FP613ASSOCIATION OF BLOOD PRESSURE AND SYMPTOMS WITH END ORGAN ISCHAEMIA DURING HAEMODIALYSIS

Abstract

Introduction and Aims: Proving a causative link between intradialytic hypotension (IDH) and hard outcomes has been problematic. Only one definition of IDH has been independently linked with increased mortality (intradialytic systolic nadir <90mmHg regardless of symptoms), in a retrospective review of datasets with limited blood pressure and symptom data. Increased mortality would likely be due to repeated end-organ ischaemia: blood pressure is a surrogate for downstream tissue perfusion. Difficulty in defining the threshold for harm may be due to patient heterogeneity, reliance on non-invasive blood pressure (NIBP) which has a large measurement error and is intermittent, and the rarity of microcirculation monitoring in renal literature. We have taken a complementary, physiological approach by continuously monitoring blood pressure, symptoms and cerebral near-infrared spectroscopy (NIRS) in 635 dialysis sessions. Our objectives were: to identify which if any blood pressure threshold best predicted inadequate oxygen delivery during dialysis on the population level; to identify whether presence of symptoms improved this prediction. Method(s): Physiological data was prospectively gathered from 635 dialysis sessions in 59 patients, including continuous blood pressure (finometer); continuous cerebral oxygenation (NIRS); type/timing of symptoms, electronically recorded in real time. Staff and patients could view NIBP only. Cerebral ischaemia was defined as a drop of 15% of baseline value. We identified onset of ischaemia and symptoms within -10 to+20 minutes of a sustained drop in MAP (according to predefined absolute and relative thresholds), and investigated relationships with univariate linear regression. The sensitivity /precision of MAP thresholds for predicting temporally related cerebral ischaemia and symptoms were calculated. In separate analyses, multilevel generalized linear models were used to identify significant predictors of dialysis sessions with ischaemia. Result(s): There was a very significant but moderate relationship between change in MAP from baseline (DELTAMAP) and onset of ischaemia (3% increase in incidence per 10mmHg drop, p<0.001, R2 0.87). A DELTAMAP of 10mmHg, the KDOQI IDH definition, was associated with ischaemia in 38/889 incidences (4.3%), rising to 20/122 (16.4%) for DELTAMAP 40mmHg. There was a similar relationship between DELTAMAP and symptoms (4% increase per 10mmHg drop, p<0.001, R2 0.889). Of control segments (DELTAMAP<10mmHg), 2.3% were associated with ischaemia, and 2.4% symptoms (largely cramp). Absolute MAP had slightly improved precision: 10.5% of cases of MAP<60mmHg (a threshold used in critical care) were associated with ischaemia, rising to 29.8% for MAP<50mmHg. However, sensitivity/precision combinations were suboptimal for all tested thresholds of DELTAMAP and absolute MAP on the population level. After adjustment for repeated measures, significant predictors of ischaemia during dialysis were maximum DELTAMAP (OR 1.04 per mmHg, p<0.001) and non-cramp symptoms (OR 2.15, p 0.023). Absolute MAP and SBP nadir were inferior to DELTAMAP. Patient level intercepts were driven by duration of diabetes and haemodialysis vintage, which may be surrogates for loss of autoregulation. Conclusion(s): Non-cramp symptoms and DELTAMAP were significant predictors of ischaemia during dialysis; symptoms are important. However, we did not find a MAP threshold with sufficient sensitivity/precision for ischaemia at population level. Thresholds which are personalised, or stratified for diabetes and dialysis vintage, may have improved performance, and this will be our next analysis.