Discontinuation of Continuous Kidney Replacement Therapy in Patients with Acute Kidney Injury: A pilot study

Abstract

Background

Data on discontinuing continuous kidney replacement therapy (CKRT) in acute kidney injury (AKI) remain limited. We investigated whether a standardized approach improved successful CKRT discontinuation.

Methods

This was a pilot prospective study of patients with AKI requiring CKRT in the intensive care unit between 7/2021-3/2024. In the intervention arm, a standardized form was completed daily to guide discontinuation of CKRT. Successful discontinuation was defined as being free from kidney replacement therapy for seven consecutive days. A 2-hour creatinine clearance and cystatin C were done at the time of CKRT discontinuation in both arms.

Results

We screened 373 patients, 32 were included in the standard care and 35 in the intervention arm after satisfying the inclusion criteria. Patients in the intervention arm were more likely to be older (65 vs. 57 years) and require more norepinephrine (0 vs. 0.11 mcg/kg/min) on the first day of CKRT compared to standard care. Among survivors at day 7 after CKRT discontinuation, successful discontinuation occurred in 16/23 (69.6%) in the standard care arm versus 19/26 (73.1%) in the intervention arm. When discontinuation was encouraged, CKRT was stopped in 26/35 (74%); when not encouraged, CKRT continued in 53/59 (90%). Performances of cystatin C alone or in combination with 2-hour creatinine clearance in predicting successful discontinuation was fair. The best performance was observed in the intervention cohort, AUROC of 0.8 (95% CI, 0.57-0.99).

Conclusion

A standardized approach did not improve the rate of successful CKRT discontinuation. Future studies are needed before biomarkers are used in decision-making.

Introduction

Acute kidney injury (AKI) in critically ill patients is common and carries unfavorable prognosis. The timing of kidney replacement therapy (KRT) initiation has been the focus of many investigations [1–4]. For most patients without an urgent clinical indication, a delayed strategy might be preferable [5]. The decision of when to discontinue KRT in patients with AKI remains contentious but far less studied. The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines suggest to “discontinue KRT when it is no longer required, either because intrinsic kidney function has recovered, or because KRT is no longer consistent with the goals of care” [6]. Kelly et al. proposed a set of criteria that included clinical criteria, the ability to manage volume status, acidemia and hyperkalemia without dialysis, timed urine creatinine clearance exceeding 15 ml/min, and spontaneous urine output of exceeding 400 ml/day or more than 2000 ml/day with diuretics [7]. The incorporation of such objective criteria can help clinicians identify the timing of liberating patients from CKRT.

The concepts of de-escalation and de-resuscitation have revolutionized medical care during critical illness. Fluid administration is one of the most common therapeutic interventions in critical care, and it carries high potential for harm, as overzealous fluid resuscitation may lead to glycocalyx degradation and endothelial injury[8]. Hence, the concept of de-resuscitation was established to encourage healing, including mechanical ventilator weaning or limiting the negative impact of fluid accumulation on organ function. While mechanical ventilators is a lifesaving intervention, prolonged dependence on mechanical ventilation can lead to ventilator-associated pneumonia, barotrauma, and lung injury[9]. Therefore, patients with respiratory failure requiring invasive mechanical ventilation undergo spontaneous breathing trials to assess their readiness for extubation. Similarly, prolonged KRT can paradoxically lengthen the time for which patients remain dialysis-dependent and can increase the risk of adverse events from such intervention [7, 10]. Thus, in this study, we evaluate the impact of a standardized approach to discontinue CKRT in patients with AKI.

Methods

Study Design

We conducted a prospective quasi-experimental (before-after) study of patients with AKI requiring CKRT in the ICU. A quasi-experimental study design was employed due to the nature of the intervention and the processes of care that led to the clinical decision. Patients recruited during the first half of the study were in the standard care arm, where the usual care process was applied. During the second half of the intervention, CKRT discontinuation was guided by an a priori-designed set of criteria. A decision tree form was completed by one of the team members each day for patients during the intervention period. This decision aid was designed to enhance adherence to the protocol and facilitate the tracking of reasons for deviations. The decision criteria were based on the patient’s hemodynamic stability each day, non-oliguric status and whether patients are expected to have a succesfull medical management of fluid and electrolyte derangements, (Supplement 1).

A randomized clinical trial was not feasible at our center, as the intervention’s effect would be diluted if the same physician or team were concurrently managing two patients who would have been otherwise randomized to different arms, leading to a Hawthorne effect [11].

