Each year, the Spanish Registry of Cardiac Catheterization and Interventional Cardiology, the Spanish Registry of Implantable Cardioverter-Defibrillators, and the Spanish Registry of Pacemakers, maintained by various associations within the Spanish Society of Cardiology (SEC),4–6 are published in the Revista Española de Cardiología.

The yearly activity related to the following techniques was analyzed for the 2011 to 2019 period: coronary angiography (CA), percutaneous coronary intervention (PCI), primary PCI (pPCI), implantable cardioverter-defibrillator (ICD), cardiac resynchronization therapy (CRT), and transcatheter aortic valve replacement (TAVR). The list of the annual registries reviewed is included in the supplementary data.

For TAVR, only the 2017, 2018, and 2019 registries were reviewed, as data from the previous years had not been broken down by AC. For the purposes of data analysis and presentation, all techniques except TAVR have been grouped into 3-year subperiods (2011-2013, 2014-2016, and 2017-2019).

Data from 2020, the year of the COVID-19 pandemic, were excluded from the analysis because of the huge change occurring in the activity related to these techniques, which was uneven between the ACs.

To estimate the strength of the clinical evidence supporting indications for the techniques included, we consulted the European Society of Cardiology clinical practice guidelines, reviewing the classes of recommendation and levels of evidence in force at the study start and the modifications occurring over the study period for each technique.7–14

The population size, per capita gross domestic product (GDP), and per capita health expenditure (HE) values for each AC and each year were obtained from official sources.15,16

Serial incidence data for the various cardiovascular diseases in each community are lacking. Hence, we used a source that can provide an estimated incidence of infarction and heart failure based on hospital admissions: the Resources and Quality in Cardiology (RECALCAR) reports of the SEC.17 The hospital admission rate (HAR) refers to the number of admissions per 100,000 population per year for cardiology in general and specifically, for infarction and heart failure. Through this variable, the activity of each AC for each technique can be contextualized by adjusting values by admissions, which would account for the disease incidence and effects of population mobility.

Data from the RECALCAR reports were used to relate (although in a relative manner) the use of the techniques studied to cardiovascular mortality. These reports describe the risk-standardized mortality ratio (RSMR), with multilevel adjustment for overall cardiovascular disease, myocardial infarction, and heart failure. The RSMR is the ratio of the predicted mortality (which considers the operation of each hospital attending a patient individually) to the expected mortality (which considers the standard operation according to the mean of all hospitals), multiplied by the crude mortality rate.

With regard to aortic stenosis, there is no available information on incidence by year and AC. Therefore, we used the national Registry of Interventions of the Spanish Society of Cardiovascular and Endovascular Surgery, which has provided the number of surgical aortic valve replacements (SAVR) by AC in some of its recent yearly editions. This information is available for 2017 and 2018,18,19 but not for 2019.

The information sources, variables analyzed, and types of analyses carried out are summarized in figure 1.

Figure 1.

Study design, and information sources and analysis. ACs, autonomous communities; CA, coronary angiography; CRT, cardiac resynchronization therapy; GDP, gross domestic product; ICD, implantable cardioverter-defibrillator; PCI, percutaneous coronary intervention; pPCI, primary PCI; RECALCAR: Resources and Quality in Cardiology reports; RSMR, risk-standardized mortality rate; SEC, Spanish Society of Cardiology; TAVR, transcatheter aortic valve replacement.

(0.22MB).

Data set distribution was analyzed with the Shapiro-Wilk test. The coefficient of variation, determined as an indicator of variability between the ACs in technique use, is presented as the ratio of the standard deviation to the mean, multiplied by 100. Coefficients of variation were compared using the Forkman test. Associations between the volume of activity of ACs for each technique and the HAR, GDP, HE, and RSMR were assessed using the correlation coefficient and the partial correlation (adjusted by variables with a potential impact). Correlations between technique use and economic variables were adjusted taking into account hospital admissions related to each technique analyzed. A similar partial correlation was made by determining the correlation between technique use and RSMR rates. For TAVR analysis, correlations between the activity and economic variables were adjusted by the total number of aortic valve procedures and the number of surgical procedures alone. A P value of <.05 was considered statistically significant. All statistical analyses were carried out using the MEDCALC V19 and SPSS V25 software packages.

