IJCS | Volume 31, Nº6, November / December 2018

566 Lacerda et al. Paradoxical choice for radial access in ACS Int J Cardiovasc Sci. 2018;31(6)562-568 Original Article revascularization surgery (0.5% versus 19.5%; p < 0.001) was also different between the groups, whereas no difference was observed in the presence of diabetes mellitus, peripheral artery disease, smoking, previous history of heart failure or bleeding (Table 1). The variables described above as significant in the univariate analysis were inserted into the logistic regression model, with radial access as dependent variable. In this analysis, the variables with independent associationwith radial access were age (OR = 0.98; 95%CI = 0.96 – 0.99), creatinine (OR = 0.54; 95%CI = 0.3 – 0.98), signs of left ventricular failure (OR = 0.45; 95%CI = 0.22 – 0.92) and previous myocardial revascularization surgery (OR = 0.022; 95%CI = 0.003 – 0.16), all with a discouraging effect on the use of the radial access (Table 2). Discussion In the present study, patients treated with radial approach for coronary procedures had lower baseline risk of bleeding as compared with the femoral access group. This finding contrasts with the logical expectation that the access related to lower incidence of bleeding is the one more commonly used in patients at higher risk for this complication, characterizing a risk-treatment paradox. This paradoxical result raises the need for discussing potential causes of this phenomenon in a critical perspective of the cognitive process of the medical decision-making process. This, in turn, is presumedly influenced by several factors. One may expect that such decision is based on the main objective of the radial approach, i.e. to prevent bleeding; however, other factors may be determinant in this process. Interventionists have a natural sense of achieving success with their techniques. By intuition, the chance of success is expected to be lower from procedures considered technically more difficult. Hence, the operator tends to avoid the access considered more difficult in attempt to reduce the challenge. Nevertheless, that would be a biased view, since the risk of failure in the radial approach is lower than the risk of increased bleeding in femoral approach (eight times greater in the present study). Besides, a migration from radial to femoral vascular access when needed is also possible. Although the results of this study were exploratory, they suggest that the physician’s decision may be more strongly influenced by a sense of self- protection rather than a protection of the treated patients. This is quite possible, since while bleeding tends to be seen as a natural complication, failure in the intervention tends to be considered a medical failure. Further studies should explore these potential mechanisms. Intuitive estimation of probabilities in conditions of uncertainties is influenced by cognitive biases. 13-15 For example, when we opt to treat less complex patients, we are seeking cognitive comfort; and in search of this, we underestimate the risk of more complex patients, intuitively reducing the magnitude of the benefits that these patients could obtain from the procedure. In consequence, patients with more complex conditions receive less treatment than needed. This generates a risk-treatment paradox, typical of this intuitive process of decision making. Tounderstand themechanisms of this paradox, we built a propensity model to identify potential determinants to the choice for the radial access. In thismodel, we identified variables that had a negative association with the radial access only, not including variables that may increase the chance for this choice. This propensity score allows us to make interpretations of the decision-making process. It is possible that our interventionist had the radial access as the first-choice option (in fact, this approach was the most frequent in the study) and then used other criteria for secondary options. These criteria were represented in our model by independent predictors of the radial access. Analysis of these predictors showed that all of them concerned more complex patients, with predictors representing each of the domains: patient’s baseline constitution (age), comorbidities (creatinine), severity of ACS presentation (acute heart failure) andprevious history (surgery). These observations suggest that the physician Table 2 - Multivariate analysis that generated the propensity model of radial access Odds ratio 95% CI p-value Creatinine 0.54 0.3 – 0.98 0.041* Age 0.98 0.96 – 0.99 0.037* Killip class > 1 0.45 0.22 – 0.92 0.029 Previous revascularization 0.022 0.003 – 0.166 0.001 Stroke 0.366 0.13 – 1.07 0.066 Previous coronary disease 0.75 0.43 – 1.31 0.313 GRACE score 1.0 0.99 – 1.01 0.543* *Numerical variables.