Patient Risk, Follow-up Duration Drive SAVR, TAVI Outcomes: Meta-analysis

Major outcomes appear similar over the longer term, putting the focus on other factors for clinical decision-making.

Patient Risk, Follow-up Duration Drive SAVR, TAVI Outcomes: Meta-analysis

For patients with symptomatic severe aortic stenosis, TAVI has an early advantage over SAVR in terms of death and death/disabling stroke in lower-risk cohorts, but over the longer term, and in higher-risk groups, patients do similarly well with either procedure, an updated meta-analysis of randomized trials shows.

More divergence between TAVI and SAVR was seen for secondary outcomes, with higher risks of new-onset atrial fibrillation, major bleeding, and acute kidney injury with surgery and greater risks of new permanent pacemaker implants, major vascular complications, and paravalvular leak with transcatheter treatment.

Those findings emerged from a meta-analysis published online Friday in the European Heart Journal, led by Yousif Ahmad, BMedSci, BMBS, PhD (Yale School of Medicine, New Haven, CT). Ahmad presented an earlier version of the analysis last year at TVT 2022.

Having this updated meta-analysis incorporating key randomized trials is helpful when discussing treatment decisions with patients, he told TCTMD, adding that a unique feature of this new effort is the consideration of lower-risk and higher-risk trials separately.

“I think that we can at the moment pretty fairly explain to patients that on the basis of all of the randomized trial data that we have and at the longest follow-up available that we have, there’s no evidence of a significant survival difference between the two therapies,” he said. The decision of TAVI versus SAVR, he added, then comes down to factors like complications or bleeding, which affect early recovery, or paravalvular leak and pacemaker implantation, which might take on greater relevance over the long term.

“Different patients are going to have different philosophies and different values, and I think all of this can feed into that conversation with patients in clinic,” Ahmad said, with the caveat that there are not a lot of long-term data in lower-risk cohorts.

Totality of the Evidence

The meta-analysis included a total of 8,698 patients from eight RCTs comparing TAVI and SAVR, which had a weighted mean follow-up duration of 46.5 months. The investigators divided those trials into two groups:

In the lower-risk trials, TAVI was associated with lower risks of death and death/disabling stroke through 1 year, with no difference in stroke alone; beyond 1 year, there were no significant differences in those endpoints between TAVI and SAVR. In the higher-risk trials, no significant differences in those main outcomes were observed between the groups either early on or through longer-term follow-up.

Major Outcomes: TAVI vs SAVR


1 Year

Beyond 1 Year

Lower-Risk Trials


    Death/Disabling Stroke



RR 0.67 (95% CI 0.47-0.96)

RR 0.68 (95% CI 0.50-0.92)

RR 0.91 (95% CI 0.46-1.80)


HR 0.90 (95% CI 0.69-1.17)

HR 0.85 (95% CI 0.61-1.15)

HR 0.93 (95% CI 0.66-1.31)

Higher-Risk Trials


    Death/Disabling Stroke



RR 0.93 (95% CI 0.81-1.08)

RR 0.90 (95% CI 0.79-1.02)

RR 0.93 (95% CI 0.68-1.27)


HR 1.04 (95% CI 0.96-1.13)

HR 1.04 (95% CI 0.96-1.13)

HR 0.94 (95% CI 0.75-1.18)

An additional reconstructed individual patient data meta-analysis revealed some nuances to the findings regarding these main outcomes. For all-cause mortality, there were no significant differences between TAVI and SAVR over the entire follow-up period, regardless of patients’ risk. In the higher-risk trials, however, TAVI was associated with a lower mortality risk in the first 6 months (HR 0.68; 95% CI 0.56-0.82), but a higher risk later on (HR 1.17; 95% CI 1.05-1.29). A similar pattern was seen for death/disabling stroke.

As for secondary outcomes, assessed up to 1 year, Ahmad et al say that “the profile of events that occurred more frequently after TAVI tends to be outcomes that may assume greater relevance during long-term follow-up (paravalvular leak, reintervention, and new pacemakers). Conversely, the events that occurred more frequently after SAVR tend to be outcomes that may be of greater short-term relevance (new-onset atrial fibrillation, acute kidney injury, major bleeding).”

