Cardiologists: Thumbs Down To SPRINT

[Updated, August 29, August 30]

–SPRINT should not be used in guidelines to lower blood pressure targets.

Should the SPRINT trial be used by guideline committees to lower systolic blood pressure targets? After listening to a high-powered debate at the European Society of Cardiology meeting in Rome on Sunday, most audience members gave thumbs down to the proposal.

The audience appeared to be swayed by a strong argument from Sverre Kjeldsen (Oslo, Norway), who said the trial was difficult to interpret at best and, in at least one important respect, unethical.Screen Shot 2016-08-28 at 6.07.19 PM

Kjeldsen’s strongest point was that the trial adopted an unprecedented and novel method to measure blood pressure, making it impossible to compare with previous trials. In all the major recent hypertension trials blood pressure had been measured three separate times with an automatic monitor in the presence of a healthcare professional. In SPRINT, however, the healthcare professionals were trained to leave the room before the measurements started. This method has not been validated and Kjeldsen presented multiple lines of evidence suggesting that it would lead to much lower blood pressure readings, as much as 10-20 mm Hg lower than usual. He said that the systolic BP target of 120 mmHg in SPRINT when examined in this light is actually closer to the prevailing standard of 140 mm Hg in other trials. He was also strongly critical of the SPRINT investigators for not reporting this significant fact in the original New England Journal of Medicine paper or in its supplement. He said it was unethical not to report this since many clinicians have tried to apply the findings to their patients but were not aware of this important change.

Kjeldsen also noted that nearly 500 people discontinued treatment during the trial and more than 250 were lost to followup, further limiting any interpretation of the trial.

One of the striking findings of SPRINT was the reduction in heart failure in the intensive treatment arm. But, said Kjeldsen, given their age and other risk factors the SPRINT population had a very high risk of heart failure. He said that differences in the main outcomes may have been due to greater and more effective use of diuretics in the treatment group than in the control group. “Thus, the main finding is likely an artifact caused by increased risk of heart failure in the standard treatment group because of less diuretic treatment.”

Further, the intensified diuretic regimen in the aggressive treatment group also likely led to a significant increase in adverse events, including hypotension, syncope, electrolyte imbalance, and acute kidney injury or acute renal failure. Kjeldsen cited the Annals of Internal Medicine article that calculated that for every 1,000 persons treated for 3.2 years with the more aggressive target “an average of 16 persons will benefit, 22 persons will be seriously harmed, and 962 will not experience benefits or harms.”

“SPRINT should have no implications for guidelines or clinical practice,” he concluded.

Kjeldsen also raised the ethical issue— which I wrote about here— concerning the early and potentially misleading release of the results two months before the publication of the trial in NEJM.“In the view of this reviewer it appears unethical that a closed and limited group of SPRINT investigators under the leadership of the NIH Director makes a press release 2 moths prior to study publication,” he said.

William Cushman (Memphis, TN) a SPRINT investigator, was charged with defending the proposition that SPRINT should be used by guideline committees to lower blood pressure targets. Cushman chose a less aggressive position, explaining the trial and defending it from some of the criticisms launched by Kjeldsen. The main substance of his talk was a detailed presentation of the main findings of SPRINT.

He took issue with the charge that there were more serious adverse events in the intensive treatment group. Overall there was no significant difference in SAEs. Moreover, he said, it is often possible to successfully manage SAEs, making these of lesser concern than a mortality event which, obviously, can not be managed.

Cushman also addressed the idea that decreased use of diuretics may have explained the increase in heart failure in the control group. In fact, he said, patients in the control group had more diuretic use after they came in the study, but he acknowledged that it may have been possible that some patients were removed from some therapies during the study as part of their treatment strategy.

He suggested that the optimal target may lay within the range of 120-135 mm Hg. He also pointed out that in practice a lower target would lead to more people reaching a higher target. (Bryan Williams (London, UK), a program chair, ended the session by emphasizing that even now more than half of hypertensive patients don’t reach the current goal of 140 mm Hg.)

Earlier, setting the stage for the debate, Giuseppe Mancia (Milan, Italy) raised questions about the trial and its interpretation, perhaps putting Cushman at a disadvantage. Mancia said that the SPRINT evidence was limited to high risk patients and might not apply to younger or lower risk populations. He also said that it is possible that optimal targets may vary for each individual, and may even vary at different times for individual patients) and said that this may “limit the practical value of trial information.”

Mancia also said that the question about unattended BP measurement was “quite disturbing,” but argued that the difference between observed and unobserved blood pressure in high quality research is much smaller, more like 5 mm Hg than 15 mm Hg. “The unattended measurement makes it very difficult to make comparisons with other trials,” he said.


[August 29]–Sanjay Kaul offered the following comment in response to the SPRINT debate:

“I have given 2 invited lectures at AHA meetings talking about consumer report guide to interpretation of trials. I use a scale form 1 to 5 stars (5 being the highest) based on the quality and quantity of evidence. Even though the effect size of the primary endpoint was large (25% RRR) and statistically persuasive (p=0.004), the quality was moderate (open label study, prematurely stopped with 5.5% missing data and no exploration of missing data on the PEP), thereby earning a 3 star rating. In my book, only 4 or 5 star-rated trials are practice changing!
The fact that stroke (which is very sensitive to BP reduction) was not favorably impacted, but HF and death was, raises additional questions! In my book, “quality” trumps “quantity”!”

[August 30]– I have been informed that there have been questions raised about the precise details about the way blood press monitoring was conducted in SPRINT. According to one source, BP measurement in SPRINT was very different from the way it was portrayed in the debate. However, it should be made clear that both participants in the debate– Cushman on behalf of SPRINT and Kjeldsen in opposition– were in agreement about these details. The slides for each are presented below. It should also be noted that Kjeldsen’s larger point remains uncontested, which is that blood pressure measurement in SPRINT was not performed in the same way as previous trials, thereby making it difficult to compare the SPRINT results with those trials. I will update this post as more information becomes available.

Here is Cushman’s slide (click to enlarge):
Cushman slide


And here is Kjeldsen’s slide (click to enlarge):

Kjeldsen slide


sprint vote

Followup story:

SPRINT: More Controversy And Confusion About ‘Landmark’ Trial

Previous coverage of SPRINT:



  1. I’m not sure if your picture is representative of the entire room but the thumbs up and thumbs down by my count seem about the same.

  2. Richard Cooper says:

    I have been involved in measuerment of 10’s and thousands of BP’s in research studies and the idea that 10-20 mmHg lower than other trials is ridiculous. Patients reach a basal pressure after no more than 3-5 readings, whether someone is in the room or not. If SPRINT participants were really 20 mmHg lower than NHANES values episodes of hypotension would have been much more common. The question needs to be studied but the strong assertion that the study is invalidated should not be taken seriously unless there is evidence that the criticism is correct.

Leave a Reply

%d bloggers like this: