Introduction
AKI is common and harmful, with clinical outcomes contingent on optimal management. AKI complicates 30%–50% of intensive care unit (ICU) admissions (1 ), and those treated with KRT have higher mortality (2 ). Attributed costs of AKI in the United States are approximately $10 billion annually due partly to resource utilization, including longer ICU length of stay and hospital length of stay (3–5 ).
Delays in KRT for AKI can result in complications and death (6 ). However, initiating KRT carries procedural risks for patients who may recover kidney function without requiring KRT. Recent randomized controlled trials have not found a lower risk of mortality associated with early versus late KRT initiation (7–9 ). To address uncertainty and variation in the care of patients with AKI requiring KRT, we previously implemented a decision-making algorithm (the Standardized Clinical Assessment and Management Plan [SCAMP]) over a 13-month period for nephrologists managing patients with AKI in the medical ICU (10 ). The AKI Standardized Clinical Assessment and Management Plan (AKI-SCAMP) provided recommendations about optimal indications for initiating KRT, including pH<7.2, potassium >6.5 mmol/L or with EKG changes, toxin ingestion, massive anasarca, FiO2 >0.7, urine output <100 ml/24 h, and uremic symptoms. We found that patients whose clinicians adhered to the AKI-SCAMP recommendation to start KRT had lower in-hospital and 60-day mortality.
To better understand the effect of AKI-SCAMP on clinical outcomes, such as inpatient mortality, hospital length of stay, and ICU length of stay, we implemented a controlled design that involved alternating AKI-SCAMP with a “sham” SCAMP, which included only questions related to estimated risk of mortality and futility. In this study, we hoped to ascertain the utility of AKI-SCAMP in a broad ICU population.
Materials and Methods
Description of the Standardized Clinical Assessment and Management Plan
A SCAMP is a quality improvement approach involving a standardized care pathway to an area of care with uncertainty and variation to optimize outcomes; AKI-SCAMP development was previously described (10 ).
AKI-SCAMP exists as a point-of-care electronic or paper form used by the nephrologist during an encounter to document patient information and guide decision making. AKI-SCAMP provides indications for initiating KRT and questions related to treatment futility that recommend that KRT should not be initiated under those circumstances. Deviations are captured.
Study Design
We conducted a controlled study of time periods with a 1:1 allocation ratio to AKI-SCAMP or control between July 2017 and July 2018. We utilized 12 time periods (each time period a priori designed to be approximately 4-weeks long) assigned as either SCAMP or sham (Supplemental Figures 1 and 2 ). An initial randomization scheme on the basis of time periods was achieved and then modified to ensure (1 ) balanced allocation of sham and SCAMP across providers and (2 ) minimized seasonal effects (i.e. , higher mortality during winter months). Ultimately, the first two time periods were randomized, followed by an alternating sequence determined a priori before study activation. Critical care attending nephrologists were required to complete the SCAMP or sham form on the basis of the time period they were attending. The number of weeks (mean of 4 weeks) and the number of nephrologists varied by time period due to rotation schedule variation (not all attending nephrologists completed full 2-week blocks of a standard rotation). The goal was equal allocation of patients to arms, but because allocation was not at the patient level, there were ultimately slight differences in patient numbers between the two arms. The split between SCAMP and sham forms was approximately 60:40 for each critical care nephrologist. Assignment was on the basis of time intervals. Blinding was not possible as nephrologists either received SCAMP or sham. In each treatment period, the mean number of nephrologists was three (range, two to four), and the mean number of study participants was 19 (range, 9–28) (Supplemental Figure 2 ).The AKI-SCAMP form (Supplemental Figure 3 ) consists of questions related to comorbidities, reasons for AKI, physician-estimated mortality and futility, an algorithm for KRT initiation, and algorithm for choice of tunneled versus nontunneled dialysis catheter for KRT initiation. The sham form consisted of questions related to comorbidities, reasons for AKI, estimated mortality, and futility estimation (Supplemental Figure 4 ). We hypothesized that use of a sham would address whether the additional step of completing a form as opposed to AKI-SCAMP, similar to the Hawthorne effect, could affect outcomes (11 ). The primary outcome was inpatient mortality between the AKI-SCAMP and sham groups. Prespecified secondary outcomes included hospital length of stay, ICU length of stay, 30-day mortality, 60-day mortality, dialysis dependence at hospital discharge, and 30-day hospital readmission. There were no changes to study outcomes after commencement, and no adverse events were reported. We did not have a data safety monitoring board as this study was a quality improvement initiative. No interim analyses were planned, and no stopping guidelines were outlined as safety concerns have not been associated with AKI clinical decision support systems at our institution. Patients did not crossover between sham or SCAMP after they were assigned. They remained in their study arm even if they were assigned on the last day of that block (<1%). As we illustrate in Table 1 , mean time from study enrollment to kidney replacement initiation was 2 days, and SCAMP and sham were not continued beyond the decision to start KRT (which is our intent with SCAMP—to guide clinical decision making related to KRT initiation). As a result, few patients were affected by transitions between time periods.
Table 1. -
Patient demographics and clinical characteristics (accounting for clustering by the treating nephrologist)
Patient Characteristics
Intervention, n =122
Control, n =102
Men
85/122 (70%)
58/102 (57%)
Age, yr, mean (SD)
62 (14)
63 (14)
Race, N (%)
White
101/122 (83%)
80/102 (78%)
Black
10/122 (8%)
15/102 (15%)
Hispanic
8/122 (7%)
4/102 (4%)
Other
3/122 (3%)
3/102 (3%)
Insurance status, N (%)
Medicare
66/122 (54%)
56/102 (55%)
Commercial insurance
51/122 (42%)
37/102 (36%)
No insurance
3/122 (3%)
6/102 (6%)
Medicaid
2/122 (2%)
3/102 (3%)
Primary reason for hospital admission, N (%)
Cardiac
27/122 (22%)
30/102 (29%)
Respiratory
14/122 (12%)
13/102 (13%)
Gastrointestinal
13/122 (11%)
7/102 (7%)
Neurologic
7/122 (6%)
8/102 (7%)
Vascular
2/122 (2%)
3/102 (3%)
Kidney: AKI
6/122 (5%)
3/102 (3%)
Kidney: AKI on CKD
2/122 (2%)
1/102 (1%)
Hematologic
4/122 (3%)
8/102 (8%)
Infectious
28/122 (23%)
14/102 (14%)
Musculoskeletal
0/122 (0%)
4/102 (4%)
Dermatologic
1/122 (1%)
1/102 (1%)
Metabolic/endocrine
6/122 (5%)
3/102 (3%)
Trauma
2/122 (2%)
0/102 (0%)
Intensive care unit location, N (%)
Medical
48/122 (39%)
42/102 (41%)
Cardiac medical
25/122 (21%)
19/102 (19%)
Cardiac surgical
19/122 (16%)
13/102 (13%)
Surgical
15/122 (12%)
15/102 (15%)
Thoracic surgical
10/122 (8%)
3/102 (3%)
Neurologic
4/122 (3%)
8/102 (8%)
Premorbid kidney function, mean (SD)
Serum creatinine prehospitalization, mg/dl
1.2 (0.5)
1.2 (0.7)
eGFR prehospitalization, ml/min per 1.73 m2
53 (23)
50 (22)
Maximum serum creatinine during intervention, mg/dl
4.3 (2.4)
4.4 (2.6)
Reasons for AKI, N (%)
Hypotension
76/122 (62%)
64/102 (63%)
Sepsis
39/122 (32%)
45/102 (44%)
Other
a
28/122 (23%)
13/102 (13%)
Contrast
21/122 (17%)
14/102 (14%)
Cardiorenal syndrome
16/122 (13%)
18/102 (18%)
Prerenal azotemia
16/122 (13%)
7/102 (7%)
Other nephrotoxins
6/122 (5%)
8/102 (8%)
Hepatorenal syndrome
5/122 (4%)
1/102 (1%)
Obstruction
2/122 (2%)
1/102 (1%)
Rhabdomyolysis
1/122 (0.8%)
4/102 (4%)
Tubulointerstitial nephritis
1/122 (0.8%)
1/102 (1%)
Thrombotic microangiopathy
1/122 (0.8%)
1/102 (1%)
GN
0/122 (0%)
1/102 (1%)
Hemolysis
0/122 (0%)
0/102 (0%)
Vasculitis
0/122 (0%)
0/102 (0%)
Chronic health conditions, N (%)
Cardiovascular disease
60/122 (49%)
53/102 (52%)
Postsurgery
39/122 (32%)
30/102 (29%)
Malignancy
33/122 (27%)
25/102 (25%)
Immunosuppressive therapy
26/122 (21%)
25/102 (25%)
Chronic hypoxemia
18/122 (15%)
15/102 (15%)
Hospital admission to study enrollment, d, median (IQR)
4 (2–10)
4 (2–12)
Study enrollment to kidney replacement initiation, d, mean (SD)
2 (4)
2 (4)
ATN mortality risk score, median (IQR)
23 (18–28)
22 (17–27)
Vitals at enrollment, mean (SD)
Heart rate, bpm
92 (22)
95 (21)
Mean arterial pressure, mm Hg
82 (19)
80 (15)
Urine output, ml/24 h
239 (626)
250 (546)
Laboratory values at enrollment, mean (SD)
Arterial partial oxygen pressure, mm Hg
120 (63)
118 (62)
Arterial pH
7.35 (0.1)
7.36 (0.1)
INR
1.8 (1.6)
1.6 (0.7)
Serum albumin, g/dl
3.0 (0.8)
2.9 (0.7)
Serum alkaline phosphatase, IU/L
123 (96)
123 (94)
Serum bicarbonate, mmol/L
21 (6)
21 (6)
Serum bilirubin, mg/dl
1.7 (2.5)
2.1 (4.3)
Serum creatinine, mg/dl
2.8 (2.0)
2.7 (1.5)
Serum phosphate, mg/dl
5.2 (1.8)
5.3 (2)
Platelet count, ×109 /L
169 (99)
179 (113)
Ventilation requirements at enrollment, N (%)
Mechanical ventilation with FiO2 <0.6
57/122 (47%)
49/102 (48%)
Mechanical ventilation with FiO2 >0.6
20/122 (16%)
10/102 (10%)
No mechanical ventilation; FiO2 >0.6
5/122 (4%)
3/102 (3%)
Vasopressor use, N (%)
97/122 (80%)
89/102 (87%)
SOFA score, median (IQR)
8 (5–10)
7 (5–10)
Types of KRT during enrollment, N (%)
None
55/122 (45%)
41/102 (40%)
CVVH
32/122 (26%)
27/102 (27%)
CVVH and HD
20/122 (16%)
18/102 (18%)
HD
15/122 (12%)
16/102 (16%)
IQR, interquartile range; ATN, Acute Renal Failure Trial Network; INR, international normalized ratio; SOFA, Sequential Organ Failure Assessment; CVVH, continuous KRT; HD, hemodialysis.
a Other includes acute interstitial nephritis, tumor lysis syndrome, cast nephropathy, traumatic injury, and cholesterol emboli.
Setting and Patient Population
We implemented the study in all ICUs at Brigham and Women’s Hospital, a tertiary care academic medical center in Boston, Massachusetts. There are nine ICUs (90 beds) in total: two medical, two surgical, two neurologic, one cardiac medical, one cardiac surgical, and one thoracic. Approval was granted by the institutional review board for data collection and analysis; the need for informed consent was waived as the research represented no more than a minimal risk of harm to participants and involved no procedures for which written consent is normally required. A study coordinator identified eligible patients in the ICU who were documented to have AKI and enrolled them on the date of initial nephrology consultation (Figure 1 ); nephrology had to be consulted in the ICU for a patient to be included. Patients with previous ICU admissions during hospitalization were excluded. Additional study design details are provided in Supplemental Figure 5 . There were no changes in eligibility criteria after study commencement. There were no losses or exclusions after randomization.
Figure 1.: Consolidated Standards of Reporting Trials flow diagram for the AKI Standardized Clinical Assessment and Management Plan (AKI-SCAMP) study. ICU, intensive care unit.
Attending Nephrologists
There were eight board-certified nephrologists attending on the ICU service who participated in the study. Seven of eight (88%) were men; they had a median of 16 years postmedical school (range, 9–20 years) and a median of 8 years of critical care nephrology attending experience (range, 2–12 years).
Study Implementation
Nephrologists completed one data form per patient on each day of enrollment. Adherence to AKI-SCAMP recommendations at the patient level was defined as adhering to all recommendations every day of enrollment. Patients were considered to have completed AKI-SCAMP after initiation of KRT, death, discharge, transfer from ICU, or nephrology signing off. Supplemental Figure 5 shows the AKI-SCAMP workflow for KRT initiation. On the basis of prespecified criteria, AKI-SCAMP recommended whether to initiate KRT. Clinicians were free to ignore the recommendation but asked to provide deviation reasons.
Data Sources and Collection
The electronic health record (EHR; EPIC, Partners Version) was used to abstract data from consult notes as well as demographics, comorbidities, hospital admission reasons, and laboratory tests. Vitals, urine output, fluid balance, vasopressor use, and CKRT were obtained from the patients’ daily flow sheets. Clinician adherence or deviation, deviation reasons, and AKI causes were obtained from SCAMP forms. Adherence with AKI-SCAMP form completion was monitored by a data coordinator, who followed nephrologists daily to ensure completion. Mortality postdischarge was ascertained by the study coordinator who followed all patients, 60 days postdischarge on the basis of EHR data. Manual EHR review was conducted to obtain data regarding patients for whom futility was indicated and KRT indications for control patients, as these were not captured on the sham form.
Statistical Analyses
Demographic and clinical characteristics are reported as counts and percentages or medians and interquartile ranges (IQRs); comparisons between intervention and control characteristics were made with generalized estimating equations (GEEs) for normal and non-normal distributed covariates with clustering according to the treating nephrologist (12 ,13 ). Given the prospective data collection by study coordinators, for patients who continued as inpatients, we had no loss to follow-up before 30 or 60 days and did not have to censor when assessing secondary outcomes of 30- and 60-day mortality.
Patient characteristics in the two arms are presented as means for continuous variables and proportions for categorical variables. P values for differences between groups, for both means and proportions, are calculated using GEEs (12 ,13 ), clustering by the treating nephrologist. These models were subsequently adjusted for prespecified variables, including age, albumin, race, sex, sepsis, and AKI-specific disease severity at enrollment. Disease severity was estimated using the mortality risk equation by Demirjian et al. (14 ) derived in a randomized controlled trial of AKI KRT.
ICU length of stay was measured from ICU admission until death or ICU discharge. Hospital length of stay was measured from hospital admission until death or hospital discharge. We used GEE with the log link and negative binomial variance to estimate the effect of AKI-SCAMP on ICU length of stay and hospital length of stay. As a sensitivity analysis, we plotted Kaplan–Meier curves to display time from enrollment to ICU or hospital discharge, with censoring due to death; curves were compared using a Cox proportional hazards regression model, accounting for clustering by the treating nephrologist (15 ). Within the intervention group, we measured adherence and assessed mortality and length of stay according to AKI-SCAMP adherence subgroups.
With both AKI-SCAMP and sham forms, we asked nephrologists to assess treatment futility defined as no meaningful chance of recovery from non-kidney illness, imminent death, untreatable metastatic cancer, end stage liver disease, and severe irreversible neurologic injury.
Sample Size Calculation
Using a GEE z test to compare proportions with an intracluster correlation coefficient of 0.01 within nephrologist, we needed a sample size of 110 patients per arm to provide 80% power, with a two-sided type 1 error of 0.05, to detect a minimum inpatient mortality decrease from 60% in the sham group to 40% in SCAMP (on the basis of prior study results) (10 ).
Results
Clinical Characteristics
Table 1 shows the characteristics of the 224 study participants. Hypotension and sepsis were the most common AKI cause in both groups. Supplemental Tables 1 and 2 show the baseline clinical or treatment characteristics between the intervention and control groups for patients who received KRT. There were no significant differences in KRT modality. Finally, although not statistically significant, there were slightly more control patients who received KRT, with a trend toward a greater number of median days from KRT cessation to death.
Mortality
There was no significant difference in the primary outcome of frequency of inpatient, 30-day, or 60-day mortality between intervention and control groups (41% versus 47%, 35% versus 42%, and 39% versus 43%, respectively) (Table 2 ), and there was no significant difference in the unadjusted or adjusted odds of inpatient, 30-day, or 60-day mortality (Supplemental Table 3 ).
