Among VA medical centers with capabilities for inpatient HD, the use of prolonged or continuous renal replacement therapy in the ICU is not uncommon. Of the 55 centers that offer PIRRT and/or CRRT, more than three-quarters reported some form of continuous therapy. While not directly assessed in this study, practitioner preference, facility-level constraints related to staffing, cost, and utility, available surgical services, and coordination with ICU staff are likely some of the determinants of therapy availability. Continuous therapies offer a practical advantage of gradual solute and volume control, particularly useful in post-operative patients with acute kidney injury.
The above conveniences of CRRT comes at the cost of additional staffing and equipment, both of which require training, vigilant monitoring, and management. Accordingly, facilities with higher level of operative complexity, and therefore more complex surgical services offered, were more likely to be equipped with CRRT. Flexible staffing presents a critical constraint that often determines therapy availability. CRRT is mostly set-up and managed by ICU nurses while PIRRT, particularly SLED, appears to have more involvement of HD nurses. While not assessed in this study, this division and staffing availability from either department could possibly influence the modes of therapy made available within a center.
One-to-one nurse staffing was employed among the majority of VA facilities that offered prolonged or continuous therapy. This staffing model supports VA National Center for Patient Safety’s recommendation for continuous visual monitoring of dialysis vascular access, especially in patients at highest risk for bleeding from access dislodgement, such as those receiving RRT in the ICU setting. More thorough investigations of staffing models are needed to better understand responsiveness and ability to troubleshooting circuit failure and how well prescribed therapies are delivered. As critical care complexity grows with extracorporeal membrane oxygenation and ventricular assist devices, nursing teams dedicated to RRT initiation and troubleshooting may become valuable.
More than half of VA facilities surveyed are PIRRT-capable. Comparative advantages relative to CRRT include less depletion of water-soluble minerals that are easily filtered and reduced losses of amino acids and protein, potentially enabling achievement of recommended net protein balance[14, 15]. While PIRRT is understudied in this regard and dosing is highly variable, the extended periods without clearance could potentially result in fewer nutritional losses, which may be key to some patients’ recoveries. Additionally, the cost of staffing with additional one-to-one nursing time, the increased potential for clotting and blood loss, and reduced patient mobility all present additional setbacks to CRRT use. PIRRT fills a need of transitional therapy that limits many of the detrimental aspects of CRRT and the cost of additional staffing. However, as discussed later, drug dosing is not well understood and may limit providers’ comfort with use in patients with critical needs for accurate dosing of medications within therapeutic indices. Importantly, while no modality has demonstrated superiority in large generalizable trials, with further study there potentially exists an opportunity to outline select patients for whom PIRRT offers less complications.
Both the U.S. General Service Administration’s centralized procurement system and the choice of RRT may influence purchasing decisions among VA centers. Survey findings indicate that one brand’s therapy fluids are most commonly used, likely because its replacement fluid and dialysates are compatible with all forms of continuous therapy (CVVH, CVVHD, CVVHDF) and with both CRRT devices used within the VA system (see Supplemental Table S1). The majority of VA institutions with CRRT capabilities utilize a single vendor’s device and therapy fluid. Therapy fluids content differs minimally across the three major manufacturers reported to be used by VA facilities, and therefore, is unlikely to yield distinct clinical outcomes per solution type.
Regarding catheter care, and specifically locking solutions, limited quality evidence exists to support the use of any particular packing solution over another, and heparin use has become quite common. Our study similarly reflects that heparin is the most commonly used locking agent among VA centers. Whether it presents advantages over saline remains uncertain, and yet it is the standard by which alternative anticoagulants are assessed. In a Cochrane review of 27 studies conducted by Wang and colleagues, tissue plasminogen activator was the only agent that reduced catheter malfunction over standard of care. However, the bleeding concern associated with injection or leak is a considerable deterrent, especially in surgical intensive care units. Notably, only one VA facility reported using tissue plasminogen activator for locking catheters. While not uniformly consistent, some studies suggest a reduced incidence of catheter-related bacteremia with citrate use. Importantly, most of these studies were not conducted in AKI requiring RRT, and the largest study to date to compare heparin vs. citrate use in patients with AKI requiring RRT via non-tunneled dialysis catheter noted no difference in event-free survival of catheter, thrombosis, infection, or adverse events. Thus, currently, no uniform recommendation regarding catheter-locking can be suggested.
Slightly more than half of respondents reported using triple-lumen temporary catheters. The impact of this additional port on catheter related infection is unknown, but just as muddy is the potential for increased clearance of vital medicines, particularly antibiotics, infused through these ports. Interestingly, almost a third of respondents carried catheters of around 12-13.5 cm, which may have limited utility in a predominantly male VA population,. However, we did not assess how frequently these smaller catheters were deployed, and therefore, we cannot comment on appropriateness of use.
