Pulmonary Artery Catheter in Cardiac Surgery Revisited

To the Editor We believe that the findings of Schwann et al.1 are not applicable to patients undergoing cardiac surgery in general, title notwithstanding, because they studied outcomes of only a narrow and nonrepresentative segment: coronary revascularization with the use of extracorporeal circulatory support and without transesophageal echocardiography guidance. The presumed cause-and-effect conclusions and the discussion of various potential etiologies of detrimental outcomes in patients in whom pulmonary artery catheters (PACs) were used were based on the statistical analysis of outcomes of propensity-matched patients. The logical absurdity of this approach could be illustrated by the following “thought experiment”: substitute the procedure in question (placement of PAC) with another (for example, performance of cardiopulmonary resuscitation [CPR]). State-of-the-art statistical analysis would show higher mortality and morbidity rates in those patients to whom CPR was administered, compared with propensity-matched patients to whom CPR was not administered. Should we call for a reassessment of CPR as a procedure associated with worse outcomes? In addition to the many concerns and limitations discussed in the accompanying editorial,2 two sources of potential bias stand out: first, the absence of pulmonary hypertension and right ventricular dysfunction, arguably the most important indications for perioperative pulmonary arterial catheterization3–5 from the list of covariates used for propensity matching analysis. The second is a significant heterogeneity of perioperative surgical care introduced by noninclusion of specific institutions and surgeons as analysis covariates. It remains entirely plausible that institutions with low frequency of pulmonary artery catheterizations also have more experienced and proficient surgeons, shorter cardiopulmonary bypass times, “better” intensive care unit care, and achieve lower morbidity and mortality rates in otherwise propensity-matched populations. This possibility might explain the mysterious doubling of mortality in patients monitored with PACs despite very modest changes in fluid balance and/or use of vasoactive support statistically attributed to PACs. We applaud the efforts to establish parameters for the perioperative use of invasive hemodynamic monitoring in cardiac surgery, but caution against extrapolating the potentially biased outcomes of on-bypass coronary revascularization in low-risk population to the entire field.