Written by Jasna Trbojevic-Stankovic
Reviewed by Wim Van Biesen
Is acute kidney injury the direct cause or merely a correlate of poor outcome in critically ill patients?
Critically ill patients who develop acute kidney injury (AKI) exhibit a notably higher mortality rate. Several clinical conditions, such as sepsis, nephrotoxins and hypotension, may cause AKI eventually resulting in fatal outcome. In such cases, it might seem that simply by managing AKI it would be possible to prevent or delay mortality. In practice, however, this is not always the case and AKI may sometimes simply be a confounding factor and not the direct cause of mortality. In such circumstances, even the optimum AKI treatment would not change the final outcome. Finally, there are situations when a combination of these mechanisms is present. Thus, the importance of renal replacement therapy (RRT) for the outcome varies from necessary to fundamental, depending on the interrelationship between AKI and mortality in each particular case.
The extent to which AKI might be related to adverse short-term outcomes of critical illness has been recently investigated using the Bradford Hill criteria for causality. This study concluded that AKI is associated with substantially increased mortality, and that association is graded and persists after accounting for known confounders (1). However, this gradient is not linear. Instead, it is U-shaped, implying that both low and high serum creatinine, and low and high urine output, have an equally important effect on the outcome (2, 3).
Even though it might seem that RRT has a crucial impact on the outcome, several studies found otherwise. In the meta-analysis of the prospective multicenter observational FINNAKI study it was concluded that high propensity to receive RRT does not aleviate mortality risk (4). Conversely, witholding RRT is not necessarily associated with worse outcome, provided that urine output is preserved and creatinine is stable, regardless its level (5). These data support the impression that there is no direct causal relationship between AKI and mortality, unless certain life threatening AKI-related factors, such as hyperkalemia, anuria, extreme acidosis or intoxication, are present. In such cases RRT may have a fundamental impact on the outcome.
When is the right time to start dialysis?
Defining the optimal time to initiate RRT for AKI does not seem to be an easy task. This is especially true for critically ill patients with AKI but no potentially life-threatening complications directly related to renal failure. The ELAIN trial concluded that early compared to delayed initiation of RRT reduces mortality over the first 90 days (6). On the other hand, the AKIKI multicenter randomized trial on patients with severe AKI and requiring mechanical ventilation, catecholamine infusion, or both, and not having a potentially life-threatening complication directly related to renal failure, no significant difference was found with regard to mortality between an early and a delayed strategy for the initiation of RRT (7). Similar results were obtained in patients with AKI and sepsis in the IDEAL-ICU trial and very recently the STARRT-AKI trial (8, 18).
Such conflicting results call for further and more detailed evaulation of the methodology. These studies differ substantialy in design, number of subjects, inclusion criteria, level of kidney injury at the onset of follow-up, and comorbidities (Figure 1). Namely, in the ELAIN trial KDIGO renal failure stage 3 was the indication to start RRT, while in the other two studies the same level of renal failure was merely the baseline. This reflected in the number of days free of RRT and prevalence of patients who avoided RRT, both of which were higher in the AKIKI and IDEAL-ICU studies than in the ELAIN trial (6, 7, 8, 9).