The Optimization of Outpatient Hemodialysis Management for Acute Kidney Injury Requiring Dialysis Patients: A Quality Improvement Study

Acute kidney injury (AKI) occurs frequently in hospitalized patients with an incidence ranging from 10 to 20% [1, 2]. Among AKI patients, up to 10% require dialysis for solute and/or volume control (AKI-D) [1, 3]. Patients with AKI-D have an increased risk of morbidity and mortality when compared to patients with less severe forms of AKI or without AKI [4, 6]. Patients with AKI-D require personalized dialysis prescriptions for solute control and particularly for fluid management as they respond diversely to ultrafiltration rates (UFRs) and need closer monitoring for intradialytic hypotension [7]. Patients who survive hospitalized AKI-D have an increased risk of rehospitalization, progressive kidney disease, and developing cardiovascular disease [5, 8, 11]. Further, patients with AKI-D who progress to end-stage kidney disease (ESKD) have a greater risk of death as compared with AKI-D patients who regain kidney function [6].

Currently, there is growing evidence pertaining to kidney recovery of patients with AKI-D within hospitals; however, the same is not true for outpatient and rehabilitation facilities. For example, post-discharge kidney function recovery in AKI-D patients was assessed in few studies, with results ranging between 19 and 65%, most of them during the first 3 months after discharge [12, 15]. The likelihood of kidney recovery in these patients – defined as dialysis-free survival – was associated with hemodialysis (HD) factors such as fewer episodes of intradialytic hypotension, lower fluid removal, and fewer HD sessions [13, 15]. Nevertheless, only few studies have prospectively assessed kidney recovery in patients discharged to community dialysis or rehabilitation facilities.

The need for improving outpatient AKI-D care has grown since the policy change by the Centers for Medicare and Medicaid Services (CMS) in January 2017 that supports reimbursement [16]. However, there is a paucity of data on processes to effectively and safely treat AKI-D patients in outpatient dialysis facilities, which treatment protocols are feasible, and which risk factors could potentially be modifiable and/or used to predict recovery of kidney function. In this quality improvement study, we assess the feasibility of implementing a management protocol for AKI-D patients receiving HD in a single outpatient dialysis facility, evaluate protocol adherence, describe proponents and hindrances to implementation, and report the clinical data of the first 9 patients treated under the protocol.

This quality improvement study included adult patients with AKI-D who were admitted to an outpatient dialysis facility of a large dialysis organization (LDO) affiliated with the University of Kentucky Hospital. These patients were discharged from a recent hospitalization with the diagnosis of AKI-D and still needed HD after discharge. Patients were accepted into the LDO based on standard procedures in coordination with the medical director of the unit to confirm AKI-D diagnosis. Kidney transplant recipients in need of HD due to failing allografts and patients with a prior diagnosis of ESKD were excluded from the study. Patients were enrolled between July 2020 and November 2021 and were followed up for approximately 90 days after admission to the HD unit or until the time of kidney recovery, transfer to another HD facility, death/hospice, kidney transplant, or transition to ESKD diagnosis. The study was approved by the University of Kentucky Institutional Review Board (IRB #43159) and waiver of informed consent was granted given the quality improvement design of the study.

The “Wildcat” AKI-D protocol (online suppl. Item S1; for all online suppl. material, see https://doi.org/10.1159/000530444) was conceptualized based on the available literature and the experience of the multidisciplinary team [17]. Evidence extrapolated from ESKD populations suggests that higher UFRs are associated with more intradialytic hypotensive events and increased mortality [18, 20]. However, while the conventional recommendation is to limit UFR to 10 mL/kg/h to avoid complications, UFR thresholds are dynamic and patient-specific, and volume removal response depends on different cardiovascular and hemodynamic factors and volume status [21, 22]. Therefore, we developed a protocol centered on two principles: (1) weekly assessment of HD goals and status of kidney recovery and (2) prevention of intradialytic hypotension by adjusting dose/UFR and frequency of HD. We assembled a protocol based on three dose/UFR and three frequency tiers that were customized weekly for individual HD treatment goals. Dose/UFR and frequency tiers are specified in Figure 1, whereas tier 1 was of lowest dose/UFR intensity for patients clinically assessed at high risk of intradialytic hypotension, and frequency A was of 3 times per week HD. Of note, the protocol did not mandate specific escalation or de-escalation of HD but provided a framework for personalized AKI-D care by a multidisciplinary team led by a nephrologist. The AKI-D protocol used standard electrolyte baths for patients with ESKD (e.g., potassium 3 mEq/L and calcium 2.5 mEq/L) and HD temperature of 36°C for patients on tiers 1 and 2 (Fig. 1). The decision about the HD prescription was led by the clinician in the team.

