Care Bundle for Acute Kidney Injury in Cardiac Patients: A Cluster-Randomized Trial

Detection and timely intervention of acute kidney injury (AKI) is a major challenge worldwide. Electronic alerts for AKI may improve process- and patient-related endpoints. The present study evaluated the efficacy of an AKI electronic alert system and care bundle. This is a two-arm, prospective, cluster-randomized, controlled trial enrolling patients with AKI (KDIGO criteria) and cardiac diseases. Patients were randomly assigned to a routine care group or intervention group (DRKS-IDDRKS00017751). Two hundred patients (age 79 years, 46% female) were enrolled, with 100 patients in each group. The primary endpoint did not differ between patients in the routine care group 0.5 (−7.6–10.8) mL/min/1.73 m2 versus patients in the intervention group 1.0 (−13.5–15.1) mL/min/1.73 m2, p = 0.527. Proportions of patients in both study groups with hyperkalemia, pulmonary edema, and renal acidosis were comparable. The stop of antihypertensive medication during hypotensive periods was more frequent in patients in the intervention group compared to patients in the control group, p = 0.029. The AKI diagnosis and text module for AKI in the discharge letter were more frequently documented in patients in the intervention group (40%/48% vs. 25%/34%, p = 0.034; p = 0.044, respectively). Continued intake of RAAS inhibitors and the presence of a cardiac device were independently associated with a less pronounced decrease in eGFR from admission to the lowest value. In this RCT, electronic alerts for AKI and a care bundle improved process- but not patient-related endpoints.


Introduction
Acute kidney injury (AKI) is one of the most serious and common complications affecting inpatient admissions [1].Early detection and appropriate management of AKI are vital to aid kidney recovery and to prevent related adverse outcomes [2].A report by the National Confidence Enquiry into Patient Outcome and Death (NCEPOD) found that only 50% of patients with AKI received the appropriate standard of care [3].In an observation study among survivors of AKI, about 80% of patients were not informed about AKI or nephrotoxic medications [4].Electronic alerts in the hospitals were recommended to improve the recognition of AKI, and an AKI electronic alert system was mandated by NHS England in all Laboratory Information Management Systems across the NHS in 2015 [5].Several trials focusing on patient safety, specialist referral, and clinical management showed positive effects of AKI electronic alert systems, including more frequent medication-related recommendations per patient, a reduced progression to higher AKI stage, emergency readmission to hospital, reduced length of stay in hospital and death during admission [6][7][8], see Appendix A, Table A1.A pragmatic stepped wedge cluster randomized trial showed improved AKI recognition, increased care, and shorter length of hospital stay in the intervention group but did not reduce 30-day AKI mortality [9].However, differences in local context must be considered when study findings are interpreted.Specifically, in the UK, laboratory parameters from both the outpatient and inpatient settings are available to any treating physician at any time, enabling the detection of community-acquired and hospital-acquired AKI.Therefore, related study findings may not be transferable to regions in the world where laboratory parameters, such as serum creatinine, are not available cross-sectorally, and the patient is not being informed about their AKI.
In a cluster-randomized, controlled study, we investigated if an AKI electronic alert system and care bundle would be able to improve process-and patient-related endpoints of patients with AKI.

Setting and Design
This is a two-arm prospective, cluster-randomized, controlled trial to evaluate the efficacy of an AKI electronic alert system and care bundle for hospitalized patients with AKI.The study was conducted at the Department of Cardiology, Heart Center Brandenburg Bernau, University Hospital of the Brandenburg Medical School (MHB), Germany, between January 2019 and March 2022, including a follow-up observation over 3 months regarding renal function and 12 months regarding survival status.
Patients were assigned to the routine care or intervention group using cluster-randomized allocation, alternating at 4-week intervals.The sequence of allocation was specified prior to the start of the study using a random number generator.
This study is registered with the German Clinical Trial Register (DRKS-IDDRKS00017751).Ethical approval of the study protocol was obtained from the local ethics committee of the Brandenburg Medical School (E-01-20181101).

