Pre‐incision hypotension and the association with postoperative acute kidney injury – an opportunity to improve peri‐operative outcomes?

It is well established that intra-operative hypotension is associated with acute kidney injury (AKI), which is in turn associated with postoperative mortality. Recently, Maheshwari et al. investigated the influence of hypotension following induction of anaesthesia, and before the first surgical incision, on the incidence of postoperative AKI 1. From a large retrospective database analysis of more than 42,000 cases, the authors concluded that one-third of hypotensive readings occurred before incision which is arguably a time period under direct control of the anaesthetist. These readings were four times more frequently observed than after the first surgical incision. Hypotension at any time was associated with postoperative AKI. Before anaesthetists change their practice, this instalment of ‘Clinical Consequences’ will critique and explore the impact of this evidence in depth. The incidence of postoperative AKI can vary depending on the type of surgery, but up to 40% of hospital-acquired AKI may occur in surgical patients 2. Postoperative AKI is associated with increased mortality, as well as increased financial costs and other complications, such as the need for long-term renal replacement therapy 3-5. Since postoperative AKI is easy to define and measure, peri-operative strategies for its prevention seem to be a reasonable target for intervention. The question is, should we place more focus on better monitoring blood pressure and optimising haemodynamic parameters during the period between induction of anaesthesia and the first surgical incision? In 2012, the Kidney Disease Improving Global Outcomes (KDIGO) group released their AKI clinical practice guidelines 6. Acute kidney injury is classified in stages by the degree of increase in serum creatinine and reduction in urine output. It follows that renal function is not a dichotomous (normal vs. deranged), but a continuous parameter. Similarly, the decrease in blood pressure pre-incision (or specifically the drop in renal perfusion pressure) is also continuous. Yet, logistic regression analysis – the method used by Maheshwari et al. to analyse their dataset – requires that both are converted to binary (dichotomous) measures. They define hypotension as a mean arterial pressure (MAP < 65 mmHg and AKI as KDIGO stage one or higher. It would have perhaps been interesting to inspect the plot of serum creatinine against the actual decrease in pre-induction MAP to look for any suggestion of correlation across the whole range of values, but this was not reported. The blood pressure required to maintain adequate blood flow to vital organs will vary between patients. Blood flow to most organs in health is governed by autoregulatory mechanisms, which are well understood. What is less well established is the lower limit of MAP beyond which autoregulation breaks down, and when flow becomes directly proportional to the perfusion pressure. Our understanding is currently based on studies of animals 7, but there appears to be no clear point at which autoregulation ‘breaks down’ (Fig. 1) 8, 9. In some animal studies, the steeper slopes of the flow-pressure curve begin at lower limits, sometimes as low as 50 mmHg, which is much lower than the 65 mmHg cut-off employed by Maheshwari et al. However, the regulation of blood pressure itself is a complex interaction of cardiac output, peripheral resistance, venous return, blood volume and renal function 10. There is consensus that the safe lower limit of autoregulation increases in patients with acute and chronic comorbidities. These points were acknowledged in a recent consensus statement which recommends that MAP < 60 mmHg is associated with AKI, and identifies a need for more research on organ-specific autoregulation thresholds 11. We conclude, therefore, that the cut-off chosen by Maheshwari et al. is probably reasonable for this patient group. Sustained periods of intra-operative hypotension are associated with postoperative AKI 3. However, how best to define ‘hypotension’ remains unclear, and a recent review found 140 different definitions across 130 original articles 12. Consequently, the incidence of peri-operative hypotension ranges from 5% to 99%. Induction of anaesthesia and positive pressure ventilation all have detrimental haemodynamic effects that can be mitigated to some degree by the choice of anaesthetic technique 13. In contrast, hypotension during surgery may result from a wide range of surgical factors, such as intra-operative blood loss. Furthermore, surgery and anaesthesia may trigger an exaggerated inflammatory and stress response, which can independently promote AKI 14. Recently, Myles et al. reported an association between intra-operative oliguria and AKI. However, there was no difference in MAP between patients with or without oliguria, underlining the conclusion that blood pressure control is perhaps not always the most important factor for AKI prevention 15. Certain comorbidities may exacerbate peri-operative hypotension and AKI, as do other risk factors, such as pre-existing hypovolaemia 16. Chronic hypertension increases the risk of intra-operative hypotensive episodes, and tissues may be more vulnerable to ischaemia in such patients 17. Conversely, pre-existing propensity to hypotension, such as that seen in patients with aortic stenosis, may contribute 18. In such patients, where significant changes in blood pressure may be anticipated, it is important to ensure adequate pre-operative fluid resuscitation. Obstetric patients undergoing surgery are at risk of hypotension due to aorto-caval compression compromising venous return 19. This may be mitigated by the use of left lateral tilt or manual displacement of the gravid uterus. Additionally, it is important to consider insertion of appropriate intravenous access before induction of anaesthesia as well as considering the use of invasive blood pressure monitoring. Together with these