Thursday, April 7, 2016

Walk, don't run, to implement SPRINT findings in primary care

This the first part of a two-part commentary that I wrote in collaboration with Dr. Stephen Martin and Dr. John Mandrola. The second part is posted here.


When the Systolic Blood Pressure Intervention Trial (SPRINT) results were published last November, a New York Times headline announced “Data on Benefits of Lower Blood Pressure Brings Clarity for Doctors and Patients.” On the same day, in the same newspaper, writing in an editorial, Dr. Harlan Krumholz detailed “many nuances” of the study’s findings. Clarity or nuance, which is it?

Front-line clinicians are well aware of the dangers of overlooking nuance when applying clinical trial data to patient care. In this Commentary, we will make the case that implementing SPRINT in primary care warrants slow thinking, and that efficacy of setting lower blood pressure targets shown in this trial is not the same as effectiveness in the real world.

SPRINT enrolled a high-risk population of hypertensive adults age 50 years and older who also had actual blood vessel disease, chronic kidney disease (CKD) (estimated glomerular filtration rate 20 to less than 60), Framingham risk greater than 15%, or were older than 75. In fact, the average Framingham risk score was 20%. Patients with diabetes, prior stroke and those who had “any factors judged to be likely to limit adherence” were excluded by the trial design. Two specific features of SPRINT’s design would be difficult or even impossible to duplicate in real world practice. The first is that blood pressure was measured in an idealized way. After the study was suspended in September, principal investigator Dr. Suzanne Oparil noted that

We have to remember that in a randomized controlled trial, the clinic blood pressures are pretty accurate in the sense that these people are coming back visit after visit after visit, they're seen by the same coordinators who are very nice to them and don't upset them or anything, they have their blood pressure measured with an automated device according to AHA and international guidelines—sitting quietly with feet on the floor for 5 minutes, no smoking, no drinking coffee, no talking, no disturbance in the room. … So if you try to push … [artificially] high routine clinic pressures down to 120, you may get into trouble.

The second is that anti-hypertensive medications were provided directly to SPRINT participants for free. This extraordinary convenience, combined with a trial run-in period which allowed the investigators to exclude patients who may not take medications properly, assured the enrollment of participants most likely to adhere to the many drug adjustments made in the trial. In the real world, patients travel to pharmacies to pay for and pick up their prescriptions, and clinicians can’t ignore patients who forget to take pills or have trouble following directions.

We view SPRINT as a proof-of-concept and efficacy trial. It asked the question: in high-risk patients, does further lowering of blood pressure with drugs help, harm, or have no effect? The answer is all three. Although the intervention group in SPRINT had better outcomes than the control group, the absolute benefits of treating to a systolic blood pressure goal of 120 mm Hg versus 140 mg Hg were modest (Table). 62 people needed to be treated more intensively to prevent one cardiovascular event or stroke, and 85 needed to be treated to prevent one death from all causes. These benefits came with costs: more intensive treatment—three blood pressure drugs on average compared with two in the control group—was associated with one additional episode of severe hypotension, one case of syncope, and two more episodes of acute kidney injury for every 100 participants. We note that for 98 to 99 percent of participants, there was no difference in outcomes.


ARR (%)
PSR (%)
Primary Outcome*
CV Mortality
Serious adverse event
Serious adverse event^
* First occurrence of myocardial infarction, acute coronary syndrome, stroke, heart failure, or death from cardiovascular causes
^ Possibly or definitely related to the intervention, including hypotension, syncope, electrolyte abnormalities, and acute kidney injury in the invasive group

Table courtesy of (