The AKI-D outpatient protocol included a multidisciplinary team comprised of nephrologists, pharmacists, nurses, psychologists, nutritionists, social workers, and technicians. The team was assembled considering the infrastructure available for the management of patients with ESKD but was trained to support the specific needs of patients with AKI-D (e.g., frequent monitoring of kidney recovery and dynamic HD prescriptions, etc.). Prior to implementing the protocol, the team reviewed the protocol and its compliance with safety standards in the LDO. The team was in constant communication with patients and their care partners to incorporate their feedback.

The team evaluated the patients weekly and adjusted the HD prescriptions and medications according to the varying clinical status of the patients. A spreadsheet was developed to collect relevant patient information from their hospitalization, HD prescription, medications, laboratory data, and pre-, post-, and intradialytic hemodynamic parameters. The AKI-D protocol incorporated standard items related to hepatitis screening, management of anemia, and heparin use for anticoagulation of ESKD patients receiving outpatient HD [23]. The AKI-D protocol did not incorporate mineral metabolism management of ESKD patients. The step-by-step protocol implementation is summarized in Figure 2 and described below:

1. 1.

   Outpatient dialysis facility enrollment.

   • –

     Hospital provider refers a patient with the diagnosis of AKI-D to the outpatient HD facility.

   • –

     Clinical manager and medical director review eligibility criteria for AKI-D to determine acceptance.

   • –

     If eligibility criteria are met, the patient provides authorization for HD treatment and is provided with a first appointment.

2. 2.

   Multidisciplinary team evaluation to determine individual patient needs.

3. 3.

   Protocol-based AKI-D care.

   • –

     Clinician determines the initial HD prescription with a focus on dose/UFR and frequency tier allocation.

   • –

     Dialysis nurse proceeds to activate the protocol.

     • Monitors the patient’s clinical status and tolerance during the procedure: anaphylactic reactions, hemodynamic changes, mental status, HD parameters, etc.

   • –

     Clinician and pharmacist perform personalized medication management that includes medication reconciliation and refills with a focus on blood pressure and volume management including the use of disease-modifying agents such as renin-angiotensin-aldosterone inhibitors and diuretics, as well as avoidance of potential nephrotoxic drugs such as nonsteroidal anti-inflammatory drugs and proton-pump inhibitors.

   • –

     Clinician assesses the patient on a weekly basis for adjustments in prescription.

     • Clinical trajectory (e.g., patient signs and symptoms, medication access and adherence, etc.).

     • HD prescription (e.g., change in the tier of dose/UFR and/or frequency, tolerance to HD, etc.).

     • Protocol adherence.

     • Kidney recovery based on serum creatinine, BUN, electrolyte panel, and 24-h urine measurements of creatinine and urea clearance (urine testing for solute clearance is only performed when 24-h urine output volume measurement is at least 500 mL). Patients provide weekly 24-h urine collections at the beginning of the week which are used for clinical evaluation later in the week.

4. 4.

   Patient empowerment.

   • –

     Effective and constant communication between the patient/care partners and the team to address concerns and inquiries about patient care.

   • –

     Provision of kidney care education (e.g., avoidance of nephrotoxins, blood pressure and glycemic control, monitoring of kidney recovery, medication compliance, etc.).

   • –

     Provision of psychological and nutritional counseling.

   • –

     Patient as the key stakeholder in AKI-D care.

The following outcomes and definitions were determined a priori. Intradialytic hypotension events were measured for each HD session according to the following definitions [24]:

1. 1.

   Definition A: Systolic blood pressure (SBP) decreases greater than or equal to 20 mm Hg, or mean arterial pressure decreases greater than or equal to 10 mm Hg. The change in the SBP and mean arterial pressure was measured in three independent ways: pre- to post-HD, pre-HD to nadir during HD, and peak to nadir during HD (the nadir must occur after the peak).

