Starting and Stopping Renal Replacement Therapy in the Critically Ill

Predictors for Dialysis Requirement

When renal replacement or renal support is required, there is a far worse prognosis than with lesser degrees of renal injury. There is little consensus among nephrologists and intensivists regarding indications for starting renal replacement therapy (RRT) in a specific patient. In a survey by Ricci et al., 560 attendees of a critical care nephrology conference listed 90 combinations of indications for initiating treatment. Oliguria/anuria was the most frequent choice, at only 27%. This is less surprising when one considers that acute kidney injury (AKI) has multiple causes and clinical manifestations. The unique combination of factors in a particular case should dictate the timing and type of therapy offered. Although the balance between risks and benefits of invasive procedures must be considered carefully in the individual patient, evidence in favor of early intervention has grown steadily. In 1998 Bellomo and Ronco proposed 12 indications for initiating dialysis in critically ill patients. In their schema, combinations of two or more indications make initiation of RRT “urgent and mandatory.” The indications for RRT replacement and RRT support are presented in Boxes 144.1 and 144.2 .

Hyperkalemia, severe hyperphosphatemia, severe hyperuricemia, severe acidemia, and uremia-related complications (coma, pericarditis, seizures) are accepted indications for starting dialysis. However, there is wide variability regarding the timing of initiation of dialysis, even when these indications are present. Aside from situations in which there are severe derangements, most nephrologists have a tendency to avoid dialysis for as long as possible. Two major factors contribute to the decision to delay dialysis. First, the dialysis procedure is not without risk. Hypotension, arrhythmias, and complications of vascular access placement are not uncommon. Second is the concern that dialysis may delay recovery of renal function. Therefore in general, dialysis in current practice is initiated when clinical features of significant volume overload and solute imbalance dictate a need for intervention. Common parameters used to define the indication and timing of dialysis in patients with AKI include the levels of blood urea nitrogen (BUN) and creatinine, presence of oliguria, evidence of heart failure and pulmonary edema, and an estimate of the catabolic state.

Neither laboratory nor clinical data alone seem to predict when dialysis should be initiated. The combination provides the basis for the decision-making process in initiating therapy with dialysis. AKI involves a complex physiologic milieu that requires early aggressive collaborative management to provide appropriate therapy in a timely manner. Careful clinical assessment of the patient, review of laboratory parameters for trends, and knowledge of anticipated events are key to the appropriate management of these intensive care unit (ICU) patients. The following parameters should be considered.

Blood Urea Nitrogen

BUN generally is considered to be nontoxic, except for its impact on platelet function and, rarely, when initiating dialysis in the face of elevated intracranial pressures. However, many observational and retrospective studies have shown improved survival for patients who start dialysis at lower levels of BUN. In one such study among 100 posttraumatic AKI patients, investigators showed higher survival among those who were initiated on dialysis when BUN was less than 60 mg/dL as compared with those who were started on dialysis when BUN exceeded 60 mg/dL (39% vs. 20.3%; p = .041). Among the critically ill, a study of 243 patients with AKI, Liu et al. found that 60-day mortality was significantly greater when dialysis was started with BUN levels of at least 76 mg/dL, despite an apparently lower burden of comorbidities than those patients with lower BUN (relative risk [RR] 1.85; 95% CI 1.16–2.96). In contrast, a randomized controlled trial conducted in oliguric critically ill patients in Europe revealed no significant difference in hospital mortality with early (mean BUN, 46 mg/dL) versus late (105 mg/dL) initiation of hemofiltration. These findings have been disputed because they did not account for nonrenal dependent processes that influence BUN. Volume depletion, hyperalimentation, gastrointestinal bleeding, and exogenous glucocorticoids raise BUN and are seen commonly in AKI patients in critical care settings. Further, the low-BUN groups may have included patients who would have improved without dialysis. Similarly, in a multicenter retrospective study among 1847 ICU patients with AKI, no correlation was found between serum urea levels at the time of dialysis and mortality. Recently, in an online survey presented to 172 nephrologists, Thakar et al. showed that proportion of physicians considering early dialysis at a BUN level of 75 mg/dL or less increased from 17% to 30% to 40% with an increased severity of patient illness. In conclusion, BUN concentration is influenced by volume of distribution, production, and catabolic rate and tubular reabsorption, and one should be prudent to consider the level of BUN carefully in the decision to initiate dialysis.