What are the right numbers for JUPITER?

Results of clinical trials can sometimes not only change our understanding of the condition studied, but may also affect the way we practice medicine. The Justification for the Use of Statins in Primary Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) is such a study.

JUPITER was investigating the prevention of major cardiovascular events (myocardial infarction, stroke, hospitalization for unstable angina, arterial revascularization, death from cardiocascular causes) by treatment with rosuvastatin compared to placebo. Many clinical trials before have looked at the reduction of cardiovascular events by statin treatment in people with elevated cholesterol levels. But in the JUPITER trial, 17,802 apparently healthy men and women with normal levels of LDL cholesterol (< 130 mg/dl) were included based on age (≥ 50 years for men, ≥ 60 years for women) and elevated levels of high-sensitivity C-reactive protein (≥ 2,0 mg/l), a marker for inflammation. The trial was stopped early in March of this year (after a median follow-up of 1.9 years instead of the planned 4 years) because treatment with rosuvastatin significantly reduced the number of cardiovascular events: 0.77 per 100 person-years compared to 1.36 per 100 person-years for placebo. In other words, treating 120 people for 1.9 years with rosuvastatin (at a cost of about $287.000) would prevent one cardiovascular event.

The JUPITER trial is important not only because it shows that a statin can have a beneficial effect in people with normal LDL cholesterol. Which complicates our understanding on how statins work. More importantly, we now have to reconsider who should be treated with statins. What do we do with a small but significant effect in a large group of apparently healthy people? The treatment is expensive (cardiovascular events can probably be prevented for less than $287.000 by other means, e.g. changes in diet and exercise) and can have side effects (e.g. an increase in newly diagnosed diabetes in the JUPITER trial). And, like most important studies, the JUPITER trial raises a number of important new questions. But instead of discussing some of these questions I rather want to look at how we can obtain information about the JUPITER trial.

Clinical Trial Databases
Since 2005 all clinical trials have to be registered in publicly available databases, and information about JUPITER is available.1 Since September 2008, newly registered trials also have to report their main outcomes.2 This will become important in the future, as the main outcomes of unpublished trials (most likely trials with negative results) will become publicly available no later than 12 months after data for the last subject were received.

Press Releases
JUPITER was stopped early in March 2008 and AstraZeneca (the sponsor of the trial and manufacturer of rosuvastatin) issued a press release.3 Another press release was issued on November 9 when the JUPITER results were first presented publicly.4 As can be expected from press releases, there is selective reporting of the trial results. No absolute numbers for risk reductions were reported, and emphasis was put on relative risk reductions. Instead of the actual number needed to treat (NNT, 120 patients treated for 1.9 years), a projected NNT (25 patients treated for 5 years) was reported. And a significant increase in newly diagnosed diabetes (3.0% vs. 2.4%, p=0.01) was reported as “there was a small increase in physician reported diabetes consistent with data from other large placebo controlled statin trials.”

Presentation at a Meeting
The JUPITER results were presented on November 9 at the annual meeting of the American Heart Association (AHA) in New Orleans. More than 6000 people were listening to this presentation according to James Butcher on the Nature in the Field blog5. The abstract of the presentation is available here.6 The abstract also lists the potential conflicts of interest of the study authors (the senior author Paul Ridger Ridker is co-inventor of patents on CRP testing in cardiovascular disease). The AHA issued a press release on that day7 and commented primarily on the role of CRP testing in the trial.

Journal Paper
The JUPITER study was published in the New England Journal of Medicine (NEJM) on November 208, but the paper was preleased on the day of the presentation at the AHA meeting. The fulltext PDF of the paper is available without subscription. The full paper is of course the best source to all the primary data. As is typical for many medical journals, it contains a structured abstract, which is a nice summary of the paper. Interestingly, the NEJM is conducting a poll and asks the readers two questions on how the JUPITER results changed their clinical practice.9 And there are over 400 reader comments as of today.

