Acute kidney injury (AKI) is associated with short and long-term mortality and morbidity.1,2 More than 300 million people undergo surgery yearly3,4 and perioperative AKI is common, with a recent study of major abdominal surgery showing an incidence of 13%.5
Intraoperative hypotension (IOH) has been suggested as one of the causes of perioperative AKI, but consensus on the definition of IOH is lacking. A number of studies have used a population-based binary cut off point, such as a mean arterial pressure (MAP) below 55 mmHg, and shown an association with AKI.6,7 However, these population-based binary cut-offs may introduce a distortion and individual patient-based IOH definitions have been proposed: both our group and others have tested individual patient-based IOH definitions when investigating risk for perioperative myocardial damage.8,9 The argument that individual patient-based IOH definitions may be beneficial in this field of study was strengthened by a multicentre randomised controlled trial, evaluating low and high blood pressure (BP) targets in patients with septic shock, where renal outcomes were improved by high BP only in patients with known hypertension.10 However, there are also studies showing that absolute and relative thresholds are comparable in their ability to discriminate patients with myocardial or kidney injury from those without.11
The aetiology of AKI is complex. A common feature of many processes causing AKI is a reduction in regional renal oxygen delivery, leading to inflammation, ischaemia and possibly even necrosis.12,13 Surgical inflammatory trauma likely increases the risk of AKI. Factors such as hypotension, bleeding and hypoxia may add insult to injury. Haase et al.14 showed that a decrease of haemoglobin concentration was associated independently with postoperative AKI. The same study found that in patients with severe anaemia, the independent effect of hypotension on AKI was more pronounced, supporting the pathophysiological theory above, with additive harmful events.14 Some of these events are modifiable in the perioperative setting. In addition, fluid overload seems to be an independent risk factor for AKI.15 Increased renal venous pressure could reduce the transrenal pressure gradient for renal blood flow and elevated interstitial and tubular pressure might diminish the net glomerular filtration pressure gradient. There is much observational data supporting the association between fluid overload and AKI,15 but as bolus fluid infusion is the standard treatment of low cardiac output and hypotension this makes investigating this confounder especially difficult. Nevertheless, a recent review suggested that the haemodynamic management of the elderly surgical patient should focus on avoiding ‘hypotension’ and high central venous pressures to minimise risk of postoperative AKI.16
We hypothesised that there was an association between individual patient-based hypotension (defined as a decrease from a patient's own preoperative baseline BP) and perioperative AKI. The aim of this study was to examine this association in patients undergoing major elective noncardiac surgery. In addition, we studied the impact of other potential risk factors, including comorbidities, blood loss, fluid overload and so on.
The study protocol (2014/1306–31/3) was approved by the Regional Ethics Committee of Stockholm, Sweden (chairperson: Pierre Lafolie) on 24 September 2014, which also waived informed consent.
Study population and design
The study population consisted of a previously described observational cohort study of all adult patients undergoing major elective noncardiac surgery at the Karolinska University Hospital, Stockholm, Sweden, between October 2012 and May 20139 augmented by similar patients recruited during the period October 2015 to April 2016. To be included, all patients had to be scheduled electively for overnight admission to the postoperative unit. There were no changes to guidelines regarding management of aforementioned intraoperative events during these periods. Exclusion criteria were patients undergoing phaeochromocytoma surgery; these patients commonly show extreme BP variability.
Patient characteristics, medical history and perioperative data were obtained from electronic medical records: age, sex, BMI, smoking status, history of cardiovascular disease (hypertension, atrial fibrillation, congestive heart failure or ischaemic heart disease; previous myocardial infarction or coronary intervention/bypass graft), antihypertensive medication (angiotensin-converting enzyme inhibitors, β-blockers or calcium channel blockers) and insulin-dependent diabetes mellitus. Furthermore, we recorded data on baseline SBP and preoperative creatinine concentration. To ensure baseline BP was representative of a patient's habitual values, all BPs documented within the 2 months before surgery were averaged. Patients were classified according to the American Society of Anaesthesiologists (ASA) physical status classification.
