Introduction: The mortality benefit of transradial primary PCI has been shown by several trials. However, it is less known whether this holds true for patients with hemodynamic instability. We studied the effect of a possible interaction between access site and systolic blood pressure (SBP) at admission on all-cause mortality at one year in patients treated with primary PCI.
Methods: We analyzed data of 1255 consecutive cases from a prospective unicenter registry using multivariable Cox modeling. One-year mortality was 14.7% (185 patients). Restricted cubic splines were used to explore presence of non-linear relationships of the continuous predictors to log relative hazards. To avoid overfitting and determine the most relevant ones from the 24 candidate predictors including the non-linear interaction between access site and SBP, backward variable selection was combined with bootstrap resampling. Variables selected in at least 60% of the 10000 samples were included in the final model, which has been validated internally by bootstrapping.
Results: Thus, in the final model, the effect of the interaction term (selected in 93.2% of the samples) was adjusted for the following 10 statistically significant (p<0.05) variables: age, SBP and heart rate at admission, final TIMI flow grade, left ventricular ejection fraction, cardiac arrest on or prior to admission, access site, baseline serum creatinine level, use of intra-aortic balloon pump, and diabetes (in order of predictor ranks). The interaction between access site and SBP was independently associated with mortality (p=0.03); i.e., the effect of access site may be a function of SBP (Figure). Both the apparent and optimism-corrected discriminatory power and model fit were good (c-statistic: 0.86 and 0.85; calibration slope: 1.00 and 0.92, respectively).
Conclusions: Our results suggest that the beneficial effect of radial access on all-cause mortality may not be present among patients with hemodynamic instability.