Diuretic strategies in acute heart failure: a systematic review and network meta-analysis of randomized clinical trials

Several diuretic strategies, including furosemide i.v. boluses (FB) or continuous infusion (FC), are used in acute heart failure (AHF).

We systematically searched phase 3 randomized clinical trials (RCTs) evaluating diuretic regimens in admitted AHF patients within 48 h and irrespective of clinical stabilization. We calculated the odds ratio (OR) of FC or FB plus another diuretic (sequential nephron blockade, SNB) compared to FB alone on 24 h weight loss (WL) and worsening renal function (WRF), with a random-effects model with inverse variance weighting. Urine output, hypokalaemia, hyponatremia, and all-cause mortality/rehospitalization were secondary endpoints. In 25 selected RCTs (7149 patients, mean age 68.9 ± 8.7 years, mean left ventricular ejection fraction 38.2 ± 10.7%), FC [OR 1.55 (95% confidence interval 1.39–1.63)], FB plus tolvaptan [OR 1.57 (1.39–1.77)], FB plus SGLT2i [OR 1.23 (1.06–1.42)], and FB plus thiazide [OR 1.63 (1.37–1.94)] were associated with greater WL than FB. FB plus SGLT2i [OR 1.52 (1.19–1.94)] and FB plus acetazolamide [OR 1.81 (1.31–2.49)] were associated with WRF. FB plus thiazide was associated with both WRF [OR 1.78 (1.43–2.21)] and hypokalaemia [OR 1.69 (1.32–2.16)]. Results were consistent in sensitivity analyses considering urine output, RCTs protocol-established furosemide doses, or daily furosemide dose. Congestion/decongestion scores and clinical outcomes were reported in around 50% of RCTs. In an underpowered exploratory analysis, mortality/rehospitalization was non-significantly lower with SGLT2i [OR 0.45 (0.19–1.07)].

FC and SNB improve surrogates of response to FB in AHF. SNB is also connoted by WRF and may induce hypokalaemia. The endpoints of diuretic RCTs should be revised and harmonized.