Surgical procedures included gastrointestinal, urological, gynaecological, vascular, head and neck, reconstructive and orthopaedic surgery. The types of anaesthesia were general and/or regional including epidural anaesthesia. The following intraoperative events were recorded by reviewing anaesthetic charts: hypotension (we recorded decreases in SBP which were >40 or 50% relative to each patient's baseline BP and which lasted >5 min), hypoxaemia (SpO2 <90% for >5 min), tachycardia (increase in heart rate of >30 beats min−1 from baseline for >5 min), blood loss and cumulative fluid balance (FLB) during the first 24 h after surgery. All intraoperative BP measures were recorded from invasive, intra-arterial BP monitoring.
As intraoperative data, including BP, were collected from nonelectronic anaesthetic charts, a validation trial was performed to test how well hypotensive events during surgery and anaesthesia were recorded in paper anaesthesia records (supplementary Table 1, http://links.lww.com/EJA/A133). For 30 major surgical cases, we photographed both the anaesthesia monitor screen with its minute-by-minute trends for arterial BP, and the corresponding paper record where BPs are entered at 5-min intervals.
Blood was obtained for routine laboratory testing on the first postoperative morning on all patients. Levels of high-sensitivity cardiac troponin T and creatinine were analysed in serum samples collected 22 h after surgery using the Elecsys 2010 system (Roche Diagnostics GmbH, Mannheim, Germany), which has been available at Karolinska University Hospital since 10 December 2010. This method has a lower detection limit of 2 ng l−1, a 99th percentile cut off point of 14 ng l−1 and a coefficient of variation of less than 10% at 13 ng l−1.17 Plasma creatinine (anticoagulated with Li-heparinate) was determined with a modified Jaffe method from Beckman Coulter (CREm, Kreatinin, Ref 472525) on a Beckman DXC800 instrument, and was measured the day before and on the first, second and third morning after surgery. AKI was defined according to the kidney disease: improving global outcomes (KDIGO) guidelines.18
The incidence of AKI, defined by the KDIGO criteria,18 was evaluated during the first two postoperative days using baseline preoperative creatinine obtained from routine laboratory testing performed within the week before surgery. The highest creatinine on postoperative mornings 1 to 3 was used for the AKI staging.
We followed a preset statistical analysis plan according to the above a priori defined hypotheses. Data were analysed by STATA version 14.2 (Stata Corp., College Station, Texas, USA). Continuous data are presented as medians with 25th to 75th percentiles and categorical variables as number (percentage). For comparison of linear variables, the Mann–Whitney U test was used. Fisher's exact test or the χ2 test were used for categorical values.
In the bivariate analysis, we observed a significant association between hypotensive events, both more than 40 and 50% decreases in BP and AKI. The observed P values were 0.013 and 0.024, respectively. We divided hypotensive events into two categories (>40 to ≤50% and >50%) and the analyses indicated a dose–response relationship. Based on the P values in the bivariate analysis and clinical consideration, multivariable analyses were performed. These assessed whether the association was because of confounding from the following factors: age, preoperative creatinine (both as continuous variables), sex, ASA more than two, hypertension and chronic antihypertensive medication (all as dichotomous variables). Furthermore, the log-linear effect of preoperative creatinine, blood loss and FLB was assessed by categorising them according to quartiles. These variables were entered into a logistic model with stepwise selection (significance level of P ≥ 0.05 for both entering and staying in the model). In the resulting model, the estimates for hypotensive events were essentially unchanged and included the following significant adjustment variables: sex (male), ASA more than two, preoperative creatinine, treated hypertension and FLB categorised into quartiles. The influence of blood loss on the association between a hypotensive event more than 40 to 50% or less or more than 50 and AKI was further explored by adjusting for blood loss categorised in quartiles.
Sensitivity analyses included restriction of the cohort to patients with a preoperative creatinine below 90 μmol l−1, excluding patients in the 75th percentile, those who had a pre-existing diagnosis of hypertension (untreated and those who were on antihypertensive medication). Further analyses of interaction were conducted between preoperative creatinine and hypotensive event by combining preoperative creatinine below or above the 75th percentile with either hypotensive event more than 40 to 50% or less or more than 50% or no hypotensive event (≤40%; supplementary Table 2, http://links.lww.com/EJA/A133). The combination of preoperative creatinine below the 75th percentile and no hypotensive event (≤40%) was considered to represent the lowest risk and therefore used as a reference. As a diagnosis of both pre-existing hypertension and intraoperative blood loss may be linked to the exposure, we tested the (multiplicative) interactions between IOH and treated hypertension (Yes vs. No) and the combination IOH and blood loss.
