Postoperative acute kidney injury in adult non-cardiac surgery: joint consensus report of the Acute Disease Quality Initiative and PeriOperative Quality Initiative

1The development of postoperative complications after major non-cardiac surgery — particularly acute kidney injury (AKI) — has consistently been associated with substantial long-term morbidity and mortality–. AKI can be considered a sentinel postoperative complication that is strongly associated with increased risk of death, co-occurrence of other postoperative complications, increased length of hospital stay and the development of progressive chronic kidney disease (CKD), which results in a considerable health-care and societal burden–.

2Considerable research and previous consensus meetings have focused on AKI after cardiac surgery. However, the role of AKI after other forms of major surgery, particularly gastrointestinal surgery, has received less attention, even though such procedures are associated with higher 30-day mortality than many forms of cardiac surgery. Despite the importance of AKI as a complication after all forms of major surgery, lack of consensus exists with regard to the definition, prevention and treatment of postoperative AKI (PO-AKI) and the pathophysiology of PO-AKI in the non-cardiac setting has not been well described. Furthermore, the understanding of and approach to this condition can differ between specialists in nephrology, anaesthesia and/or perioperative medicine.

3In 2019, a joint meeting of the Acute Disease Quality Initiative (ADQI-24) and the PeriOperative Quality Initiative (POQI-7) was convened to address PO-AKI after major non-cardiac surgery. Here, we present our recommendations for clinical practice as well as for further research in this important area.

4The Conference Chairs of the 24th ADQI/7th POQI consensus committee (J.R.P., J.A.K., A.D.S. and L.G.F.) convened a diverse panel of clinicians and researchers representing relevant disciplines (internal medicine, nephrology, critical care, anaesthesia, perioperative medicine, pharmacy and surgery) from Europe, North America and Australia, to discuss the issues relating to PO-AKI. The consensus meeting was held at Emmanuel College, Cambridge, UK, on 4–7 September 2019, and followed the established ADQI process and POQI protocols, including use of a modified Delphi method to achieve consensus,. The broad objective of ADQI 24 was to produce expert-based statements and a summary of current knowledge pertaining to the definition and management of PO-AKI in the setting of non-cardiac surgery for use by clinicians and researchers, based on review of evidence by and professional judgment of the faculty as well as to identify evidence gaps to establish research priorities.

5Conference participants were divided into five work groups: group 1 was tasked with the epidemiology and pathophysiology of PO-AKI. Group 2 was tasked with developing a definition for PO-AKI. Group 3 developed recommendations for the prevention of PO-AKI. Group 4 focused on treatment options for patients with PO-AKI, and group 5 explored outcomes after PO-AKI (Supplementary Information). Members of the work groups performed comprehensive literature searches and developed a consensus of opinion, backed by evidence where possible, to distil the available literature and articulate a research agenda to address important unanswered questions. Members assessed the level of evidence and strength of recommendation for all consensus statements using the GRADE evidence to decision frameworks. A number of ungraded research recommendations were also identified in areas where high-quality evidence is lacking. Strength of recommendations and aggregated conclusions were established by consensus of all of the participants, who formally voted and approved the consensus recommendations.

6Like all forms of AKI, PO-AKI represents a clinical syndrome rather than a single disease and limited histological data from kidney biopsy samples are available to identify specific pathology, (Box 1). Animal models that are used to study AKI pathophysiology tend to employ a single insult to the kidneys rather than explore the multifactorial insults that are common in clinical practice–. However, commonly implicated mechanisms for PO-AKI include ischaemia–reperfusion injury, endogenous or exogenous nephrotoxins, physical causes such as obstruction, inflammatory factors, vasoconstriction, and oxidative stress (Fig. 1). Thus, a variety of exposures that are encountered in the surgical setting are involved in the development of PO-AKI. Many of these exposures might also have a role in other forms of AKI, but their relative importance, severity and timing in PO-AKI may differ from that of other AKI settings. Within this framework, the contribution of individual mechanisms is often difficult to determine; for instance, nephrotoxic drugs have been implicated in 20–30% of all AKI episodes and are commonly encountered in the surgical setting. Moreover, agents such as radiocontrast and antibiotics are commonly used in patients who are already at risk of AKI and discerning their relative contribution to the overall AKI course is often difficult. Future research including studies aimed at understanding the interplay between concomitant perioperative complications might help to identify the underlying pathophysiological mechanisms of PO-AKI.

7Consensus Statement 1a

8Postoperative acute kidney injury (PO-AKI) is a clinical syndrome rather than a single disease. The majority of cases of PO-AKI are multifactorial in cause (ungraded).

9Consensus Statement 1b

10The incidence of PO-AKI (defined by changes in serum creatinine) varies with the characteristics and urgency of surgery. The incidence of AKI after day or ambulatory surgery is uncertain (ungraded).

