We retrospectively selected all patients older than 75 years undergoing TAVI from January 2010 to April 2019 at the Hospital Universitario Central of Asturias (Principality of Asturias Ethics Committee, number 159/19). This tertiary hospital is the only public or private center that offers TAVI in this region and is the referral hospital for all patients needing this technique.

We constructed the reference population using mortality table data for the Principality of Asturias from the Spanish National Institute of Statistics (INE) stratified by age and sex. These data are available on the official INE website.12 More information on the INE can be consulted in “Methods” of the supplementary data.

We retrospectively collected all data relating to the preoperative, intraoperative, and postoperative periods from a digital database prospectively entered by the treating physician. The postoperative period was defined as the first 30 days of follow-up or as the discharge day in the event that this occurred later.

One of the researchers collected mortality-related data during follow-up by analyzing the information available in the medical records of all health centers and hospitals in the Principality of Asturias.

Primary objectives: a) to compare the survival curves of patients older than 75 years who underwent TAVI with those of the general population of the same age, sex, and geographic region; and b) to compare the survival curves of the patients in the TAVI sample who survived the postoperative period.

Quantitative and categorical variables are expressed as mean±SD and No. (%), respectively.

To compare the survival curves of the TAVI sample with those of the general population of the same age and sex, we calculated observed survival, expected survival, and relative survival (RS): a) observed survival was defined as the real survival of the TAVI sample calculated using standard Kaplan-Meier analysis; b) expected survival was defined as survival of the reference population of exactly the same age, sex, and geographic region as the TAVI sample (ie, survival that the TAVI sample would have had in the absence of aortic valve disease). It was calculated using the Ederer II method, which has been the preferred method to calculate expected survival since 2011,13 using information on mortality rates provided by the INE for different age, sex, and geographic ranges.12 If expected survival was within the confidence interval of observed survival, we considered that there were no significant differences between the two; and c) RS was defined as the estimated survival that patients in the TAVI sample would have had under the theoretical and imaginary assumption that they could only die from a problem associated with their aortic valve disease.14,15 RS was obtained by calculating the ratio between the observed survival rate and the expected survival rate. An RS of 100% during the first year would imply that the aortic valve disease had been completely resolved using TAVI: however, an RS of 80% in the first year would imply that 20% (80%-100%) of patients would have died due to a problem associated with aortic valve disease.15 Thus, an RS confidence interval that included 100% of the patients would imply that there was no evidence of mortality caused by aortic valve disease and that TAVI had been completely effective in resolving the problem.14

One of the main advantages of using RS is that it identifies mortality exclusively due to the disease under study without requiring information on the cause of death.14

All analyses were conducted using the “strs”16 command of the STATA v.15.1 software package (STATA Corp, Texas, USA).

During the study period, 526 patients older than 75 years underwent TAVI. Mean age was 84.15±3.86 years, 300 (57.03%) were women, and 319 patients (60.65%) had a EuroScore II of more than 8. The preoperative characteristics of the patients are shown in table 1. Table 1 of the supplementary data shows the prevalence in the Spanish population of some common comorbidities in patients with aortic stenosis. Until 2017, all patients were implanted with the CoreValve prosthesis (Medtronic, Minnesota, USA). Subsequently, the Evolut PRO prosthesis (Medtronic, Minnesota, USA) was used.

In total, 26 patients (4.94%) in the TAVI group died within the first 30 procedural days. The main characteristics and perioperative and postoperative complications are shown in table 2.

No patients were lost to follow-up. The mean follow-up of the censored patients in the TAVI group was 33.06 months. Of the total of 185 deaths, 159 occurred after the postoperative period. The main causes of death are shown in table 3.

In the TAVI sample, cumulative survival at 1, 3, 5, and 8 years of follow-up were 86.1% (95%CI, 82.75-88.85), 68.92% (95%CI, 63, 90-73.39), 50.60% (95%CI, 44.19-56.66), and 33.96% (95%I, 26.35-41.71), respectively. In the reference population, cumulative survival was 91.96%, 75.63%, 59.65%, and 37.67%, respectively. Cumulative survival for each of the follow-up years is shown in table 4. Both survival curves are shown in figure 1.

Figure 1.

Survival curves of the TAVI sample compared with those of the general population. During the first year, the survival curve of the TAVI sample undergoes a marked decrease compared with that of the general population. Subsequently, they are very similar. 95% CI, 95% confidence interval; TAVI, transcatheter aortic valve implant.

(0.08MB).

During the first year of follow-up, RS showed that there was high mortality due to valvular heart disease: RS, 93.63% (95%CI, 89.98-96.61). In the remaining years, RS showed that there was no mortality due to aortic valve disease: in other words, expected survival was similar to observed survival. Figure 2 and table 4 show RS for each year of follow-up.

Figure 2.

95% confidence intervals of relative survival in the TAVI group during each year of follow-up. During the first year, relative survival is lower in the TAVI group. Subsequently, they are very similar. RS, relative survival; TAVI, transcatheter aortic valve implant.

(0.05MB).

In the group of patients who survived the first 30 days, the survival curves were similar during most of the follow-up period. The survival curves are shown in figure 3. The probability of survival at 1, 3, 5, and 8 years in this TAVI group were 90.58% (95%CI, 87.54-92.91), 72.51% (95%CI, 67.38-76.97), 53.23% (95%CI, 46.52-59.48), and 35.73% (95%CI, 27.72-43.80), respectively. In the reference population, the probability of survival was 91.93%, 75.63%, 59.6%, and 37.47%, respectively. Cumulative survival for the TAVI sample and the reference population is shown in table 5.

Figure 3.

Survival curves of patients in the TAVI sample surviving the first 30 days compared with those of the general population. The survival curve of the reference population is within the confidence interval of the survival curve of the TAVI group during most of the follow-up. This implies that the probability of survival is similar in both groups during most of the follow-up. 95%CI, 95% confidence interval; TAVI, transcatheter aortic valve implant.

(0.07MB).

During the first year of follow-up, RS showed low mortality due to valvular heart disease: RS, 98.53% (95%CI, 95.22-101.07). In the remaining years, RS showed no mortality due to aortic valve disease: in other words, expected mortality was the same as observed mortality. Table 5 shows the survival curves stratified by age groups (≤ 85 years and > 85 years) (see figure 1 and 2 of the supplementary data). Figure 3 of the supplementary data shows survival curves stratified by EuroSCORE ≥ 8 or < 8.

Data from the TAVI sample were collected retrospectively. However, ideally, the TAVI sample should be compared with a random population of thousands of residents in the Principality of Asturias followed up over several years; subsequently, the individual data should be analyzed controlling for risk factors, comorbidities, socioeconomic factors, or even municipality. However, this scenario is not very probable and likely explains the lack of studies on these characteristics. In this study, we attempted to address this difficulty by using a novel method to calculate expected survival using data from the INE. Although the possibility of selection bias is reduced by the fact that there is only 1 implant center using TAVI in the Principality of Asturias, we recognize that this aspect could make it difficult to extrapolate the results.

Finally, we are aware that the patients in the TAVI sample may have had a higher rate of comorbidities than that in the general population, which could have led to reduced survival. Nevertheless, patients who are candidates for TAVI, although probably frail and with comorbidities, go through a specific selection process in which patients with end-stage disease are excluded for reasons of futility. These circumstances make it very difficult, if not impossible, to compare the probability of survival in a sample of patients with that of the general population. In this regard, the calculation of RS can shed light on the situation.14 This measure, which is frequently calculated in the setting of cancer, shows that TAVI prevents any death associated with severe aortic stenosis after the end of the postoperative period.