What the trial itself answered
The DECISION trial randomised 1,001 heart-failure patients with a left ventricular ejection fraction of 50 percent or less, double-blind, to low-dose digoxin or placebo, and followed them for a median of 36.5 months. The primary endpoint combined total hospitalisations and urgent hospital visits for worsening heart failure with cardiovascular death. There were 238 events in 131 of 500 patients on digoxin and 291 events in 152 of 501 on placebo: a rate ratio of 0.81, 95 percent confidence interval 0.61 to 1.07, p value 0.133. The authors' own conclusion is plain: no significant reduction was shown on that endpoint.[1]
Who it was measured in is part of the measurement. Participants were 72 years old on average with a standard deviation of 9; 28 percent were women and 29 percent had atrial fibrillation. Dosing aimed at a serum digoxin concentration between 0.5 and 0.9 nanograms per millilitre. On that regimen the drug was generally well tolerated and safe, and results were similar in men and women. No serious safety signal over more than three years in an elderly, largely male group of patients is a narrow but real gain.[1]
Where does the headline figure come from?
In the picture the UMCG team announced, the trial does not stand alone. Cardiovascular death and worsening heart failure fell by 19 percent on digoxin, and that did not reach statistical significance on its own. When the researchers combined the finding with two earlier studies the patient count grew and heart-failure hospitalisations came out about 25 percent lower, a significant result that held in patients already taking all four standard heart-failure drugs. For a drug costing less than ten cents a day, that is a number with weight for the guidelines.[1]
The most economical reading of the distance between the two figures is that the trial was underpowered for its composite. DECISION counted 238 primary-outcome events on digoxin and 291 on placebo, and if that confidence interval stretches from 0.61 to 1.07, both a real benefit and no benefit can sit inside it; as the pool grows the interval narrows and a smaller effect starts to look significant. There is another explanation: the combined studies ran in different years and in different patient groups, so the larger effect may come from those trials' patients rather than from DECISION's. Which explanation holds is something the pooled analysis's heterogeneity measures and subgroup breakdown can tell.[1]
What changed for a patient?
What a trial establishes depends on the comparison it chose. My earlier column on people over 65 tied the failure to show a clinical value for mFlusiva to the fact that it had not been compared against the high-dose vaccine that age group receives. In the DECISION trial the comparison is sound: placebo, and on top of four-drug standard care. This time the question moves one step along, into which analysis carries the number.[1], [2]
For someone living with heart failure today, the concrete thing the trial delivers is this: low-dose digoxin added on top of four-drug therapy was generally well tolerated and safe over an average of three years of follow-up. The claim that it lowers admissions comes from the analysis pooling two earlier studies rather than from the trial's primary endpoint. More than 500,000 people in the Netherlands are estimated to have heart failure and digoxin costs less than ten cents a day; those two figures make it all the more necessary to know how much of the result each analysis carries.[1]