11Consensus Statement 1c

12AKI defined by transient oliguria is more common in the intraoperative and postoperative period than AKI defined by serum creatinine. Severe oliguria and anuria, even in the absence of changes in serum creatinine, are associated with increased long-term morbidity and mortality (ungraded).

13Consensus Statement 1d

14The majority of observational studies focus on AKI in the immediate postoperative period. Little evidence is available regarding the epidemiology of acute kidney disease in the perioperative setting (ungraded).

15Consensus Statement 1e

16Risk factors for PO-AKI include age >50 years, male sex, glomerular filtration rate <60 ml/min/1.73 m2, diabetes mellitus, heart failure, ascites, hypertension, emergency surgery, intraperitoneal surgery, number of medications, use of angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs), high American Society of Anesthesiology Physical Status classification score and albuminuria. Patients with chronic kidney disease (CKD) and/or diabetes are at a particularly high risk of AKI (ungraded).

17A variety of criteria for changes in serum creatinine levels and urine output have been used to define AKI; however, the perioperative medicine community has largely adopted the 2012 Kidney Disease: Improving Global Outcomes (KDIGO) consensus criteria to define PO-AKI in clinical trials and epidemiological studies. Adoption of this consensus definition has resulted in a greater appreciation of the incidence, varied presentation and outcomes of PO-AKI (Box 1). Although changes in serum creatinine are fairly common in the postoperative period, the incidence of PO-AKI varies according to the characteristics and urgency of surgery as well as the exact AKI definition employed, with a higher incidence in emergency surgery,,– (Supplementary Table 1). The incidence of AKI in day or ambulatory surgery is unknown — but presumed to be low — owing to a lack of published data on the detection of AKI among patients who are discharged to the community immediately following surgery.

18More is known about PO-AKI defined by change in serum creatinine than that defined by perioperative decrements in urine output. The prognostic implications of intraoperative transient oliguria are controversial, with some studies reporting a higher risk of subsequent adverse outcomes and others suggesting no increase in risk–. One study suggested that discrimination for subsequent adverse outcomes could be enhanced by testing for novel biomarkers. In this study, which included some patients with PO-AKI, the sum of the concentration of tissue inhibitor of metalloproteinases 2 (TIMP2) multiplied by that of insulin-like growth factor binding protein 7 (IGFBP7) added prognostic information to AKI defined by urine output alone ([TIMP2]×[IGFBP7] > 2 was associated with progression from stage 1 AKI).

19Many, but not all, of the available studies suggest that PO-AKI defined by oliguria alone is more common than PO-AKI defined by changes in serum creatinine levels,–. In a study of PO-AKI after non-cardiac surgery, the incidence of AKI increased from 8% to 64% if urine output was included in the definition. Some postoperative oliguria is assumed to be an appropriate physiological response to decreased intravascular volume, vasodilatation or non-osmotic release of arginine vasopressin in response to tissue injury; however, sustained decrements in urine output, even without accompanying changes in serum creatinine, have been associated with longer-term morbidity and mortality,.

20The concept of acute kidney disease (AKD) is intended to bridge the gap between AKI, which refers to an acute alteration in kidney function over a 7-day period, and CKD, which refers to a sustained alteration in kidney function over a 3-month period. AKD therefore refers to AKI that persists for more than 7 days after the initial insult, or a progressive decline in kidney function that does not meet AKI criteria on a week-by-week basis. Few epidemiological data are available on the incidence and prognosis of postoperative AKD, despite its likely importance as a bridge to development of sustained kidney dysfunction and CKD. A study that evaluated substantial decline in kidney function, defined as a 30% decrease in estimated glomerular filtration rate at 60 days after surgery, suggested that the incidence of this outcome was only 2% among patients without PO-AKI (defined by change in serum creatinine only), but 10%, 17% and 29% among patients who experienced PO-AKI stages 1, 2, and 3, respectively. The incidence of AKD in the general population is likely lower than in the post-surgery population; in a study of Canadian electronic health records, 5.2% of patients in a given period had AKD with or without AKI.

21Several patient-related, surgery-related, and anaesthetic-related risk factors for developing early PO-AKI during the first 48 h after surgery have been described,. The three validated risk scores for PO-AKI include male sex, age >50 years, diabetes mellitus, hypertension, ascites, heart failure, emergent surgery, intraperitoneal surgery, poly-pharmacy, use of an angiotensin-converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB), and increasing American Society of Anesthesiologists physical status classification score,,. Albuminuria and hypoalbuminaemia have also been described as risk factors for PO-AKI–. Risk factors for PO-AKI occurring in patients in the ICU more than 48 h after surgery reflect risk factors for AKI in critical illness including the need for new nephrotoxic medication, sepsis and shock,. Whether targeting modifiable risk factors for early PO-AKI, for example, using perioperative haemodynamic interventions, will impact the incidence of AKI later in the postoperative period, >48 h after surgery, remains uncertain.

