Upcoming events

Survival probability loss from percutaneous coronary intervention compared with coronary artery bypass grafting across age groups.

Selected in The Journal of Thoracic and Cardiovascular Surgery by S. Head



Umberto Benedetto, Mohamed Amrani, Toufan Bahrami, Jullien Gaer, Fabio De Robertis, Robert D. Smith, Shahzad G. Raja, on behalf of the Harefield Cardiac Outcomes Research Group.


J Thorac Cardiovasc Surg 2015;149:479-84


March 2014


Access the abstract

My Comment


Both percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) are treatment options for multivessel coronary artery disease. Whether one treatment is better than the other has been the topic of many randomised trials. However, these trials are limited by the relatively low number of patients, which is why observational studies are complementary to provide evidence from real-world practice and can find significant differences in events occurring at a lower rate. Moreover, several patient groups, such as the elderly, may be underrepresented in randomised trials and the best treatment option for these patients can more easily be evaluated from observational studies.


  • Records of 1,097 patients that underwent PCI and 5,626 patients that underwent CABG for multivessel coronary artery disease, between April 2001 and May 2013 in London.
  • Drug-eluting stents were used in 83.5% of patients.
  • Propensity-matched groups were constructed, maintaining 1,097 matched patients.
  • In unmatched groups, the PCI group as opposed to the CABG group had more females, less previous MI or previous PCI, less diabetes, more renal failure, and more smokers, while the extend of coronary disease less often involved a left main lesion and three-vessel disease. After propensity-matching, no significant differences were found and the standardised mean difference was <10% for all variables.
  • After a mean-follow of 5.5 ± 3.2 years, 301 deaths occurred (PCI: 208, CABG: 93) in the matched groups. Overall survival was 95% for both groups at 1 year, but for PCI and CABG it was respectively 84% vs 92% at 5 years, and respectively 75% vs 90% at 8 years (P<0.001).
  • The early hazard within 12 months was similar for PCI and CABG (deaths PCI: 47, CABG: 48), while the longer hazard beyond 12 months was in favour of CABG (deaths PCI: 161, CABG: 45).
  • The hazard function for CABG declined within the first 2 years after CABG, after which it remained steady. For PCI, the hazard function increased up to 1 year after PCI and stayed at a similar rate up to 5 years, but showed an increase beyond 5 years.
  • In subgroup analyses, CABG was superior to PCI in patients with DES implanted, those with 2-vessel, 3-vessel, and left main disease, as well as across all age groups of <60, 60-69, 70-79, and ≥80 years of age.

My comments

Observational studies are required to assess PCI and CABG in the real world and evaluate its safety and efficacy in larger populations. From this study, it is interesting to see that the hazard function of PCI increased again beyond 5 years of follow-up, particularly since randomised trials on PCI versus CABG usually extend follow-up only to 5 years and therefore these trials do not provide the complete picture. This highlights the importance of such studies, providing hypotheses for trial (design). However, it should be noted that there are also several limitations to observational studies, even when propensity matching is applied. It remains important to interpret both trial and registry data for guideline recommendations and clinical practice.

No comments yet!