Reply to the editor on "Impact of type of procedure and surgeon on EuroSCORE operative risk validation".

Dear Mr. Editor:

We appreciated the thoughtful comments by Spiliopoulos et al. regarding the article entitled "Impact of type of procedure and surgeon on EuroSCORE operative risk validation", recently published in the Brazilian Journal of Cardiovascular Surgery[.

We agree with their comments on the relevance of this issue due to the public scrutiny and demand for increasingly better quality of care. Moreover, risk stratification models in cardiac surgery are important to adjust outcomes to certain clinical profiles, being therefore useful in patient consent, quality assurance programs, as well as being used in patient selection for controlled randomized trials.

We recognized that an outdated risk model (EuroSCORE I) was used to assess hospital mortality in our study, because that was the most accepted and worldwide used system available at the time for most of the patients studied. As you stated in your letter, our study confirmed some of the limitations that have been previously shown. However, that was not our primary purpose to validate the EuroSCORE I in a Brazilian single cardiovascular surgery center, since other Brazilian cardiac surgery groups[ had already done so. Certainly, we will commit to validate our results to the new EuroSCORE II system in the nearest future, in order to determine its performance on a non European population.

On the other hand, the primary objective of our study was to demonstrate that, regardless of the risk stratification model used, there are unmeasured factors that significantly influence its validation and performance. Besides the fact that the patient population is different, hospital protocols may vary among different centers, diverse surgeon's background may influence patient's management, regardless of strict protocols. We proved that there is significant variability in outcomes in the same hospital, using a standardized patient care, adjusting to patient's severity.

Since it is impossible to control all the issues of concern, all risk models tend to be imprecise, subject to error and less than perfect. We then totally agree with Spiliopoulos and colleagues that we, as clinicians, should be extremely careful on interpreting a patient's condition based on a scoring model that will never replace the good individual clinical judgment, which is the foundation of our profession.