ENHANCE Study: Frustration With Trial and Coverage of It

Despite all the press attention dedicated to the ENHANCE study results, which were finally presented at the ACC's annual meeting and printed in the NEJM yesterday, there's been little-to-no mention of the "missing" or "implausible" ultrasound data that were revealed originally in Schering-Plough's study timeline from January. The effect of these missing or implausible data on the validity of study has now been, in essence, inexplicably ignored. (See previous Pathophilia entries on the ENHANCE study here and here.)

The NEJM article devotes exactly one sentence to the subject:

The exclusion from the statistical analyses of patients with missing data or biologically implausible measures of the carotid-artery intima-media thickness (defined as a difference of >0.1 mm between visits) did not change the primary or secondary outcome results (data not shown).

Ok, the exclusion of these data, the volume of which is not provided, did not change the negative study outcomes. However, if the numbers of patients with missing or implausible data are considerable—and the S-P timeline suggests that they were*—they could very well undermine the validity and/or statistical power of the ENHANCE study and any subgroup analyses that might otherwise reveal significant treatment differences.

The authors do consider that the ultrasound measurement technique that was used to assess arterial intima-media thickness may not accurately reflect changes in atherosclerotic burden; but they dismiss this consideration on the basis of the precision of the measurements, "as exemplified by the high intraclass correlation coefficient and the small standard deviations." However, this explanation only addresses the use of arterial intima-media thickness as a surrogate marker for atherosclerotic burden; it is not relevant to the plausibility of the specific thickness measurements obtained by the ENHANCE study investigators and certainly doesn't address missing data.

In an accompanying NEJM editorial, Drazen et al write, "Only with the full data set and an independent analysis on the scientific record is it possible to properly interpret the trial and formulate follow-up studies to further pursue the intriguing but puzzling results." Amen. But neither of the 2 NEJM editorials that discuss the ENHANCE study mention the missing or implausible data and their potential effect on the trial's results.

Update: In its coverage of the ENHANCE study presentation, heartwire talked to Roger Blumenthal, MD, of Johns Hopkins, who said that 2 limitations "bias ENHANCE toward the null." One is that ENHANCE investigators did not consistently measure carotid intima-media thickness during cardiac contraction (ie, they did not use ECG gating), a method often employed to control the variability of images. The other limitation is that investigators "used single-frame technology instead of putting the images on a cine loop," which would have improved image quality. These comments imply that a sizeable percentage of the ENHANCE study images were, in fact, suboptimal and therefore undermined the validity of the study results.

*These are the data believed to have delayed the publication of the ENHANCE study results in the first place.

Photo: iStockPhoto