During the study time frame, a total of 470 patients fulfilled the inclusion criteria. No patients were lost to follow-up. Table 1 presents data on the patients’ characteristics. The average age was 67 years, 47% were women and the most prevalent preoperative comorbidity was hypertension. The most common surgical procedures were gastrointestinal surgery (51%), urological (29%) and gynaecological (11%). General anaesthesia was used for 96% of the patients and two-thirds also received an epidural or a spinal anaesthetic. IOH was frequent with 286 patients (61%) having a more than 40% and 68 patients (14%) having a more than 50% reduction from preoperative baseline.
The accuracy of the recordings of IOH in our validation trial showed that 13 out of 30 patients had a fall in BP of more than 50% captured by the electronic monitors but only nine patients had the hypotensive events recorded on their anaesthetic study records (supplementary Table 1, http://links.lww.com/EJA/A133). Thus, there were no overestimated hypotensive events, rather a trend that the anaesthetic staff was underestimating the lower limit of intraoperative BP recordings.
During the perioperative phase, 127 (27%) patients developed AKI (Table 1). More patients with AKI were men, and AKI patients also had a higher ASA class, a greater frequency of treated hypertension and a greater preoperative creatinine concentration. During anaesthesia and surgery, hypotensive events were more frequent in the AKI subgroup, as were higher levels of intraoperative blood loss. In the postoperative phase, positive-FLB was more common among AKI patients, as was myocardial damage. All-cause mortality at 30 days was 4% in patients with AKI, compared with 1% in those without (P = 0.046). Neither the type of surgery nor the choice of anaesthesia was significantly associated with perioperative AKI.
As presented in Table 2, IOH of more than 50% was associated with a more than doubled risk of postoperative AKI, odds ratio 2.27; 95% CI, 1.20 to 4.30, P = 0.013. This was after adjustment for the covariates sex (male), ASA more than two, treated hypertension, preoperative creatinine more than 90 μmol l−1 and FLB. The association was similar to the crude result, odds ratio (OR) 2.38; 95% CI, 1.30 to 4.36, P = 0.005 (Supplementary Table 3, http://links.lww.com/EJA/A133). When further adjusting for blood loss in quartiles, as detailed in Table 2, the association was affected minimally (OR 2.02; 95% CI, 1.05 to 3.89, P
As preoperative creatinine concentration, treated hypertension and intraoperative blood loss seemed to be associated with an elevated risk of AKI, we performed a sensitivity analysis, detailed in the supplementary material, http://links.lww.com/EJA/A133. This entailed using the adjusted model with exclusion of patients in the quartile with the highest preoperative creatinine (n = 122). This yielded an OR of 2.85, 95% CI, 1.31 to 6.23 (supplementary Table 2, http://links.lww.com/EJA/A133), which indicated that the impact of an intraoperative hypotensive event on risk of AKI could be somewhat higher for patients without lowered glomerular filtration rate. This theory was strengthened in a further interaction analysis (supplementary Table 4, http://links.lww.com/EJA/A133), where we created five risk groups based on one or a combination of the two risk factors; preoperative creatinine more than 90 μmol l−1 (75th percentile) and a hypotensive event (>40 to ≤50% or >50%). There was a clear risk gradient of AKI in the presence of greater hypotension when creatinine was less than 90 μmol l−1. Among patients with preoperative creatinine more than 90 μmol l−1, there was a consistently high risk, with or without a hypotensive event, but no difference between the levels of hypotension.
When testing for interaction between the preoperative risk factors treated hypertension and intraoperative blood loss, in combination with IOH we found no evidence of a significant interaction (P = 0.82 and P = 0.32, respectively).
We found perioperative AKI occurred in almost one-third of high-risk elective surgery patients in this university hospital cohort study. A number of unalterable characteristics, such as male sex, elevated preoperative creatinine, treated hypertension and ASA class more than two were associated with a risk of AKI. An intraoperative hypotensive event, as defined by a more than 40 or 50% decrease from a patient's SBP baseline for a minimum of 5 min, was associated with an elevated risk of AKI, even after adjustment for the aforementioned characteristics. This finding is important, as individualised anaesthetic management to minimise hypotension is possible. Large-scale clinical trials are needed to confirm if tailored BP targets could reduce risk of AKI and other adverse events.