The mean number of CA procedures performed per million inhabitants/triennium in each AC during the 2011 to 2019 period, per capita GDP, per capita HE, and the overall cardiology HAR and RSMR are shown in figure 2.

Figure 2.

Mean CA values per million population/triennium by autonomous community (2011-2019), per capita GDP, per capita HE, and overall cardiology HAR and RSMR. CA, coronary angiography; HAR, hospital admission rate; HE, health expenditure; GDP, gross domestic product; RSMR, risk-standardized mortality rate.

(0.31MB).

The coefficients of variation and partial correlations obtained for the total study period and each triennium are shown in table 1. There were no changes in coefficients of variation, and the overall value was 9%. Correlations with the HAR were significant in the total period and in the last triennium. HAR-adjusted partial correlations of activity with economic indices and RSMR were nonsignificant.

The mean number of PCIs performed per million inhabitants/triennium in each AC during the 2011 to 2019 period, per capita GDP, per capita HE, and the overall cardiology HAR and RSMR are shown in figure 3.

Figure 3.

Mean PCI values per million population/triennium by autonomous community (2011-2019), per capita GDP, per capita HE, and overall cardiology HAR and RSMR.

HAR, hospital admission rate; HE, health expenditure; GDP, gross domestic product; PCI, percutaneous coronary intervention; RSMR, risk-standardized mortality rate.

(0.3MB).

The coefficients of variation obtained were all comparable, and that of the total period was 11%. Correlations with the HAR were significant in all periods, as were correlations with the CA results. Partial correlations of activity with the economic indices and RSMR were nonsignificant (table 1).

The mean number of pPCIs performed per million inhabitants/triennium in each AC during the 2011-2019 period, per capita GDP, per capita HE, and the HAR and RSMR, both for infarction, are depicted in figure 4.

Figure 4.

Mean pPCI values per million population/triennium by autonomous community (2011-2019), per capita GDP, per capita HE, and HAR and RSMR, both for infarction.

GDP, gross domestic product; HAR, hospital admission rate; HE, health expenditure; pPCI: primary percutaneous coronary intervention; RSMR, risk-standardized mortality rate.

(0.29MB).

The coefficients of variation showed a significant decrease from 36% to 17% (P=.007). Correlations with the HAR for infarction were nonsignificant. Partial correlations between volume of activity and economic indicators were moderate and individually significant in 1 triennium for the GDP and in the total period for the HE. The partial correlation with the RSMR for infarction was significant and negative only in the first triennium (table 1).

The mean number of ICD procedures performed per million population/triennium in each AC during the 2011 to 2019 period, per capita GDP, per capita HE, HAR for infarction and heart failure, and RSMR for heart failure are depicted in figure 5.

Figure 5.

Mean ICD values per million population/triennium by autonomous community (2011-2019), per capita GDP, per capita HE, HAR for infarction and heart failure, and RSMR for heart failure. AMI, acute myocardial infarction; GDP, gross domestic product; HAR, hospital admission rate; HE, health expenditure; HF, heart failure; ICD, implantable cardioverter-defibrillator; RSMR, risk-standardized mortality rate.

(0.3MB).

The coefficients of variation showed no significant changes, and yielded a value of 22% for the total study period. Correlations were nonsignificant with the HAR for infarction, but were significant with the HAR for heart failure. Partial correlations between volume of activity and economic indicators were nonsignificant (table 1). A correlation analysis between RSMR values for infarction and acute in-hospital mortality was considered inappropriate for ICD procedures.

The mean number of CRT procedures performed per million population/triennium in each AC during the 2011 to 2019 period, per capita GDP, per capita HE, and the HAR and RSMR, both for heart failure, are shown in figure 6.

Figure 6.

Mean CRT values per million population/triennium by autonomous community (2011-2019), per capita GDP, per capita HE, and HAR and RSMR, both for heart failure. AC, autonomous community; HAR, hospital admission rate; HE, health expenditure; HF, heart failure; GDP, gross domestic product; RSMR, risk-standardized mortality rate; CRT, cardiac resynchronization therapy.