Patients
To be included in the study, patients admitted to the Department of Cardiology had to have evidence of renal insufficiency according to KDIGO criteria [10], be over 18 years of age, and be able to sign a written informed consent form.KDIGO criteria were applied as creatinine increase (hospital-acquired AKI) or creatinine decrease (community-acquired AKI).Hospital-acquired AKI was defined as 0.3 mg/dL or 50% increase, respectively, in serum creatinine between hospital admission and highest serum creatinine within 48 h or 7 days, respectively.To be defined as community-acquired AKI, patients had to have, on admittance to the hospital or within 48 h of admission, a 33% decrease in serum creatinine from baseline within 7 days.
Patients who had already undergone kidney transplantation or who were receiving chronic or acute renal replacement therapy at the time of the study, pregnant patients, patients with HI/Hepatitis virus infections, or patients with a life expectancy below 3 months were excluded.

Care Bundle (Intervention Group)
Prior to the study, an algorithm to detect relevant changes in serum creatinine was developed and embedded in the hospital information system.AKI was detected by an electronic alert system based on the Kidney Disease: Improving Global Outcomes (KDIGO) AKI criteria [10].In addition, a notification system was added to the hospital's SAP work system in the form of a lamp icon to indicate critical patient features and alert the ward physician to the problem of AKI.
When a patient with AKI was identified, the study investigator received an internal email with information about the patient's identity, department, creatinine increase, and the AKI stage, forwarding this information to the ward physician.In educational activities before study initiation, physicians were made aware of the AKI lamp icon, risk factors for AKI, and the KDIGO AKI care bundle to prevent further renal function deterioration and support renal recovery.
Patients in the intervention group were interviewed in detail by a medical colleague, their pre-existing conditions and risk factors were assessed, and their current medication regimen was checked.Measures of the care bundle [10] included the identification of the cause(s) of AKI in the clinical context, measures to achieve euvolemia (through fluid administration or negative balance), pharmacological intervention including discontinuation of nephrotoxic drugs or switch to less nephrotoxic drugs of the same substance class or monitoring of plasma levels, and the adjustment of drug doses to renal function.Also, optimizing hemodynamics, detection, and treatment of electrolyte and acid-base disorders, as well as monitoring of heart and kidney function during the index hospital stay (including blood pressure, heart rate, serum creatinine, diuresis, weight, and fluid balance) and nephrology consultation of patients with AKI stage 3 were part of the care bundle.Additionally, all patients received an information flyer and a kidney passport to record, together with the general practitioner, renal function values, enabling monitoring of these parameters.The passport could also be used to be presented in other medical institutions, such as pharmacies, to advise on alternatives to nephrotoxic medications.Finally, the discharge letter included information about participation in the study, AKI cause and severity, advice on subsequent kidney function tests in the outpatient setting three and twelve months after AKI, avoidance of NSAID, and sick day advice.

Routine Care Group
In the study control period, patients were also contacted by the investigator-but later in the course of AKI-primarily to ask for permission to collect data on renal function within 3 months after AKI.In addition, in patients of the routine care group, a lamp icon in the hospital's SAP work system (SAP Inc., Boston, MA, USA) indicated potential AKI.Patients in the routine care group received standard care for AKI, including the possibility of the ward physician consulting a nephrologist.

Outcomes
The primary outcome was the loss of kidney filtration function from hospital admission to three months after AKI (change of eGFR).Secondary outcomes were the length of stay in the hospital, peak serum creatinine during an index hospital stay, AKI complications (hyperkalemia, kidney-related pulmonary edema, and renal acidosis), chronic renal replacement therapy, major adverse cardiac events (MACE) and rehospitalization within 90 days of the index admission as well as 3-and 12-month mortality.

