Renal outcomes according to renal replacement therapy modality and treatment protocol in the ATN and RENAL trials

Abstract

Background: In critically ill patients with acute kidney injury, renal replacement therapy (RRT) modality and treatment protocols may affect kidney recovery. This study explored whether RRT modality and treatment protocol affected RRT dependence in the 'Randomized Evaluation of Normal versus Augmented Level of RRT' and the 'Acute Renal Failure Trial Network' (ATN) trials.

Methods: Primary outcome was 28-day RRT dependence. Secondary outcomes included RRT dependence among survivors and in different SOFA-based treatment protocol groups. We used the Fine-Gray competing-risk model sub-distribution hazard ratio (SHR) to assess the primary outcome. Analyses were adjusted for confounders.

Results: Of 2542 patients, 2175 (85.5%) received continuous RRT (CRRT) and 367 (14.4%) received intermittent hemodialysis (IHD) as first RRT modality. CRRT-first patients had greater illness severity. After adjustment, there was no between-group difference in 28-day RRT dependence (SHR, 0.96 [95% CI 0.84-1.10]; p = 0.570) or hospital mortality (odds ratio [OR], 1.14 [95% CI 0.86-1.52]; p = 0.361) However, among survivors, CRRT-first was associated with decreased 28-day RRT dependence (OR, 0.54 [95% CI 0.37-0.80]; p = 0.002) and more RRT-free days (common OR: 1.38 [95% CI 1.11-1.71]). Moreover, among CRRT-first patient, the ATN treatment protocol was associated with fewer RRT-free days, greater mortality, and a fourfold increase in RRT dependence at day 28.

Conclusions: There was no difference in RRT dependence at day 28 between IHD and CRRT. However, among survivors and after adjustment, both IHD-first and the ATN treatment protocol were strongly associated with greater risk of RRT dependence at 28 days after randomization. Trial registration NCT00221013 registered September 22, 2005, and NCT00076219 registered January 19, 2004.

Keywords: Acute kidney injury; Continuous renal replacement therapy; Dialysis dependence; Intermittent hemodialysis; Mortality.

Fig. 1

Cumulative Incidence Plot of Renal Replacement Therapy Independence at Day 28 and 60 before and after Matching. Panel A, renal replacement therapy independence at day 28 in the cohort before the covariate-balancing propensity score matching. Panel B, renal replacement therapy independence at day 60 in the cohort before the covariate-balancing propensity score matching. Adjusted models in panel A and B included for age, gender, weight, type of admission (medical, surgical or other), APACHE III, cardiovascular SOFA, hours between randomization and therapy, use of mechanical ventilation, presence of oliguria, presence of hyperkalemia, presence of sepsis, last bicarbonate, urea and creatinine before randomization, premorbid estimated glomerular filtration rate, and intensity of treatment (as allocated in the original trials). SHR denotes sub-distribution hazard ratio and CI confidence interval

Fig. 2

Forest Plot for the Analyses of the Primary Outcome. Multivariable analysis in the primary analysis is adjusted for age, gender, weight, type of admission (medical, surgical or other), APACHE III, cardiovascular SOFA, hours between randomization and therapy, use of mechanical ventilation, presence of oliguria, presence of hyperkalemia, presence of sepsis, last bicarbonate, urea and creatinine before randomization, premorbid estimated glomerular filtration rate, and intensity of treatment (as allocated in the original trials). IPTW inverse probability of treatment weighting, SIPTW stabilized inverse probability of treatment weighting, CI confidence interval, SHR sub-distribution hazard ratio, and HR hazard ratio

Fig. 3

Renal Replacement Therapy-Free Days at Day 28. Renal replacement therapy-free days at day 28 as horizontally stacked proportions according to study group. Red represents worse outcomes, and blue represents better outcomes. COR is common odds ratio

Fig. 4

Effect of Continuous Renal Replacement Therapy on Renal Replacement Therapy Dependence at Day 28 Compared to Intermittent Hemodialysis According to the Different Cohorts Assessed. Panel A, RRT dependence at day 28 considering death as a competing risk. Panel B, RRT dependence at day 28 among survivors at the longest follow-up only. Initial mode considered the mode used after randomization. Mode for the first three days considered patients receiving CRRT or IHD exclusively in the first three days of follow-up (patients who died or do not have information of modality in the first three days were excluded from this analysis). Exclusive mode for the follow-up considered patients receiving CRRT or IHD exclusively during the whole follow-up available for each patient (including those who died early). Adjusted models included for age, gender, weight, type of admission (medical, surgical or other), APACHE III, cardiovascular SOFA, hours between randomization and therapy, use of mechanical ventilation, presence of oliguria, presence of hyperkalemia, presence of sepsis, last bicarbonate, urea and creatinine before randomization, premorbid estimated glomerular filtration rate, and intensity of treatment (as allocated in the original trials)