The full paper is discussed in an editorial in the same issue10, and is also discussed in an editorial in the British Medical Journal11. Only the editorial in the NEJM is available without subscription, but both critically discuss the paper and put it in perspective.

Traditional News
The JUPITER trial was of course discussed in many traditional news media such as newspapers, radio and television. The New York Times had a long article12 where the author had interviewed not only the study authors but also several experts in the field. The article discusses several issues surrounding the study, but failed to report the absolute risk reduction or the number needed to treat (important numbers for the reasons discussed above). National Public Radio discussed the story with two cardiologists, including the author of the editorial in the NEJM (available as transcript and audio file).13 The interview is again short on actual numbers, but puts JUPITER in perspective for the typical radio listener. The Nature News article14 also didn't mention the number needed to treat. Stopping a trial early can be controversial, because the numbers for risks and benefits might look different at the planned end of the trial. Nature News should have talked to someone that was not involved in that decision in the JUPITER trial.

Scintilla has aggregated the blog posts on this study15. Harriet Hall on the Science-Based Medicine Blog not only has the story with the best title (Statins Are Better on JUPITER), but gives a detailed analysis of the study results. Ben Goldacre on Bad Science has a shorter blog post that is more of a discussion of absolute risk, relative risk and number needed to treat (as the title of the blog post suggests).16

There are many ways we can learn more about the JUPITER trial, and most of this information is freely available, including the fulltext of the paper. It is not the access to the primary data that is the problem, but rather the many different ways the results can be put into perspective. And for that we need not only a basic understanding of cardiovascular risks, but also clinical trials. And we should not forget the financial and other interests that are always connected to large trial like this. I didn't do a systematic analysis of the newspaper articles and blog posts about the study, but it is clear to me that science blogs can add an important perspective.

fn1. JUPITER – Crestor 20mg Versus Placebo in Prevention of Cardiovascular (CV) Events. ClinicalTrials.gov NCT00239681 October 13, 2005

fn2. FDAAA: Push to open data in clinical medicine

fn3. Crestor Outcomes Study JUPITER Closes Early Due To Unequivocal Evidence Of Benefit. AstraZeneca Press Release March 31, 2008

fn4. CRESTOR Demonstrates Dramatic CV Risk Reduction in a Large Statin Outcomes Study. AstraZeneca Press Release November 9, 2008

fn5. AHA 2008: Should statins be put in the water? James Butcher. In the Field November 11, 2008

fn6. Late-breaking abstracts. News Conference November 9, 2008

fn7. American Heart Association Comment on JUPITER trial. AHA Press Release November 9, 2008

fn8. P. M Ridker, E. Danielson, F. A.H. Fonseca, J. Genest, A. M. Gotto, J. J.P. Kastelein, W. Koenig, P. Libby, A. J. Lorenzatti, J. G. MacFadyen, B. G. Nordestgaard, J. Shepherd, J. T. Willerson, R. J. Glynn (2008). Rosuvastatin to Prevent Vascular Events in Men and Women with Elevated C-Reactive Protein New England Journal of Medicine, 359 (21), 2195-2207 DOI: 10.1056/NEJMoa0807646

fn9. The JUPITER Trial: Will You Change Your Practice? Clinical Directions

fn10. M. A. Hlatky (2008). Expanding the Orbit of Primary Prevention — Moving beyond JUPITER New England Journal of Medicine, 359 (21), 2280-2282 DOI: 10.1056/NEJMe0808320

fn11. N. Donner-Banzhoff, A. Sonnichsen (2008). Statins and primary prevention of cardiovascular events BMJ, 337 (nov14 2) DOI: 10.1136/bmj.a2576

fn12. Cholesterol-Fighting Drugs Show Wider Benefit. Pam Belluck. New York Times November 10, 2008

fn13. Study Finds Statins Benefit Patients With no History of Heart Problems. PBS Newshour November 10, 2008

fn14. Should healthy people take statins too? Katharine Sanderson. Nature News November 10, 2008

fn15. Search for term rosuvastatin. Scintilla

fn16. Statins Are Better on JUPITER. Harriet Hall. Science-Based Medicine November 11, 2008

fn17. You are 80% less likely to die from a meteor landing on your head if you wear a bicycle helmet all day. Ben Goldacre. Bad Science November 15, 2008

This entry was posted in Snippets. Bookmark the permalink.

