Skip to main content
Dryad logo

Retrospective cohort analysis of Spanish national trends of coronary artery bypass grafting and percutaneous coronary intervention from 1998 to 2017


Carnero Alcazar, Manuel et al. (2021), Retrospective cohort analysis of Spanish national trends of coronary artery bypass grafting and percutaneous coronary intervention from 1998 to 2017, Dryad, Dataset,



Spain is one of the countries with the lowest rates of revascularization and highest ratio of percutaneous coronary intervention (PCI) to coronary artery bypass grafting (CABG).


To investigate the changes and trends in the two revascularization procedures between 1998 and 2017 in our country.


Retrospective cohort study. Analysis of in-hospital outcomes.


Minimum Basic Dataset from the Spanish National Department of Health: mandatory database collecting information of patients who are attended in the Spanish public National Health System.


603,976 patients who underwent isolated CABG or PCI in the Spanish National Health System. The study period was divided in four 5-year intervals. Patients with acute myocardial infarction on admission were excluded.

Primary and Secondary Outcomes:

We investigated the volume of procedures nationwide, the changes of the risk profile of patients and in-hospital mortality of both techniques.


We observed a 2.2-fold increase in the rate of any type of myocardial revascularization/million inhabitants-year: 357 (1998) to 776 (2017). 93,682 (15.5%) had a coronary surgery. PCI to CABG ratio rose from 2.2 (1998-2002) to 8.1 (2013-2017). Charlson´s index increased by 0.8 for CABG and 1 for PCI. The median annual volume of PCI/hospital augmented from 136 to 232, while the volume of CABG was reduced from 137 to 74. In the two decades, we detected a significant reduction of CABG in-hospital mortality (6.5% Vs 2.6%, p<0.001) and a small increase in PCI (1.2% Vs 1.5%, p<0.001). Risk adjusted mortality rate was reduced for both CABG (1.51 Vs 0.48, p<0.001), and PCI (1.42 Vs. 1.05, p<0.001).


We detected a significant increase in the volume of revascularizations (particularly PCI) in Spain. Risk-adjusted in-hospital mortality was significantly reduced.


Sources of information and patient selection:

Data was obtained from the MBDS from the Department of Health of Spain (National Institute of Healthcare Information).  This research was carried out according to the STROBE (Strengthening the Reporting of OBservational studies in Epidemiology) recommendations. This study was approved by the Institutional Review Board and Ethics Committee at Hospital Clínico San Carlos (Madrid, Spain). 

 We investigated all the outpatient or hospitalization episodes of the Spanish NHS from 1998 to 2017 in which a CABG or PCI procedure had been carried out. Those episodes during which patients underwent concomitant procedures were excluded (See next Table 1 ICD9 and ICD10 codes). 









00.66, 36.03, 36.06, 36.07, 36.09

0270xxx, 0271xxx,0272xxx,0273xxx, 02C0xxx, 02C1xxx, 02C2xxx, 02C3xxx, 02C4xxx

Excluded Concomitant procedures

35.xx, 37.3x, 37.51, 38.44, 38.45, 39.1x, 39.2x, 39.3x & 37.90

027Fxxx, 027Gxxx, 02NFxxx, 02NGxxx, 02Vxxxx, 027Jxxx, 02NJxxx, 02Nxxxx, 02Rxxxx, 02Qxxxx, 028xxxx, 02Bxxxx, 02Cxxxx (different from 02C0xxx, 02C1xxx, 02C3xxx and 02C4xxx), 02Fxxxx, 02Hxxxx, 02Jxxxx, 02Kxxxx, 02Nxxxx, 02Pxxxx, 02Uxxxx, 02Wxxxx, 02Yxxxx, 025xxxx



I21.x9, I21.x1, I21.x, I21.4, I21.3, I21.9

Likewise, all episodes with an acute myocardial infarction/acute coronary syndrome with ST segment elevation as the primary diagnosis on admission (See supplementary Table 1) were excluded, as those with both types of revascularization. In addition, to avoid possible coding errors, patients younger than 18 or older than 100-year-old, and patients operated on CABG in centers without CABG or who underwent PCI in centers without PCI were also discarded. Patients discharged alive earlier than two days after CABG were also considered as coding errors. The episodes corresponding to patients who were transferred to another center and consecutive planned revascularizations episodes were consolidated into a single episode(1). Each episode corresponds to a single patient, but a patient might have more than one episode. Given that we analyzed in-hospital outcomes, different consolidated episodes will be considered as different patients for the purpose of this study. 