Table 2. -
Effect of Standardized Clinical Assessment and Management Plan on patient outcomes
Patient Outcome
Unadjusted Results
Difference (95% Confidence Interval)
Adjusted Results, Mean (95% Confidence Interval)
Intervention, n =122
Control, n =102
P Value
Intervention, n =122
Control, n =102
P Value
Primary outcome
Mortality, N (%)
Inpatient mortality
50/122 (41%)
48/102 (47%)
0.42
−6% (−20% to 8%)
37% (26% to 48%)
45% (35% to 55%)
0.40
Secondary outcomes (prespecified)
Mortality, N (%)
30-d mortality
43/122 (35%)
43/102 (42%)
0.34
−7% (−18% to 5%)
29% (21% to 40%)
32% (27% to 39%)
0.64
60-d mortality
48/122 (39%)
44/102 (43%)
0.59
−4% (−18% to 7%)
28% (19% to 38%)
45 (39% to 50%)
0.34
Length of stay, mean (95% CI)
ICU length of stay, d
8 (8–9)
12 (12–13)
<0.001
−4 (−6 to −2)
7 (7 to 8)
12 (10 to 13)
0.04
Hospital length of stay, d
25 (22–29)
30 (27–34)
0.02
−5 (−9 to −1)
22 (20 to 24)
27 (25 to 29)
0.03
Mortality risk prediction
Probability of 60-d mortality by ATN risk score, %, mean (SD)
30% (26)
31% (27)
0.76
−1% (−7% to 5%)
24% (23% to 24%)
23% (22% to 23%)
0.17
Dialysis dependence at hospital discharge, N (%)
10/122 (8%)
11/102 (11%)
0.65
−3% (−7% to 1%)
6% (3% to 11%)
9% (4% to 20%)
0.54
30-d readmission, N (%)
Readmission to the hospital within 30 d
8/122 (7%)
15/102 (15%)
0.05
−8% (−16% to 0%)
9% (6% to 13%)
13% (9% to 18%)
0.32
Readmission to the ICU within 30 d
3/122 (3%)
6/102 (6%)
0.31
−3% (−7% to 3%)
5% (3% to 8%)
5% (3% to 8%)
0.86
Vascular access, N (%)
Nontunneled dialysis catheter placed
60/122 (49.2%)
54/102 (52.9%)
0.79
−4% (−15% to 12%)
41% (29% to 55%)
47% (34% to 60%)
0.34
Tunneled dialysis catheter placed
13/122 (10.7%)
20/102 (19.6%)
0.33
−9% (−17% to 2%)
9% (7% to 12%)
17% (14% to 20%)
0.13
Patients with line infection
8/122 (7%)
11/102 (11%)
0.34
−4% (−11% to 3%)
3% (1% to 6%)
6% (3% to 11%)
0.25
Blood cultures taken, mean (SD)
3 (5)
5 (7)
0.003
−2 (−3 to −1)
2 (2 to 3)
4 (3 to 5)
0.04
Patients with positive blood cultures
8/122 (7%)
11/102 (11%)
0.34
−4% (−12% to 3%)
3% (1% to 6%)
6% (3% to 11%)
0.25
KRT
KRT initiated
67/122 (54.9%)
61/102 (59.8%)
0.93
−5% (−10% to 16%)
48% (34% to 61%)
53% (39% to 67%)
0.22
KRT, d per patient, median (IQR)
5 (2–15)
8 (2–15)
0.81
−3 (−9 to 6)
11 (8 to 14)
10 (7 to 14)
0.85
Enrollment to KRT, d, median (IQR)
0 (0–1)
0 (0–2)
0.87
0 (−1 to 1)
1 (1 to 2)
1 (1 to 2)
0.86
For those with inpatient mortality, KRT cessation to death, d, median (IQR)
0 (0–0)
1 (0–3)
0.07
−1 (0 to −2)
1 (0 to 5)
1 (0 to 7)
0.54
95% CI, 95% confidence interval; ICU, intensive care unit; ATN, Acute Renal Failure Trial Network; IQR, interquartile range.
Length of Stay (Intensive Care Unit and Hospital) and 30-Day Hospital Readmission
With respect to secondary outcomes, the median ICU length of stay in the intervention group was 8 days (95% confidence interval [95% CI], 8 to 9) compared with 12 days (95% CI, 12 to 13) for the control group (P <0.001) (Table 2 ). In a proportional hazards model censored for ICU mortality with clustering by the treating nephrologist, ICU length of stay was significantly shorter for the intervention group (P =0.05) (Figure 2 ); multivariable hazard ratio was 1.61 (95% CI, 1.13 to 2.28; the hazard ratio >1 signifies a shorter length of stay).
Figure 2.: Kaplan–Meier curves for ICU length of stay according to treatment group; censoring for ICU mortality. C, control; I, intervention.
The median hospital length of stay in the intervention group was 29 days (IQR, 16–42 days) compared with 36 days (IQR, 20–64 days; P =0.02). In a proportional hazards model censored for inpatient mortality with clustering by the treating nephrologist, hospital length of stay was significantly shorter for patients in the intervention group (P =0.01) (Figure 3 ); multivariable hazard ratio was 1.91 (95% CI, 1.28 to 2.86).
Figure 3.: Kaplan–Meier curves for hospital length of stay according to treatment group; censoring for inpatient mortality .