Participants

Patients with AKI who were started on CKRT were screened daily. Patients were approached within 24 to 36 hours of CKRT beginning. Informed consent was obtained from all participants prior to their inclusion in the study. The protocol was approved by the IRB at West Virginia University (protocol: 2101226767). The study was registered on Clinicaltrials.gov, NCT04898595. CONSORT statement was followed for reporting results.[12]

Patients were eligible to participate if they were adults (≥18 years old), diagnosed with AKI, and the treating nephrology team had already decided to start CKRT. Patients were excluded if they were diagnosed with chronic kidney disease stage 5 or end-stage kidney disease, received a kidney transplant in the past, were moribund (expected to die within 1 day of CKRT initiation), presence of an intoxication requiring extracorporeal removal, KRT within the previous one month, or presence of suspiscion of rapidly progressive glomerulonephritis. Due to challenges with COVID19 isolation policies, an amendment was made and approved by IRB to allow obtaining the informed consent from the patient’s surrogate (or power of attorney) via telephone if they were not available in the room. A written signed informed consent was subsequently obtained via mail.

Study procedure and endpoints

We defined successful discontinuation of CKRT as being free of any form of KRT for at least 7 days. In a systematic review of studies investigating predictors of successful KRT weaning, the most common definition was of not requiring re-initiation of KRT within 7 days [13]. This study was initially designed to recruit a total of 142 participants. Using 80% power, we predicted that the intervention would lower the occurrence of the outcome from 45% to 22.5%. Using a 0.05 significance level, the sample size for each group was set as 71, assuming a 5% attrition rate. This was based on previous studies showing 40-46% of the patients required re-initiation of CKRT before the 7-day mark after discontinuation [14, 15]. However, as of March 2, 2022, we had screened 137 patients since initiating the study on July 15, 2021. We were able to recruit only 30 patients, as there were more moribund patients than anticipated. This higher-than-expected mortality rate impacted the recruitment rate, partly due to the COVID-19 pandemic. At the conclusion of the study, in-hospital mortality occurred in 125/139 (90%) among moribund patients. We amended our protocol to continue the study as a pilot proof of concept and feasibility study. The last patient in the standard treatment arm was recruited on 4/8/2022. Recruitment for the intervention arm was from 4/20/2022 till 3/23/2024.

The primary outcome was successful discontinuation of CKRT. The secondary outcomes included: 1) major adverse kidney events (MAKE30) as the composite outcome of death, continued need of dialysis, and persistent kidney dysfunction (doubling of baseline creatinine) at day 30 after starting CKRT and 2) using 2-hour creatinine clearance and/or cystatin C at time of CKRT discontinuation to predict successful CKRT discontinuation.

We also investigated potential adverse events, consisiting of: 1) daily fluid balance and 2) electrolyte disturbances occuring within 7 days after discontinuation of CKRT, including hyperkalemia with potassium of >6.1 mmol/L and severe metabolic acidosis (defined as pH ≤7.20, PaCO2 ≤45 mm Hg, and bicarbonate concentration ≤20 mmol/L).

Data collection

Once the decision to discontinue CKRT was made, non-anuric patients underwent a 2-hour creatinine clearance. Creatinine clearance was calculated using the equation [Cr urine (mg/dl) × urine volume (ml) × 1.73/Cr serum (mg/dl) × minutes × body surface area]. Body surface area was calculated using the equation by Mosteller et al. [16]. A 2-hour creatinine clearance was found to correlate well with 24-hour creatinine clearance [17, 18]. Thus, due to its ease and practicality, we thought that a 2-hour creatinine clearance might be a potentially useful biomarker to aid clinical decision-making during a critical time if it is proven to predict successful discontinuation of CKRT.

Serum cystatin C was measured by Quest Diagnostics using a particle-enhanced turbidimetric immunoassay on the Beckman Coulter platform. The sample was centrifuged within two hours of collection. Serum or plasma was separated from the red cells and placed into aliquot tubes following centrifugation. Urine and serum creatinine were measured using the Alkaline picrate kinetic method at the WVU lab. For the urine specimen, the time of collection in hours was documented on the sample label.

Data Safety Monitoring Plan

This study involved a low-risk intervention. We devised a data safety monitoring plan. The primary investigator reviewed the patients accrued in the intervention arm daily. The process involved examining whether any electrolyte abnormalities occurred within the first 72 hours of stopping CKRT, as outlined above. The plan also included reporting serious adverse that were unanticipated, serious, and possibly related to the study intervention to be reported to the IRB in accordance with requirements. Data safety monitoring was conducted daily and presented to the independent data monitor personnel monthly, with the option for earlier reports depending on safety issues.

