GETTING TO THE HEART OF MEDICINE
(continued from page 17)
any observed differences were due to the treatment or differences in
selection. Bias is the issue. For instance, did the surgeon’s ligation of
the left atrial appendage at the time of cardiac surgery reduce future
stroke, or did the surgeon simply choose not to do the procedure
on sicker patients?
Much stronger on the strength of evidence is the randomized
control trials (RCTs). RCTs, especially when double-blind, reduce the
biases of observations because the randomization (mostly) balances
the differences in the two treatment groups. In a well-conducted
RCT, if the group who received treatment A had a lower death rate
than the group who had treatment B, a clinician can know that
treatment A caused the reduction.
Cardiology is replete with examples of the power of randomization.
Years ago, when I began medical training, we observed that
premature ventricular complexes (PVCs) seen after a myocardial
infarction (MI) conferred a higher risk of sudden death. We also
knew anti-arrhythmic drugs reduced PVCs. It became standard of
care to use these drugs to suppress the PVCs.
But then, a courageous group of investigators, led by Dr. Debra
Echt, decided to test the standard approach. Patients who had
PVCs after MI were randomly assigned to either placebo or an
anti-arrhythmic drug. It is hard now to convey the scariness of
giving post-MI patients with complex ectopy a placebo, but the
Cardiac Arrhythmia Suppression Trial (CAST) 2 trialists stunned
the world of clinical medicine when they reported a higher death
rate in patients who received the anti-arrhythmic drug.
CAST sped up the era of evidence-based cardiology practice.
Now, much of what we do has high-level evidentiary support.
Doctors do not recommend statins because they lower cholesterol
levels. We recommend these drugs because RCTs, and the
combination of RCTs, called meta-analyses and systematic reviews,
have reliably shown a clear-cut reduction in cardiac events with
these agents. 3
We use warfarin to prevent stroke in patients with atrial fibrillation
(AF) not because AF increases the tendency to clot and warfarin
reduces clotting, but because trials showed that warfarin reduced the
probability of stroke in patients with AF versus aspirin or placebo. 4
Then, when direct acting oral anticoagulants were invented, trials
in tens of thousands of patients showed that the more convenient
drugs did as well as or better than warfarin. 5
Cardiologists do not prefer percutaneous coronary intervention
and stents over fibrinolytic therapy in patients with acute myocardial
infarction because the procedure makes an angiogram look beautiful.
Interventionalists come in at all hours because RCTs showed acute
Percutaneous Coronary Intervention (PCI) to be superior in efficacy
and safety (less intracranial bleeding) versus fibrinolytic drugs. 6
RCTs dominate in the care of patients with heart failure due to
systolic dysfunction. Trials have shown benefits from drugs that
block receptors in the renin-angiotensin, adrenergic and mineralocorticoid
systems. Most recently, the RCT called DAPA-HF found
that the sodium-glucose co-transporter-2 inhibitor, dapagliflozin,
has shown substantial efficacy in patients with heart failure. 7
Coronary artery bypass surgery, internal cardiac defibrillators,
transcatheter aortic valve replacement, mechanical thrombectomy
of acute stroke and many other modern-day procedures are backed
by evidence from randomized controlled trials. Cardiology trials
are also known for their hard endpoints—outcomes like MI, stroke
and death.
Granted, we have had decades to study heart disease and only
months to study COVID-19. But as of this writing, more than 6
million people have tested positive for this virus and yet the vast
majority of science has come from the lowest rung of evidence—
anecdote and observational data.
I searched the term “COVID-19 treatment” on PubMed on June
1 and found nearly 5,000 publications. Only 9 were RCTs. My friend
Dr. Bishal Gyawali and his colleague Dr. Aakash Desai published a
review of trials of COVID-19 listed on www.clinicaltrials.gov and
found only 49 ongoing RCTs of which only 6% had overall mortality
as an endpoint. 8
The hydroxychloroquine (HCQ) controversy has occurred in
large part because doctors failed to apply the tenets of EBM. The
original “study” of HCQ in COVID-19 from Southern France had
no randomization and the authors simply removed patients who
did worse on the drug from their analysis. 9 A colleague aptly called
this effort an unstudy.
Through good intentions, the spin of mainstream media and
therapeutic momentum, HCQ actually became codified as standard
of care in many institutions. Not only did we not know whether
the drug worked, but normalization of the drug made enrolling
patients in trials difficult because patients did not want to be in
the placebo arm.
Months later, however, The Lancet published a large observational
trial of HCQ use in patients with COVID-19 that found a
much higher risk of death in patients who took the drug. 10 Mainstream
media took off with news that the now politicized drug was
harmful. Momentum had swung against the drug. But this study,
like the first one from France, was also a flawed, non-randomized,
observational analysis.
In the same way that low-level evidence should not have estab-
18 LOUISVILLE MEDICINE