Exploratory analyses were conducted to determine if there were possible carryover effects, as two of eight nephrologists were exposed to intervention-control-intervention periods consecutively, which demonstrated that hospital length of stay and ICU length of stay were significantly longer in the control period following the first intervention period (P =0.02) and were significantly shorter in the second intervention period than in the prior control period (P <0.001). The results of this subset analysis suggest that there were not any carryover effects.
Fewer patients in the intervention group required hospital readmission within 30 days (7% versus 15%; P =0.05) (Table 2 ). The adjusted odds of 30-day hospital readmission was significantly lower for patients in the intervention group (odds ratio, 0.38; 95% CI, 0.15 to 0.99).
There was no significant difference in tunneled dialysis catheter placement between the intervention and control groups, although significantly greater numbers of blood cultures were obtained in the control group.
Adherence
There was a 9% nonadherence rate to AKI-SCAMP recommendations in the intervention group (23 of 272 patient-days for 21 of 122 patients). Reasons for deviation are provided (Supplemental Table 4 ). In adjusted analyses, there were no significant differences in inpatient mortality, 30-day mortality, or 60-day mortality on the basis of adherence to AKI-SCAMP recommendations (Table 3 ). Hospital length of stay was 24% (95% CI, 16% to 31%) shorter and ICU length of stay was 32% (95% CI, 29% to 34%) shorter for patients with AKI-SCAMP adherence.
Table 3. -
Standardized Clinical Assessment and Management Plan adherence association with patient outcomes
Patient Outcome
Univariable Odds Ratio or Relative Risk (95% Confidence Interval)
Multivariable Odds Ratio or Relative Risk (95% Confidence Interval)
Inpatient mortality
1.48 (0.55 to 4.0)
a
1.54 (0.43 to 5.5)
a
30-d mortality
1.93 (0.65 to 5.7)
a
2.93 (0.78 to 11.1)
a
60-d mortality
2.37 (0.81 to 7.0)
a
3.33 (0.83 to 13.3)
a
Hospital length of stay
0.86 (0.78 to 0.94)
b
0.76 (0.69 to 0.84)
b
ICU length of stay
0.85 (0.82 to 0.88)
b
0.68 (0.66 to 0.71)
b
For example, a relative risk of 0.86 indicates a 14% reduction in length of stay. ICU, intensive care unit.
a Odds ratio.
b Relative risk.
Treatment Futility
Treatment was considered by the attending nephrologists to be futile for 14 of 122 (12%) in the intervention group and 15 of 102 (15%) in the control group. Significantly fewer KRT treatments were performed when physicians deemed the treatment to be futile in the intervention group compared with the control group (2% versus 7%; P =0.003) (Table 4 ). Inpatient mortality was significantly higher for patients in the control group for whom KRT was thought to be futile compared with patients in the intervention group. There was no significant difference between the intervention and control groups in actions taken when futility was indicated (e.g. , code status change or palliative care consultation).
Table 4. -
Standardized Clinical Assessment and Management Plan effect on KRT decision making when futility indicated
Patient Outcome
Intervention, n =272 patient-d (%)
Control, n =262 patient-d (%)
P Value
KRT proceeded when physician felt that treatment was futile
5/272 (2%)
19/262 (7%)
0.003
KRT proceeded against physician wishes due to primary team/family preference
2/272 (0.7%)
5/262 (2%)
0.28
ATN score 60-d risk of mortality for treatment futility subgroup, median + IQR
38 (24–65)
36 (22–75)
C statistic 0.72 (P =0.44)
Patients for whom KRT was determined to be futile
14/122 (12%)
15/102 (15%)
0.70
60-d mortality for treatment futility subgroup
5/14 (36%)
11/15 (73%)
0.07
Code status change after futility indicated
6/14 (42.9%)
11/15 (73.3%)
0.32
Palliative care consulted after futility indicated
3/14 (21.4%)
4/15 (26.7%)
0.62
Inpatient mortality
3/14 (21.4%)
10/15 (66.7%)
0.05
For those with inpatient mortality, from futility indicated to death, d, median (IQR)
22 (5–71)
3 (1–19)
0.29
For those with inpatient mortality and KRT started, KRT cessation to death, d, median (IQR)
1 (0–387)
0 (0–0)
0.13
ATN, Acute Renal Failure Trial Network; IQR, interquartile range.
Discussion
In this controlled study involving patients with AKI in the ICU, we found that use of a clinical decision support SCAMP did not affect the primary outcome of in-hospital, 30-day, or 60-day mortality, but it was associated with significantly reduced ICU length of stay and hospital length of stay. We also found that intervention patients with perceived treatment futility were less likely to receive KRT.
The risk of mortality and length of stay associated with AKI remain staggeringly high—with a 6.5-fold higher risk of mortality and a 3.5-fold longer hospital length of stay (16 ). Studies to date have focused on recognition and management to prevent AKI progression among inpatients, with a variable effect on mortality and length of stay (17–19 ). In contrast, our study focused on patients with severe AKI frequently requiring KRT who are in the ICU, providing an algorithm for KRT initiation.
Guidelines related to AKI KRT timing can have limitations. Trial populations from which they are derived may not reflect real-life populations due to specific inclusion and exclusion criteria (20 ,21 ). In contrast, SCAMPs offer a clinician-designed approach to promoting care standardization that accommodates patients’ individual differences and keeps pace with the growth of medical knowledge. SCAMPs have been shown to increase adherence with evidence-based practices, reduce unnecessary utilization, and improve patient outcomes and provider satisfaction (22–25 ).