22Use of a validated clinical risk score is a potentially useful approach to risk stratify patients for targeted interventions or to facilitate the conduct of randomized controlled trials (RCTs). However, existing validated risk scores for PO-AKI were developed using a variety of older (pre-KDIGO) definitions of AKI or were applied to select surgical settings,,. More generalizable prognostic clinical tools are needed to accurately risk-stratify patients preoperatively, particularly to distinguish those patients at the highest levels of risk (e.g. >20–30%). In the absence of such tools, patients with CKD and those with diabetes can reasonably be considered to be at increased risk of PO-AKI. The KDIGO guideline for the diagnosis, evaluation and management of AKI suggests that age >60 years, emergency surgery, elevated American Society of Anesthesiologists physical status classification score and preoperative comorbid illnesses, including diabetes and any chronic disease of the heart, lung or liver, are risk factors for PO-AKI.

23Understanding the pathophysiology and epidemiology of PO-AKI is central to the identification and development of novel preventative therapies and treatments. This understanding may be improved by conducting prospective studies that include collection of biological samples for histological and biomarker examination as well as retrospective analyses that examine shared pathogenesis across a spectrum of perioperative complications. The incidence of PO-AKI in the day and ambulatory surgery settings is uncertain and should be investigated in epidemiological studies. Development of better risk prediction tools for PO-AKI could help to risk-stratify patients and set thresholds of predicted risk that merit adjustments in perioperative care.

24Although the KDIGO definition of AKI is now widely employed in the perioperative literature, the exact implementation of this definition varies considerably in terms of timeframe of diagnosis. We recommend defining PO-AKI as occurring when existing KDIGO criteria for AKI are met within 7 days of an operative intervention (Box 2). AKI occurring de novo ≥7 days after surgery may arise in a variety of contexts that are not necessarily related to the surgery itself; we recommend that such AKI should be evaluated and managed as hospital-acquired AKI. Our recommendation for using a 7-day window is somewhat arbitrary but maintains important consistency with the KDIGO definitions for AKI in other clinical contexts and is supported by evidence from numerous clinical studies,–. Furthermore, within the 7-day window, different timing of AKI may be indicative of the nature of the kidney insult, that is, intra-operative events versus post-operative insults.

25Notably, AKI could potentially be present before surgery (for example, in patients with sepsis requiring surgical intervention, acute trauma or preoperative contrast exposure), highlighting the importance of a risk-based kidney health assessment (KHA). Use of a KHA was previously recommended by ADQI in the context of quality improvement processes for monitoring patients at a high risk of AKI before surgery. A KHA comprises a structured evaluation that includes previous history of AKI, current medications, cardiovascular health, haemodynamic status and markers of kidney function (i.e. serum creatinine) and kidney damage (i.e. urine dipstick). Use of a KHA in the perioperative period involves integrated assessment of patient-specific and procedure-specific risk factors, clinical context and resource setting.

26Importantly, the KHA approach can be used to perform context-dependent evaluation both before and after surgery, when AKI is diagnosed and during follow-up (Fig. 2). A perioperative KHA may range from a simple history and physical examination conducted as part of a routine low-risk preoperative evaluation, to an evaluation that includes more specific measures such as urinalysis to identify proteinuria, biomarkers of kidney damage and kidney imaging in higher-risk patients (Box 2). Measurements other than serum creatinine and urine output are not currently part of the KDIGO AKI definition, but likely provide important context for enhanced interpretation of these criteria. Although we strongly support routine use of KHAs before and after surgery, the potential clinical benefit of this approach has not been formally evaluated and so remains uncertain. However, an accepted advantage of routine perioperative KHAs is improved recognition of preoperative kidney dysfunction, which establishes a baseline for subsequent diagnosis of PO-AKI.

27Conventional diagnostic criteria for AKI might be influenced by multiple factors specific to the perioperative period, rendering their predictive utility for PO-AKI less certain. Serum creatinine concentration is subject to abrupt changes in total body water, which may spuriously increase or decrease measured values, as well as perioperative changes in creatinine production that alter serum creatinine levels independently of changes in GFR. Urine output is also likely to be influenced by multiple physiological factors specific to the perioperative period, including intravascular volume status, relative or absolute hypotension and neurohormonal response to surgery. Data suggest that although intraoperative oliguria is associated with PO-AKI, the positive predictive value is low. We therefore recommend that urine output is interpreted in the context of severity of oliguria, clinical setting and other corroborating evidence of PO-AKI. Thus, isolated perioperative oliguria meeting AKI stage 1 criteria in the context of transient hypotension should not be viewed as being the same as more severe or prolonged oligo-anuria unresponsive to haemodynamic interventions. The presence of new evidence of kidney tubular damage (e.g. on urine microscopy or from novel biomarkers) or evidence of evolving sepsis in the context of early oliguria would be suggestive of AKI; these factors can be identified and integrated within a timely postoperative KHA.

