Paradigms can turn on a dime,or maybe a data-dredge,Case in point digitalis

With in memory (ok, a long memory) we did not worry about should we give digitalis to heart failure patients, the only issue was how much of which preparation.Some of us even remember digitalis leaf being talked about.
The earlier-day paradigm envisioned the heart as a failing pump and drugs to increase contractility seemed to make sense and digitalis was considered an inotrope and it still is but is now thought to do more good as a renin inhibitor and to tone down over- exuberant sympathetic activity.
Decades of observed improvement from digoxin were in part validated by two clinical trials that gave enough data for the FDA to approve it as a treatment for CHF. But there were problems and the results were not nearly as conclusive as those demonstrating the value of ACE inhibitors. The ACE inhibitors did it all, decreased symptoms, increased exercise tolerance and improved survival. Digitalis did not increase survival in the large DIG trial but did decrease the need for hospitalizations.The RADIANCE trial interestingly showed that patients on digoxin worsened when it was discontinued.
The data seemed sound enough for the AHA/ACC 2001 guidelines to recommend its use,either early on while the ACE inhibitors were taking their time to work or later in the progression of disease when the patients had already been treated with diuretics, ACE inhibitors and beta blockers.And-after 200 years of use-the FDA had approved its use as a treatment for CHF.
Then in October 2002, a retrospective analysis of the DIG trial was published in NEJM. There was a 4.2 % increase in mortality in women in the DIG trial. The AHA/ACC guidelines had made no distinction based on gender, but the authors of the analysis suggested perhaps they should rethink that.
In the same issue a "Perspective" was offered pointing out that perhaps higher digitalis levels may have played a role in the findings as after one month of treatment they were higher in women.However, an adjustment was not made in their analysis for digitalis levels. The authors cautioned against abandoning a therapy that may help women in heart failure but that greater attention was needed in doses and monitoring.
I do not really know how much the digitalis paradigm was turned around by this one article; there are other arguments for using it sparingly if at all. But when a paper really turns around a practice or a mindset, I hope it will usually be something more substantive that a retrospective sub set analysis of data. Perhaps something like the Woman's Health Initiative which took HRT from a means of decreasing heart attack risk, maintaining bone density,healthy skin and urogenital structures to posing a risk of heart attack, stroke blood clots and breast cancer. On the other hand,just because sub set analysis is more vulnerable to biases it does mean that the conclusions based on it could not be correct. This seems to be still another instance in which we have large clinical trial-the DIG study (randomized,etc,etc)-and various analysis of the data, and we still are not sure what to do. Do you give digitalis to women with CHF or not?