Statistical analysis

Continuous variables were presented as mean and standard deviation or median and interquartile range (IQR), depending on the distribution of the variables. Frequencies and percentages were used to describe categorical variables. The difference in the primary outcome was compared using chi-square test. Logistic regression was done when necessary to account for potential confounding factors, specifically age, norepinephrine dose, COVID19 presence and SOFA scores. A sensitivity analysis was conducted to evaluate the primary outcome, taking death as a competing risk. This was done using Fine and Gray model. Another sensitivity analysis was done by considering successful CKRT discontinuation as being free of any form of KRT in the first 72 hours. The ability to predict successful CKRT discontinuation using cystatin C and/or 2-hour creatinine clearance was assessed using logistic regression, and the area under the receiver operating characteristic curve (AUROC) was reported. Age was included in these models as a covariate. DeLong test was performed to compare the performance of different predictive models. We used the model with age and urine output as the reference. The difference in daily fluid balance between the two arms was determined using generalized estimating equations (GEE) with repeated measures analysis of variance. All analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Participants

Between July 15, 2021, and March 23, 2024, 373 patients with AKI who required CKRT were screened. A total of 67 patients met the eligibility criteria and consented to participate in the study. There were 32 patients in the standard care arm and 35 in the intervention arm (Figure 1). Patients in the intervention arm were older (65 vs. 57 years), were more likely to require norepinephrine at the time of CKRT initiation (0.11 vs. 0 mcg/kg/min) and did not have COVID-19 (Table 1).

Figure 1.

Flow chart

Table 1.

Patient Demographic

		Standard care (N=32)	Intervention (N=35)	P-value
Age, years [mean (SD)]		57 (18)	65 (12)	0.03
SOFA score on CKRT initiation day [mean (SD)]		10 (4)	10.8 (4)	0.4
Ethnicity, not Hispanic or Latino, [N(%)]		32 (97%)	35 (100%)	0.5
White/Caucasian race, [N(%)]		31 (97%)	33 (94%)	0.6
Female sex, [N(%)]		18 (58%)	25 (71%)	0.3
APACHE II [mean (SD)]		26.9 (7)	24.7 (7)	0.2
Charlson Comorbidity Index [median (IQR)]		4 (2;5)	5 (2;7)	0.05
Mean arterial pressure at CKRT initiation, mmHg, [mean (SD)]		78 (16)	76 (13)	0.5
CKRT initial dose, ml/Kg/hour, [mean (SD)]		28.8 (6)	26.7 (6.3)	0.14
Baseline serum creatinine, mg/dl [mean (SD)]		1.21 (0.5)	1.37 (0.5)	0.24
Creatinine at time of CKRT initiation, mg/dl [mean (SD)]		5.55 (2.8)	5.28 (2.7)	0.7
Sodium at time of CKRT initiation, mmol/L [mean (SD)]		135 (6)	135 (4)	0.9
Potassium at time of CKRT initiation, mmol/L [mean (SD)]		5.4 (1)	5.2 (1.2)	0.6
Bicarbonate at time of CKRT initiation, mmol/L [mean (SD)]		16.5 (5)	14.1 (6)	0.09
pH [mean (SD)]		7.24 (0.1)	7.27 (0.1)	0.3
Cumulative fluid balance before CKRT initiation, [median (IQR)]α		1.2 (−0.3;4.2)	2 (0.26;5.75)	0.25
Norepinephrine dose at CKRT initiation, mcg/kg/min, [median (IQR)]α		0 (0; 0.09)	0.11 (0; 0.25)	0.004
Vasopressin dose at CKRT initiation, units/minute[median (IQR)]α		0 (0;0.01)	0 (0;0.03)	0.33
Epinephrine dose at CKRT initiation, mcg/kg/min, [median (IQR)]α		0 (0;0)	0 (0;0)	0.6
Phenylephrine dose at CKRT initiation, mcg/kg/min, [median (IQR)]α		0 (0;0)	0 (0;0)	0.5
Weight at hospital admission, Kg, [mean (SD)]		106 (37)	104 (31)	0.8
Weight at CKRT initiation, Kg, [mean (SD)]		108 (38)	106 (35)	0.8
Diagnosis of septic shock, [count (%)]		23 (72%)	27 (77%)	0.8
Mechanical Ventilation [count (%)]		19 (59%)	16 (46%)	0.3
Access used, [count (%)] b	Right internal jugular	17 (53%)	18 (51%)	1
	Left internal jugular	2 (6%)	3 (8%)
	Right femoral vein	10 (31%)	10 (29%)
	Left femoral vein	3 (10%)	3 (9%)
	Right subclavian	0	1 (3%)
Presumed contributing factors to AKI * , N (%)	Hemodynamic derangements	18 (56%)	9 (26%)	0.013
	Acute tubular injury	5 (16%)	1 (3%)	0.09
	Sepsis	20 (63%)	25 (71%)	0.6
	Hypovolemia	4 (13%)	2 (6%)	0.4
	Pigment-associated AKI	2 (6%)	1 (3%)	0.6
CKRT modality, [count (%)] b	CVVHD	27 (84%)	30 (86%)	1
	CVVHDF	5 (16%)	5 (14%)
Intensive care unit type, [count (%)] b	Medical	23 (72%)	30 (86%)	0.25
	Surgical	7 (22%)	5 (14%)
	Neurocritical care	2 (6%)	0 (0%)
Indications for starting CKRT, [count (%)] * b	Hypervolemia	3 (9%)	5 (14%)	0.7
	Acidemia	13 (41%)	15 (43%)	1
	Hyperkalemia	14 (44%)	15 (43%)	1
	Uremia	1 (3%)	1 (3%)	1
	Severe AKI without other reasons mentioned	6 (19%)	6 (17%)	1
	Oliguria/anuria	7 (22%)	5 (14%)	0.5
COVID19 diagnosis b		10 (31%)	0	0.001