28Consensus Statement 1a

29We recommend defining postoperative acute kidney injury (PO-AKI) as occurring when existing Kidney Disease Improving Global Outcomes (KDIGO) criteria for AKI are met within 7 days of an operative intervention (ungraded).

30Consensus Statement 1b

31We recommend performing a risk-based Kidney Health Assessment (KHA) preoperatively and postoperatively, with the specific methods and frequency of assessment determined by clinical judgement based on individual patient-specific and procedural-specific risk factors and the available resources (grade D evidence; strong recommendation).

32Consensus Statement 1c

33We recommend using the term postoperative acute kidney disease (PO-AKD) when KDIGO criteria for AKI are present after postoperative day 7 and objective evidence of injury initiation was present within the first 7 postoperative days (ungraded).

34Consistent with the definition of AKD as a condition in which AKI stage 1 or greater is present ≥7 days after an AKI initiating event, postoperative acute kidney disease (PO-AKD) occurs in patients with PO-AKI who continue to meet KDIGO criteria for AKI >7 days after surgery. Patients whose rise in serum creatinine level begins before postoperative day 7 but whose serum creatinine level does not increase ≥50% above baseline until 7–90 days after surgery would also meet PO-AKD criteria based on a slow progressive deterioration in kidney function in the postoperative period (Box 2). However, the greater the time interval between surgery and identification of kidney injury the less likely that the injury could be attributed to the perioperative process. Patients who meet criteria for AKI for the first time after postoperative day 7 without any previous increase in serum creatinine should therefore be referred to as having AKI or AKD without the postoperative prefix. With this proposed nomenclature (Fig. 3), it sought to combine a level of consistency with existing recommendations for use of AKI and AKD terminology,. However, revision may be warranted as our understanding of the pathophysiology and prognostic implications of temporal variations in the postoperative disease course evolves.

35Limited information exists regarding the prognostic implications of the nature and timing of the AKI diagnosis within 7 days after surgery, and these important aspects of the PO-AKI diagnostic criteria warrant better characterization. Furthermore, to what extent indices of kidney injury other than serum creatinine and urine output, such as urinary or plasma biomarkers, are important for prognostication in PO-AKI is unknown and should be the subject of further study. Whether or not important differences exist in the epidemiology and prognostic significance of PO-AKD that arises as persistent PO-AKI compared with PO-AKD that develops after a gradual rise in creatinine after day 7 is also unknown.

36The impact of routine, structured KHA on the detection of PO-AKI remains to be determined. An expanded perioperative urinalysis conducted as part of a detailed KHA might identify previously unknown pathology and could potentially lead to additional, potentially expensive and/or invasive investigations that could delay surgery. The presumed benefits and cost-effectiveness of a KHA therefore require confirmation. Finally, the optimal interpretation of intraoperative and postoperative oliguria remains unclear and should be the subject of further study.

37Surgery is sometimes performed in patients with established AKI; however, common perioperative interventions focus on prevention of PO-AKI (Box 3). As nephrotoxic drugs are an important preventable contributor to the risk of PO-AKI, exposure to these drugs should be limited whenever possible and the benefits of the intervention weighed against the risk of developing or worsening AKI. Although the avoidance of potential nephrotoxins is considered to be a cornerstone of AKI prevention, there is a relative paucity of data supporting this approach. However, routine use of gentamicin for surgical prophylaxis was associated with an increased risk of AKI following orthopaedic surgery and, in the paediatric setting, systematic screening for nephrotoxin use, with a focus on close monitoring in patients exposed to multiple nephrotoxins and nephrotoxin cessation, has been shown to reduce rates of AKI,.

38At present, common practice is to discontinue use of ACE inhibitors and ARBs before surgery to avoid perioperative hypotension. This approach is assumed to reduce the risk of perioperative AKI based on physiological rationale. In line with this established practice, we recommend discontinuing ACE inhibitors and ARBs for a minimum of 24 h before surgery to minimize the risks of perioperative hypotension and/or postoperative AKI (Box 3). This recommendation is weak owing to limited data in support of this practice and is likely to be revised as more evidence becomes available. For instance, an analysis of 949 patients undergoing major gastrointestinal or hepatobiliary surgery failed to show a difference in the rates of AKI between those who did and those who did not have ACE inhibitors and ARBs withheld before surgery.

39NSAIDs are generally avoided in surgical patients at a high risk of AKI, but might be beneficial as opiate-sparing agents in patients who are at a low risk of AKI. Although NSAIDs are widely used in patients with normal kidney function, their impact on risk of AKI in the postoperative period is unclear.

40To date, no preoperative pharmacological intervention has been shown to reduce the risk of PO-AKI in RCTs. However, attention should be paid to ensuring euvolaemia prior to surgery and addressing or avoiding preoperative fluid losses caused by excessive fasting, bowel preparation and/or acute illness.

