Outcomes of septic medical intensive care unit patients who develop renal failure requiring renal replacement therapy

Acute kidney injury (AKI) is prevalent in the critically ill with an incidence of greater than 50%. ¹ Despite advances, the outcomes of patients requiring acute renal replacement therapy (RRT) are poor, often with 60- and 90-day mortality rates of greater than 50%. ² Sepsis-associated acute kidney injury (SA-AKI) is distinct from other causes of AKI and is associated with worse outcomes than non-sepsis-associated renal failure.^3,4While SA-AKI is the leading cause of renal failure in the intensive care unit (ICU), responsible for approximately 20% of AKI, characteristics and outcomes of septic patients requiring acute RRT are not known.

The goal of this study was to retrospectively analyze patients with sepsis who required RRT and compare outcomes with patients with sepsis who did not require RRT at Hahnemann University Hospital in Philadelphia, Pennsylvania between January 2013 and September 2017. Inclusion criteria for the RRT-sepsis group included adults with AKI as defined by Kidney Disease Improving Global Outcomes (KDIGO) guidelines requiring initiation of RRT. Any patient who had received RRT previously was excluded.

One hundred subjects in each arm were included. Baseline differences between the RRT group and non-RRT group include an average age of 59.4 vs. 63.3 (p = 0.08), percent male 57 vs. 52 (p = 0.47), APACHE II score of 22.2 vs. 19.5 (p = 0.02), and a mean SOFA score 10.2 vs. 7.2 (p<0.01). Differences in chronic conditions between the RRT and non-RRT groups included chronic kidney disease in 26% vs. 19% (p = 0.24), hypertension in 57% vs. 55% (p = 0.78), chronic liver disease in 25% vs. 7% (p<0.01), and congestive heart failure in 23% vs. 12% (p = 0.04). For the groups combined, the most common site of infection was the blood stream (33.5%), followed by the lung (32.5%), genitourinary (22.5%), and gastrointestinal tract (13.5).

A Kaplan-Meier curve displays the unadjusted association between RRT groups and survival time (Figure 1). No difference in mortality was seen in the first 10 days, after which there was a sharp separation. This is demonstrated by the hazard ratios and 95% confidence interval for mortality, which in the RRT group was 1.017 (0.549, 1.884) for the first 10 days compared with the non-RRT group, and 8.547 (4.465, 16.361) for the remaining 80 days. Overall 90-day mortality was 23% in the non-RRT group vs. 68% in the RRT group.

Figure 1.

Survival probability.

The divergence of mortality after day 10 affirms the concept of persistent critical illness. The largest study documenting this effect was by Iwashyna et al. in 2016, where a large cohort of critically ill patients was shown after exactly 10 days in the ICU admitting diagnoses failed to predict outcomes more accurately than antecedent patient characteristics such as age, sex, or chronic health status. ⁵ This appears to be also true in this subset of septic patients, where the need for RRT was not associated with worse outcomes in the acute phase of the illness, but increased the likelihood of persistent critical illness and subsequent mortality. The higher APACHE II score and elevated levels of chronic illness in the RRT group also likely contributed to worse outcomes, but again no difference was noted until after day 10, demonstrating that we are improving at handling acute illness, while the chronic aspect remains a challenge.