8 Responses to What are the right numbers for JUPITER?

  1. Maxine Clarke says:

    Very interesting post, Martin, it must have taken a while to put this together. As you say, it is the filtering and the intepreting that are more time-consuming issues than the direct availability of the data themselves – because there are so many possible perspectives on the story. And I wonder how many of the articles and blog posters did read the original study in one of the forms you mention before writing about it?
    BTW, I have featured James Butcher’s post in my From The Blogosphere column next week, but owing to the constraints of the print medium, my piece won’t come out for several days 😉 Unfortunately I can’t add a link to your valuable blog post into the print version as it has already passed for press, but I’ll do so in the Nature Network archive.

  2. Martin Fenner says:

    Maxine, I didn’t find the presentation slides from the AHA meeting online, that would also be very interesting material.
    This study indeed has many perspectives that could be interesting to blog about, e.g.:
    # Why is rosuvastatin effective? Is it a direct effect on LDL cholesterol (bringing it to very low levels) or is it an effect on inflammation in the blood vessels?
    # Who should receive blood tests for high-sensitive CRP?
    # Are we willing to pay for treating 120 people for 1.9 years (or a smaller number of people for a longer period of time) to prevent one cardiovascular event?
    # Would we expect to see the same results with a generic (i.e. much cheaper) statin?
    # Did conflicts of interests influence the trial results?
    # Was it right to stop the trial prematurely?
    I believe that one very good role for science blogging is the interpretation of trial results like this one. Or other science stories that are complex but have important implications.
    I’m still learning how to use the new *Nature Blogs* portal, but it looks like all blog posts discussing the JUPITER trial are summarized “here”:http://blogs.nature.com/stories/279.

  3. Maxine Clarke says:

    Good collection, Martin – this is what Euan Adie is aiming at with the blogs portal – I am not sure if one has to “claim” one’s blog before it is included in the aggregation – we’ll have to ask Euan. Do you think the role of blogs is important because of their spontenaity and speed? A significant result would get covered in the newspapers and/or science press, but not as quickly as on blogs. But with more reflection?

  4. Martin Fenner says:

    Blogs are speedy. They can also cover a certain perspective of a science story, e.g. only discuss the role of inflammation and CRP in the JUPITER trial or only discuss the premature termination of the trial. And they can be persistent, e.g. pick up a story 6 months later when there is a new angle to it. But most importantly, science bloggers are sometimes experts of a particular topic and know more about the science in that story than the average science journalist. This also means that they will catch an interesting story (e.g. a published paper) that was missed by the traditional media.
    Now I probably sound like Ben Goldacre at the Science Blogging conference in London.

  5. Maxine Clarke says:

    James has mentioned to me another of his posts from the AHA meeting which addressed the JUPITER trial, “see here”:http://blogs.nature.com/news/blog/2008/11/what_can_labbased_scientists_l.html. The post title is ‘What can lab-based scientists learn from clinical researchers?’ 😉

  6. Maxine Clarke says:

    “This is probably a more relevant blog post”:http://blogs.nature.com/news/blog/2008/11/should_statins_be_put_in_the_w.html from James on JUPITER specifically, but the post I refer to in the comment immediately above is definitely worth reading.

  7. Martin Fenner says:

    I enjoyed reading the post about _What can lab-based scientists learn from clinical researchers?_. And of course I agree 100%, as I spend most of my time doing clinical research. And I have a wife who is a PhD basic science researcher.
    One important concept of clinical research (even suggested for this year’s Nobel Price by Thomson Scientific) is *meta-analysis*, a standardized pooled analysis of all research addressing a specific question. I haven’t yet seen this kind of systematic analysis for basic research, e.g. in review papers.