41Consensus Statement 3a

42We recommend that when considering use of medications or interventions that are associated with an increased risk of acute kidney injury (AKI), the relative benefits and/or necessity of the medications or interventions should be balanced with the risk of postoperative AKI (PO-AKI). This recommendation applies to outpatient medications, new medications and exposures including contrast media (grade C evidence; strong recommendation).

43Consensus Statement 3b

44We recommend discontinuing angiotensin-converting enzyme (ACE) inhibitors and angiotensin receptor blockers (ARBs) for a minimum of 24 h before surgery, dependent on the specific medication, to minimize the risks of perioperative hypotension and/or PO-AKI (grade D evidence, weak recommendation).

45Consensus Statement 3c

46We recommend assessing the relative risks and benefits of use of NSAIDs (including COX2 inhibitors) in individual patients. In general, in patients at a high risk of PO-AKI, the risk of AKI associated with use of NSAIDs might outweigh the benefits with regard to opioid minimization, whereas in low-risk patients, the benefits of use of NSAIDs might outweigh the risks. The risk–benefit profile in patients at a moderate risk of AKI is particularly uncertain (grade D evidence; strong recommendation).

47Intravenous fluid and haemodynamic management in the intraoperative and early postoperative period has a major impact on the development of PO-AKI (Box 4). The RELIEF RCT in 3,000 patients undergoing major elective non-cardiac surgery compared a zero fluid balance target with a “moderately liberal” approach to fluid administration and demonstrated an increased risk of AKI in the restrictive, zero fluid balance group. Algorithms that use cardiac output monitoring to optimize cardiac stroke volume and/or increase global oxygen delivery, known as goal-directed therapy, have been the subject of multiple RCTs. A Cochrane review and meta-analysis suggested that goal-directed therapy can reduce the risk of PO-AKI,. Large observational studies have shown strong associations between intraoperative hypotension and PO-AKI with the risk of organ injury a function of the severity and duration of the hypotension–. Both absolute mean arterial blood pressure (MAP (e.g. 60–70 mmHg)) and relative MAP (e.g. <30% reduction from baseline) thresholds are associated with PO-AKI. The results of an interventional trial of use of a MAP target of >65 mmHg as part of a goal-directed strategy have added weight to these observational data, demonstrating fewer postoperative complications (including AKI) and shorter length of hospital stay with this approach. Similar to other settings such as sepsis, a higher MAP target may be beneficial in reducing the incidence of AKI in patients with poorly controlled pre-existing hypertension.

48Fluid composition may also affect the risk of PO-AKI; 0.9% saline is associated with increased risk of PO-AKI compared with balanced crystalloids– (Box 4). High-quality evidence has shown that use of starch-based colloids increases the risk of AKI and other adverse outcomes in patients with sepsis and other critically ill patients,. Although caution is advised when using starch-based colloids, the limited data in the elective perioperative setting, where lower fluid volumes may be given than in treatment of septic shock, suggest little difference in kidney outcomes when starch-based colloids are used in preference to crystalloids or non-starch colloids,. Ongoing research is investigating use of starch-based colloids versus crystalloids in the perioperative setting. One such trial published in 2020 reported no significant difference in a composite outcome of death or postoperative complications within 14 days of major abdominal surgery between patients who received low-molecular-weight hydroxyethyl starch compared with those who received 0.9% saline.

49Consensus Statement 4a

50We recommend not using restrictive or zero-balance perioperative fluid regimens in major elective surgery (except in specific circumstances). Such regimens are associated with increased risk of postoperative acute kidney injury (PO-AKI) (grade B evidence; strong recommendation).

51Consensus Statement 4b

52We recommend the use of goal-directed haemodynamic therapy in high-risk patients to optimize volume status, blood pressure and cardiac output and reduce the risk of PO-AKI (grade B evidence; strong recommendation).

53Consensus Statement 4c

54We recommend maintaining an intraoperative mean arterial blood pressure (MAP) >65 mmHg to reduce the risk of PO-AKI. The risk of AKI is a function of both the magnitude of hypotension and its duration. For selected patients, including those with pre-existing systemic hypertension, a higher MAP target should be considered (grade C evidence; weak recommendation).

55Consensus Statement 4d

56We recommend the perioperative use of balanced crystalloids rather than 0.9% saline to reduce the risk of PO-AKI (grade C evidence; strong recommendation).

57Consensus Statement 4e

58In the setting of elective surgery outside of the ICU, insufficient evidence exists to make a recommendation regarding the choice of crystalloids or colloids for volume expansion to modify the risk of PO-AKI (ungraded).