2. 2.

   Definition B: Definition A or SBP nadir below a predefined threshold (SBP below 100 mm Hg if pre-HD SBP is > 160 mm Hg or below 90 mm Hg if pre-HD SBP is ≤ 160 mm Hg).

Blood pressure variability was assessed using the following calculations [25]:

1. 1.

   Within-subject variability: Coefficient of variability = 100 × (ASBP/mean SBP); whereas ASBP is the mean absolute difference between SBP adjacent readings.

2. 2.

   Within-subject overall variability: Coefficient of variability = 100 × (SD of SBP readings/mean SBP); whereas SD is the standard deviation. In addition, sensitivity analyses were performed comparing the highest SBP reading with the mean of SBP readings in each session and assessing diastolic blood pressure (DBP) variability.

Kidney recovery was defined as being alive and no longer needing HD due to regain of kidney function. The status of “No recovery” consisted of death, transition to hospice, receipt of kidney transplantation, or declaration of ESKD. Protocol adherence was calculated according to the total treatment hours of HD. When the dose/UFR prescription was maintained at or below the corresponding tier threshold, it was considered adherent to the protocol. Protocol adherence per HD session was calculated by dividing the adherent treatment hours by the number of total treatment hours with available data.

Medians and interquartile ranges were used for describing continuous variables and counts and percentages for categorical variables. Mann-Whitney U test and χ² test were used to compare continuous and categorical variables, respectively, when appropriate. The trajectories in the figures were evaluated individually by autoregressive integrated moving average. In these analyses, the p value denotes the F-test comparison of autoregressive integrated moving average model parameters between the trajectories of the two kidney recovery groups. The correlation between delta weight and UFR was assessed with Person correlation coefficient. Statistical analyses were performed using R Programming, and p < 0.05 was considered statistically significant.

Twelve patients were referred to the LDO with AKI-D diagnosis during the study period. Three patients were then excluded from the study: two because of advanced baseline chronic kidney disease (CKD) status (eGFR <15) and the inability to differentiate AKI on CKD versus CKD progression and one because the patient did not have AKI, but advanced heart failure and the referral were specifically for fluid management in preparation for heart transplantation.

Data pertaining to 9 patients treated under the AKI-D protocol are presented. Patient demographics and hospitalization characteristics are detailed in Table 1. The median age was 53 (IQR: 44–60) years, and 6 patients were male. The median length of hospital stay was 18 (IQR: 13–26) days, and 6 patients were admitted to the intensive care unit for a median of 7 (IQR: 5–7) days. Regarding AKI etiology, all patients had a clinical pattern of acute tubular necrosis. Baseline CKD before AKI was present in 5 patients. Six patients received only HD and three received continuous kidney replacement therapy and then HD. The median time on continuous kidney replacement therapy was 4 (IQR: 3–10) days, and the overall length of inpatient dialysis exposure was 15 (IQR: 8–17) days.

Among the 491.7 h of total outpatient HD time in these 9 patients, 457.3 (93%) hours were interrogated for protocol adherence due to data availability. Overall protocol adherence was 53.8%, which was not different according to kidney recovery groups: 54.5% for those that recovered kidney function and 53.7% for those that did not. Protocol adherence below 50% was more frequent in patients who did not recover kidney function (42.9% vs. 33.3%) (online suppl. Fig. S1). Protocol adherence was associated with less frequent intradialytic hypotension events, specifically when pre- to post-HD BP was used (28.0% vs. 41.7% for definition A, p = 0.002) and peak to nadir BP during HD was interrogated in both definitions (69.6% vs. 82.5%, p = 0.001 for definition A and 74.1% vs. 85.8%, p = 0.002 for definition B) (Fig. 3). Protocol adherence up to the first seven HD sessions for each patient stratified by recovery group is presented in online supplementary Figure S2.

In Table 2, we describe the outpatient HD prescription, and intradialytic hemodynamics stratified by kidney recovery status. Patients who recovered kidney function had higher intradialytic SBP (135 [128–145] vs. 118 [103–136] mm Hg, p < 0.001) and DBP (89 [87–95] vs. 75 [65–85] mm Hg, p < 0.001) during HD sessions. The trajectories of SBP, DBP, and heart rate up to the first seven HD sessions are presented in Figure 4. In addition, the trajectory of SBP and DBP of each individual patient is presented in online supplementary Figure S3.