In this controlled study, there was no significant effect of AKI-SCAMP on mortality in contrast to our 2017 observational study, which found a significant reduction in mortality associated with SCAMP adherence. Some explanations include a limited medical ICU population in the original study, a higher rate of AKI-SCAMP nonadherence at 57% (9% in this study), the lack of a control group, and inclusion of fellows and attendings. This study utilized an intention-to-treat design, whereas the 2017 study analyzed adherence versus nonadherence. Furthermore, given that the mortality rate in the original study was higher than the observed mortality in both groups, this likely affected our power calculation for the primary mortality outcome. Finally, our mortality findings are likely limited by the limited number of patients included in the study. AKI-SCAMP was associated with a significant reduction in ICU length of stay and hospital length of stay; however, these were secondary outcomes of the study. Use of AKI-SCAMP may have led to more judicious overall use of KRT. Notably, fewer patients in the intervention arm received KRT in the setting of physician-perceived treatment futility. These results are important in the context of the recent STARRT-AKI multicenter trial, which showed that earlier initiation of KRT was associated with a higher risk of 90-day dialysis dependence and adverse events (9 ). Similarly, our study evinces that timing of KRT initiation may not affect mortality but suggests that standardization of KRT initiation may affect other outcomes, like length of stay.
The reduced length of stay observed in the SCAMP cohort is likely due to multiple factors. First, patients in the control arm had a slightly higher rate of KRT initiation, and of those who had KRT, higher median days of KRT (not statistically significant). Second, there was significantly more KRT use in those for whom KRT was thought to be futile and a trend toward a greater number of median days from KRT cessation to death. Finally, although not significant, there was greater nontunneled dialysis catheter use and there were significantly more blood cultures obtained in the control group. SCAMP may, therefore, encourage standardization of decision making related to timing of KRT, KRT initiation despite treatment futility, and tunneled dialysis catheter use, all of which can affect length of stay.
Strengths of our study are that AKI-SCAMP was designed iteratively on the basis of published data and expert consensus. We examined outcomes that are clinically relevant and inform practice. In addition, this was a cluster-assigned controlled study of time periods that ensured balance in patient characteristics and included a large heterogenous patient population in the ICU. The adjudication step allowed for balance in allocating critical care nephrologists exposure to both SCAMP and sham forms.
Limitations include the fact that this was a single-center study in a large academic teaching tertiary hospital with a previous prospective implementation of SCAMP, which may limit the generalizability. Our study encountered challenges in incorporating a clinical decision support system into daily AKI management. A relatively small number of providers participated in the study, and a study coordinator was required to ensure daily form completion. In some cases, similarly in both arms, nephrologists needed reminders to ensure form completion, and forms were completed retrospectively. We were unable to integrate the forms within EHR as intended. It was not possible to blind the nephrologists as the SCAMP and sham forms were clearly distinguishable. There were practical limitations to alternating AKI-SCAMP and “sham” control in 4- to 6-week blocks, resulting in additional adjudication to ensure balance between SCAMP and sham. There was also heterogeneity in the causes of AKI, including a trend toward higher rates of sepsis in the control group and higher rates of prerenal azotemia in the intervention group, which may have acted as an unintended confounder. Finally, mortality was limited to EHR-based data, which could have resulted in missed events due to postdischarge mortality.
In summary, use of an AKI-SCAMP clinical decision support tool for assessment and management of AKI KRT did not affect inpatient mortality but was associated with reductions in ICU length of stay and hospital length of stay and reduced KRT use in cases of physician-perceived treatment futility.
Disclosures
S. Ahmed reports ownership interest in The Kidney Health and Preventive Medicine Institute. D.M. Charytan reports consultancy agreements with Allena Pharmaceuticals (Data Safety Monitoring Board), Amgen, AstraZeneca, Eli Lilly/Boehringer Ingelheim, Fresenius, Gilead, GlaxoSmithKline, Janssen (steering committee), Medtronic, Merck, Novo Nordisk, and PLC Medical (clinical events committee); research funding from Amgen, BioPorto (clinical trial support), Gilead, Medtronic (clinical trial support), and Novo Nordisk; serving as an associate editor of CJASN ; and expert witness fees related to proton pump inhibitors. S. Desai reports research funding from a Society for Diagnosis in Medicine grant. D.E. Leaf reports consultancy agreements with Sidereal Therapeutics and research funding from BioPorto Diagnostics. E.I. Mandel reports employment with Hebrew SeniorLife and serving as an American Society of Nephrology (ASN) Kidney Self-Assessment Program/Nephrology Self-Assessment Program editor. E.I. Mandel's spouse reports employment at a law firm that includes DaVita, DCI, and Fresenius Kidney Care among its clients. G. McMahon reports research funding from Alexion Pharmaceuticals and Allena Pharmaceuticals; serving as a scientific advisor or member of ASN Kidney Disease Screening and Awareness Program and as a scientific coordinator for the United States and Canada for the Anemia Studies in Chronic Kidney Disease: Erythropoiesis via a Novel Prolyl Hydroxylase Inhibitor Daprodustat (ASCEND) trial (GlaxoSmithKline); and other interests/relationships with the Irish Society of Nephrology. M.L. Mendu reports consultancy agreements with Bayer AG. S.S. Waikar reports consultancy agreements with Allena, BioMarin, CVS, GlaxoSmithKline, JNJ, Mallinckrodt, Mass Medical International, Metro Biotechnology, Oxidien, Pfizer, Regeneron, Roth Capital Partners, Sironax, Strataca/3ive, Venbio, and Wolters Kluewer; research funding from Vertex; serving on a Kantum scientific advisory board; and serving as an expert witness for litigation related to the GE product Omniscan, an expert witness for litigation related to the Fresenius product Granuflo, an expert witness for litigation involving cisplatin toxicity, an expert witness for litigation related to the Gilead product tenofovir, and an expert witness for litigation related to DaVita laboratory testing. All remaining authors have nothing to disclose.