Data Collection
Patient demographics and hospital data, comorbidities, medications, serum creatinine values during the index hospital stay, and other laboratory values were collected.
The following variables were collected when available: admission serum creatinine (baseline), peak serum creatinine, and discharge serum creatinine, as well as serum creatinine 3 months after discharge.Discharge serum creatinine was defined as serum creatinine measured nearest to the date of hospital discharge.Urine output was not available.
Also, the cause of AKI and AKI-related complications were documented.The surrogate marker of AKI management included information about AKI in the discharge letter (text module and AKI cause) and recommendations for outpatient follow-up.Three months after study enrolment, the primary care physician was contacted, and intercurrent renal function, need for chronic renal replacement therapy, rehospitalization within 90 days of the index admission, and possible cardiovascular events (MACE) were recorded.MACE was defined as acute myocardial infarction (AMI), stroke, or cardiovascular death.

Statistical Analysis
Using data from a published randomized controlled trial [11], we estimated that 86 patients per group would be needed to have a 90% power to detect an absolute difference of 5 mL eGFR loss of patients in the intervention group between hospital admission and 3 months after discharge compared to the control group at a two-sided test with an alpha of 0.05 and a standard deviation of 10 mL eGFR.Assuming a 15% loss to follow-up in the primary endpoint, we aimed to enroll 100 patients per group.
A Mann-Whitney U test was used for non-parametric two-group comparisons.The chi-squared test and Fisher's exact test were used for dichotomous variables for two groups.Multivariable linear regression analysis for the decrease in GFR from admission to the lowest value included clinically relevant variables affecting eGFR.Patients with missing data (missing follow-up data or similar) were excluded from further examination.We report values as median with 25th to 75th percentiles or as a proportion of patients (%) as appropriate.A planned subgroup analysis for patients with community-acquired AKI was performed.Posthoc, additional subgroup analyses were performed for patients with diabetes, female gender, patients aged >70 years, patients with a cardiac device, or those using ACE inhibitors/AT-1 blockers.
A two-tailed p-value of <0.05 was defined as significant.Analysis was performed using SPSS version 27 (SPSS Inc., IBM, Chicago, IL, USA).

Patient Characteristics
Two hundred patients (aged 79 (68-84) years, 46% female) were prospectively randomized to a routine care group or intervention group.Patient flow is shown in Figure 1.The study groups were similar in terms of demographics, laboratory parameters, and most comorbidities (Table 1, Appendix A, Table A2).A higher proportion of patients in the routine care group presented with atrial fibrillation and were admitted electively compared to those in the intervention group.In the intervention group, more patients were admitted with acute coronary syndrome compared to patients in the routine care group (Table 1).Also, more patients in the routine care group received peri-interventional antibiotics (Table 2).

Interventions
The proportions of patients receiving heart-and kidney-function-related interventions, including fluid administration, stopping potentially nephrotoxic medications, and nephrology consultation, were similar in both groups.However, stopping antihypertensive medication during hypotensive periods was more frequent in patients in the intervention group compared to patients in the control group (6% vs. 0%, p = 0.029); see Table 3.

Characteristics of AKI
The course of serum creatinine and eGFR was similar in the study groups (Figure 2a,b).Loss of eGFR from admission to the lowest value was similar in both study groups (−8 vs. −9 mL/min/1.73m 2 , p = 0.586, Table 4).In the routine care group, 59 patients had community-acquired AKI and 41 hospital-acquired AKI, and in the intervention group, 54 and 46 patients, respectively, p = 0.476.The most frequent cause of AKI was pre-renal, with 92% in the routine care group and 83% in the intervention group (Table 4).The severity of AKI was mostly stage 1, with no group differences (Table 1).In the routine care group, the AKI electronic alert was at 4 (2-8) days after hospital admission compared to 3 (2-8) days in the intervention group, p = 0.244.