Thus, Ahmad said, the analysis starts to provide a glimpse of why the relative outcomes of TAVI versus SAVR vary depending on the time horizon.

Remaining Questions and the Need for Longer-term Data

Commenting for TCTMD, Patrick Myers, MD (Lausanne University Hospital, Switzerland), secretary general of the European Association for Cardio-Thoracic Surgery (EACTS), said this is a timely meta-analysis, but he had some critiques.

The conclusion of the paper mentions “therapy decisions” when that process was not explicitly addressed in the analysis, and the methods diverged in some respects from the plan preregistered in the PROSPERO database, he said, citing issues related to which data were included, which outcomes were studied, and how the trials were divided into risk groups. “I’m not saying that there’s anything nefarious behind that, but it just makes you wonder why there were so many changes,” Myers said.

On the basis of all of the randomized trial data that we have and at the longest follow-up available that we have, there’s no evidence of a significant survival difference between the two therapies. Yousif Ahmad

Moreover, he said, the current meta-analysis did not take into account biases that have been found to favor TAVI in randomized comparisons to surgery, as highlighted in a recent paper for which Myers was a co-author.

Despite these issues, much of the information is consistent with what’s been shown in prior meta-analyses, such as one published last year in the European Journal of Cardio-Thoracic Surgery, according to Myers.

“For patients who have a very short estimated survival, TAVI is an excellent option, and even for those who have a short-to-intermediate [expected] survival, but when we get into longer survival, SAVR does have better outcomes,” he said, pointing out that following the survival curves out to 5 years reveals better outcomes for SAVR.

Still, the biggest remaining question when it comes to comparing TAVI and SAVR is what the data will look like with even longer follow-up, which will provide information on the impact of things like paravalvular leak and pacemaker implantation, Myers said. “I’m a TAVI implanter, I believe in it completely,” he said. “But I think it has to be in the right patient. And for the moment, for patients who have a very long expected survival, we don’t have the data.”

Impact on Patient Choice

Ahmad pushed back on some of Myers’ critiques, stating that the way the meta-analysis was conducted was consistent with the stated analysis plan in PROSPERO and also—referring to the biases in the RCTs highlighted by Myers— “to my knowledge, there is no way to statistically account for issues pertaining to concomitant procedures, loss to follow-up, or deviation from allocated treatment within the confines of a study-level meta-analysis.”

He agreed, however, that the top remaining need when comparing TAVI and SAVR is longer-term data. “Based on our analysis and based on the longer-term data that we have from the intermediate-risk trials, I don’t think there’s yet any suggestion that there’s going to be a hazard of death long term with transcatheter valves, but we need to follow them up,” he said. Other key questions revolve around the treatment of patients with bicuspid anatomy, the prognostic impact of pacemaker implantation, the right way to handle concomitant coronary disease, and the durability of transcatheter versus surgical valves.

Asked how patients weigh the risks of different outcomes when choosing between TAVI and SAVR, Ahmad said they care most about stroke because of its impact on independence and quality of life, although survival is important to them as well. Some patients value a quicker recovery more than other factors, and for them, TAVI might be the better choice, he indicated. “For patients that value that above everything else, then we have a very good treatment that doesn’t seem to be sacrificing anything in terms of main outcomes like long-term survival and stroke.”

Myers said the decision is strongly dependent on the patients and who they’ve seen before. “Depending on how everything has been laid out, you will get a very, very different response, and I think the research that we have on getting informed consent is really, really lacking,” he said.

Myers said he’s not convinced that patients today are making fully informed decisions, with a firm grasp on all the data. “We know that patients discount a long-term risk much more than one that’s immediate. That’s the way it is, and that’s not something that we can work against,” he explained. “But it’s just something that we need to integrate into how we integrate shared decision-making into the algorithm in choosing what option. The patient is the boss, so ultimately they’re the ones who decide, but I think the way we get to that decision for the moment is still a little bit murky for me.”

Todd Neale is the Associate News Editor for TCTMD and a Senior Medical Journalist. He got his start in journalism at …

Read Full Bio
  • Ahmad reports no relevant conflicts of interest.