SUN-182 THE ROLE OF PERFUSION PRESSURE AND FLOW IN CARDIOPULMONARY BYPASS ASSOCIATED ACUTE KIDNEY INJURY

Abstract

Acute kidney injury (AKI) after cardiopulmonary bypass (CPB) is associated with morbidity and mortality. The effect of perfusion flow and pressure on CPB-associated AKI is not well understood. Using multicenter registry data, this study aims to evaluate the role of perfusion pressure and flow in the development of CPB- associated AKI. The hypothesis of the study is that low flow is more strongly predictive of AKI than low pressure. Data from 31,749 adult patients who underwent surgery that employed the use of CPB over a 10-year period from January 2008 to January 2018 was retrospectively analysed. Data was obtained from the Perfusion Downunder Collaborative Database, which collates data from 9 participating hospitals in Australia and New Zealand. Primary predictor variables of interest were: time elapsed while perfusion flow was less than 1.6L/min/m2 and time elapsed while perfusion pressure was less than 30mmHg. The primary outcome variable was new post-operative AKI defined by the RIFLE criteria. Secondary outcomes included stroke and all cause mortality at time of discharge. The influence of flow and pressure on the primary outcome of AKI was estimated using separate multivariate models, with log transformation of both predictor and outcome required to meet model assumptions. The influence of flow and pressure on secondary outcomes of stroke and death were also estimated using separate multivariate models. The median age was 66 years. 72.4% (n=22,997) of patients were male and 27.6% (n=8,748) were female. The overall incidence of AKI of any RIFLE classification was 13.8% (n=4,071). The overall incidence of stroke was 1.5% (n=458) and death at time of discharge was 2.8% (n=818). In a multivariate linear regression model (Model 3‡) there is no association between low flow and AKI (β 0.00, CI 0.00-0.01, P=0.300). Likewise, in a multivariate linear regression model (Model 3‡) there is no association between low pressure and AKI (β 0.00, CI 0.01-0.02, P=0.098). Model 3|| demonstrates that in a multivariate logistic regression model, both low flow and low pressure are comparably predictive of stroke with an odds ratio (OR) of 1.48 (CI 1.19–1.84, P=0.000) and 1.57 (CI 1.44 – 1.72, P=0.000) respectively. Likewise, in a multivariate logistic regression model (Model 3◊), both low flow and pressure are predictive of death with OR of 1.26 (CI 1.07-1.49, P=0.005) and 1.17 (CI 1.06-1.30, P=0.002) respectively. In summary, neither flow less than 1.6L/min/m2, or pressure less than 30mmHg, had any power to predict post-operative creatinine in a fully adjusted model. Other non-operative factors were better predictors of AKI. These factors include a critical pre-operative state, ejection fraction and diabetes. By contrast, low flow and low pressure were both predictive in models for death and stroke. Neither the time elapsed while perfusion flow was less than 1.6L/min/m2 nor the time elapsed while perfusion pressure was less than 30mmHg were predictive of post- operative AKI. They were, however, predictive of both stroke and death.