Characteristics of AKI
The course of serum creatinine and eGFR was similar in the study groups (Figure 2a,b).Loss of eGFR from admission to the lowest value was similar in both study groups (−8 vs. −9 mL/min/1.73m 2 , p = 0.586, Table 4).In the routine care group, 59 patients had community-acquired AKI and 41 hospital-acquired AKI, and in the intervention group, 54 and 46 patients, respectively, p = 0.476.The most frequent cause of AKI was pre-renal, with 92% in the routine care group and 83% in the intervention group (Table 4).The severity of AKI was mostly stage 1, with no group differences (Table 1).In the routine care group, the AKI electronic alert was at 4 (2-8) days after hospital admission compared to 3 (2-8) days in the intervention group, p = 0.244.
Also, the proportion of patients in both study groups with AKI-related complications, including hyperkalemia, pulmonary edema, and renal acidosis, were comparable (Table 5).Also, the proportion of patients in both study groups with AKI-related complications, including hyperkalemia, pulmonary edema, and renal acidosis, were comparable (Table 5).

Process-Related Endpoints
In patients of the intervention group, the AKI diagnosis and text module for AKI in the discharge letter of the index hospital stay were more frequently documented compared to those in the routine care group (40%/48% vs. 25%/34%, p = 0.034; p = 0.044, respectively, Figure 3).

Process-Related Endpoints
In patients of the intervention group, the AKI diagnosis and text module for AKI in the discharge letter of the index hospital stay were more frequently documented compared to those in the routine care group (40%/48% vs. 25%/34%, p = 0.034; p = 0.044, respectively, Figure 3).

Independent Modifiers of Change in eGFR
Continued intake of RAAS inhibitors (regression coefficient −7.42, p = 0.032) and presence of a cardiac device (regression coefficient −5.25, p = 0.037) were independently associated with a less pronounced decrease in eGFR from admission to the lowest value (Table 5).

Other Patient Outcomes
The rehospitalization within 90 days was 21.5% (24% the routine care group vs. 19% in the intervention group), Figure 4. Sixteen patients died within 12 months, seven in the routine care group and nine in the intervention group.The proportion of patients developing MACE or requiring chronic renal replacement therapy within 90 days did not differ between both study groups (Figure 5).

Independent Modifiers of Change in eGFR
Continued intake of RAAS inhibitors (regression coefficient −7.42, p = 0.032) and presence of a cardiac device (regression coefficient −5.25, p = 0.037) were independently associated with a less pronounced decrease in eGFR from admission to the lowest value (Table 5).

Other Patient Outcomes
The rehospitalization rate within 90 days was 21.5% (24% in the routine care group vs. 19% in the intervention group), Figure 4. Sixteen patients died within 12 months, seven in the routine care group and nine in the intervention group.The proportion of patients developing MACE or requiring chronic renal replacement therapy within 90 days did not differ between both study groups (Figure 5).

Independent Modifiers of Change in eGFR
Continued intake of RAAS inhibitors (regression coefficient −7.42, p = 0.032) and presence of a cardiac device (regression coefficient −5.25, p = 0.037) were independently associated with a less pronounced decrease in eGFR from admission to the lowest value (Table 5).

Other Patient Outcomes
The rehospitalization rate within 90 days was 21.5% (24% in the routine care group vs. 19% in the intervention group), Figure 4. Sixteen patients died within 12 months, seven in the routine care group and nine in the intervention group.The proportion of patients developing MACE or requiring chronic renal replacement therapy within 90 days did not differ between both study groups (Figure 5).