(0.32MB).

Changes in the coefficients of variation were nonsignificant, with a decrease from 43% to 37%. Correlations with the HAR for heart failure were nonsignificant. All partial correlations between volume of activity and economic indicators, as well as the partial correlation with the RSMR for heart failure, were nonsignificant (table 1).

The mean number of TAVRs performed per million population/y in each AC during the 2017 to 2018 period, SAVRs per million population/y, per capita GDP, and per capita HE are shown in figure 7.

Figure 7.

Mean TAVR values per million population/y by autonomous community (2017-2018), SAVR per million inhabitants/y, per capita GDP, and per capita HE. GDP, gross domestic product; HE, health expenditure; SAVR, surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.

(0.25MB).

The coefficient of variation for TAVR activity was 42%. TAVR showed a modest correlation with SAVR activity, which was surprisingly positive, as it was expected to be negative. Correlations between TAVR activity and economic variables, presented as crude values and following adjustment for the total number of aortic valve procedures or surgical procedures alone, were nonsignificant (table 1).

An analysis was performed to examine possible relationships between TAVR activity and ICD/CRT activity in the various ACs during the most recent period (figure 8). A coefficient of 0.57 (P=.02) was obtained for the TAVR/CRT correlation, and 0.61 (P=.01) for the TAVR/ICD correlation. However, after adjusting for the overall HAR of the period, these correlations lost significance: 0.36 (P=.2) and 0.28 (P=.3), respectively.

A correlation analysis was performed between the practice of each technique in the total study period and the total population of each autonomous community, which yielded nonsignificant results in all cases: CA, r=0.18, P=.4; overall PCI, r=−0.27, P=.28; pPCI, r=−0.35, P=.2; ICD, r=−0.2, P=.3; CRT, r=−0.1, P=.7; and TAVR, r=−0.04, P=.8.

The activity and quality registries used as information sources for the study are based on data provided by the hospitals. Data collection was retrospective, through a standardized form. Submission was voluntary and was not audited; hence, quality assurance was dependent on the reliability of the professionals providing data. It would have been optimal to have strict monitoring and auditing, but in reality, these controls are applied very little. All these registries record volume of activity, not patient information; hence, it was not possible to adjust for age and sex or other variables.

The techniques analyzed were available in all Spanish ACs during the study period, with a few exceptions. These infrequent cases were attributed to the population of the region accepting the transfer. In addition, because of vacation travelling and stays in temporary residences, some patients underwent interventions in ACs other than their own (particularly pPCI, which is not an elective procedure). Although this implies smaller volume with respect to the total activity, application of the HAR enabled a more realistic adjustment regarding use of the various techniques.

The burden of disease could not be precisely estimated, as incidence and prevalence data on cardiovascular diseases by AC and specific periods are not available. The closest approximation is the HAR data provided in the RECALCAR reports, which are quite reliable. Irrespective of this consideration, the Spanish population is not as diverse epidemiologically as larger countries having a highly mixed ethnic composition and greater economic inequalities between regions. In the ICD analysis, the adjustment used only the HAR for infarction and heart failure. Obviously, a percentage of patients with an ICD indication other than these conditions would be left out, but the proportion would be small and comparable between ACs. The analysis of aortic stenosis, in which estimation of disease burden included the SAVR plus TAVR, has a relative value. It would depend on the percentage of patients with treated severe aortic stenosis, and the surgery data also included cases of aortic regurgitation. Nonetheless, considerable disparity between ACs in these variables would not be expected.

The correlation analysis with mortality is the least robust of those studied, as the RSMR refers only to in-hospital mortality, not that occurring at follow-up, which is when the techniques analyzed have the greatest impact. Hence, the correlations with in-hospital mortality should be viewed with caution. Even assuming that mortality was not affected by some techniques, there could be other potentially positive effects on symptoms, rehospitalizations, medication expenditure, and social impact, which we were unable to analyze, but should be taken into consideration.