Funding
This work was supported by Brigham and Women’s Hospital Department of Medicine Quality Program Award.
Acknowledgments
Because Dr. David M. Charytan is an associate editor of CJASN , he was not involved in the peer review process for this manuscript. Another editor oversaw the peer review and decision making process for this manuscript.
S. Desai, M.L. Mendu, S. Shaykevich, and S.S. Waikar designed the study; D.M. Charytan, P.G. Czarnecki, Y.P. Kelly, D.E. Leaf, E.I. Mandel, G. McMahon, M.L. Mendu, K. Mistry, E. Robinson, and S.S. Waikar acquired data; Y.P. Kelly, D.E. Leaf, S.R. Lipsitz, M.L. Mendu, and S. Shaykevich analyzed data; S. Ahmed, D.M. Charytan, P.G. Czarnecki, S. Desai, Y.P. Kelly, D.E. Leaf, S.R. Lipsitz, E.I. Mandel, G. McMahon, M.L. Mendu, E. Robinson, S. Shaykevich, and S.S. Waikar interpreted data; S. Ahmed, Y.P. Kelly, K. Mistry, and S. Shaykevich drafted the paper; S. Ahmed, D.M. Charytan, P.G. Czarnecki, S. Desai, Y.P. Kelly, D.E. Leaf, S.R. Lipsitz, E.I. Mandel, G. McMahon, M.L. Mendu, K. Mistry, E. Robinson, S. Shaykevich, and S.S. Waikar revised the paper critically; and S. Ahmed, D.M. Charytan, P.G. Czarnecki, S. Desai, Y.P. Kelly, D.E. Leaf, S.R. Lipsitz, E.I. Mandel, G. McMahon, M.L. Mendu, K. Mistry, E. Robinson, S. Shaykevich, and S.S. Waikar gave final approval of the version to be published and agree to be accountable for all aspects of the work with regard to its accuracy and integrity.
Data Sharing Statement
Individual deidentified patient data and data dictionaries will not be shared publicly given concerns for patient confidentiality as this is a single-center study. The data that support the findings of this study are available on request from the senior author (M.L. Mendu). Our study protocol will be available as part of the accompanying supplemental material with this paper.
Supplemental Material
This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.02060221/-/DCSupplemental .
Supplemental Figure 1 . Intervention (SCAMP) versus control (sham) form.
Supplemental Figure 2 . Schedule of intervention (SCAMP) versus control (sham) form including total number of patients and nephrologists in each time period.
Supplemental Figure 3 . Standardized Clinical Assessment and Management Plan form.
Supplemental Figure 4 . Control (“sham”) form.
Supplemental Figure 5 . Standardized Clinical Assessment and Management Plan trial work flow.
Supplemental Table 1 . Clinical characteristics of patients who received KRT in the AKI-SCAMP study.
Supplemental Table 2 . Treatment characteristics in the subgroup that received KRT.
Supplemental Table 3 . Effect of SCAMP on patient outcomes.
Supplemental Table 4 . Reasons for deviation from AKI-SCAMP recommendations.
References
1. Gallagher M, Cass A, Bellomo R, Finfer S, Gattas D, Lee J, Lo S, McGuinness S, Myburgh J, Parke R, Rajbhandari D; POST-RENAL Study Investigators and the ANZICS Clinical Trials Group: Long-term survival and dialysis dependency following acute kidney injury in intensive care: Extended follow-up of a randomized controlled trial. PLoS Med 11: e1001601, 2014
2. Star RA: Treatment of acute renal failure. Kidney Int 54: 1817–1831, 1998
3. Vandijck DM, Oeyen S, Decruyenaere JM, Annemans L, Hoste EA: Acute kidney injury, length of stay, and costs in patients hospitalized in the intensive care unit. Acta Clin Belg 62[Suppl 2]: 341–345, 2007
4. See EJ, Jayasinghe K, Glassford N, Bailey M, Johnson DW, Polkinghorne KR, Toussaint ND, Bellomo R: Long-term risk of adverse outcomes after acute kidney injury: A systematic review and meta-analysis of cohort studies using consensus definitions of exposure. Kidney Int 95: 160–172, 2019
5. Sawhney S, Marks A, Fluck N, McLernon DJ, Prescott GJ, Black C: Acute kidney injury as an independent risk factor for unplanned 90-day hospital readmissions. BMC Nephrol 18: 9, 2017
6. Bagshaw SM, Uchino S, Bellomo R, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Oudemans-van Straaten HM, Ronco C, Kellum JA; Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators: Timing of renal replacement therapy and clinical outcomes in critically ill patients with severe acute kidney injury. J Crit Care 24: 129–140, 2009
7. Gaudry S, Hajage D, Schortgen F, Martin-Lefevre L, Pons B, Boulet E, Boyer A, Chevrel G, Lerolle N, Carpentier D, de Prost N, Lautrette A, Bretagnol A, Mayaux J, Nseir S, Megarbane B, Thirion M, Forel JM, Maizel J, Yonis H, Markowicz P, Thiery G, Tubach F, Ricard JD, Dreyfuss D; AKIKI Study Group: Initiation strategies for renal replacement therapy in the intensive care unit. N Engl J Med 375: 122–133, 2016