We have previously reported an association between IOH, defined as above, to be associated with increased risk of myocardial injury.9 Our present findings confirm published data on the association between hypotension and AKI. Sun et al.6 investigated outcomes of 5127 patients and showed that time spent at different levels of lowered MAP were associated with AKI. However, this study excluded patients with baseline MAP less than 65 mmHg, making it impossible to study the effect of IOH in these patients. Possibly, they also generated selection bias, which could enhance the risk estimate. Our results show that patients with pre-existing hypertension are more vulnerable to IOH and have a higher relative risk of AKI in the presence of a hypotensive event (supplementary Table 2, http://links.lww.com/EJA/A133). In a large retrospective study, Walsh et al.7 found that an intraoperative MAP less than 55 mmHg was associated with AKI development: a duration of more than 1 min was sufficient and there was a graded increase of AKI risk with an increased duration of the hypotensive event. In an observational study of more than 15 000 patients with normal preoperative renal function, those developing AKI more often had a MAP less than 40 mmHg.19 In a recent study, Salmasi et al.11 studied both absolute and relative MAP thresholds and found comparable ability to discriminate patients with myocardial or kidney injury. MAP less than 65 mmHg, or relative decrease in BP of 20% below baseline was related to myocardial and kidney injury, with an increased risk at lower absolute thresholds and with prolonged hypotension. Notably, when the Salmasi et al.11 study used an IOH definition close to ours (MAP below 50% of preoperative values), a duration of 5 min significantly increased the risk for myocardial and kidney injury. In contrast to the Salmasi et al.11 study with a 5.6% AKI incidence, our cohort had five times that incidence. This is likely explained by our cohort being subjected to high-risk surgery, where all patients were scheduled for an overnight admission to the postoperative unit. The incidence of AKI after cardiac surgery has been studied frequently; using the KDIGO criteria, a recent retrospective analysis reported an incidence of 42%.20 In a systematic review and meta-analysis, including 91 observational studies and 320 086 cardiac surgery patients, Hu et al.21 report a pooled AKI incidence rate of 22.3% (95% CI, 19.8 to 25.1). The recently published study of global patient outcomes after elective surgery, found 16.8% of the total population developing one or more postoperative complications.22
As mentioned, our findings are consistent with previous published results and add important information. As pointed out by Sun et al.6 future studies should focus on the effect of relative hypotension on perioperative AKI. As we used an individually based definition of IOH, we were able to include patients with preoperative normal BP, increasing the generalisability of our findings. Owing to the fact that we included patients with preoperative elevated creatinine, compared with Walsh et al.,7 we were able to explore the influence of hypotension on patients with pre-existing renal insufficiency. Interestingly, we found that patients with an elevated preoperative creatinine had an elevated risk of perioperative AKI with or without hypotension. In contrast, the risk among those without elevated preoperative creatinine was only negatively affected if an intraoperative hypotensive event occurred.
Importantly, intraoperative blood loss may be a confounder and risk factor at the same time. A sudden substantial blood loss during surgery is often associated with a period of lowered BP readings before treatment with crystalloid fluids, blood transfusion, inotropes or vasopressors. It may therefore be difficult to distinguish if the increased risk of kidney damage is because of the fall in BP or to the loss of haemoglobin and the reduced capacity of oxygen transport.14 In our study, there was a strong association between the estimated intraoperative blood loss and AKI in the bivariate analyses but, as we are unable to determine the temporal relationship between the two, our findings cannot contribute to the reasoning above. The addition of this variable to the multivariable model resulted only in a slight attenuation in the relative risk, (OR 2.18 vs. 2.46), suggesting there was no effect of significant importance on the association between IOH and perioperative AKI. The relation between FLB and AKI is complex; if fluid is administered incorrectly (e.g. as a treatment for anuria), it can cause renal compartment syndrome or renal venous congestion23 with consequent fluid retention, but fluid overload may be the result of symptomatic AKI too. As for blood loss, we have insufficient data on the timing of fluid administration and thus cannot discuss the relation between the exposure or outcome any further.