Regarding intradialytic hypotension, there were more pre- to post-HD events in patients that did not recover kidney function: 37.9% versus 22.2% (definition A) and 53.6% versus 25.9% (definition B), respectively. All interrogations of intradialytic hypotension according to multiple definitions and kidney recovery status are reported in Table 2 and online supplementary Figure S4. There were no major differences in SBP variability among recovery groups, but there was greater DBP within-subject overall variability in patients who did not recover versus those that recovered kidney function (10.2 [7.2–15.1] vs. 8.8 [6.5–10.1], p = 0.03) (Table 2; online suppl. Fig. S5, S6).

A total of 3 patients were alive and liberated from HD within the first 90 days of outpatient HD. The median time from the first outpatient HD session to kidney recovery was 14 (IQR: 4–52) days. Among the 6 patients who did not recover kidney function, one received a simultaneous liver-kidney transplant, four were declared ESKD requiring chronic HD, and one died due to complications of his underlying malignancy.

Patients who recovered kidney function were younger than those who did not (44 [35–60] vs. 56 [50–68] years) and had less prevalent CKD before AKI (33.3% vs. 66.7%) (online suppl. Table 1). Patients who recovered kidney function had higher pre- and post-HD SBP and DBP (Table 3). There was a significant inverse correlation between pre- to post-HD change in weight and UFR in both kidney recovery groups (online suppl. Fig. S7). The trajectories of these variables are presented in online supplementary Figure S8.

This study demonstrates that implementing AKI-D protocols for HD management in outpatient HD units is feasible and sustainable. The keys to effective implementation are the multidisciplinary team, the optimal communication among healthcare professionals and patients, the continuous education about AKI and how to assess kidney recovery, and the empowerment of patients and their care partners. The tier-based HD prescription was based on two principles: (1) weekly assessment of HD goals and kidney recovery status and (2) prevention of intradialytic hypotension by adjusting dose/UFR and frequency of HD. Protocol adherence with the tier-based HD prescription was observed in 53.8% of HD treatment time, with fewer episodes of intradialytic hypotension during periods of protocol adherence.

Similar to prior studies focusing on AKI-D patients, our study showed a trend toward more episodes of intradialytic hypotension during outpatient HD in patients who did not recover kidney function. Pajewski et al. [15] reported that ≥3 episodes of intradialytic hypotension during outpatient HD were more frequent in patients who did not recover versus recovered kidney function (77.8% vs. 22.2%, respectively). McAdams et al. [13] found that each episode of intradialytic hypotension during HD in a rehabilitation facility was associated with a 20% decreased odds of kidney recovery. In contrast, Jordan et al. [14] did not find a significant relationship between intradialytic hypotension and kidney recovery, although they only evaluated intradialytic hypotension in HD sessions during the AKI-D hospitalization but not after discharge. In the context of AKI-D outpatient management, intradialytic hypotension could represent a modifiable risk factor, but its occurrence could be multifactorial including prescription- and patient-related factors. While the role of intradialytic hypotension in AKI-D recovery has not been fully elucidated, one should recognize that intradialytic hypotension and high UFR (i.e., >10 mL/kg/h) have been consistently associated with cardiovascular disease and mortality in the ESKD population [20, 26, 28]. Therefore, developing AKI-D protocols to minimize intradialytic hypotension – such as the one reported here – seems to be a logical approach. Further, other parameters such as intra- and inter-dialytic BP variability need to be explored, as BP variability has been associated with progressive kidney and cardiovascular disease in persons with CKD [29, 30].

In our study, one-third of AKI-D patients recovered kidney function within the first 3 months of outpatient HD. Patients who recovered were younger, had less prevalent CKD before AKI, and received fewer outpatient HD sessions. Other studies have reported variable AKI-D recovery rates within the first 3 months after discharge ranging between 19% and 65%, although most studies exhibit a common pattern of higher rates of recovery in the first versus subsequent months after discharge [12, 15]. Collectively, studies have shown that patient-specific factors such as older age, lower baseline kidney function, and the presence of specific comorbidities such as anemia or heart failure are associated with lower chances of kidney recovery [12, 15]. The all-cause mortality rate in our study was 11.1%, similar to mortality rates within the same observation period reported in AKI-D patients ranging from 9 to 15% [12, 15].