disease states are concomitant medications which may increase the risk of intra-operative hypotension and AKI 20. Particularly, it is common practice to omit angiotensin-converting enzyme inhibitor drugs on the day of surgery as they may increase the risk of AKI through a range of mechanisms, including systemic hypotension, renal artery constriction and interstitial nephritis 21. However, recent evidence has challenged these long-established practices 22. For all these clinical conditions, it is well established that careful blood pressure management during anaesthesia and surgery will help prevent AKI and other complications of hypotension. However, the new suggestion of Maheshwari et al. is that even a brief period of hypotension pre-incision is influential and that this effect encompasses even healthy patients, as ~15% of patients in their database with postoperative AKI were ASA physical status 1 or 2. Some of the results are striking and surprising. Firstly, the relevant median duration of pre-incision hypotension <65 mmHg was just four minutes. This implies either that continuous intra-arterial blood pressure monitoring was always used or that non-invasive pressures were measured almost continuously. In the UK, measurements are usually made every five minutes 23. This will not capture the relevant critical period described by Maheshwari et al. Expressed another way, if the implications of the study are to be adopted in full, then UK blood pressure monitoring guidance during anaesthesia would need to change to take recordings every 1–2 minutes during the pre-incision period. Secondly, the authors calculated the area under the curve for pre-incision hypotension and the relevant median was just 14 mmHg.min. That is, an MAP of 64 mmHg for 14 minutes or, say, a pressure of 60 mmHg for just under three minutes (or a pressure of just over 50 mmHg for one minute etc.). These seem relatively modest departures from the thresholds for brief periods and would require very prompt action if detected. Thirdly, and perhaps most importantly, is the fact that this study tells us nothing about the benefits of any likely intervention or indeed, what that optimum intervention should be. Simply because brief periods of pre-incision hypotension are associated with AKI does not mean that preventing or treating them will prevent AKI. That would require a separate clinical trial. It is also conceivable that different interventions (e.g. vasoconstrictors vs. inotropes vs. intravenous fluid boluses) each yield different outcomes. Moreover, deciding on which intervention to use may require even more invasive monitoring to identify which specific intervention is necessary. The magnitude of the overall effect of pre-incision hypotension is modest. Maheshwari et al. report mean odds ratios of 1.02–1.05 for the influence of pre-incision hypotension. For the incidence of AKI they quote as 5%, this yields a ‘number needed to treat’ (NNT) of 422–1054. Thus, even optimistically, it would require 422 patients (and more likely > 1000 patients) across all comorbidities aggressively monitored and treated for pre-incision hypotension to prevent one case of AKI. The magnitude of effect may be greater for patients with more comorbidities, but the relevant data were not presented in the paper. This, and other aspects of the Maheshwari et al's. study reflect the advantages and limitations of ‘big data’ analyses. Interrogation of large datasets enables us to detect influences that would be missed by smaller observational studies. However, this comes at the cost of detecting statistically significant yet very small and inconsequential effects. This is perhaps most evident in the study by Maheshwari et al. where they report the comparison of mean (SD) MAP 94 (10) vs. 94 (11) as having a p value of < 0.001, or median (IQR [range]) serum creatinine values of 0.8 (0.7–0.9 [0.1–1.7]) vs. 0.8 (0.7–1.0 [0.1–1.5]) as having a p value of 0.004. These clearly identical values between the two groups are only statistically significant because the sample size was almost 43,000. In a recent editorial 24, Biais and Pearce acknowledge the barriers to implementing recommendations on intra-operative blood pressure management, citing data that suggests patients may spend less than half their time within recommended limits 25. They confirmed the uncertainty, discussed above, around which monitors, vasopressors and fluids to use, and how frequently blood pressure measurements should be taken. They also identified a risk of conflicting guidelines, and of extending recommendations that increased the volume of information provided to practitioners. In that context, there is little doubt or controversy to suggest that careful blood pressure management during induction of anaesthesia and the pre-incision period in high-risk patients, guided by appropriate invasive monitoring, will prevent or reduce the incidence of postoperative AKI. However, the suggestion that benefits can extend to all patients requires more careful consideration. The clinical consequence of the study by Maheshwari et al. is that all patients should have frequent (i.e. every minute) measurements of blood pressure during induction of anaesthesia and prior to the first surgical incision, and that any departure from mean arterial pressure < 65 mmHg should be promptly managed so that any hypotensive period is limited to less than four minutes. However, even by their own analysis, the NNT makes this consequence difficult to justify for all patients. Re-analysis of results focusing on odds ratios for ASA 3 and 4 vs. ASA 1 and 2 patients would be of interest. Finally, this justifies the need for a clinical trial to test the hypothesis that preventing modest, brief pre-incision hypotension is actually beneficial. JP is a former Editor of Anaesthesia and an elected member of the Council of the Royal College of Anaesthetists and Chair of the Safe Anaesthesia Liaison Group. SEG is President of the Association of Anaesthetists Trainee Committee. No other competing interest declared.