Primary strategies for reducing blood loss from filter clotting include pre-emptive filter changes or anticoagulation. Anticoagulation has the added benefit of limiting time away from therapy, as changing out filters requires substantial disassembly and set-up. It also does not incur the costs of additional filters, but whether this is more cost-effective is still uncertain given the additional medications and monitoring required with anticoagulation. Compared to heparin, regional citrate anticoagulation in select patients has demonstrated increases in filter life as well as decreases in complication rates, blood loss, and therapy interruptions[20–23]. The Kidney Disease: Improving Global Outcomes 2012 guideline on acute kidney injury maintains a 2B recommendation for anticoagulation use in patients requiring CRRT, but only in those who do not have an “increased bleeding risk or impaired coagulation” or liver dysfunction or shock, and only in centers with an established protocol. The use of regional citrate anticoagulation is limited to 0–20% of patient treatments whereas heparin remains the most common form of anticoagulation worldwide in CRRT. Heparin use is likewise favored among VA centers with CRRT capabilities, with regional citrate anticoagulation practiced by approximately 1/3rd of facilities (See Fig. 2). The complexity of coordinating across several disciplines – physicians, pharmacy, nursing, respiratory therapy, laboratory – to manage two infusions with frequent monitoring may be the main aversion to use. Despite studies endorsing safety, the potential for error-induced hypocalcemia and citrate accumulation or net citrate overload may present significant concerns for centers with limited prior experience. More evidence in the form of a large randomized, multi-center trial is imminent and could potentially further illuminate the benefits and harms of RCA in CRRT.
Antimicrobial dosing for patients on RRT in the ICU is immensely challenging. Variability among intrinsic patient characteristics, clinical condition, and preferred prescriptions by different practitioners renders valid clearance studies in CRRT rather difficult; however, theoretical estimates of clearance can be made from the total delivered therapy fluid dose and the protein-binding of that drug. Unfortunately, real-world evidence for dosing is limited to studies using more antiquated methods of CRRT. More often than not, patients requiring CRRT are administered inadequate doses of antimicrobials[27–29]; similar concerns exist with SLED. Limited guidance exists for PIRRT as the therapy results in both periods of substantial clearance and negligible clearance. The only available guidance for dosing antimicrobials on AVVH and SHIFT are derived from in silica studies[31–33], limiting many providers’ comfort with use in patients with infections or substantial concerns for infections. Even in SLED, a particular form of PIRRT that has historically been in use longer than other modes (AVVH and SHIFT), pharmacists do not agree on the dosing of commonly used antimicrobials. Thus, sole reliance on pharmacy input may be insufficient for accurate dosing. The dynamic nature of patient courses and complementary therapy chosen necessitates frequent interdisciplinary communication regarding medication dosing. The majority of VA centers, as reported by nephrologists or their designees, endorse input from both pharmacy and nephrology regarding antimicrobial dosing. Mechanisms that ensure that providers who order antimicrobials are frequently (at least daily) educated of the renal replacement therapy plan is paramount to ensuring our best estimations of drug doses.
Limitations of this study include reliance on providers’ recall. We likely garnered reliable responses regarding modes of therapy, machine type, solution type, and other variables related to therapy that are commonly encountered on a daily basis. However, the actual use of more detailed aspects of care such as initiation and stopping points for RRT, nutritional support, diuretic use, deployment of catheter of specific lengths (as compared to stock availability) are not captured. Additionally, questions regarding frequency of anticoagulation and provider input on antimicrobial dosing are more subjective impressions of the respondents. Regardless, these questions still served the purpose of underscoring the preferential equipment and supply use, anticoagulation strategies, nurse staffing models, and multidisciplinary input into into antimicrobial dosing for RRT within the VA. Unfortunately, comparable data outside the VA system was not available. Additionally, in order to keep the questionnaire brief enough to maximize the response rate, we were limited in how comprehensive our study could be. Other than noting the prevalence of peritoneal dialysis, we did not assess any characteristics or equipment related to peritoneal dialysis therapy. We did not request responses regarding French size of catheters, or whether catheters were straight, pre-curved, or with curved extensions. We also did not capture adherence to other VA required RRT practice standards (e.g. mandatory time-out pre-RRT, dedicated water hook ups, adherence to current ISO water standards, completion of life sustaining treatment directive). We did not capture whether pharmacy assisted with therapy fluid preparation. Lastly, we did not collect responses from 11 centers with inpatient HD capabilities. Our use of a simpler online tool with less flexibility allowed for non-responses among CRRT-capable surveyees, an option intended to be reserved for respondents that were exclusively HD-capable. Nonetheless, with a response rate of 87.4% and only a few inappropriate non-responses among seven questions, we believe that we received a sizable sample that serves to adequately describe the variety of strategies and equipment surrounding RRTs in Veteran Affairs ICUs.
Within the largest integrated healthcare system in the United States, we described diverse approaches to RRT. While we documented significant variation in modality and supply use, this study uncovers significant opportunities to explore measures of quality care which are agnostic of RRT modality. With the exception of catheter size and adequate length per location, limited evidence exists to distinguish particular equipment or solutions. The broader, primary objective of continuous and prolonged renal replacement therapy is to deliver precise and consistent treatment that minimizes harm to patients. Aligned with this aim, practitioners should ideally deliver adequate and steady solute and volume control, while concomitantly limiting blood loss, nutritional deficiencies, and errors in drug dosing. Given the variety of treatment, broadly applicable quality measures should be an initial area of focus for improving CRRT and PIRRT. Previous suggestions include ratios of prescribed to delivered therapy, time devoted to malfunction-related suspension of therapy, assessments of target solute clearance in effluent, filter clots, and blood loss. As highlighted in this study, facility and organizational level variation exists in a number of other processes of care related to RRT in the ICU. By virtue of their RRT modality independence, parameters such as care coordination with antimicrobial dosing, nursing models for RRT delivery, RRT prescription authority, vascular access insertion and management, institutional safety standards, may also be well suited for assessment of cross cutting quality measures in the future. The landscape of RRT in the ICU has never been described at this level, and this report serves to educate further on options for addressing ICU-related care.