The KDIGO acute kidney injury guidelines for cardiac surgery patients in critical care: a validation study

Acute Kidney Injury (AKI) occurs in up to 50% of patients following cardiac surgery [1, 2]. Even in its mildest form, AKI is associated with increased mortality and prolonged Critical Care Unit stay [3‐6]. AKI requiring renal replacement therapy (RRT) occurs in 2–5% of patients following cardiac surgery and is associated with mortality of up to 60% [1, 7, 8]. The Kidney Disease: Improving Global Outcomes (KDIGO) AKI guidelines were designed to standardise the criteria for AKI based on serum creatinine and urine output (Table 1) [9]. Patients are assigned the same stage of AKI regardless of which criteria (urine output, serum creatinine or both) for that stage are met. However, concerns have been raised that the guidelines’ urine output criteria are poorly calibrated [10].

Studies validating the KDIGO guidelines following cardiac surgery have frequently stratified patient risk based on serum creatinine alone as reliable urine output data are difficult to collect [1, 11‐13]. Studies that have had access to urine output data have tended to be relatively small and have disagreed on the importance of urine output when identifying those at risk of adverse outcomes. [2, 14, 15]. A recent study with access to urine output data after cardiac surgery did demonstrate that patients with AKI diagnosed on oliguria alone had increased long-term mortality but due to the relatively small sample size they were unable to assess the importance of urine output within each AKI level [6].

The objective of this study was to validate the KDIGO guidelines for AKI by assessing the outcomes of patients meeting different criteria for each stage of AKI after cardiac surgery.

Data from 2284 patients were available. Seven patients who required RRT preoperatively and ten patients with no preoperative creatinine values were excluded leaving 2267 patients for the analysis (Fig. 1). Patient characteristics are shown in Table 2. There were 1448 patients who did not develop AKI during their CICU stay. A total of 819 (36.1%) developed AKI and 147 (6.5%) required RRT. There were 177 (7.8%) patients who died within two years of surgery. Of the 1448 patients who did not develop AKI, 255 (17.7%) had a prolonged LOS on CICU and the 2-year mortality rate for this group was 3.9%.

This represents the first study to explore the associations between different criteria for multiple stages of KDIGO-defined AKI and morbidity and mortality after cardiac surgery. The frequency of AKI in our population was broadly similar to that previously reported in cardiac surgery patients [1, 19, 20]. RRT was more frequent than reported elsewhere [1, 6‐8, 14] possibly as a substantial number of high risk tertiary referral, transplant and mechanical circulatory support patients are referred to our centre.

Univariable analyses identified stepwise increases in the risk of prolonged LOS, RRT and 2-year mortality within each stage of AKI where patients met urine output criteria, creatinine criteria or both. The increases in risk of poor outcomes were confirmed on multivariable analysis. This trend was previously reported in general ICU patients [21] and in cardiac surgery patients [6] although in the latter study the differences were not statistically significant.

In this study, when comparing outcomes for patients diagnosed with the same stage of AKI, all patients who met any criteria for that stage were included. This approach was chosen to improve the clinical relevance of our findings. In practice, clinicians cannot know whether a patient with AKI-1 will subsequently develop AKI-2 until the higher stage is diagnosed or the patient has left CICU without meeting AKI-2 criteria. As a consequence of this approach, some patients appear in both the AKI-1 and AKI-2 analyses. However, for comparison of survival outcomes between patients suffering different stages of AKI, each patient was placed retrospectively into a subgroup according to the maximum AKI stage reached during their CICU stay to avoid double counting.

The KDIGO AKI guidelines were developed using expert opinion at a conference of the Kidney Disease: Improving Global Outcomes group [9]. The guidelines drew on previous diagnostic criteria such as AKIN [22] and RIFLE [23] using urine output and creatinine to grade severity of renal dysfunction. The guidelines referenced several studies which demonstrated that patients with increasing stages of AKI had greater subsequent need for RRT and increased mortality risk. These findings were used to support the stratification of AKI within the KDIGO guidelines. However, none of the validation studies cited assessed the risk associated with AKI diagnosed by hourly urine output measurement [24‐30]. The authors acknowledged that the urine output thresholds were less well substantiated than those related to serum creatinine concentration. Indeed, they noted that “the influence of urinary output criteria on AKI staging needs to be further investigated” and these concerns have been echoed in subsequent work validating the guidelines in their entirety [10]. The calibration of urine output thresholds might be particularly poor for cohorts such as those who have undergone major surgery for whom oliguria may be an appropriate response to surgery.

Many studies assessing the KDIGO guidelines in cardiac surgery patients used creatinine criteria alone to classify AKI and explore its association with subsequent adverse outcomes [12, 20]. Those which have included urine output disagreed on whether UO diagnosed by AKI was associated with adverse outcomes [2, 6, 14]. The current study contains more than three times the number of participants as the largest similar study into this topic in cardiac surgery patients. It is also the first to explore differences between with rates of adverse outcomes in those who meet different criteria within the same AKI stage for both AKI-1 and AKI-2.

Our findings support conclusions drawn from earlier studies that the risk of adverse outcomes associated with AKI diagnosed by UO alone is relatively low [2, 14]. In many cases the oliguria which resulted in diagnoses of AKI-1 and AKI-2 by urine output alone was almost certainly an appropriate physiological response to the stress of surgery. It is understandable that physiologically appropriate oliguria was associated with a prolonged LOS on CICU, but equally it is unsurprising that such oliguria was not associated with significantly increased 2-year mortality rates. In contrast, the risk of adverse outcomes in patients who met both urine output and creatinine criteria for AKI-1 or AKI-2 was markedly higher. Patients who suffered AKI-1 meeting both criteria had worse 2-year survival than those who suffered AKI-2 by urine output alone. This is highly relevant to treating clinicians; in these instances an AKI-1 diagnosis may be falsely reassuring. To ensure that the direct relationship between increasing AKI stages and risk of adverse outcomes is maintained, the AKI classification criteria may need to be adjusted to reduce the importance placed on isolated oliguria and increase the risk attributed to fulfilling both urine output and creatinine criteria. While we have studied patients undergoing cardiac surgery, similar findings may be reproduced in other cohorts, particularly those undergoing other types of major surgery and this should be the focus of further investigation.

The current KDIGO guidelines assign the same AKI stage to patients with markedly different risks of adverse outcomes after cardiac surgery. This study identified a stepwise increase in the frequency of poor outcomes following cardiac surgery where urine output criteria, creatinine criteria or both criteria for each AKI stage were met. Moreover, 2-year mortality for some subgroups within AKI-1 clearly overlapped that of some subgroups within AKI-2. The KDIGO criteria need to be revisited before they can be used to reliably stratify the severity of AKI in cardiac surgery patients on the critical care unit. The applicability of these criteria to other patient groups within the critical care unit, particularly those in whom oliguria may be an appropriate physiological response also requires further evaluation.