Intended for healthcare professionals

Rapid response to:

Practice Uncertainties

How to prescribe loop diuretics in oedema

BMJ 2019; 364 doi: (Published 21 February 2019) Cite this as: BMJ 2019;364:l359

Rapid Response:

Loop diuretic dose response on a linear scale and monitoring for adverse effects

There have been many excellent Rapid Responses posted regarding Anisman et al.’s article entitled “How to prescribe loop diuretics in oedema”1 but two critical points have not yet been addressed. The first is that the underlying concept promoted by the article, that “Loop diuretics respond in an all-or-none fashion”, is not accurate. To support this concept the article displays an idealized “Dose response curve (logarithmic) . . . adapted from Brater 19982 and Brater 2011.3” However, both of these references are review articles rather than original research. More revealing is Brater’s original investigation of furosemide pharmacodynamics published in 1980.4 The latter paper displays, also with a logarithmic scale on the x-axis, the pharmacodynamics for 40 mg of furosemide given intravenously (IV) in ten heart failure patients compared to the mean results in eight normal controls. Because clinicians may have more difficulty interpreting a graph with a logarithmic scale and because the use of “log(dose) . . . can visually imply a threshold dose when in fact there is none”,5 I digitized the points and curves on Brater’s plot using ImageJ 1.52i and replotted them on a linear x-axis scale using R version 3.5.2. Because Brater’s study addresses concentration-response rather than total dose-response, I also digitized and replotted the dose-response plots from another study.6 Unfortunately the Rapid Response format does not support images, so I submitted the images and discussion to a peer-reviewed journal specializing in pharmacology. They have recently been published at DOI 10.1093/ehjcvp/pvz026.7

Figure 1 in my article shows the illusion of a large threshold in the semilog plot from normal subjects that is not apparent in the linear plot. The linear plot also shows that the curves for all but one of the ten heart failure patients are shifted to the right and flattened, confirming varying resistance to diuretic effect in these patients. Figure 2 shows that the dose-response for bumetanide is virtually linear in the oral dose range 0.5 to 2 mg. Dose-response for furosemide in the oral dose range 20 to 80 mg was also close to linear (although not shown in my article). There is no evidence of an all-or-none effect. This issue is a prime example of the adage that one picture is worth a 1000 words, so I urge clinicians to examine these plots of original study data.

The second critical point missed is that adverse effects of loop diuretics are almost completely ignored. In the Anisman et al. article there is only one sentence stating that “With any loop diuretic, monitor electrolytes and renal function routinely and when dose or symptom changes.” There is no advice regarding how often to do the monitoring or what to do if any results are concerning. For most prescription drugs advice regarding how to monitor for and address adverse effects is just as important as what dose to prescribe. Loop diuretics affect potassium, magnesium, calcium, uric acid, and glucose metabolism. All of these effects have to be addressed to convey a complete understanding of how to prescribe loop diuretics in edema.

For loop diuretics monitoring for adverse effects is critical: One of the potential adverse effects, hypokalemia, can be deadly. Hypokalemia increases the risk of ventricular arrhythmias that can be instantly fatal. I have become convinced of the importance of hypokalemia in heart failure survival based on my experience as a U.S. Food and Drug Administration (FDA) reviewer in analyzing the raw data from the four major trials of mineralocorticoid receptor antagonists (MRAs) in heart failure. In all four trials the beneficial impacts of MRAs upon cardiovascular mortality is greatest in the subgroups of patients with low serum potassium at baseline. The impact is attenuated as baseline serum potassium increases. MRAs increase serum potassium so some, or all, of the beneficial impact of MRAs upon heart failure survival may be mediated through their effects upon potassium metabolism. The relationship between baseline potassium and cardiovascular mortality in two of the trials is shown in a publication.8 Unfortunately the results for the other two trials, included in an FDA review posted on the FDA website, were censored by an FDA lawyer. Finally, I believe that monitoring for hypokalemia and taking appropriate action if it is found (e.g., prescribing an MRA, potassium-sparing diuretic, or potassium supplement) is not only essential for the use of loop diuretics but also the most underutilized maneuver for improving the survival of heart failure patients.


1. Anisman SD, Erickson SB, Morden NE. How to prescribe loop diuretics in oedema. BMJ. 2019;364:l359. doi:10.1136/bmj.l359

2. Brater DC. Diuretic therapy. N Engl J Med. 1998;339(6):387-395. doi:10.1056/NEJM199808063390607

3. Brater DC. Update in diuretic therapy: clinical pharmacology. Semin Nephrol. 2011;31(6):483-494. doi:10.1016/j.semnephrol.2011.09.003

4. Brater DC, Chennavasin P, Seiwell R. Furosemide in patients with heart failure: shift in dose-response curves. Clin Pharmacol Ther. 1980;28(2):182-186.

5. Wikipedia contributors. Dose–response relationship. In: Wikipedia. ; 2019. Accessed March 12, 2019.

6. Ramsay LE, McInnes GT, Hettiarachchi J, Shelton J, Scott P. Bumetanide and frusemide: a comparison of dose-response curves in healthy men. Br J Clin Pharmacol. 1978;5(3):243-247.

7. Marciniak TA. What are the pharmacodynamics of loop diuretics? Eur Heart J Cardiovasc Pharmacother. July 2019. doi:10.1093/ehjcvp/pvz026

8. Marciniak TA. Eplerenone treatment of aortic stenosis. [letter]. Am Heart J. 2009;157(3). doi:10.1016/j.ahj.2008.10.023

Competing interests: No competing interests

05 August 2019
Thomas Marciniak
39344 Hatteras Dr