March 2nd, 2011

Questioning the DOSE

Although widely used for decades, the best way to use loop diuretics in patients with acute decompensated heart failure (ADHF) has never been well studied. The Diuretic Optimization Strategies Evaluation (DOSE) study, published in the New England Journal of Medicine, randomized 308 ADHF patients to a bolus every 12 hours or a continuous infusion of furosemide at either a high or low dose. No significant differences were observed in either patient symptoms or the change in creatine from baseline to 72 hours. In an accompanying editorial, Gregg Fonarow writes that the study “has not solved the problem of the poor prognosis for patients hospitalized with acute decompensated heart failure, nor has it modified the substantial expenditures for this disease.” However, Fonarow argues that the study introduces “the new concept of comparative-effectiveness studies into the field of heart-failure research.”

As a heart-failure clinician, these are some of my thoughts about the DOSE trial:

What did I like about DOSE?

In conducting DOSE, the NIH-sponsored Heart Failure Consortium has taken an important step back to evaluate age-old, evidence-devoid practice patterns, instead of focusing solely on novel therapies.

What did DOSE show?

The results suggest that high-dose, compared to low-dose, diuretics improve patient symptoms, and increase net diuresis and weight loss within 72 hours of hospitalization. However, they do so at the expense of a greater proportion of patients developing worsening renal function during their hospital stay. While the trial was not powered to look for differences in re-admission or mortality, high-dose diuretics — despite their early benefit — were not associated with a decrease in median length of stay or a decrease in the rate of death or re-hospitalization within 6 months, compared to a low-dose strategy.

What did DOSE claim to show, but didn’t?

The results suggested that there was no difference in 72-hour or 6-month outcomes between the bolus and continuous infusion strategies. However, patients randomized to the bolus strategy received substantially higher doses of diuretics than those randomized to the continuous infusion strategy (median dose: 592 mg vs. 480 mg, p=0.06). Therefore, before we can conclude that continuous infusion is no better than bolus dosing — shouldn’t we compare the bolus and continuous strategies within the low- and high-dose groups? Unfortunately, the trial was not adequately powered to conduct these subset analyses.

What would I have liked to know from DOSE?

I wish the Heart Failure consortium had powered the study to look for differences in heart-failure death or re-hospitalization between the different diuretic strategies. The plethora of prior literature suggesting that high-dose diuretics lead to bad outcomes, both because of and irrespective of worsening renal function, has led to the systematic under-treatment of patients with ADHF. Even in DOSE, less than 20% of patients were free from congestion after 72 hours of treatment in the hospital. While we do not know the proportion of patients who were free from congestion by the time of discharge, it is safe to assume that the majority were still “wet” given that the median length of stay was approximately 5 days in each treatment group. We know that 43% of patients were either dead, re-hospitalized, or presented to the Emergency Department within 60 days of randomization. What we don’t know is whether this was because they were discharged prematurely, because there truly was no difference between the different diuretic strategies, or whether the initial benefit gained with high-dose diuretics was negated by the higher incidence of worsening renal function.

So, in the end, is DOSE really going to change how we give diuretics? As it stands, all that DOSE tells us is that in a small trial of 308 patients, the way you administer diuretics has no impact on patient outcomes. In my opinion, the failure to power DOSE to see a difference in a conclusive hard end-point will make it a valiant research attempt that has virtually no impact on clinical practice.

For more on the DOSE study, check out Harlan Krumholz’s journal club.

6 Responses to “Questioning the DOSE”

  1. William DeMedio, MD says:

    I always give the dose of furosemide in ADHF initially as 20 mg plus the patient’s age. DOSE isn’t going to change it yet. This the art, not science, of medicine.

    Competing interests pertaining specifically to this post, comment, or both:

    • However, in diuretic naive patient, I saw a dramatic response to a small dose of IV furosemide(e.g. 10-20 mg.) I think the best approach should be to follow urine output closely. Quickly repeat/increase the dose if no response is seen in an appropiate time.

  2. Has anyone studied Age + Bun as an approach to furosemide dosing in a randomized, controlled, double-blind, multinational, trial?

  3. Nice to see people still reading House of G-D even if fellow in training. I would argue that the high dose arm was generally age plus BUN and the low dose arm was lower therefor the age old wisdom is correct

  4. Umar Shakur, D.O. says:

    You are, perhaps, a little harsh on the DOSE investigators. They are investigating a problem which receives little funding and that would limit both size and scope of the study. Nonetheless, it is curious that the investigators did not at least look at readmission from ADCHF, a criteria by which Medicare evaluates hospitals, if not death.

    I struggle with “de-escalation” of diuretic therapy. In that sense, DOSE helps me by saying that starting off high for 72 hours is relatively safe, but how do I slow down? Cut dose by 25%, 50%? How long should I leave them on a higher dose? What should I tell the PCP about when to go back to their “home dose,” if that’s appropriate?

    Given the work DOSE has done, I think a cross-over trial would be the next step, one that looks at starting high and then different strategies of de-escalation, for the entire length of stay and perhaps 1 week post-discharge.

    What is your typical practice on this issue?

    Competing interests pertaining specifically to this post, comment, or both:

  5. At P.06, there is 94% chance that high dose is better- p values are arbitrary numbers. A slight increase in creatnine does indicate “renal failure”- it is an iconsequential lab abnormality as no change in renal function long term We all know clinically that many pateints with CHF require a higher than usual BUN/CR ratio to maintain homeostasis- same me be true and a slightly elevated Cr may bea abtter endpoint than a lower one for a specfic patient.