Pulmonary artery catheter trials in cardiac surgery and haemodynamic decision states

Abstract

Perry et al. show that allocating low-risk cardiac surgical patients randomly to pulmonary artery catheter or central venous catheter monitoring is feasible [1]. The near absence of crossover, low clinician refusal and near-complete follow-up are useful because device preference has often been substituted for trial evidence in this setting. The pilot also clarifies the problem facing the subsequent trial. PUMA did not use a protocolised goal-directed treatment algorithm: the catheter model, insertion site and data use were left to clinicians, and other advanced haemodynamic monitoring was not restricted [1]. Although this pragmatic approach preserved clinical relevance, allocation assigned access to monitoring rather than a defined haemodynamic decision strategy. The issue is not that clinicians used judgement, but that the trial did not capture the decision states in which judgement was informed by pulmonary artery catheter data. A definitive trial should therefore specify three elements: who might benefit; when pulmonary artery catheter data are expected to add value; and what changes to management occur. Baseline characteristics should identify patients in whom incremental information is plausible, without treating the low-risk exclusions as a flaw. If eligibility is broadened, such characteristics might include operative category; surgical urgency; ventricular function; pulmonary hypertension; or right ventricular risk. Shared clinical triggers should then mark when pulmonary artery catheter data might add information in both groups, beyond transoesophageal echocardiography, central venous catheter data and routine observations. These include failure to separate from cardiopulmonary bypass; escalating vasoactive or inotropic support; discordant pressure–flow information; suspected pulmonary hypertension or right ventricular failure; consideration of pulmonary vasodilators or mechanical circulatory support; and uncertainty at intensive care handover. Process outcomes should record whether those triggers changed use of fluids; vasopressors; inotropes; pulmonary vasodilators; additional monitoring; or mechanical support. The acute kidney injury imbalance in PUMA is a useful illustration, not evidence of mechanism. Acute kidney injury occurred in 26/76 patients allocated to pulmonary artery catheter and 14/73 allocated to central venous catheter (RR 1.80, 95%CI 1.03–3.14), and the authors treated this appropriately as hypothesis generating [1]. Future trials should therefore measure not only whether a catheter was placed, but when its information changed management, what changed and whether those changes improved or harmed recovery.

The authors' abstract, as published at the source. Anaesthesia, 2026 · DOI ↗