Читать книгу 40 Years of Continuous Renal Replacement Therapy - Группа авторов - Страница 43

There is no consensus on how to define ”timing” relative to starting CRRT in critically ill patients with AKI. Observational studies have employed a spectrum of definitions for “early,” “delayed” and “late” start of CRRT [5]. These studies have incorporated physiological measures (e.g., urine output), biochemical measures (e.g., serum creatinine, urea), start relative to AKI onset, start relative ICU admission, and start relative to the occurrence of a complication associated with AKI (i.e., a conventional indication). However, reference to “early,” “delayed,” or “late” is relative and susceptible to bias. What may represent an “early” start of RRT in one context may constitute “late” in another, where the spectrum of clinical characteristics, diagnoses, and illness severity vary. This heterogeneity in definitions for “timing” or “thresholds” or “criteria” across observational studies has contributed to lack of clarity to guide clinical practice and practice variation [2].

A common feature of observational data has been a general focus only on patients who received RRT, with omission of patients also with severe AKI not treated with RRT [6]. Although clinicians may have difficulty prospectively identifying such patients, it is well known that a subgroup of these patients will survive and recover kidney function, despite severe AKI, without ever receiving RRT [7]. The exclusion of the subgroup from observational data has contributed to “delayed” groups disproportionately comprised of patients with less favorable outcome. This result may be an overestimation of the relative survival advantage of earlier RRT initiation [5].

Survey data suggest the decision by clinicians to start CRRT in critically ill patients with AKI is largely subjective based on a wide spectrum of clinical information and the relative perception of benefit, and likely further modified by both patient-specific (i.e., age, multimorbidity, kidney reserve, response to a diuretic challenge, acuity of illness) and health-system-specific factors (i.e., prescribing service, time of day, day of week) [3]. Moreover, survey data also showed substantial variation in the minimum severity of indications that would prompt clinicians to start CRRT. Recently, the Acute Dialysis Quality Initiative performed a consensus meeting focused on “Precision CRRT,” with one of the workgroups tasked with making recommendations on when to start CRRT [8]. The workgroup proposed that CRRT initiation should be individualized and not based solely on stages of AKI or kidney function, without broader consideration of the patient’s clinical context. Moreover, the workgroup articulated a conceptual model of a dynamic “demand-capacity” relationship for a patient’s given kidney function (Table 1). As such, the optional timing of initiation of CRRT could be considered when a patient is confronted with a mismatch in kidney capacity to manage ongoing metabolic and fluid demands (Fig. 1).

Table 1. Factors to consider while initiating CRRT (adapted from [8])

Perhaps in recognition that practice variation is itself a marker of suboptional quality of care, a recent pilot study described the implementation of a Standardized Clinical Assessment and Management Plan (SCAMP) algorithm to guide the initiation and discontinuation of RRT in critically ill patients with severe AKI [9]. The SCAMP algorithm was implemented in a single tertiary ICU setting in a practice shared by 9 nephrologists and it evaluated 176 patients over 13 months. The algorithm was completed daily in each patient with AKI in whom RRT would be reasonably considered and integrated tiered thresholds of physiological and biochemical indications for starting RRT. The algorithm further directed clinicians, when confronted by selected indications, to initiate RRT or not. In total, the SCAMP algorithm recommended starting RRT in 31% of completed forms for 176 critically ill patients with AKI. In 57% of these cases, clinicians deviated from the recommendation (i.e., did not start RRT), most commonly due to expected kidney recovery and futility. Alternatively, when the SCAMP algorithm recommended not starting RRT, compliance was 98%. Among patients whose clinicians adhered to the algorithm for starting RRT, hospital mortality was lower (42 vs. 63%, p < 0.01); however, this apparent mortality benefit was modified by patient baseline illness severity and predicted risk of death. The mortality benefit was evident only for those with a predicted hospital mortality <50% (i.e., less severely ill). The pilot study has numerous limitations (i.e., single-center, small, non-randomized, selection bias, confounding by indication, high non-adherence); however, it shows potential for quality improvement interventions, such as the implementation of a standardized algorithm, to contribute to improvements in the reliability of care and reduce suboptimal outcomes that may be partly attributable to variability in practice.

Fig. 1. Summary of the demand and capacity relationship – a conceptual model indicating as to when to initiate CRRT (reprinted with permission from ADQI [www.ADQI.org]).