^*Multiple options apply per patient.

^aWilcoxon test

^bFisher’s exact

Primary and secondary outcomes and adverse events

Among patients who survived at least 7 days after CKRT discontinuation, there was no statistically significant difference in successful CKRT discontinuation rates between the two arms: standard care (16/23, 69.5%) and the intervention arm (19/26, 73.1%). This did not change when defining successful CKRT discontinuation for 3 days, standard care 18/27 (66.7%) vs 22/30 (73.3%). Results did not change when considering patients without COVID19. A sensitivity analysis was done using multivariable logistic regression to adjust for age, norepinephrine dose, COVID19 presence and SOFA scores. The intervention was not associated with the outcome (OR: 0.66, 95%CI: 0.2-3, p-value of 0.6).

Another sensitivity analysis was conducted using the Fine and Gray model, which accounts for death as a competing risk; the intervention did not show a statistically significant effect on KRT dependence, with a sub-hazard ratio of 0.75 (0.33; 1.7). There were no differences in other secondary outcomes (Table 2).There was no difference in adverse event rates between the two groups. The daily fluid balance on CKRT between the two groups did not differ, with a mean of 0.43 liters (95% CI: −0.34 to 1.2, p = 0.27; Table 3).

Table 2.

Secondary outcomes

	Standard care	Intervention	p-value
Initial CKRT duration, hours, [mean (sd)]	73 (54)	60.2 (35)	0.26
MAKE30	15/32 (47%)	18/35 (51%)	0.8
KRT within 3 days of stopping CKRT among survivors	9/27 (33.3%)	8/30 (26.7%)	0.7
Total KRT dependence duration during hospital stay, days*, [median (IQR)]	3.14 (1.63;8.2)	2.79 (1.45;5.95)	0.4

^*This is calculated by the addition of the initial CKRT duration plus the time difference between CKRT discontinuation and the final day of KRT session (CKRT or intermittent HD).

Table 3.

Adverse events

	Standard care	Intervention	p-value
Hyperkalemia, [count (%)]	3 (9%)	1 (3%)	0.6
Severe metabolic acidosis, [count (%)]	0	0
Catheter-related blood stream infection, [count (%)]	0	0
Mechanical ventilation days, [median (IQR)]	0 (0;4)	0 (0;2)	0.28
Need for mechanical ventilation after CKRT discontinuation a,b, [count (%)]	3/22 (13.6%)	0/29 (0%)	0.07
Fluid balance at time of ICU discharge, [Median, (IQR)]	3.14 (1.63;8.2)	2.79 (1.45;5.95)	0.42