Subgroup Analyses
Excluding patients with hospital-acquired AKI, change of eGFR from hospital admission to three months after AKI did not differ between patients in the routine care group 9.0 (−3.6-18.5)mL/min/1.73m 2 versus patients in the intensive care group 7.0 (−4.7-22.2) mL/min/1.73m 2 , p = 0.886.Also, three months after discharge, eGFR was similar in patients with community-acquired AKI in the routine care group 52.9 (33.5-66.8)mL/min/1.73m 2 compared to those in the intensive care group 50.0 (41.0-77.0)mL/min/1.73m 2 ), p = 0.647.
Also, posthoc analyses of other patient subgroups (patients with diabetes, patients aged > 70 years, female patients, and those with a cardiac device or using an ACE inhibitor/ AT-1 blocker) revealed no significant intervention effect on the primary study endpoint, Table 6 (all p > 0.05).
Table 6.Analyses of potential intervention effects of the primary study endpoint in subgroups of patients.

Discussion
This study randomized two hundred cardiac patients to the intervention or routine care groups.The intervention consisted of an AKI electronic alert system combined with a care bundle, including education of physicians according to the KDIGO recommendations [10], medication intervention, and information about AKI to the patient, the attending physician, and the primary care physician.The primary study endpoint, loss of eGFR from admission to three months after AKI, did not differ between the study groups.Also, secondary endpoints, including loss of eGFR in hospital, proportions of patients with AKI-related complications, and magnitude of kidney recovery until three months after AKI, were comparable between the study groups.Proportions of patients receiving heart and kidney function-related interventions were similar in both groups.In the intervention group, the general physician was more frequently provided with more comprehensive information regarding AKI in the discharge letter.Finally, continued intake of RAAS inhibitors and the presence of a cardiac device were independently associated with a less pronounced decrease in eGFR from admission to the lowest value.
Recent studies of AKI electronic alert and clinical decision support systems demonstrated variable results, which likely result from differences in study design, patient population, local context, and implementation strategies.Non-randomized studies evaluating AKI eAlerts enrolled heterogenous hospitalized patients, frequently used pre-and postdesign, and reported a reduction of higher AKI stages, requirement of renal replacement therapy, length of stay, and in-hospital mortality [12][13][14].Randomized controlled trials also included heterogeneous patient populations of hospitalized patients from all wards; however, they did not demonstrate patient benefit regarding mortality, renal replacement therapy requirement, or renal function recovery (Appendix A, Table A1).Most RCTs reported improvement in process-related parameters, including discontinuation of nephrotoxic medications, involvement of a nephrologist, and documentation in patient medical records [15].
Patients admitted to a cardiology ward may be different from patients admitted to other departments regarding etiology, timing, actionability, and recovery of AKI.Cardio-renal syndrome appears to be the major cause of AKI in cardiac patients, potentially requiring specific work-up and treatment [16,17].None of the recent studies evaluated the impact of AKI electronic alert systems exclusively in a cardiac patient population.Therefore, a study investigating the efficacy of an AKI electronic alert system in patients exclusively admitted for cardiac diseases was needed.
In this study, we observed a protective effect of RAAS inhibitors on the kidney.This is in line with experimental and clinical studies showing, in most cases, that RAAS inhibitors reduce proteinuria renal fibrosis, slow the decline of renal function, and protect against cardiovascular events.However, there are also data from an observational cohort study proposing that discontinuation of RAAS inhibitors in patients with advanced CKD may increase eGFR or slow its decline [18].Findings of a recent multicenter randomized study assigning 411 patients with advanced CKD to discontinuation or continuation of RAAS inhibitors found that discontinuation was not associated with a significant between-group difference in the long-term rate of a decrease in the eGFR [19].Also, a recent meta-analysis found that the continuation of RAAS inhibitors may benefit patients with CKD [20].Overall, data on the use of RAAS inhibitors and the course of renal function suggest that renal function is protected, at least in the non-acute state.Also, we found an inverse association between the use of cardiac devices and eGFR decline.Cardiac implantable electronic devices may preserve central venous circulation and improve left ventricular function [21].
The present study reported the first-time impact of an AKI electronic alert system and care bundle on cardiac patients.Patient-related endpoints were not different from RCTs in general hospital patient populations.
There may be several explanations why our intervention failed to demonstrate a patient benefit.In the present study, most patients developed AKI stage 1 or presented with community-acquired AKI with already recovering renal function during the days after admission, even without further intervention.Both factors may have diminished the chance of an intervention effect.Also, medication interventions used to treat AKI or cardio-renal syndrome on a cardiology ward were similar in both study groups except for the more frequent stop of antihypertensive medication in the intervention group, however, being a rare event with 6% of cases.Intake of RAAS inhibitors-being kidney protective also in this study-was continued as is current practice in many cardiology departments and may have reduced the intervention gradient between the study groups.
In addition, patient characteristics slightly differed between study groups, with more patients presenting with acute coronary syndrome and fewer patients developing infections in the intervention group, potentially diminishing the intervention gradient.
This study aimed to reduce selection bias using time-period clustered randomization and focused on a typical patient cohort with cardiorenal syndrome, which is widely spread among hospitalized patients.Patient follow-up was extended to 3 months for kidney function as recommended by KDIGO [10].However, the study care bundle appeared to be not fully applied in the intervention group, including restriction of nephrology consultation to patients with severe AKI.Early nephrology consultation of patients with AKI may lead to better patient outcomes, as previously shown in a retrospective study by Meier et al. [22].The lack of stratified patient randomization and monitoring of adherence regarding measures of the care bundle limited this study.Although we left the treating physicians unaware of whether the patient developed AKI during an intervention or control period, we acknowledge the potential carryover effect in treatment from the intervention to the control period as a study limitation.
The inclusion of patients with all stages of AKI inherently resulted in a greater proportion of patients with mild AKI versus those with severe AKI.Therefore, this study cannot exclude the effects of the study intervention in patients with severe AKI.Finally, we cannot exclude carryover effects from the intervention to the routine care group with measures of the study care bundle also applied in the control group.
In sum, this study informs the cardiologist about the effects of an AKI electronic alert system in a typical cardiac patient cohort, separated for community and hospital-acquired AKI.
Subsequent studies are preferred to be multicentric and may focus on patients with more severe AKI or routinely include nephrology consultation.

