Sensitivity and subgroup analyses
| RCT (n) | p Value for heterogeneity | 95% CI | p Value for overall effect | |
| Ischaemic HF | 15 | p=0.0009, I2=63% | 7.37 (6.05 to 8.70) | <0.00001 |
| Non-ischaemic HF | 3 | p=0.48, I2=0% | 8.72 (5.51 to 11.92) | <0.00001 |
| Ischaemic HF with diabetes | 5 | p= 0.93, I2=0% | 6.19 (4.16 to 8.23) | <0.00001 |
| Excluding crossover effect | 14 | p=0.002, I2=57% | 7.61 (6.31 to 8.92) | <0.00001 |
| Trials that did not include class IV HF | 13 | p= 0.62, I2=0% | 6.39 (5.52 to 7.27) | <0.00001 |
| Trials that include class IV HF | 3 | p= 0.60, I2=0% | 10.87 (9.39 to 12.35) | <0.00001 |
| Trials with concomitant us of ACE inhibitor/ARB and β-blocker | 15 | p=0.05, I2=38% | 7.14 (5.99 to 8.30) | <0.00001 |
| Trials without concomitant use of ACE inhibitor/ARB and β-blocker | 1 | NA | 11.0 (8.53 to 11.47) | <0.00001 |
| Jadad scale ≥3 | 13 | p=0.03, I2=43% | 7.52 (6.22 to 8.82) | <0.00001 |
| Follow-up interval ≥3 | 13 | p=0.006, I2=55% | 8.12 (6.68 to 9.55) | <0.00001 |
ARB, angiotensin receptor blocker; HF, heart failure; NA, unable to assess; RCT, randomised controlled trial.