Intraoperative hypotension and change in estimated glomerular filtration rate after major abdominal surgery

Editor, A reduced renal blood flow because of intraoperative hypotension may contribute to acute kidney injury. Recently, two retrospective studies showed that intraoperative hypotension was associated with an increased risk of acute kidney injury after noncardiac surgery.1,2 Pre and postoperative serum creatinine values were not routinely measured, however, and this may have introduced bias. Also, the use of the estimated glomerular filtration (eGFR) rate may provide a more accurate assessment of renal function than serum creatinine.3 The aim of this study was to determine whether intraoperative hypotension is associated with a change in eGFR in patients undergoing abdominal surgery with routine perioperative creatinine measurements. The local Medical Research Ethics Committee (trial number W15.032) approved this secondary analysis of the prospective observational Myocardial Injury and Complications after major abdominal surgery (MICOLON) study (ClinicalTrials.gov Identifier NCT02150486). In the MICOLON study, the association between high-sensitive cardiac troponin T levels and noncardiac complications after major abdominal surgery was investigated in patients at risk for coronary artery disease.4 Serum creatinine was routinely measured on the day of surgery and on the first, third and seventh postoperative day. eGFR was calculated using the Modification of Diet in Renal Disease 4 equation.3 Primary outcome was change in eGFR, defined as the difference between baseline eGFR and lowest postoperative eGFR, expressed as a percentage of baseline eGFR value. In the sensitivity analysis we used the change in creatinine as the outcome variable. Intraoperative hypotension was expressed as the total duration below several absolute and relative mean arterial pressure (MAP) threshold values and the area under the curve, combining depth and duration, for several MAP threshold values. MAP was recorded every minute in case of invasive blood pressure (BP) monitoring, and every 1 to 3 min in case of noninvasive monitoring. When the MAP was not measured or in case of an artefact, the prior MAP was carried forward to the next MAP measurement. Linear regression analysis was used to assess the relation of intraoperative hypotension with change in eGFR, before and after adjustment for potential confounders. Potential confounders for the association of intraoperative hypotension and change in eGFR were age, sex, renin–angiotensin–aldosterone system inhibitors, congestive heart failure, American Society of Anesthesiologists classification, type of surgery, duration of surgery and blood loss.2,5 Effect estimates are expressed as unstandardised coefficients (βs) with their accompanying 99% confidence interval (CI). As multiple thresholds are compared, we used a more stringent level of significance of P < 0.01 and hence present effective estimates with 99% CIs. For statistical analyses IBM SPSS, Chicago, IL version 22 was used. In total, 202 patients were included in the analysis. Creatinine was available in all 202 patients at baseline, and in 99% (201/202), 96% (181/188) and 88% (106/121) of hospitalised patients on the first, third and seventh postoperative day, respectively. Invasive BP monitoring was performed in 120 patients (59%). In the univariable analysis, intraoperative hypotension, defined as a MAP below 75 mmHg, was associated with a change in eGFR (eGFR decreased with 0.05% for each minute spent below a MAP of 75 mmHg, 99% CI: −0.09 to −0.00, P = 0.009; Fig. 1a). A similar association was observed for intraoperative hypotension defined as a decrease in MAP of 20% (−0.05%, 99% CI: −0.10 to −0.01, P = 0.004) and 25% from baseline (−0.05%, 99% CI: −0.09 to −0.01, P = 0.004, Fig. 1b). The area under the curve for intraoperative hypotension thresholds was not associated with a change in eGFR (Fig. 1c). In the multivariable analysis, none of the intraoperative hypotension definitions were associated with change in eGFR (Fig. 1d, e and f). In the sensitivity analysis with a change in creatinine as the outcome variable, similar results were found.Fig. 1: Uni (a, b, c) and multivariable (d, e, f) analysis of absolute and relative MAP threshold values and AUC for MAP thresholds and change in eGFR. AUC, area under the curve; eGFR, estimated glomerular filtration; MAP, mean arterial pressure.The potential influence of intraoperative arterial perfusion pressure on organ function preservation is an ongoing debate. Walsh et al.2 and Sun et al.1 found, in two retrospective studies, that intraoperative hypotension, defined as a MAP below 55 mmHg, was associated with acute kidney injury. In contrast to these studies, we had highly detailed information on perioperative renal function for our study patients. Instead of using creatinine values that were requested by treating physicians on medical indication (potentially leading to confounding by indication and information bias), creatinine measurements were systematically conducted. In doing this, we may however, have missed the ‘peak’ creatinine value, potentially leading to an underestimation of the true incidence of acute kidney injury. We used eGFR as the outcome variable because it represents renal function better than creatinine.3 A drawback of the eGFR, however, is that it has not been validated for patients with unstable creatinine values. However, the perioperative use of eGFR is recommended by others and the use of a change in creatinine as the outcome variable showed similar results.6 Typically, a decline in renal function is dichotomised at a creatinine level of 1.5 to 1.9 times baseline or an increase of at least 0.3 mg dl−1 (≥26.5 μmol l−1). Although categorisation of continuous variables is often done to simplify statistical analysis, dichotomising a continuous variable increases the chance of a type 1 error (‘false positive’).7 By including the outcome as a continuous variable we increased statistical power to detect a true association. Still, our study may be underpowered to find an association between intraoperative hypotension and change in eGFR. For example, the change in eGFR we found in patients with MAP less than 55 mmHg (−0.074% for each minute below this threshold, 99% CI: −0.228 to 0.079) would have required the inclusion of 876 patients to reach statistical significance. A closer investigation of the effect estimates suggests a trend towards renal injury as BP declines, which may become statistically significant in larger sample sizes. In summary, we did not observe an association between intraoperative hypotension and change in eGFR, although we cannot exclude that our study was underpowered. Nevertheless, the effect estimates of change in eGFR per minute intraoperative hypotension had not been studied before and could serve as a basis for sample size calculations in future studies. Acknowledgements relating to this article Assistance with the study: none. Financial support and sponsorship: support was provided from institutional and departmental sources. Conflict of interest: none.