As we are entering an era of individualised medicine,24,25 where the aim is to optimise clinical decisions about a patient's care by utilising all available data, it seems reasonable to apply baseline electronic medical record information to physiological parameters. An elderly person with known hypertension is likely to have a higher risk for adverse events at a different BP threshold than a patient with normal baseline BP. Research supporting this exists. In a 2015 study, Monk et al.26 found BP measurements of 50% below baseline to be associated with increased 30-day mortality. Similarly, a recent investigation by van Waes et al.8 showed that a relative MAP decrease of 40% below preinduction BP for more than 30 min was associated with an elevated incidence of myocardial injury. In the ward setting, a decrease in SBP relative to the baseline value was a significant independent predictor of the development of severe AKI.27 As both our present study and multiple previous studies show that certain levels of hypotension for certain at-risk patients exists, the logical step is to determine if goal-directed anaesthesia can minimise these risks. In a review of 20 studies (4220 participants), patients receiving perioperative haemodynamic optimisation were indeed at decreased risk of renal impairment28 However, large-scale multicentre randomised studies are lacking.
Our study has strengths. With high-quality pre-, intra- and postoperative data, including baseline BP, we could determine the impact of an individually based decline in SBP. The cohort was homogenous, with an a priori decision to have an overnight postoperative admission. No patients were lost to follow-up and all had creatinine measurements performed at the same time. Our definition of AKI within the first two postoperative days, instead of the extended acute kidney injury network definition within seven days, is more likely to be related to the intraoperative course. Increases in serum creatinine seven days after surgery are possibly confounded by other factors but, as we lack these data, we cannot investigate this. Importantly, all BP readings in our study were monitored invasively, reducing the risk of mismeasurement and gives us a less biased effect estimate. Furthermore, a validation trial was performed to evaluate the correctness in BP recordings. Sensitivity analyses were done to correct for preoperative renal risk and other confounders were controlled with logistic regression. However, because of the observational nature of this study, we cannot determine causality and, like all single-centre studies, we acknowledge inherent limitations regarding generalisability. Importantly, as the intraoperative BP data came from paper charts we may have underestimated the risk of IOH, as seen in the validation trial. However, if ‘cases’ with SBP below 50% of baseline are hidden among the ‘controls’, our risk estimates would be higher. Another paper chart drawback is that it forces us to use SBP instead of MAP or DBP. It would have been of interest to observe the impact of MAP decrease from baseline. A much larger cohort would have allowed us to investigate smaller changes in creatinine than that which defines the KDIGO criteria. Finally, there are other important unmeasured factors that could introduce bias. We lack information on inflammatory markers, haemoglobin levels and duration of surgery.
In conclusion, AKI is frequent among high-risk surgical patients. IOH, defined as a more than 50% reduction in SBP from baseline for more than 5 min, may be an important factor associated with this adverse event. Future randomised trials focusing on individually based avoidance of IOH may minimise the risk of AKI and other organ damage.
Acknowledgements relating to this article
Assistance with the study: none.
Financial support and sponsorship: in 2013, the corresponding author received 110 000 Swedish Crowns, approximately equalling 11 000€, from the Swedish Associations of Local Authorities and Regions (SKL, Sveriges Kommuner och Landsting). This agency strives to promote the research use of the national Swedish quality registers, and this funding was won in national competition.
Conflicts of interest: none.
Presentation: preliminary data for this study were presented at a poster session at the European Society of Anaesthesiology Euroanaesthesia, 3 to 5 June 2017, Geneva, Switzerland.
1. Rimes-Stigare C, Frumento P, Bottai M, et al. Evolution of chronic renal impairment and long-term mortality after de novo acute kidney injury in the critically ill; a Swedish multicentre cohort study. Crit Care
2. Rimes-Stigare C, Frumento P, Bottai M, et al. Long-term mortality and risk factors for development of end-stage renal disease in critically ill patients with and without chronic kidney disease. Crit Care
3. Weiser TG, Haynes AB, Molina G, et al. Estimate of the global volume of surgery in 2012: an assessment supporting improved health outcomes. Lancet
2015; 385 (Suppl 2):S11.