59The early postoperative period has not been extensively studied in relation to kidney outcomes. However, therapeutic strategies used in the hours after surgery could likely modify the risk of AKI (Box 5). Clinically important postoperative hypotension (defined as systolic blood pressure <90 mmHg and requiring medical intervention) affects over 30% of patients, often goes uncorrected for longer than intraoperative hypotension and is associated with an increased risk of all-cause mortality and myocardial injury. These findings suggest that maintenance of fluid and haemodynamic status in the early postoperative phase is important to avoid organ injuries including AKI. A trial that used stroke volume optimization and vasopressors to target a MAP within 10% of the preoperative baseline value during and for 4 h after surgery showed that this strategy was associated with a reduction in postoperative organ injury, but was not powered to assess kidney-specific outcomes. Further research is warranted to evaluate postoperative monitoring strategies and ideal clinical settings to reduce the risk of PO-AKI in moderate- to high-risk patients.

60As postoperative hyperglycaemia is strongly associated with AKI, avoidance of perioperative hyperglycaemia (>180 mg/dl) is recommended. Failure to restart chronic ACE inhibitor or ARB therapy that was suspended preoperatively is associated with increased 30-day mortality,. However, the optimal time point to restart these medications has not been clearly established. Given their effects on the kidney, a reasonable approach would be to conduct a focused KHA before restarting ACE inhibitors or ARBs.

61Enhanced recovery after surgery pathways are typically multimodal, multidisciplinary interventions that have been promoted to facilitate recovery from surgery. Use of NSAIDs to minimize opiate analgesia requirements is a common component of such pathways, along with avoidance of markedly positive fluid balance. Accordingly, in some settings such as colorectal surgery, implementation of an enhanced recovery after surgery pathway could be associated with a higher, rather than lower, incidence of AKI.

62Consensus Statement 5a

63We recommend maintenance of adequate organ perfusion, including attention to euvolaemia and treatment of hypotension in the early postoperative period, to reduce the risk of postoperative acute kidney injury (PO-AKI). Monitoring and targeting haemodynamic parameters may require enhanced levels of monitoring and/or care depending on the clinical context and patient risk profile (grade D evidence; strong recommendation).

64Consensus Statement 5b

65We recommend treatment of postoperative hyperglycaemia (target glucose <180 mg/dl [<10 mmol/l]) to reduce the risk of PO-AKI (grade C evidence; strong recommendation).

66Consensus Statement 5c

67We recommend considering restarting angiotensin-converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB) therapy within the first 48 h postoperatively in patients who are haemodynamically stable without evidence of PO-AKI. This decision should be individualized and context-specific (grade C evidence; strong recommendation).

68Consensus Statement 5d

69We recommend that enhanced recovery after surgery (ERAS) pathways (for example, use of NSAIDs and avoiding markedly positive fluid balance) are individualized based on the baseline risk of PO-AKI. The impact of ERAS pathways on the incidence of PO-AKI is unclear owing to the potential for both beneficial and harmful effects (grade C evidence; strong recommendation).

70As a general principal, treatment of AKI should be initiated as early as possible, including in patients with suspected AKI or who are at a substantially increased risk of AKI. Management of PO-AKI involves specific approaches, for example, avoidance of nephrotoxins, as well as general management principles that are common to all postoperative complications, such as haemodynamic optimization. Similarly, treatment of PO-AKI shares many features with the treatment of AKI in other settings (Table 1). Treatment goals include reducing kidney injury and complications related to decreased kidney function.

71A full KHA should be performed, including drug history, which should focus on potential nephrotoxins, including antibiotics and radiocontrast agents. In addition to classical evaluation techniques, a number of biomarkers of cellular damage and functional change are available for the early diagnosis, risk assessment and prognosis of AKI–. Nephrology or urology consultation should be sought if intrinsic causes of AKI or obstruction are suspected based on the KHA results. The frequency and duration of monitoring should be individualized based on patient risk, exposure and clinical course (Box 6).

72Consensus Statement 6a

73We recommend that the underlying causes of postoperative acute kidney injury (PO-AKI) should be determined and treated whenever possible (ungraded; strong recommendation).

74Consensus Statement 6b

75In patients with PO-AKI, we recommend that criteria for implementation of kidney replacement therapy (KRT) should be based on directly anticipated or observed life-threatening AKI-related complications (grade D evidence; strong recommendation).

76Consensus Statement 6c

77We suggest that AKI phenotype is taken into account when deciding what type of treatment to implement (ungraded; weak recommendation).

78The treatment and prevention bundle outlined in the 2012 KDIGO clinical practice guideline for AKI consists of supportive measures, including volume management, maintenance of adequate blood pressure and judicious avoidance of nephrotoxins. Two RCTs demonstrated that implementation of this bundle in high-risk patients identified by biomarkers after surgery significantly reduced the occurrence of PO-AKI, but were not powered to demonstrate differences in longer-term patient outcomes,. However, observational studies have shown that the severity of AKI correlates with short-term– and longer-term adverse events, suggesting that interventions that reduce the severity of AKI could improve these outcomes.