8. Barbar SD, Clere-Jehl R, Bourredjem A, Hernu R, Montini F, Bruyère R, Lebert C, Bohé J, Badie J, Eraldi JP, Rigaud JP, Levy B, Siami S, Louis G, Bouadma L, Constantin JM, Mercier E, Klouche K, du Cheyron D, Piton G, Annane D, Jaber S, van der Linden T, Blasco G, Mira JP, Schwebel C, Chimot L, Guiot P, Nay MA, Meziani F, Helms J, Roger C, Louart B, Trusson R, Dargent A, Binquet C, Quenot JP; IDEAL-ICU Trial Investigators and the CRICS TRIGGERSEP Network: Timing of renal-replacement therapy in patients with acute kidney injury and sepsis. N Engl J Med 379: 1431–1442, 2018
9. Bagshaw SM, Wald R, Adhikari NKJ, Bellomo R, da Costa BR, Dreyfuss D, Du B, Gallagher MP, Gaudry S, Hoste EA, Lamontagne F, Joannidis M, Landoni G, Liu KD, McAuley DF, McGuinness SP, Neyra JA, Nichol AD, Ostermann M, Palevsky PM, Pettilä V, Quenot JP, Qiu H, Rochwerg B, Schneider AG, Smith OM, Thomé F, Thorpe KE, Vaara S, Weir M, Wang AY, Young P, Zarbock A; STARRT-AKI Investigators; Canadian Critical Care Trials Group; Australian and New Zealand Intensive Care Society Clinical Trials Group; United Kingdom Critical Care Research Group; Canadian Nephrology Trials Network; Irish Critical Care Trials Group: Timing of initiation of renal-replacement therapy in acute kidney injury. N Engl J Med 383: 240–251, 2020
10. Mendu ML, Ciociolo Jr. GR, McLaughlin SR, Graham DA, Ghazinouri R, Parmar S, Grossier A, Rosen R, Laskowski KR, Riella LV, Robinson ES, Charytan DM, Bonventre JV, Greenberg JO, Waikar SS: A decision-making algorithm for initiation and discontinuation of RRT in severe AKI. Clin J Am Soc Nephrol 12: 228–236, 2017
11. McCambridge J, Witton J, Elbourne DR: Systematic review of the Hawthorne effect: New concepts are needed to study research participation effects. J Clin Epidemiol 67: 267–277, 2014
12. Liang KY, Zeger SL: Longitudinal data analysis using generalized linear models. Biometrika 73: 13–22, 1986
13. Lipsitz SR, Fitzmaurice GM, Orav EJ, Laird NM: Performance of generalized estimating equations in practical situations. Biometrics 50: 270–278, 1994
14. Demirjian S, Chertow GM, Zhang JH, O’Connor TZ, Vitale J, Paganini EP, Palevsky PM; VA/NIH Acute Renal Failure Trial Network: Model to predict mortality in critically ill adults with acute kidney injury. Clin J Am Soc Nephrol 6: 2114–2120, 2011
15. Wei LJ, Lin DY, Weissfeld L: Regression analysis of multivariate incomplete failure time data by modeling marginal distribution. J Am Stat Assoc 84: 1065–1073, 1989
16. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW: Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol 16: 3365–3370, 2005
17. Kolhe NV, Staples D, Reilly T, Merrison D, Mcintyre CW, Fluck RJ, Selby NM, Taal MW: Impact of compliance with a care bundle on acute kidney injury outcomes: A prospective observational study. PLoS One 10: e0132279, 2015
18. Hodgson LE, Roderick PJ, Venn RM, Yao GL, Dimitrov BD, Forni LG: The ICE-AKI study: Impact analysis of a clinical prediction rule and electronic AKI alert in general medical patients. PLoS One 13: e0200584, 2018
19. Selby NM, Casula A, Lamming L, Stoves J, Samarasinghe Y, Lewington AJ, Roberts R, Shah N, Johnson M, Jackson N, Jones C, Lenguerrand E, McDonach E, Fluck RJ, Mohammed MA, Caskey FJ: An organizational-level program of intervention for AKI: A pragmatic stepped wedge cluster randomized trial. J Am Soc Nephrol 30: 505–515, 2019
20. Coyne DW: Influence of industry on renal guideline development. Clin J Am Soc Nephrol 2: 3–7, 2007
21. Norris SL, Holmer HK, Ogden LA, Burda BU: Conflict of interest in clinical practice guideline development: A systematic review. PLoS One 6: e25153, 2011
22. Farias M, Friedman KG, Powell AJ, de Ferranti SD, Marshall AC, Brown DW, Kulik TJ: Dynamic evolution of practice guidelines: Analysis of deviations from assessment and management plans. Pediatrics 130: 93–98, 2012
23. Verghese GR, Friedman KG, Rathod RH, Meiri A, Saleeb SF, Graham DA, Geggel RL, Fulton DR: Resource utilization reduction for evaluation of chest pain in pediatrics using a novel Standardized Clinical Assessment and Management Plan (SCAMP). J Am Heart Assoc 1: jah3-e000349, 2012
24. Porras D, Brown DW, Rathod R, Friedman K, Gauvreau K, Lock JE, Esch JJ, Bergersen L, Marshall AC: Acute outcomes after introduction of a standardized clinical assessment and management plan (SCAMP) for balloon aortic valvuloplasty in congenital aortic stenosis. Congenit Heart Dis 9: 316–325, 2014
25. Farias M, Ziniel S, Rathod RH, Friedman KG, Colan S, Newburger JW, Fulton DR: Provider attitudes toward Standardized Clinical Assessment and Management Plans (SCAMPs). Congenit Heart Dis 6: 558–565, 2011