Diuretic treatment in high-risk acute decompensation of advanced chronic heart failure – bolus intermittent vs continuous infusion of furosemide

Frea. Clinical Research in Cardiology 2019; Published online 29 June 2019. doi:10.1007/s00392-019-01521-y

Clinical Question

  • In adults with decompensated heart failure who are at high-risk of diuretic resistance, does furosemide administered by infusion compared to intermittent bolus increase the incidence of freedom from congestion at 72 hours?


  • Diuretic resistance is a common issue in patients with acute decompensation of advanced chronic heart failure (ACHF)
  • Controversies exist about the best modality of furosemide administration
  • The largest prospective double-blind randomized trial (DOSE study) comparing continuous infusion and intermittent boluses in acute decompensated heart failure demonstrated no significant differences in global assessment of symptoms or in the mean change in the creatinine level
  • A recent meta-analysis found continuous infusion superior to boluses administration in terms of weight reduction, total urine output, and brain natriuretic peptide reduction


  • Randomised, controlled trial
  • Computer generated randomisation sequence with 1:1 ratio
  • Allocation was concealed by using sequential numbered cases with pre-prepared drug and dummy placebo
  • Double blind, double dummy design
    • Both groups received infusion and bolus, with one being active diuretic and the other being 5% dextrose
  • Two different doses of furosemide were given depending upon patient characteristics
    • This was defined prior to randomisation
    • This dose was then given by the randomly allocated method
  • Designed to assess superiority of continuous infusion over intermittent bolus therapy
  • Sample size of 78 patients required
    • Powered at 70% (False positive rate 30%)
    • Significance level 10% (False negative rate 10%)
    • Expected absolute difference of 23% (NNT 5)
      • Estimated outcome in bolus group 48%
      • Estimated outcome in infusion group 25%
  • Prespecified subgroups (renal function, haemodynamic profile and diuretic dose) was analysed with Bonferroni correction


  • Intensive Cardiac Care Unit of University Hospital in Turin
  • May 2013 to December 2016


  • Inclusion: adults over 18 years with advanced chronic heart failure; ejection fraction < 30%; all of the following risk factors for diuretic resistance:
    • Systolic BP < 110 mmHg
    • Serum sodium < 135 mmol/L
    • Severe congestion (wet score > 12/18)
  • Exclusion: Reversible cause of acute heart failure (e.g. acute coronary syndrome, myocarditis, pulmonary embolism, valve disease, arrhythmias); cardiogenic shock (INTERMACS class < 2); eGFR < 15 ml/min/1.73 m2
  • 217 screened; 80 randomised
  • Baseline characteristics were similar between groups (Intermittent Bolus vs Continuous Infusion):
    • Age: 58.7 vs 63 years
    • Male: 88% vs 93%
    • Ejection Fraction: 19.2% vs 19.4%
    • Furosemide daily dose pre-randomisation: 242 mg vs 198 mg
    • Proportion that received high-dose furosemide in trial: 80% vs 63%
      • P = 0.14 so not sufficiently different to be considered significant


  • Continuous Infusion of Furosemide
    • The assigned dose of furosemide was given by continuous infusion
    • A dummy administration of 5% glucose placebo solution was given by intermittent bolus every 12 hours
    • Treatment continued for 72 hours, after which ‘open label’ treatment was given at the clinician’s discretion


  • Intermittent Bolus of Furosemide
    • The assigned dose of furosemide was given by intermittent bolus every 12 hours
    • A dummy administration of 5% glucose placebo solution was given by continuous infusion
    • Treatment continued for 72 hours, after which ‘open label’ treatment was given at the clinician’s discretion

Management common to both groups

  • Patients were given ‘low dose’ (120 mg/day) unless the following criteria were met, in which case they received ‘high dose’ (240 mg/day) furosemide:
    • Severe systemic congestion (wet score > 14/18)
    • Pre-hospital furosemide dose of > 125 mg
    • Supplemental metolazone therapy
  • Furosemide 40 mg or 60 mg was given in the Emergency Department
  • Maximum time from admission to diuretic treatment was 2 hours
  • Treatment was protocolised
    • If Systolic BP > 90 mmHg, sodium nitroprusside (vasodilator) was infused and titrated
    • If hypoperfused state (cold shock), dopamine or dobutamine were infused and titrated
      • If high doses were required and shock progressively evolved, vasopressors were started and the patient left the study protocol
    • Thiazide diuretics were not allowed
    • Fluid restriction of 1000 ml/day
    • Strict bed rest


