Background
Methods
Literature search
Study selection
Data extraction
Quality and risk of bias assessment
Statistical analysis
Results
Study selection
Study | Year | Country | Design | Population | Sample Size | Mean age (years) | Sex (% females) | % of patients with LVEF ≥ 50% | Follow-up (years) |
---|---|---|---|---|---|---|---|---|---|
Masoudi et al. [20] | 2005 | USA | Observational, retrospective cohort | T2DM, age ≥ 65 years, hospitalisation with HF | 13,930 | 76 | 58 | 23 | 1.0 |
Romero et al. [21] | 2011 | Spain | Observational, prospective cohort | New-onset T2DM, HF | 1519 | 72 | 54 | 51 | 4.7 |
Facila et al. [19] | 2017 | Spain | Observational, retrospective cohort | T2DM, discharged with diagnosis of acute decompensated HF | 835 | 72 | 49 | 49 | 2.4 |
Aguilar et al. [18] | 2011 | USA | Observational, retrospective cohort | T2DM, prior diagnosis of HF | 6185 | 68 | 7 | 45a | 2 |
Study characteristics
Study | Mean age (years) | Female gender n (%) | EF ≥ 50% n (%) | CAD (%) | HTN (%) | PVD (%) | ACEi/ARB (%) | Beta-blocker (%) | Insulin (%) | Sulfonylurea (%) | |
---|---|---|---|---|---|---|---|---|---|---|---|
Metformin treatment | Masoudi [20] (n = 1861) | 75.8 | 1060 (57) | 447 (24) | 1135 (61) | 1321 (71) | 56 (3) | 1191 (64) | 707 (38) | 354 (19) | 1042 (56) |
Romero [21] (n = 926) | 70.6 | 500 (54) | 478 (52) | 349 (38) | 555 (60) | 141 (15) | 751 (81) | 324 (35) | 409 (44) | 185 (20) | |
Facila [19] (n = 275) | 71.0 | 120 (44) | 144 (52) | 115 (42) | 159 (58) | 22 (8) | 207 (75) | 179 (65) | 65 (24) | 71 (26) | |
Aguilar [18] (n = 1561) | 67.3 | 122 (8) | 727 (47) | 514 (33) | 1202 (77) | 312 (20) | 1350 (87) | 929 (60) | 523 (34) | 923 (59) | |
Pooled (n = 4623) | 71.2±4 | 1802 (39) | 1796 (39) | 2113 (46) | 3237 (70) | 531 (11) | 3499 (76) | 2139 (46) | 1351 (29) | 2221 (48) | |
Non-Metformin treatment | Masoudi [20] (n = 12,069) | 77.0 | 7000 (58) | 2776 (23) | 7966 (66) | 8448 (70) | 482 (4) | 6879 (57) | 3862 (32) | 6638 (55) | 6155 (51) |
Romero [21] (n = 593) | 72.3 | 317 (54) | 302 (51) | 218 (37) | 335 (57) | 98 (17) | 480 (81) | 218 (37) | 394 (66) | 257 (43) | |
Facila [19] (n = 560) | 74.0 | 291 (52) | 264 (47) | 278 (50) | 475 (85) | 95 (17) | 386 (69) | 329 (59) | 278 (49) | 117 (21) | |
Aguilar [18] (n = 4624) | 70.0 | 282 (6) | 2048 (44) | 1665 (36) | 3648 (75) | 1147 (25) | 3801 (82) | 2839 (61) | 2452 (53) | 2719 (59) | |
Pooled (n = 17,846) | 73.3±3 | 7890 (44) | 5390 (30) | 10,127 (57) | 12,906 (72) | 1822 (10) | 11,546 (65) | 7248 (41) | 9762 (55) | 9248 (52) |
Study | Treatment | Total mortality events | Adjusted effect size [95% CI] | Covariates in effect size adjustment |
---|---|---|---|---|
Masoudi et al. [20] | Metformin vs. No Insulin sensitiser | 4805 | HR 0.86 [0.77, 0.97] | Patient factors (such as age and gender), treating physician and hospital characteristics |
Romero et al. [21] | Metformin vs. No metformin | 1045 | HR 0.88 [0.83, 0.93] | Propensity matched: age, gender, educational level, social situation, occupation situation, type of HF, aetiology of HF, clinical signs, radiological data, ECG data, comorbidity, bloods (lipids, albumin, creatinine, eGFR, ACR, Hb, electrolyte levels), medication received, time to follow-up, place of diagnosis, income, early readmissions, time of hospitalisation, visits to medical professionals |
Facila et al. [19] | Metformin vs. No metformin | 382 | HR 0.68 [0.53, 0.87] | Age, gender, prior admission for AHF, prior history of stroke, PVD, HTN, QRS > 120 ms, LVEF, eGFR, Hb, NT pro-BNP, treatment with beta-blockers |
Aguilar et al. [18] | Metformin vs. No Metformin | 1423 | HR 0.76 [0.63, 0.92] | Propensity matched: Age, gender, race, BMI, SBP, DBP, LVEF, history of hypertension, PVD, CVD, AF, MI, cancer, COPD and diabetic complications, HF hospitalisation within the last 2 years, bloods (Hb, HbA1c, sodium, urea, eGFR, total cholesterol and triglycerides) and treatment with beta-blockers, sulfonylurea, thiazolidinediones, insulin, statins, ACEi/ARB |
Risk of bias and quality assessments
Patient characteristics
Mortality outcomes
Subgroups | β-coefficient for the interaction with metformin and mortality [95% CI] | P-value | R-squared |
---|---|---|---|
Female Gender | 4.9 [3.7–6.1] | < 0.001 | 0.993 |
LVEF≥50% | − 2.3 [− 3.3, − 1.3] | 0.003 | 0.954 |
LVEF < 50% | 0.2 [− 0.2, 0.6] | 0.279 | 0.863 |
CAD | − 2.0 [− 10.3, 6.2] | 0.530 | 0.801 |
PVD | − 3.4 [− 4.4, − 2.3] | 0.001 | 0.969 |
Hypertension | − 1.4 [− 2.8, 0.1] | 0.060 | 0.909 |
ACEi/ARB | − 1.1 [− 1.3, − 0.8] | < 0.001 | 0.974 |
Beta blocker | − 1.6 [− 2.7, − 0.5] | 0.017 | 0.934 |
Insulin | − 1.8 [− 2.5, − 1.0] | 0.002 | 0.958 |
Sulfonylurea | 2.7 [− 0.5, 5.9] | 0.077 | 0.915 |
Metformin treatment adjusted for LVEF, ACEi/ARB and beta-blocker therapy | − 0.2 [− 0.3, − 0.1] | 0.020 | 0.996 |