Renal Denervation Trials Have Largely Fallen Victim to Unintentional Bias
A new analysis of renal denervation trials confirms what many experts in the field have suspected for years: that trial designs more resistant to bias—namely, blinded and sham-controlled—will report substantially smaller blood pressure reductions. Thus, conducting unblinded studies has failed to provide useful information while squandering valuable resources for research, one researcher argues.
The findings have implications
for the design of future trials in the field, which had a major setback with
the failure of SYMPLICITY
HTN-3. 
“The horrible truth is that just adding a randomized control group, without blinding, is almost useless,” study author Darrel Francis, MD, of Imperial College London, told TCTMD in an email. “This is because most of the bias in these trials came from patients changing their medication adherence or doctors making measurements in a biased way. In randomized but unblinded trials, this carried on happening.”
Unblinded data should never be accepted at face value, he continued, because in the case of renal denervation they “were as wrong as they could possibly be. The attempt to save money by doing unblinded trials has led to a colossal waste of research resources.”
Preliminary studies of catheter-based renal denervation suggested that large reductions in blood pressure—as high as 59 mm Hg—were possible, sparking the development of numerous devices before completion of blinded, randomized trials. That enthusiasm was halted, however, when SYMPLICITY HTN-3 failed to show that renal denervation was superior to a sham procedure, forcing device manufacturers to reconsider their approaches.
Because of the large number of renal denervation studies, it is possible to explore links between study design and blood pressure results, leading Francis and colleagues to quantify the impact of 3 main biases on the results of renal denervation studies. The types of bias were:
Regression to the mean: This is caused by the tendency to select patients for studies of novel hypertension therapies when their blood pressures at the time of enrollment are higher than their individual long-term averages. A control group protects against this type of bias.
Asymmetrical data handling: This is the result of physicians remeasuring blood pressure after an intervention believed to be effective if the initial reading does not show a change. Either blinding physicians to treatment allocation or removing them from data collection protects against this source of bias.
Confounding or nondenervation effect: This occurs when true blood pressure drops are caused by something other than the procedure, such as an increase in drug adherence or placebo effect.
The analysis, published in the January 2016 issue of Circulation: Cardiovascular Quality and Outcomes, included 148 studies that enrolled at least 5 patients: 140 nonrandomized, 6 randomized but open-label, and 2 randomized and blinded.
The researchers found that trial design was associated with reported effect size (P < .001).
The drops in office systolic blood pressure were 23.6 and 29.1 mm Hg in nonrandomized and randomized but open-label studies, respectively. The reduction in ambulatory pressure was lower (11.2 mm Hg) in unblinded studies.
In the blinded trials, there was no significant reduction in systolic blood pressure following denervation.
When looking at the influence of bias, regression to the mean did not have an effect (P = .71) on the blood pressure reductions. However, asymmetrical data handling contributed to a 10.8-mm Hg drop in reported blood pressure and confounding or nondenervation effect contributed 8.3 mm Hg to the reported reductions (P < .0001 for both).
“These data indicate that despite the extra effort involved, instituting blinding may be essential if the effect size of the intervention is to be quantified reliably,” the authors write. “This principle likely holds for all device therapies for hypertension.”
Study author James Howard, MA, also of Imperial College London, told TCTMD in an email that “the real value of the renal denervation escapade is to teach our generation of cardiologists that honest doctors making objective measurements can still introduce enormous bias. This is the old lesson from the pharmaceutical world that we have to relearn in the device world, that earnest belief in one’s own honesty doesn’t prevent you from producing a wildly biased and unrepresentative result.”
Non-Drug Options Still Needed for Hypertension
Commenting on the study, Ajay J. Kirtane, MD, SM, of NewYork-Presbyterian Hospital/Columbia University Medical Center (New York, NY), said that “while some of the specific conclusions drawn about the quantification of individual biases may be limited (as the authors themselves state), the emphasis upon ambulatory blood pressure measurements and blinded study designs/endpoint assessments are very important learning points.”
The findings, he told TCTMD in an email, “lend credence to the designs of several of the important ongoing trials in the device-based treatment of hypertension.”
He pointed out that medications are not the best way to treat high blood pressure for all patients.
“While we in the medical profession are very apt to prescribe medications, if one were to ask the patients how they feel about this (and whether they would consider other options other than daily pills) they would certainly support other therapies,” Kirtane said. “That fact is very important to stress, when considering the future of device-based therapies to treat hypertension.”
Source:
Howard JP, Shun-Shin MJ, Hartley A, et al. Quantifying the 3 biases that lead to unintentional overestimation of the blood pressure-lowering effect of renal denervation. Circ Cardiovasc Qual Outcomes. 2016;Epub ahead of print.
Disclosures:
Francis reports serving as a consultant to Medtronic.
Kirtane reports receiving institutional research grants from Abbott Vascular, Abiomed, Boston Scientific, Eli Lilly, Medtronic, St. Jude Medical, and Vascular Dynamics.
Howard reports no relevant conflicts of interest.
Related Stories:
Todd Neale is the Associate News Editor for TCTMD and a Senior Medical Journalist. He got his start in journalism at …
Read Full Bio
Comments