^a)Among patients who were ventilator-free at the time of CKRT discontinuation

^b)Fisher’s exact test

Analyzing the daily forms

There were 94 forms filled as part of daily assessments. Only two patients did not have a form, as their code status changed to comfort measures only. The forms were completed 53 (56%) times by the nephrology fellows, 36 (38%) times by the advanced practitioner, and 5 times (6%) by attending physicians. The form encouraged CKRT discontinuation in 35 instances, vs. 59 cases in which the decision aid did not encourage CKRT discontinuation. CKRT was discontinued 26 (74%) out of the 35 times it was encouraged to be discontinued. CKRT was continued 53 (90%) out of the 59 times it was not encouraged to be discontinued. Discordance between the form recommendation and team action occurred in 15 instances (16%). Intubation or needing more ultrafiltration were among the top reasons not to discontinue CKRT when it was encouraged to do so, Table S1. Hemodynamic stability and non-oliguric status were the most common reasons for encouraging CKRT to be discontinued, accounting for 20 (57%) cases, as shown in Table S2.

Cystatin C and 2-hour creatinine clearance

Cystatin C was available in 58 patients (29 from each arm). The reasons for not ordering cystatin C in the other patients were primarily because the patients expired on CKRT, or their code status changed to comfort measures. Two-hour creatinine clearance was available on 47 patients (22 in the standard care and 25 in the intervention arm). The lower number of measurements available for 2-hour creatinine clearance was due to their anuric status. Among survivors at day 7, cystatin C levels were lower and 2-hour creatinine clearance was higher among patients who achieved the primary outcome, 2.1 vs. 2.7 mg/L and 24.2 vs. 11 mL/min/body surface area, compared to those who required CKRT within 7 days, with a p-value of 0.04 for both. In the overall cohort, while non-statistically significant, the model including cystatin C, 2-hour creatinine clearance, age, and urine output on last day of CKRT had the highest AUROC of 0.76 (95% CI: 0.54-0.977) in predicting successful discontinuation of CKRT. A similar non-statistically significant finding was noticed when comparing the models’ performance in each of the intervention and standard care arm separately. While the model including the biomarkers, age and urine output seemed to perform better in predicting successful CKRT discontinuation in the intervention arm with AUROC 0.835 (95%CI: 0.58-0.99) compared to the standard care AUROC 0.7 (95%CI: 0.436-0.99), this difference was not statistically significant, DeLong test p-value=0.5. Table 4.

Table 4.

Performance of Cystatin C and 2-hour creatinine clearance in predicting successful CKRT discontinuation among survivors at day 7 as reported by AUROC with 95% confidence interval

	Cystatin C	2-hour Creatinine Clearance	Cystatin C and 2-hour Creatinine Clearance	Urine output on last day of CKRT	All Combined
Overall cohort	0.711 (0.54-0.97)	0.725 (0.52-0.93)	0.757 (0.53-0.97)	0.677 (0.45-0.901)	0.76 (0.543-0.977)
Standard care	0.7 (0.39-0.99)	0.683 (0.29-0.99)	0.75 (0.45-0.99)	0.58 (0.2-0.99)	0.7 (0.36-0.99)
Intervention	0.8 (0.53-0.99)	0.811 (0.59-0.99)	0.835 (0.58-0.99)	0.74 (0.45-0.99)	0.835 (0.58-0.99)

All models contained age in addition to the variable specified in each column. Using the model with age and urine output as the reference, the other models were not statistically significantly different within each stratum.

Discussion

In this prospective pilot feasibility study, using a standardized form to encourage timely CKRT discontinuation in patients with AKI was feasible with high protocol adherence. This intervention is of low cost as it is incorporated as part of the team’s daily clinical patient assessment. CKRT was discontinued in 74% of the times it was encouraged to be discontinued and continued in 90% of the time it was not encouraged to be discontinued. However, it did not improve the rate of successful CKRT discontinuation. This may suggest that we need to better define what constitutes successful discontinuation of CKRT. This study also showed that current criteria that can successfully predict when the patients are ready to be weaned off CKRT remain deficient. This goal might be achieved through a multicenter study.