79Kidney replacement therapy (KRT) may be required in patients with severe AKI who develop medically refractory fluid and/or metabolic complications. In the absence of conventional indications, the best timing for initiating KRT in PO-AKI is uncertain. Potential rationales for initiating KRT during PO-AKI are avoidance of fluid accumulation, maintenance of acid-base and electrolyte control, provision of space to accommodate nutritional and medication needs, and abrogation of the interaction between the kidney and other organs. Theoretically, early initiation of KRT could facilitate better fluid, electrolyte and acid–base homeostasis. However, KRT is associated with increased risk of complications, healthcare costs and clinical workload.

80The 2012 KDIGO guidelines recommend starting KRT “when life threatening changes in fluid, electrolyte, and acid-base balance exist” and considering the “broader clinical context, the presence of conditions that can be modified with [KRT], and trends of laboratory tests — rather than single BUN and creatinine thresholds alone — when making the decision to start [KRT]”. Contradictory trial results, have stimulated debate regarding the best timing for KRT in patients with established AKI and differences in study design, patient populations, definitions of early and late RRT initiation, and inclusion criteria make head-to-head comparisons difficult.

81Following our consensus conference, the results of the STARRT-AKI study provided high-quality evidence that early commencement of KRT based on AKI criteria in the absence of conventional indications does not improve survival, may needlessly expose some patients to KRT and results in a higher risk of KRT dependence at 90 days compared with a conservative strategy for starting KRT. Importantly, STARRT-AKI included a large cohort of patients with PO-AKI (n = 965) and showed no evidence of a benefit of early commencement of KRT in this subgroup (odds ratio for mortality with earlier KRT: 1.20; 95% CI, 0.91–1.59). Thus, decisions regarding timing of KRT initiation remain a complex issue that should be individualized based on the clinical state of the patient, in particular imminently anticipated or observed life-threatening complications of AKI.

82AKI is a heterogeneous condition consisting of distinct endotypes and phenotypes based on its aetiology, prognosis and molecular pathways. Different subtypes might require and benefit from different therapeutic strategies. However, current AKI definitions based on serum creatinine and urine output provide no information on AKI subtypes. Studies using various damage and functional AKI biomarkers suggest that these new tools might overcome the limitations of current AKI definitions to improve AKI phenotyping. Until these strategies are better embedded in clinical practice, the potential benefits of individualized therapy for AKI subtypes cannot be clearly established.

83We recommend further trials to assess the optimal choice and volume of fluids for intravascular expansion and vasopressor therapy in patients with PO-AKI as well as the use of protocol-based management of haemodynamic and oxygenation parameters to prevent worsening of PO-AKI. Investigation of individual components of the KDIGO guideline to identify those that are the most effective in treating PO-AKI is important given that these were largely assembled based on expert opinion. Better understanding of AKI phenotypes is required to enable further trials to assess the response to implementation of the KDIGO bundle and other treatment options according to different AKI endotypes and phenotypes.

84Regardless of the contributing mechanisms, PO-AKI has repeatedly been shown to be associated with a high risk of complications, including short-term and long-term mortality, high hospital costs and resource utilization,,,,,– (Fig. 4). Data from patients admitted to an ICU for at least 24 h following major surgery show that those who experienced PO-AKI had persistently worse survival over 10 years, even if kidney function recovered completely. By contrast, a retrospective analysis of patients undergoing repair of ruptured abdominal aneurysm, a very high risk of acute exposure, suggested long-term survival was not significantly different between those with and without AKI, perhaps as a result of the overwhelming influence of immediate surgical outcome in this setting.

85Patients with PO-AKI or AKD are at an increased risk of recurrent episodes of AKI and progressive deterioration of kidney function, including CKD and dialysis dependency. In patients undergoing major surgery, the incidence of kidney failure at 1 year was 0.94% among those who experienced AKI versus 0.05% in those who did not experience AKI. Furthermore in a competing risk model accounting for death, adjusted progression to kidney failure during 10 years of follow-up was 0.4%, 2.3%, 7.3% and 15.7% for patients with no kidney disease, AKI with no CKD, CKD with no AKI and AKI with CKD, respectively (P < 0.001).

86Other postoperative complications, including infection, prolonged mechanical ventilation, tracheostomy and cardiovascular events are also more common in patients who develop AKI after surgery than in those who do not. Among 1,200 patients who underwent non-cardiac and non-vascular surgery, of whom 6.7% met the RIFLE criteria for AKI (2), the risk of non-cardiovascular complications (such as pneumonia and stroke) was similar in those with and without AKI over the first 2 weeks after surgery, but thereafter, patients with AKI experienced a significantly higher rate of cardiovascular events, including acute coronary syndrome, acute heart failure and arrhythmias. Similarly, AKI after abdominal surgery has been associated with an increased risk of in-hospital complications, including perioperative acute myocardial infarction, pneumonia and sepsis (P < 0.001).

