Pulmonary artery catheter trials in cardiac surgery and haemodynamic decision states: a reply

We thank Zhou et al. [1] for their correspondence regarding the PUMA pilot trial [2]. Two important points are raised by the authors: first, the implications of using a pragmatic intervention in a clinical trial (which avoids using a goal-directed treatment algorithm); and second, future opportunities to explore how haemodynamic management with a pulmonary artery catheter in patients undergoing cardiac surgery influences clinical decisions and whether these decisions improve patient outcomes. Following the successful pilot, the PUMA trial has been funded by the Medical Research Future Fund of Australia. It will be an international, multicentre, Bayesian, pragmatic, non-inferiority randomised trial comparing allocation to a pulmonary artery catheter or no pulmonary artery catheter in 1600 adults undergoing low-risk cardiac surgery [2]. The primary estimand is the median treatment effect on days alive and at home on postoperative day 30, with a non-inferiority margin of 1 day. The design of the full PUMA trial [3] largely mirrors that of the PUMA pilot, with two key population differences that improve external generalisability: the eligibility criteria have been expanded to include additional surgical indications including mitral valve surgery; and the EuroSCORE 2 threshold has been increased to 3%. Accordingly, the eligible population now includes most patients presenting for cardiac surgery. Like the pilot, the PUMA trial is pragmatic in that it does not enforce a goal-directed therapy algorithm. This decision maximises external generalisability for several reasons. Goal-directed therapy approaches can differ widely between individual centres; are subject to poor adherence if too complex; do not reflect patient-level physiology or pathophysiology if not complex enough; disempower clinicians from making complex and nuanced decisions based on diverse sources of information; and could discredit the results of the trial if the wrong algorithm is chosen. The trade-off, we considered, was non-uniformity in the intervention and the potential for the pulmonary artery catheter to go ‘underused’ in some settings, which could dilute treatment effects. Importantly, the heterogeneity of treatment effects, influence of clinician and site knowledge and protocols and a range of mechanistic outcomes will be explored as part of the full PUMA trial to determine how (if at all) access to pulmonary artery catheter data influences patient outcomes.