33 / 0.34
TC (mmol/l) / 0.11 / 0.22 / -0.11
S.E. / 0.02 / 0.20 / 0.20
HDL (mmol/l) / 0.18 / 0.16 / 0.02
S.E. / 0.01 / 0.16 / 0.16

*Data synthesized from extracted dapagliflozin trials, methods regarding data synthesize can be found here [70].

** From Table B.1

*** Data calculated based on Dapagliflozin + “Dapa vs Gli”

Figure B.1 Forest graph for synthesizingclinical efficacy on HbA1c (%)

Figure B.2 Forest graph for synthesizing clinical efficacy on SBP(mmHg)

Figure B.3 Forest graph for synthesizing clinical efficacy on Weight (Kg)

Figure B.4 Forest graph for synthesizing clinical efficacy on TC (mmol/l)

Figure B.5 Forest graph for synthesizing clinical efficacy on HDL (mmol/l)

1

Table B.3Module output results for dapagliflozin (both 5mg/d and 10mg/d compared to glimepiride)
Total Events Predicted / Glimepiride / Dapagliflozin / Difference / Total Costs (RMB) / Glimepiride / Dapagliflozin
Macrovascular / Non-Fatal / Fatal / Non-Fatal / Fatal / Macrovascular
IHD / 133.7 / 0.0 / 133.4 / 0.0 / -0.3 / IHD / 9,794,617 / 9,745,233
MI / 130.9 / 160.8 / 130.6 / 160.6 / -0.5 / MI / 17,615,200 / 17,535,331
CHF / 58.8 / 6.0 / 58.9 / 6.0 / 0.1 / CHF / 3,168,598 / 3,162,360
Stroke / 66.2 / 17.7 / 65.9 / 17.7 / -0.2 / Stroke / 3,982,271 / 3,942,450
Microvascular / Microvascular
Blindness / 67.8 / 0.0 / 67.8 / 0.0 / -0.0 / Blindness / 4,605,385 / 4,606,778
Nephropathy / 23.3 / 2.3 / 23.2 / 2.3 / -0.1 / Nephropathy / 9,067,965 / 8,960,786
Amputation / 42.6 / 4.4 / 42.4 / 4.4 / -0.1 / Amputation / 3,130,897 / 3,105,202
Hypoglycemia / 1,221,171 / 1,221,037
Treatment / 69,586,481 / 67,232,681
BMI Costs / 146,154,092 / 101,178,985
Total / 268,326,677 / 220,690,844
Cost-Effectiveness
Per patient / Glimepiride / Dapagliflozin / Difference
Discounted Cost / RMB 268,326.68 / RMB 220,690.84 / - RMB 47,636
Discounted QALYs / 12.54 / 13.52 / 0.98
Discounted LY / 16.58 / 16.59 / 0.01
Cost per QALY / - RMB 48,796
Cost per LY / - RMB 8,378,153

1