Everyone keeping blood pressure low because of the SPRINT trial is practicing bad medicine
SPRINT is a flawed study
SPRINT is a large randomized trial that has changed practice and led many providers to lower blood pressure beyond prior targets. Here is the conclusion of the paper:
But is this trial reliable?
I always say that a clinical trial can only change your practice if the control arm is your practice. SPRINT messed this up. The trial didn’t just target less than 140 in the control arm — it did something beyond this. If, by chance, you got lucky and controlled someone’s BP easily — taking them lower than 140, the trial design made you dose reduce their pills. This was a non-standard maneuver that might cause harm. Marc Pfeffer at Harvard saw it, and asked how often it happened. Check it out:
If it happened a lot, it would be a big problem.
And here was the reply.
Wow! 87% of people in the control arm of SPRINT had BP pills dose reduced even though they felt fine. This is not consistent with clinical practice, and a huge group of people. As such, it isn’t just a trial of aiming for <140 vs <120, but a trial of sabotaging people to keep them >135 vs aiming for less than 120.
Of course, the latter is better, but that’s not what doctors were doing.
It might still be best to aim for <140 and not <120. We don’t know.
Sadly SPRINT was poorly designed and interpreted poorly.
This is a huge problem and renders the trial useless IMO.
Wow. I am shocked. I have never picked up on that issue with the control arm. Disappointing. Those of us in primary care have got to be able to "trust the science". When a landmark study comes out and every body that gives guidance is in agreement and gives you specific advice for these targets, you generally accept that the science is good! It is a very unsettling situation that we've really got to question everything told to us and go back to the study design to parse it out for ourselves. The bigger take home point I've gotten from Sprint is that the method by which we are even measuring BP is extremely important! I can attest to the high variability in blood pressures obtained in daily clinical practice. Health systems that continue to use auscultatory methods are not using what was done in Sprint. The entire procedure used to measure BP has so many areas where things can be done "wrong" - taking BP over shirts/sweaters/jackets, while talking to the patient, patient legs crossed, patient sitting on exam table with arm not at correct level, not waiting 5 minutes, etc etc. This is how BP is monitored in the vast majority of places. I'll consistently get a certain BP range on a patient and the very next day in the cardiology office, a drastically different result is obtained. Not terribly helpful. Before we can hit targets, it would make most sense to get the data point in correct fashion! It would also be useful to do a "real world" study to see how big health system "quality metrics" affect BP data collection, overall BP control, and patient outcomes. When your health care system takes "the last BP measurement for the year" as the data point on which your "quality bonus" will be determined, it drives various providers to do different things to make sure their patients hit that target. Again, if you focus on blood pressure for a living, you realize how labile this particular data point can be and the myriad factors that affect the measurement and the treatment of such. In all honesty, I don't think this realization of Sprint design will change my practice. Several years ago I decided to focus on the actual measurement of blood pressure in my office and to make sure we were doing it correctly. Having patients able to do home BP monitoring and instructing them on the ideal method is also helpful. The first step is making sure the quality of your "data" is as accurate as it can be and then make the adjustments from there.
Always amazes me : the inherent variability of BP measurement in different settings, using different devices and different personnel makes any study of BP treatment suspect. Suggesting that you can interpret differences in the magnitude of 5 mm seems a bit silly.