Blood pressure thresholds for the administration of balanced crystalloids and the effect on intra-operative hypotension: A proof-of-concept causal inference analysis featuring idealised dynamic treatment regimens.

Background Haemodynamic management and fluid administration constitute a challenging cornerstone of intra-operative care to counteract hypotension. Observational studies remain a vital source of evidence regarding the associations of fluids and vasopressors with peri-operative outcomes.Objective To evaluate dynamic treatment regimens (DTRs) of balanced crystalloids on the incidence of intra-operative hypotension. DTRs constitute a statistical framework to evaluate the causal impact of different treatment strategies (e.g. fluid administration) on an outcome (e.g. hypotension) whilst accounting for time-dependent treatments and treatment-confounder feedbacks.Design Analysis of observational data using modern causal inference methods.Setting Tertiary institution in South Korea during January 2011 to December 2020 (INSPIRE dataset).Patients N = 23 305 patients undergoing elective surgery under general anaesthesia.Interventions The DTRs refer to thresholds of mean invasive arterial blood pressure (MAP) to guide a hypothetical administration of 250 ml of balanced crystalloids. As example, a DTR implies administering 250 ml of balanced crystalloids if MAP drops below a 70 mmHg threshold, and to administer no crystalloids otherwise. MAP thresholds of 60, 65, 70 and 75 mmHg were evaluated.Main Outcome Measures Primary endpoint was intra-operative hypotension defined as a MAP below 65 mmHg averaged in 15-min windows.Results The DTRs resulted in clinically similar incidences of intra-operative hypotension for surgeries shorter than 2 h. For surgeries lasting longer, the DTRs with early fluid administration resulted in a reduction of hypotension of more than 4% towards the end of the surgery. As higher MAP-thresholds imply higher amounts of fluid, our findings suggest there is in general a positive effect of fluid administration on intra-operative hypotension.Conclusions Using modern causal inference methods, we demonstrated the clinical utility of idealised DTRs to study the impact of time-dependent haemodynamic management on the incidence of intra-operative hypotension in idealised settings. Future studies are required to investigate more complex DTRs mirroring clinical practice and to assess the robustness of the findings with respect to the different surgical populations.