Conclusions
In this RCT, electronic alerts and a care bundle for AKI improved the process-but not patient-related endpoints in cardiac patients.

Figure 2 .
Figure 2. (a) Median serum creatinine concentration over time in both study groups.(b) Median eGFR concentration at admission, lowest during hospitalization, and 3 months after discharge in both study groups.

Figure 2 .
Figure 2. (a) Median serum creatinine concentration over time in both study groups.(b) Median eGFR concentration at admission, lowest during hospitalization, and 3 months after discharge in both study groups.

Figure 5 .
Figure 5. Survival analysis.COX'S proportional hazards regression model for 12-month mortality adjusting for acute coronary syndrome, atrial fibrillation, cardiac devices, and the use of ACE inhibitors or AT-1 blockers.

Figure 5 .
Figure 5. Survival analysis.COX'S proportional hazards regression model for 12-month mortality adjusting for acute coronary syndrome, atrial fibrillation, cardiac devices, and the use of ACE inhibitors or AT-1 blockers.

Figure 5 .
Figure 5. Survival analysis.COX'S proportional hazards regression model for 12-month mortality adjusting for acute coronary syndrome, atrial fibrillation, cardiac devices, and the use of ACE inhibitors or AT-1 blockers.

Table 2 .
Medication during hospital stay.

Table 5 .
Multivariable linear regression analysis for eGFR decrease from admission to lowest value.

Table 5 .
Multivariable linear regression analysis for eGFR decrease from admission to lowest value.

Table A1 .
Cont.The sensitivity, specificity, Youden Index, and accuracy of the AKI e-alert system were 99.8, 97.7, 97.5 and 98.1%, respectively AKI, acute kidney injury; RRT, renal replacement therapy; LOS, length of stay in hospital.* All studies used a (modified) KDIGO-guideline-related AKI care bundle.