4. Molina G, Weiser TG, Haynes AB. Maternal and neonatal mortality after cesarean delivery: reply. JAMA
5. O’Connor ME, Kirwan CJ, Pearse RM, et al. Incidence and associations of acute kidney injury after major abdominal surgery. Intensive Care Med
6. Sun LY, Wijeysundera DN, Tait GA, et al. Association of intraoperative hypotension with acute kidney injury after elective noncardiac surgery. Anesthesiology
7. Walsh M, Devereaux PJ, Garg AX, et al. Relationship between intraoperative mean arterial pressure and clinical outcomes after noncardiac surgery: toward an empirical definition of hypotension. Anesthesiology
8. van Waes JA, van Klei WA, Wijeysundera DN, et al. Association between intraoperative hypotension and myocardial injury after vascular surgery. Anesthesiology
9. Hallqvist L, Martensson J, Granath F, et al. Intraoperative hypotension is associated with myocardial damage in noncardiac surgery: an observational study. Eur J Anaesthesiol
10. Asfar P, Meziani F, Hamel JF, et al. SEPSISPAM Investigators. High versus low blood-pressure target in patients with septic shock. N Engl J Med
11. Salmasi V, Maheshwari K, Yang D, et al. Relationship between intraoperative hypotension, defined by either reduction from baseline or absolute thresholds, and acute kidney and myocardial injury after noncardiac surgery: a retrospective cohort analysis. Anesthesiology
12. Evans RG, Ince C, Joles JA, et al. Haemodynamic influences on kidney oxygenation: clinical implications of integrative physiology. Clin Exp Pharmacol Physiol
13. Gomez H, Ince C, De Backer D, et al. A unified theory of sepsis-induced acute kidney injury: inflammation, microcirculatory dysfunction, bioenergetics and the tubular cell adaptation to injury. Shock
14. Haase M, Bellomo R, Story D, et al. Effect of mean arterial pressure, haemoglobin and blood transfusion during cardiopulmonary bypass on post-operative acute kidney injury. Nephrol Dial Transplant
15. Prowle JR, Kirwan CJ, Bellomo R. Fluid management for the prevention and attenuation of acute kidney injury. Nat Rev Nephrol
16. Martensson J, Bellomo R. Perioperative renal failure in elderly patients. Curr Opin Anaesthesiol
17. Giannitsis E, Kurz K, Hallermayer K, et al. Analytical validation of a high-sensitivity cardiac troponin T assay. Clin Chem
18. Kellum JA, Lameire N. KDIGO AKI Guideline Work Group. Diagnosis, evaluation, and management of acute kidney injury: a KDIGO summary (part 1). Crit Care
19. Kheterpal S, Tremper KK, Englesbe MJ, et al. Predictors of postoperative acute renal failure after noncardiac surgery in patients with previously normal renal function. Anesthesiology
20. Machado MN, Nakazone MA, Maia LN. Prognostic value of acute kidney injury after cardiac surgery according to kidney disease: improving global outcomes definition and staging (KDIGO) criteria. PloS One
21. Hu J, Chen R, Liu S, et al. Global incidence and outcomes of adult patients with acute kidney injury after cardiac surgery: a systematic review and meta-analysis. J Cardiothorac Vasc Anesth
22. International Surgical Outcomes Study group. Global patient outcomes after elective surgery: prospective cohort study in 27 low, middle and high-income countries. Br J Anaesth
23. Martensson J, Bellomo R. Does fluid management affect the occurrence of acute kidney injury? Curr Opin Anaesthesiol
24. Farrugia G, Weinshilboum RM. Challenges in implementing genomic medicine: the Mayo Clinic Center for individualized medicine. Clin Pharmacol Ther
25. Gottesman O, Scott SA, Ellis SB, et al. The CLIPMERGE PGx Program: clinical implementation of personalized medicine through electronic health records and genomics-pharmacogenomics. Clin Pharmacol Ther
26. Monk TG, Bronsert MR, Henderson WG, et al. Association between intraoperative hypotension and hypertension and 30-day postoperative mortality in noncardiac surgery. Anesthesiology
27. Liu YL, Prowle J, Licari E, et al. Changes in blood pressure before the development of nosocomial acute kidney injury. Nephrol Dial Transplant
28. Brienza N, Giglio MT, Marucci M, et al. Does perioperative hemodynamic optimization protect renal function in surgical patients? A meta-analytic study. Crit Care Med