The full period of time (1998-2017) was divided in four 5-year intervals (1998-2002, 2003-2007, 2008-2012 and 2013-2017).

Patient and Public Involvement:

No patient was actively involved in the study

National volume of revascularization procedures and risk profile of the patients:

We investigated the absolute number of CABG and PCI per year, the number of procedures per million of inhabitants and the changes in the PCI/CABG ratio. Further analyses to investigate the trends in the indexed volume of each type of procedure were also performed according to sex and age. To estimate the nationwide population, data was extracted from the National Institute of Statistics.

Healthcare centers were classified according to the volume of procedures per year. Thus, for both CABG and PCI, hospitals were divided into four groups according to the quartile of the volume of PCI or CABG interventions that they performed in each year: Low volume (quartile 1), Low-Intermediate Volume (quartile 2), High-Intermediate Volume (quartile 3) and High Volume (quartile 4). 

Patients were classified into four groups according to their age (≤60,>60 & ≤70,>70 & ≤80, and >80-year-old). We analyzed the evolution of the prevalence of various comorbidities Age-modified Charlson´s Index was calculated (3,4). In addition, the individual components of this score (previous history of myocardial infarction, kidney disease, diabetes, ...) and other procedural variables were analyzed throughout the study period (see Table 1).


We analyzed in hospital non-adjusted and adjusted mortality for PCI and CABG and its changes over the study period.

Statistical Analysis:

Categorical variables were represented with absolute and relative frequencies (%) and were compared with the chi-squared test. The normality of the quantitative variables was analyzed with PP- plots, and they were expressed with mean and standard deviation or median and interquartile range. Imputation ​​was not made for missing values. Statistics were estimated using available data. Quantitative variables were compared among the periods of the study with an analysis of variance or non-parametric comparison of medians. Contrasts were performed to investigate the presence of a linear trends (LT). The relative risk reduction (RRR) and odds ratio (OR) were used to represent the strength of association between different variables and mortality.

We investigated factors associated to mortality for each type of revascularization. For this purpose, we created multivariable models including variables with theoretical value and variables related to mortality (statistical significance p<0.1) in an univariable analysis. The best models were selected based on the value of the Akaike information criterion, R2 and their area under the curve.

Subsequently, we estimated 2 new models to predict mortality after PCI and CABG, respectively, excluding the time period. We divided the observed mortality in each year for PCI and CABG by that expected according to the corresponding model. In this way, we analyzed the evolution of risk- adjusted mortality rate (RAMR) over time. (14).

Statistical analysis was performed with Stata v 15.0 (StataCorp. 2017. Stata Statistical Software:Release 15.College Station,TX: StataCorp LLC.).

1. Goicolea Ruigómez FJ, Elola FJ, Durante-López A, Fernández Pérez C, Bernal JL, Macaya C. Coronary artery bypass grafting in Spain. Influence of procedural volume on outcomes. Rev Esp Cardiol (Engl Ed).2020[Epub ahead of print]. 

2. INEbase [Internet]. Madrid: Instituto Nacional de Estadística (Spain); [cited 2019, July, 20]. Available from:

3. Charlson ME, Szatrowski TP, Peterson J, Gold J. Validation of combined comorbidity index. J Clinical Epidemiol.1994;47:1245-51.

4. Sun JW, Rogers JR, Her Q, Welch EC, Panozzo CA, Toh S, et al. Validation of the combined comorbidity index of Charlson and Elixhauser to predict 30-day mortality across ICD 9 and ICD 10. Med Care.2018;56:812.


Usage Notes

See README text.

This dataset differs from the one originally  obtained from the Spanish Department of Health (National Institute of Healthcare Information) in order to comply with the Spanish Personal Data Protection Law (LOPD 3/2018).