87Unsurprisingly, the development of PO-AKI is associated with substantial increases in hospital costs and resource utilization, as well as longer stays in the ICU and in hospital,,. Furthermore, the risks of 30-day re-admission and discharge to a nursing facility or rehabilitation centre are increased. PO-AKI has also been associated with important, but difficult to quantify adverse outcomes, including reduced quality of life and perceived well-being as well as with increased risk of malignancy, possibly due to interactions with the immune system.

88Given the increased risk of short- and long-term adverse outcomes following PO-AKI, appropriate monitoring and follow up of this cohort is a potential approach to improving patient outcomes. Although follow-up schemas for all-cause AKI have been proposed, the optimal timing of follow-up after a postoperative episode of AKI or AKD is unknown, and limited evidence is available to support any specific monitoring paradigm. However, adverse outcomes are increased in patients with prolonged, more severe AKI, and in those whose kidney function does not recover to baseline levels, suggesting that use of these factors to inform the initial frequency and nature of follow-up would be a reasonable approach,.

89Monitoring pathways should be developed in conjunction with nephrologists, but may be directly provided by perioperative physicians, primary care physicians or other specialists (Fig. 5). As a minimum, we recommend that all postoperative patients with AKI or AKD have a KHA within 30 days of hospital discharge. To facilitate this assessment, accurate information transfer from the hospital team to the primary care team and the patient or care giver is essential. At the very least, a post PO-AKI KHA should consist of checking glomerular filtration (e.g. serum creatinine levels) as well as other persistent signs of kidney injury (e.g. albuminuria). Screening for albuminuria 3 months after the development of AKI has been shown to identify those patients who are at a greater risk of CKD progression. When available, further work-up and monitoring of kidney function with other diagnostics may be appropriate. Changes in management based on this monitoring should be communicated with the patient as well as with other treating physicians.

90In survivors of PO-AKI or AKD, the key priorities of long-term management are to facilitate kidney recovery, prevent progressive CKD and mitigate long-term comorbidities, including cardiovascular risk. However, whether any effective strategies beyond supportive CKD care exist is unknown. ADQI recommends that a post-AKI or AKD care bundle should include assessment of kidney function, advocacy and education, review and adjustment of medications, blood pressure management and a sick day protocol. A multi-disciplinary follow-up approach (involving nephrology, primary care, pharmacy, a dietician, a social worker and non-nephrology subspecialists) combined with patient education is recommended but has not been formally studied. Undoubtedly, drug selection, dosing and monitoring are important and should be guided by personalized clinical decision making and regular assessment of kidney function. Both the use of nephrotoxic drugs and the inappropriate exclusion of potentially beneficial medications (for example, ACE inhibitors or antidiabetic medications) need to be avoided,. In patients with AKD, in particular, a substantial risk exists of potential therapeutic failure caused by underdosing or avoidance of effective cardiovascular drugs. Close collaboration between the relevant clinical teams is paramount and future research efforts should clarify the ideal timing and method for providing post-AKI and AKD care (Box 7).

91Consensus statement 7a

92The incidence of kidney and non-kidney adverse events is increased following postoperative acute kidney injury (PO-AKI) or postoperative acute kidney disease (PO-AKD). In particular, patients with surgery-associated AKI or AKD are at an increased risk of new or worsening chronic kidney disease (CKD), which is associated with increased long-term mortality (grade B evidence).

93Consensus Statement 7b

94The prevention of new or worsening CKD following an episode PO-AKI or PO-AKD is a public health priority. We recommend that all patients undergo at least one kidney health assessment once the acute phase of AKI or AKD is complete. This approach will require optimal communication and AKI education between health-care professionals and the patient or caregiver (grade B evidence; strong recommendation).

95A need exists for epidemiological research to describe the frequency, nature and severity of PO-AKI and PO-AKD progression to CKD. The effectiveness and cost-effectiveness of supplementing a simple KHA with enhanced kidney health care in the prevention of CKD progression after surgery-associated AKI or AKD should be evaluated in appropriately selected patients. Finally, research is needed to explore a possible common aetiology between surgery-associated AKI or AKD and other chronic organ injuries as the longer-term effects of AKI do not only involve the kidney.

96PO-AKI is a sentinel postoperative event that is strongly associated with both short-term surgical complications and long-term adverse outcomes. In this Consensus Statement, we provide an overview of PO-AKI and management. However, PO-AKI is a heterogeneous syndrome with a variable clinical course. In many areas the strength of evidence is poor and accordingly our recommendations are weak. We provide research recommendations for these areas to enable greater clarity to be achieved in the future. Overall, we believe that an integrated approach to PO-AKI management requires serial evaluations of AKI risk, diagnosis, progress and outcomes throughout the perioperative period, which can be constructed as a series of focused KHAs.