November 22nd, 2010
SHARP: A Tale of Two Press Releases
Two press releases were issued about the SHARP trial : one from Merck, the much-maligned and criticized manufacturer of Vytorin (the combination of ezetimibe and simvastatin), and one from the group running the trial, the highly-respected Clinical Trials Service Unit (CTSU) at Oxford. Here’s the surprise: compared to the Oxford press release, the Merck release is a paragon, fully disclosing important details about the trial. The Oxford release is also a paragon, but of a different type. It’s a model of a press release that tries to manipulate the reader to adopt its view of the trial and ruthlessly suppresses all information and perspective that doesn’t support its view.
Let’s review the deficiencies of the Oxford press release:
The press release headline touts the “big benefits” shown in the trial, and the first sentence begins: “Around a quarter of all heart attacks, strokes, and operations to open blocked arteries could be avoided in people with chronic kidney disease by using the combination of ezetimibe and simvastatin. . . .” The problem here is simple: the trial showed no such thing. The actual reduction in events was 16% and not 25%. The larger reduction in events comes from an extrapolation by the study investigators based on their estimates of the reduction that would have been achieved if every patient who was supposed to take Vytorin took the drug for the entire trial period. But, of course, that never happens, either in clinical trials or in real life, so the “big benefit” touted in the headline is pure fiction and manipulation. Such a calculation might be appropriate in the discussion section of a paper, or at the bottom of a press release or other communication if properly explained and qualified. But as the lead of a press release conveying the initial results of an important clinical trial, it is inexcusable.
The Oxford press release touts the benefits of adding ezetimibe to simvastatin, but in fact, SHARP is completely unable to make any assessment of the relative effects of ezetimibe and simvastatin versus simvastatin alone. With the exception of a small group of patients who were first randomized to simvastatin monotherapy during the first year of treatment and were then randomized to either placebo or Vytorin, all patients in the trial received either the Vytorin pill or placebo. It is therefore completely inappropriate for them to state:
“The SHARP results are also relevant to people who don’t have chronic kidney disease. The combination of ezetimibe and a statin produced similar benefits to those resulting from the same LDL cholesterol reduction achieved with a high dose of a statin. Since the lower the cholesterol the bigger the risk reduction, these results suggest that patients who remain at high risk of major atherosclerotic events despite maximal statin therapy may benefit further from adding ezetimibe to their current statin regimen.”
SHARP has no relevance to patients without chronic kidney disease, and it tells us almost nothing about the value of adding ezetimibe to simvastatin.
It is astonishing that the Oxford press release has so much to say about things that weren’t tested in the trial and has nothing to say about a number of major issues that were addressed in the trial. In particular, the press release doesn’t report the primary endpoint of the trial. As has been discussed on CardioBrief and elsewhere, earlier this year, near the completion of the trial, the SHARP investigators sought to change the primary endpoint (from first major vascular event to first major atherosclerotic event) and the planned statistical analysis of the trial to avoid producing a false-negative result. The trial sponsor, Merck, did not endorse their plan, and officially, the primary endpoint remains unchanged. The SHARP investigators, however, chose to ignore the actual primary endpoint, or any discussion about the change in the primary endpoint, in the press release.
The Oxford press release is severely deficient in another way: it presents no hard numbers for the results, including the actual number or percentage of events of either the primary or alternate endpoint or the individual components of the composite endpoints. The press release also includes no discussion about the statistical power of the trial or any of the findings. In essence, the investigators ask the readers of the press release to take their conclusions on faith.
By contrast, the Merck press release is both more detailed and more restrained. It explains (in a slightly confusing manner) the issue of the attempted change in endpoint and presents the actual percentages and P values for both the original primary endpoint and the alternate Oxford endpoint. The Merck press release also includes the information, completely absent in the Oxford press release, that in this study of patients with kidney disease, treatment with Vytorin had absolutely no effect on the progression of kidney disease.
In the past, I’ve been as critical as anyone of Merck’s handling of the Vytorin controversy. The interesting thing here is that Merck appears to have played it straight, while the Oxford CTSU doesn’t seem to have learned any of the lessons from earlier episodes of this long-running imbroglio.