Real-world outcomes for transcatheter aortic valve replacement (TAVR) patients are likely similar to those observed in randomized controlled trials (RCTs), despite some key differences in important baseline characteristics between those treated in trial and clinic, researchers comparing U.S. CoreValve trial results with Medicare data said.
The study is the first to estimate the real-world treatment effects of TAVR trials, they added, noting that their methods could provide the basis for future research into the applicability of cardiovascular RCTs in real-world practice.
The findings were published online Monday, ahead of the Oct. 11 issue of JACC: Cardiovascular Interventions, by Neel M. Butala, MD, MBA, of Beth Israel Deaconess Medical Center and Massachusetts General Hospital, Boston, together with colleagues.
RCTs remain the “gold standard” to understand the effectiveness of an intervention.
However, because these results are based on “highly selected” populations that give the intervention “the best chance of demonstrating an effect,” whether they truly apply to patients in real-world clinical practice – where there are many more variables – often remains unknown, the researchers said.
There have been concerns that large RCTs that have rocketed uptake of TAVR over the last decade “may not have adequately represented important patient subgroups, thus raising questions as to their generalizability,” the researchers added.
The current study, therefore, set out to find differences between the real-world U.S. TAVR population of Medicare beneficiaries who underwent TAVR in U.S. clinical practice between Nov. 2, 2011, and Dec. 31, 2017 and patients in U.S. CoreValve pivotal trials. Inverse probability weighting was applied in a bid to extrapolate what trial results might have looked like had they been performed in the real-world population.
The analysis included all patients aged 65 years and older and included in the U.S. CoreValve Extreme Risk, U.S. CoreValve High Risk, and SURTAVI (Surgical Replacement and Transcatheter Aortic Valve Implantation) pivotal trials who could be successfully linked to Centers for Medicare and Medicaid Services (CMS) Medicare Provider and Review (Med-PAR) database (as part of the EXTEND study) – a total of 2,026 patients in the CoreValve pivotal trials and 135,112 in the Medicare cohort, who underwent TAVR.
The trial patients were mostly similar to the real-world patients at baseline, the researchers noted; however, those in RCTs were more likely to have hypertension (50% vs. 39%), coagulopathy (25% vs. 17%), fluid and electrolyte disorders (32% vs 20%), and weight loss (7.6% vs 4.2).
Real-world patients were more likely to have congestive heart failure (75% vs. 68%), renal failure (37.0% vs 32.3%), metastatic cancer (0.6% vs 0.0%), solid tumor without metastasis (2.3% vs 0.8%), obesity (17.0% vs 13.3%), alcohol abuse (0.8% vs 0.1%) and frailty (mean frailty index of -0.1 ± 1 vs. 0 ± 1).
To assess whether imbalances in the distribution of characteristics between the trial and real-world cohort would affect anticipated cumulative incidence of outcomes in the real-world population, the researchers reweighted the trial cohort to represent the characteristics of the real-world population.
Despite “some notable differences” in the populations, the cumulative incidence of death or stroke at 1 year among patients undergoing TAVR in the trial cohort (20.9%; 95% confidence interval [CI]: 18.2% to 23.6%) was similar in the real-world cohort after reweighting (19.6%; 95% CI: 16.6% to 22.7%).
The cumulative incidence of death at 1 year among patients undergoing TAVR in the trial cohort (18.2%; 95% CI: 15.6% to 20.7%) was also found to be similar to death rates after reweighting the trial cohort on the basis of characteristics of the real-world cohort (17.0%; 95% CI: 14.1% to 19.8%).
The estimated real-world treatment effect of TAVR was an 11.37% absolute reduction in death or stroke (95% CI: 7.50% to 14.92%), whereas the trial treatment effect was an 8.38% absolute reduction (95% CI: 4.62% to 11.91%; difference in risk differences 2.99% [95% CI: 0.82% to 5.53%]).
The estimated real-world treatment effect for death at 1 year following TAVR (95% CI: 5.20% to 12.32%) compared with conventional therapy of surgical aortic valve replacement in intermediate- and high-risk patients, or medical therapy for extreme-risk patients, was an 8.74% absolute reduction (95% CI: 5.20% to 12.32%), compared to a 6.95% absolute reduction for the trial population (95% CI: 3.73% to 10.31%; difference in risk differences 1.79% [95% CI: -0.43% to 4.12%]).
“Reweighted trial estimates of death or stroke, death, stroke, and aortic reintervention among patients undergoing TAVR on the basis of characteristics of the real-world cohort were similar to the observed rates of these outcomes in the real-world cohort,” the researchers noted.
“Although rates of pacemaker placement were different (reweighted trial 19.9% [95% CI: 16.8%-23.0%] vs observed real-world 14.1% [95% CI: 13.9%-14.2%]), presence of a pre-existing pacemaker, which is an important determinant of this particular outcome, was not included in the propensity score model, because of inconsistent capture in claims data.
“Overall, this suggests that there were minimal unobserved differences in baseline characteristics between the trial and real-world cohorts affecting the majority of trial outcomes that were not accounted for by the covariates included in the propensity score model.”
The researchers concluded that the methods from the current study could be used to estimate trial effects in real-world populations and “evaluate how novel therapies are introduced into clinical practice more broadly”.
Validation of ‘novel approach’
In an accompanying editorial, Bahira Shahim, MD, PhD, from New York’s Cardiovascular Research Foundation and Karolinska University Hospital, Stockholm, and David J. Cohen, MD, MSc, also of the Cardiovascular Research Foundation, and St. Francis Hospital, Roslyn, New York, lamented that the study “does not provide much help” in understanding TAVR vs. medical therapy or surgical aortic valve replacement.
“These limitations derive from the ‘long and winding road’ required to combine trials across different patient cohorts and with different comparator strategies when the key patient-stratification variable (surgical risk) is not directly observable in the real-world population,” the editorialists added.
“As such, the greatest value of this study is in providing proof of concept for this novel approach in the field of interventional cardiology. Ultimately, there may be far greater value in applying this approach to other cardiovascular devices, where the differences between the trial population and those treated in general practice may be greater and the comparator strategies are better defined.”
Butala NM, Secemsky E, Kazi DS, et al. Applicability of Transcatheter Aortic Valve Replacement Trials to Real-World Clinical Practice: Findings From EXTEND-CoreValve. JACC Cardiovasc Interv 2021;14:2112-2123.
Shahim B, Cohen DJ. Transporting Results of TAVR Trials to the Real World: A Long and Winding Road. JACC Cardiovasc Interv 2021; 14: 2124-2126.
Image Credit: Vadim – stock.adobe.com