Performance measurement of intraoperative systolic arterial pressure to predict in-hospital mortality in adult liver transplantation

Abstract

Profound hypotension during liver transplantation is aggressively treated with vasopressors thus frequently unrevealed in a retrospective study. The relationship between concealed intraoperative hypotension and in-hospital mortality after liver transplantation was evaluated using performance measurement (PM) of systolic arterial pressure (SAP). Median performance error (MDPE), median absolute performance error (MDAPE), and wobble of SAP were calculated using preoperative SAP as the reference value, and prereperfusion and postreperfusion SAPs as measured values. Univariable and multivariable logistic regression analyses were performed using 6 PM parameters and 36 traditional SAP-derived parameters to predict in-hospital mortality. In-hospital mortality was 3.9% (22/569 cases). Prereperfusion MDAPE and postreperfusion wobble were the only significant SAP-derived predictors of in-hospital mortality. The area under receiver operating characteristic curve of prediction model was 0.769 (95% confidence interval 0.732-0.803, P < 0.001; sensitivity = 55%, specificity = 94%). Severe hypotension during liver transplantation is concealed by proactive vasopressor treatment thus traditional measures of hypotension generally fail to detect the masked hypotension in retrospective analysis. PM analysis of intraoperative SAP including prereperfusion MDAPE and postreperfusion wobble is most likely to detect treated and therefore concealed hypotension, and was able to independently and quantitatively predict in-hospital mortality after liver transplantation with high diagnostic specificity.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Figure 1

Systolic arterial pressure and performance error during prereperfusion and postreperfusion periods in survivors and non-survivors. Multiple line scatters show the means (dots) and standard deviations (error bars) of survivors and non-survivors. Horizontal lines are averages of intraoperative systolic arterial pressure in survivors (solid line) and non-survivors (dashed line) in Fig. 1A. The graphs start from the time of reperfusion then go towards the start or end of surgery, thus both ends of plot involves only a small number of patients. On average, systolic arterial pressures are a well-maintained around the average systolic arterial pressure in both the survivors and non-survivors during surgery (Fig. 1A). However, performance errors of non-survivors are more deviated and fluctuate above and below the target 0% level, compared to those of survivors (Fig. 1B).

Figure 2

Comparison of performance measures between survivors and non-survivors. Box plot shows median and interquartile range of prereperfusion and postreperfusion MDPE, MDAPE, and wobble in survivors and non-survivors. Upper and lower whiskers are maximum and minimum values, respectively. Round symbols show the 5th/95th percentile values. Prereperfusion and postreperfusion MDPE were similar between groups. However, prereperfusion and postreperfusion MDAPE, and postreperfusion wobble were significantly larger in non-survivors compared to survivors. *P < 0.001 vs prereperfusion values, †P < 0.001 vs survivors. Abbreviations: MDPE = median performance error, MDAPE = median absolute performance error.

Figure 3

ROC curves of the prediction models. Three ROC curves were built from the propensity scores of prediction models built with performance measures (PM model), MELD score (MELD model) or both (PM + MELD model), respectively. Addition of performance measurement parameters to the MELD model increased the diagnostic specificity of prediction model from 77% to 84%. The combined model has a significantly larger area under ROC curve compared to the PM model. *P = 0.007 vs PM model. Abbreviations: ROC = receiver operating characteristic; PM = performance measurement; MELD = model for end-stage liver disease.