  • Primary outcome: freedom from congestion at 72 hours (defined as JVP < 8 cm, and no orthopnoea and with little or no peripheral oedema) occurred more frequently in the Continuous Infusion group
    • Continuous Infusion group: 47.5%
    • Intermittent Bolus group: 25%
    • Absolute Risk Reduction: 22.5% (95% CI 2.02% to 42.98%; P = 0.036)
    • Number Needed to Treat: 5 patients
    • Fragility Index: ZERO
      • P = 0.619 when analysed with Fisher’s exact test
  • Secondary outcome:
    • Total urine output was greater in the continuous infusion group
    • Treatment failure occurred less often in the continuous infusion group
    • Diuretic response was greater (as defined by kg weight loss per 40 mg furosemide) in the continuous infusion group
    • Worsening renal function was not different
    • Worsening heart failure was not significantly different

Authors’ Conclusions

  • When administered to adult patients with advanced chronic heart failure and a high risk of diuretic resistance, furosemide administered by continuous infusion achieved better decongestion than furosemide administered by intermittent bolus


  • Appropriate methodology to test well-defined hypothesis
  • Appropriate concealment of random allocation sequence
  • Excellent double-dummy strategy to maintain blinding of allocation throughout treatment period
  • Well-defined patient population
  • Consecutive screening of all admitted patients to avoid selection bias


  • Power / sample size calculation does not make sense
    • The authors describe a decrease in the primary outcome from 48% in the control group to 25% in the intervention group
    • This would be a good thing if the primary outcome was a negative event
    • In this trial, the primary outcome was a positive event (freedom from congestion) so the hypothesis is that the intervention group will have a higher incidence of the primary outcome than the control group
  • Power / sample size calculation is of concern
    • Usually the control group and intervention group event rates are estimated from prior data
    • These estimates are combined with defined Power and significance levels to calculate a necessary sample size
    • In this trial, the event rates used in the sample size calculations are exactly the same as the primary outcome results of this trial
    • Has the power calculation been performed after the trial to establish the effective power having recruited an arbitrary number of patients?
  • Power / sample size calculation has not defined Power and significance levels that are generally acceptable in clinical trials
    • Usually Power is defined as 80% or 90%, but in this trial they have defined it as 70%, which increases the potential for Type 2 error (failing to reject the null hypothesis when there is a true effect)
    • Usually significance is defined as 0.05, but in this trial they have defined it as 0.10, which increases the potential for Type 1 error (rejecting the null hypothesis when there is not a true effect)
    • In summary, this trial has a sample size too small to have sufficient power or statistical significance to adequately test the hypothesis, by usual clinical trial standards
  • Imbalance in furosemide dosing between groups
    • Although this difference was not statistically significant, it may have been relevant
    • The dosage was higher in the intermittent bolus group, which may have biased in favour of the null hypothesis, so this may have reduced the measured effect
  • Fragility Index is zero
    • When using Fisher’s exact test instead of Chi Squared test, the trial result is no longer statistically significant by the threshold the authors defined
  • Single centre design reduces generalisability
  • Predominantly male population may reduce generalisability to females
  • Tightly defined homogenous population reduces the generalisability to patients with less severe heart failure, those without features suggestive of diuretic resistance and critically ill patients with iatrogenic oedema

The Bottom Line

  • This small and under-powered trial did demonstrate a benefit from continuous infusion of furosemide compared to intermittent boluses in a highly defined subset of patients with advanced chronic heart failure
  • I will continue to administer furosemide by continuous infusion in my critically ill patients, and unfortunately this trial fails to add any evidence to support or refute this clinical practice

External Links


Summary author: Duncan Chambler
Summary date: 2 August 2019
Peer-review editor: Steve Mathieu

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.