The paper is:
The study was funded by NIH and the authors declare no conflicts of interest, so at first blush this would appear to be a poster child for reliable, unbiased research.
The research question is: If you take a group of folks who have had a stroke of one specific type (lacunar strokes, located in the lower portion of the brain and comprising about a quarter of all strokes caused by blood clots), will they do better in the future if you set a lower (less than 130) rather than a higher (130-149) target for controlling their blood pressure? To study this the research group enrolled about 1900 folks at 81 sites in North America, Latin America, and Spain and followed them for nearly 4 years. The local physicians used whatever medicines they wanted to try to reach the blood pressure goals; and the lower-pressure group did in fact achieve a lower blood pressure overall than the higher-target group (11 points difference).
If you listen to the authors of the paper, they'll tell you:
- While none of the results of the study were statistically significant, the so-called "trends" were in the direction of fewer strokes of all sorts, and fewer bad vascular events, in the lower-pressure-target group.
- The lower-pressure target group did not seem to have any more serious side effects from treatment.
- Other previous studies have tended to show benefits from lower blood pressure in similar populations
- Therefore, we should recommend setting the blood pressure targets lower for patients who have had a lacunar stroke.
- Since no results were statistically significant, that means the study failed to show any real benefit from the lower blood pressure target in these patients.
- Let's for a minute pretend that all the so-called "trends" were in fact statistically significant. We'd still have to ask how large the resulting differences were. It turns out that the differences are really tiny anyway--often in the fractions of a percentage point.
- If you are going to look at slight, nonsignificant trends, how about the slightly higher overall mortality rate in the lower blood pressure target arm of the study? Funny how the authors get so excited about "trends" that confirm their hypothesis, and yet ignore similar "trends" that go the wrong way.
If good scientists like this can publish results that are so little supported by their data, and a supposedly excellent journal lets them get away with it-- then what happens when to the natural enthusiasm of the research team, we add the incentives as well as the incredible monetary resources of the drug industry? You start to get a sense of how much misleading information can appear on the pages of the "best" medical journals, and how long it could take us to sort through it all and decide what's really best for patients.