In January 2017, the CMS changed their policy to allow coverage for patients with AKI-D to receive HD in outpatient dialysis facilities [16]. However, more than 5 years later, there are no standardized key performance indicators to monitor the quality of AKI-D care in these facilities. Similarly, kidney recovery rates from LDOs are not publicly available, and importantly, outpatient HD data are not accessible to independent investigators to evaluate risk-classification of kidney recovery and identify potentially modifiable risk factors that can be targeted in clinical trials. Further, the paucity of outcome data and the lack of evidence-based guidelines additionally contribute to the problem of heterogeneity in clinical practice and HD prescriptions in outpatient facilities.

This quality improvement study has limitations to acknowledge. First, the reported AKI-D protocol was developed and implemented in a single outpatient dialysis facility dedicated to AKI-D patients and may not be feasible or reproducible in other facilities without similar logistics and personnel. Second, the study lacked a control group for comparison, and therefore we could not rigorously evaluate if the protocol was effective in mitigating intradialytic hypotension or promoting kidney recovery as conceptualized. Similarly, the small patient sample size precluded statistical power to further interpret highlighted comparisons. Finally, we did not collect patient-reported outcomes or perform a qualitative assessment of patient’s feedback to the protocol. The latter is critical for protocol refinement and future testing. Nonetheless, our study has important strengths to highlight. This is the first study to report the development and implementation of an outpatient AKI-D protocol after the CMS change in policy to allow outpatient HD coverage for AKI patients. Second, our report adheres to the SQUIRE guidelines for quality improvement studies [31]. Third, the reported AKI-D protocol could assist clinicians in improving AKI-D outpatient care in other dialysis facilities, given that the protocol does not mandate specific escalation or de-escalation of HD but provides a framework for implementing personalized AKI-D care in outpatient HD units.

In conclusion, we demonstrated the feasibility of implementing a management protocol for AKI-D patients admitted to an outpatient dialysis facility. Our protocol was based on (1) current evidence and the expertise of a multidisciplinary team, (2) tier-based HD prescriptions, (3) frequent monitoring of pre-defined outcomes (e.g., weekly evaluation of kidney recovery), and (4) patient/care partner empowerment. Protocol adherence with the tier-based HD prescription occurred in more than half of HD treatment time and fewer episodes of intradialytic hypotension were observed in HD treatments that were adherent to the protocol. The impact of protocol-based AKI-D care on intradialytic hemodynamics and kidney recovery requires further investigation in clinical trials. Similarly, qualitative research is necessary to evaluate patient perceptions of care and incorporate patient feedback in the development of additional interventions to improve overall patient wellness in AKI-D outpatient care.

The authors would like to thank the clinicians, pharmacists, nurses, psychologists, nutritionists, social workers, and technicians for their dedication to the development and implementation of this protocol.

The University of Kentucky Institutional Review Board approved this study (IRB #43159) and waived the need of written informed consent given the retrospective nature of data collection and the quality improvement study design. The study was conducted in accordance with the Declaration of Helsinki.

JAN reports consulting honoraria from Baxter Healthcare, Vifor Pharma, Outset Medical, and Leadiant Biosciences; participation on DSMB Chair for STOP-AKI trial; and leadership or fiduciary role for ASN Training and Workforce Committee and ASN AKINow Steering Committee. All remaining authors have nothing to disclose.

No funding was received for the development or implementation of the study. JAN is supported by grants from NIDDK (R01DK128208, U01DK12998, and P30 DK079337).

Research idea and study design: Javier A. Neyra; data acquisition: Victor Ortiz-Soriano, Brian Armentrout, Durham Colohan, and Ruchir Paladiya; analysis/interpretation: Victor Ortiz-Soriano, Augusto Cama-Olivares, Lucas J. Liu, Seda Babroudi, Juan-Carlos Aycinena, and Javier A. Neyra; writing of the first draft manuscript: Augusto Cama-Olivares and Javier A. Neyra. All authors reviewed the results and approved the final version of the manuscript.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author JAN upon reasonable request.

Victor Ortiz-Soriano and Augusto Cama-Olivares contributed equally to this work.