Recent invasive CV trials fall short on certain design aspects
Contemporary randomized clinical trials assessing invasive CV interventions are often small with short follow-up and limited power to identify large treatment effects, according to a study published in JAMA Internal Medicine.
The study also found that commercially supported trials were linked to differences in results, design and reporting. In particular, commercially supported trials were more likely to report favorable outcomes than other trials, and to report findings inconsistent with the trial results.
“You must consider that randomized trials are considered the highest level of evidence not only in cardiology,” Mario F.L. Gaudino, MD, FEBCTS, Stephen and Suzanne Weiss Professor of Cardiothoracic Surgery at Weill Cornell Medicine, told Healio. “In the end, our guidelines and our practice are mostly based on the results of those trials. If those trials are not properly designed, conducted and reported, our practice may not be accurate. We may use the wrong treatment on patients, and ultimately we can also harm patients.”
Trials assessing intervention, surgery
In this cross-sectional study, researchers analyzed data from 216 randomized clinical trials focused on vascular, coronary and structural interventional cardiology and vascular and cardiac surgeries.
“There were not a lot of trials because we have 216 trials over the course of more than 10 years, so around 20 trials per year,” Gaudino said in an interview. “If you consider that we are pooling interventional cardiology, vascular medicine and cardiac surgery, it’s really not a lot of trials.”
Data recorded for each trial included trial sponsors, conflicts of interest for the first and last authors listed, single center or multicenter, sample size, statistical power, follow-up duration and primary analysis details. Trials were also analyzed by how results were classified, potential spin and distortion of the advantage of the procedure, discrepancy between the registered and published primary outcomes and fragility index calculations.
Of the trials in this study, 53.2% were commercially sponsored. Most of these trials were designed with 80% power to identify a treatment effect of 30%. In addition, 59.3% of trials utilized a primary outcome with composite endpoints. Trials had a median sample size of 502 patients and a median follow-up of 12 months.
More than half of these trials (57%) showed a statistically significant difference in the primary outcome that favored the experimental intervention. Spin was identified in 65.5% of the trials that reported nonsignificant differences in the primary outcome. Trials with commercial sponsorship were more likely to report favorable outcomes (exponent of regression coefficient beta = 2.8; 95% CI, 1.09-7.18) after adjusting for differences in trial characteristics. These trials were also more likely to have a greater extent of spin (exponent of regression coefficient beta = 4.64; 95% CI, 1.05-20.54).
Inconsistencies between the registered and published primary outcomes occurred in 38% of trials. There were no significant differences in these inconsistencies between commercially and noncommercially sponsored trials (35.7% vs. 40.6%, respectively; P = .13). To change statistically significant results to nonsignificant results, a median of five patients would need to experience a different outcome.
Commercially sponsored trials were linked to a greater number of patients at multivariable regression analysis (exponent of regression coefficient beta = 1.29; 95% CI, 1-1.66).
“This was not very reassuring,” Gaudino told Healio. “I hope that this is a wake-up call for the whole cardiovascular community. We really need to do a better job in designing randomized trials.”
For more information:
Mario F.L. Gaudino, MD, FEBCTS, can be reached at firstname.lastname@example.org.
Disclosures: Gaudino reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.