NIH, news media, need to slow down on the SPRINT hype

This post has been updated with additional expert commentary on the SPRINT trial and the NIH’s announcement regarding the results (scroll to the bottom).

The National Institutes of Health (NIH) today made an announcement about the SPRINT (Systolic Blood Pressure Intervention Trial) study that is making big waves across the world of health care.

Landmark NIH study shows intensive blood pressure management may save lives

And after reading the NIH release, you just might be convinced that the SPRINT study does indeed deserve a big red-carpet roll out.

Because the claims it makes are pretty bold:

More intensive management of high blood pressure, below a commonly recommended blood pressure target, significantly reduces rates of cardiovascular disease, and lowers risk of death in a group of adults 50 years and older with high blood pressure.

Coming from a big government-funded trial, such a finding certainly seems to qualify as a news-making, and potentially practice-changing, event.

There’s just one problem with all of this: The big announcement about these “landmark” results doesn’t actually include any “results” from the SPRINT study.

You can search the news release high and low but you’ll find nary a single statistic from the study or a number related to what the researchers found.

But there’s plenty about how pleased the researchers are to have “achieved this important milestone.”

The study apparently “provides potentially lifesaving information that will be useful to health care providers as they consider the best treatment options for some of their patients, particularly those over the age of 50,”  said Gary H. Gibbons, M.D., director of the National Heart, Lung, and Blood Institute (NHLBI).

But what exactly is that information? Apparently we’ll need to wait a few more months for those pesky specifics.

At a news conference about the study, Director Gibbons said that the lower blood pressure target reduced events by 30% and cardiovascular deaths by 25%, according to journalist Larry Husten, who tweeted these statistics.

But Gibbons apparently gave no indication of either the absolute reduction in events or the Number Needed Treat — statistics that we think are vital to understanding the magnitude of the benefit.

And there was nothing in the news release (or apparently in the news conference) about the potential harms of treating to a lower target.

Harlan Krumholz, a Yale University cardiologist, immediately took to Twitter to voice his concern.

Jeremy Sussman, described on Twitter as a primary care physician, was also a prominent early voice weighing in with criticism about how the results were communicated.

Let me be clear about one thing: It may well turn out that the study is just as important and influential as the NIH communications suggest it will be. And I certainly hope that it is. My point and my concern is that we shouldn’t have to guess at what the results will eventually show.   The horse is already out of the barn, with the New York Times headlining the study “lifesaving” and calling for blood pressure targets that are “way lower than the current guidelines.” But the Times’ incomplete report, like the incomplete news release that it’s based on, has very few specifics about the study results as of this writing.  Let’s all slow down, take a deep breath, and wait until there’s some real news to report here.

The study may be called SPRINT, but good health care is more like a marathon.


wilkesUpdate 9/14/15: Michael Wilkes, MD, PhD, Professor of Medicine at the University of California, Davis and a recent addition to the team, pointed us to his audio commentary about SPRINT, “A half-baked pie,” for NPR affiliate KCRW.




Update 9/12/15: pignone2A longtime contributor to, Michael Pignone, MD, MPH, Professor of Medicine and Chief of the Division of General Internal Medicine at University of North Carolina – Chapel Hill, said: “I agree with Harlan [Krumholz] (and others) that they should have released more data simultaneously. A larger, more general issue is whether to look at this trial in isolation, or whether to consider it in the context of other trials that have asked similar questions. Most such trials have been done in higher risk populations (particularly patients with diabetes); the HOT trial is an exception. Of course, we will need more data and details in order to do the systematic review work and meta-analyses that are the foundation for such (Bayesian) consideration. One technical, epidemiological point: I actually care less about the ARR (absolute risk reduction)/NNT (number needed to treat) for this trial alone because it is mainly a representation of the population studied in this trial, and because it is difficult to synthesize ARR or NNT across trials. Instead, I want to know whether the relative risk reduction is similar for patients across different risk and blood pressure levels (this is controversial). If it is, then I want to develop a summary estimate of the RRR (relative risk reduction) and then be able to apply it to patients at different levels of risk, as we can do with statins. Once we have a stable RRR, say 20%, then we can say:

  • If you treat people with SBP of 140 and 10 year CVD risk of 10%, you will reduce that risk to 8% (NNT 50)
  • If you treat people with SBP of 140 and 10 year CVD risk of 20%, you will reduce that risk to 16% (NNT 25) etc.

The key is to determine whether the RRR (relative risk reduction) is stable at different BP and risk levels- this new trial suggests that it is . . . but I would need to see more data to be confident in that conclusion.”

You might also like

Comments (7)

Please note, comments are no longer published through this website. All previously made comments are still archived and available for viewing through select posts.

Ignatius Brady

September 11, 2015 at 1:45 pm

Gina Kolata’s coverage in the times might be a good candidate for full review on your front page. It’s pretty fact free, doesn’t address any limitations to generalizability of findings (the study subjects’ risk category, etc) or side effects of treatment. I was surprised by it.

Steven Miles, MD

September 11, 2015 at 2:56 pm

Wow! This is a SPRINT to the flashing lights on the marquee. Fact free science? What a novel idea!
CDC says 70 million US adults have high blood pressure; 360,000 die / year. If this study shows that intensive management would cut this by 25% that is 90,000 deaths per year which seems huge. But for every 1 million treated people, 3800 will die of have high blood pressure and 1200 would be saved–roughly treating a thousand persons to save one life assuming there are no risks to treatment which is not a valid assumption.

David K. Cundiff

September 13, 2015 at 5:53 pm

I want to know what lifestyle change recommendations (diet, exercise, stress management, etc.) were given to the subjects of this trial, how compliance was monitored, and what were the results of the lifestyle change intervention. I also want to know what the NIH thinks should be the new hypertension treatment guidelines based on this trial and what they estimate it will cost to implement the new guidelines over the next decade.

    Richard Huhn

    September 15, 2015 at 12:12 pm

    I would like to know what drugs were used in what proportions of subjects. For some of the anti-hypertensive agents’ side effects, I would rather take my risk of CV/Stroke outcomes than suffer the side effects for an additional few years of (uncomfortable) life.

Lisa Cool

September 14, 2015 at 12:32 pm

Another thing that troubles me about media coverage of this study is that many news outlets, including some very well-known ones, mistakenly refer to blood pressure below 140/90 as “normal,” when prehypertension, which affects about one in three US adults according to the CDC, is defined as blood pressure between 120/80 and 129/89, and “normal” is BP below 120/80. (See

Also, I’d like news stories to provide context by reporting that there are a number of earlier studies showing a strong association between lower blood pressure and lower risk for stroke, so while this study is being called “landmark” research by the NIH, it is not the first to link lower BP to reduced risk for CV events. For example, a peer-reviewed study published several years ago reported that risk for stroke starts to rise once blood pressure reaches 115/70 and doubles for each 10 mmHg rise in systolic pressure and each 5 mmHg rise in diastolic pressure.

Antonio Machado

September 21, 2015 at 12:42 pm

Because bedtime versus at awakening administration of BP lowering drugs improves BP control and reduces cardiovascular, renal and metabolic events, it is very important to know at what time of the day were the medications given in SPRINT.

peter melcher

October 7, 2015 at 4:47 pm

The results are at odds with previous research. The Framingham raw data showed little increase in cardiac events until systolic blood pressure reached about 155 (See Figure 2 in hyperlink below).

They are also inconsistent with the conclusion of the Cochrane Collaboration that treating systolic blood pressure between 140 and 160 wasn’t beneficial–

Finally, cutting the trial short suggests that they wanted to stop before side effects started developing.

Today at 4:32 PM

Today at 4:32 PM