It remains unclear if a standardized approach improves outcomes. . . . In clinical trials of the timing of KRT initiation, standardized discontinuation of KRT was done based on various criteria. In the AKIKI trials, discontinuation of KRT was considered or recommended based on a range of urine outputs (500-1000 ml) [19]. Discontinuation of KRT was mandatory if diuresis was sufficient to allow for a spontaneous decrease in serum creatinine concentration. In the ELAIN study, KRT was discontinued if renal recovery occurred. This was defined by urine output (greater than 400-2100 mL/24 h) and creatinine clearance (greater than 20 mL/min) [4]. In another study, discontinuation of KRT was based on whether kidney function recovery occurred, as defined by a 6-hour creatinine clearance test performed once urine output exceeded 30 mL/h or when there was a spontaneous decline in serum creatinine [20]. Thresholds of 20 ml/min and later 12 ml/min were used [20]. In the IDEAL-ICU study, KRT was discontinued if kidney recovery occurred. This was defined as a decrease in creatinine level and a return of spontaneous urine output to more than 1000 ml per 24 hours [or more than 2000 ml with diuretics] [21]. However, none of these trials tested whether these criteria resulted in successful discontinuation of KRT, as they were used in both arms. In addition, the recovery of kidney function was heterogeneously defined across these studies. In our study, we utilized a set of criteria that considered urine output, hemodynamic stability, and the ability to manage acid/base and volume disorders medically. There was no difference in successful CKRT discontinuation, and no impact on relevant outcomes, such as mortality and kidney recovery.

The overarching goal of a standardized approach to discontinuing CKRT was to limit exposure to CKRT while doing so in a safe manner… . In a post hoc analysis of the ATN study, exposure to more frequent and higher intensity KRT was associated with delayed renal recovery [22]. . In a retrospective study, after propensity score matching, prolonged exposure to CKRT (>7 days) was independently associated with an increased risk of doubling serum creatinine or requiring dialysis at 90 [23]. Similarly, successful CKRT liberation (KRT independence for 72 hours) was found to be independently associated with the increased probability of kidney recovery at 90-day follow-up [24]. In our study, the initial CKRT duration (and total KRT dependence) was not statistically significantly different between the two groups. Thus, this may explain the absence of the potential benefit of the intervention on the kidney outcomes.

Successful CKRT discontinuation has not been well-defined in the literature. Available traditional or novel markers of kidney recovery are limited by study design (retrospective analyses, small sample size), variable heterogeneity, and lack of prospective validation of the cut-offs [25]. In a study by Valet et al., 24-hour urinary creatinine clearance of 15 ml/min was the most powerful parameter associated with KRT discontinuation success [26]. In a study by Fröhlich et al., a 2-hour creatinine clearance performed within 12 hours before stopping CKRT showed good predictive characteristics for CKRT discontinuation success [27]. In another study, higher serum cystatin C was associated with lower odds of successful weaning from CKRT [28]. Lower plasma cystatin C concentrations during the first 3 days of CKRT were associated with early recovery of kidney function.[29] Nonetheless, CKRT discontinuation was based on the clinical discretion of the healthcare rather than being guided by the cystatin C results in the latter study. . In our study, 2-hour creatinine clearance performed better than cystatin C. However, this difference was not statistically significant. While not statistically significantly different, the biomarkers appeared to perform better in the intervention group. One hypothesis is that a standardized approach may improve the timing of performing these tests, this remains to be confirmed in future studies.

Our study had limitations. We were underpowered to detect a difference in successful CKRT discontinuation. Although this was a pilot study, we identified some challenges that can inform future studies. It remains crucial to establish validated objective criteria to guide clinical decision-making and inform study design [7]. One potential solution is to combine biomarkers and a standardized form to guide decision-making. It is also essential to define successful CKRT discontinuation, as many definitions have been used in the literature. Another limitation is that this study reflects the experience of a single center. Also, 31% of the patients in the standard care arm had COVID-19, whereas no patients in the intervention arm were affected. Although sensitivity analyses adjusted for this imbalance, residual unmeasured confounding may still be present. Lastly, while there was no difference in MAKE30 between the two arms, longer follow-up might shed light on potential effect of such intervention. Future studies may require the involvement of multiple centers. This is because the intervention’s effect would be diluted if the same physician or team were concurrently managing two patients who would have been otherwise randomized to different arms. Thus, a future multicenter cluster-randomized clinical trial might be warranted to test this approach.

In conclusion, a standardized approach to discontinuation of CKRT did not improve the rate of successful CKRT discontinuation. A consensus regarding definition of successful discontinuation of CKRT need to be established. Individualized decision tree that considers other patient-specific factors not found in our form such as volume status and non-patient factors such as logistical ones may be needed. Future studies that also considers biomarkers in the decision-making may help achieve successful CKRT discontinuation.

Data sharing

All data are within the manuscript. A data use agreement needs to be established after a reasonable request from researchers who meet criteria to access confidential data is made.