The Assessment, Serial Evaluation, and Subsequent Sequelae of Acute Kidney Injury (ASSESS-AKI) Study: Design and Methods

Alan S Go; Chirag R Parikh; T Alp Ikizler; Steven Coca; Edward D Siew; Vernon M Chinchilli; Chi-yuan Hsu; Amit X Garg; Michael Zappitelli; Kathleen D Liu; W Brian Reeves; Nasrollah Ghahramani; Prasad Devarajan; Georgia Brown Faulkner; Thida C Tan; Paul L Kimmel; Paul Eggers; John B Stokes


BMC Nephrology 

In This Article


Despite the emerging clinical and public health importance of AKI, studies to date have traditionally focused on characterizing its short-term consequences. Recent attempts to extend focus on longer-term outcomes have primarily involved the retrospective study of administrative databases in different population and clinical settings.[9–14] While these studies have reported a link between AKI and increased risk for either advanced CKD,[9,14] ESRD,[10–13] and all-cause death,[9–12,56–59] there have been several important limitations. Specifically, all published studies were retrospective in nature. Many did not use observed changes in serum creatinine to define AKI or rigorously quantify baseline renal function or but rather relied on administrative diagnostic codes for defining "AKI" and determining presence or absence of baseline CKD and potential confounders. Most studies included only a relatively small number of outcomes. Importantly, none of the studies collected specimens to measure biomarkers both during the AKI admission and after hospital discharge. The ASSESS-AKI study will address many of these limitations by establishing a prospective, matched parallel cohort of persons (including children) with and without AKI based on serum creatinine-based criteria measured in a standardized fashion, serial collection of blood and urine for evaluating of diagnostic and prognostic markers in AKI, and systematic follow-up for changes in renal function and multiple clinical and patient-centered outcomes.

One of the key challenges of existing studies relates to defining "baseline" kidney function for determination whether an episode of AKI has occurred and the severity of AKI.[60,61] A recent examination of this issue revealed that the use of imputed or commonly used surrogate estimates of baseline function can result in substantial misclassification of AKI and hinder adequate study of its associated outcomes.[62] Enrollment into ASSESS-AKI requires a pre-admission serum creatinine performed using an IDMS-traceable assay using a time frame recent enough to the index hospitalization that is likely to represent steady-state renal function and will substantially reduce the misclassification of the main exposure (i.e., occurrence of AKI) and of a key confounder (pre-admission CKD status) for enrolled participants, but it also creates barriers to participant recruitment. Many hospitalized patients who experience or are at risk for AKI may not have serum creatinine values available before the AKI episode, available values may not be performed using an IDMS-traceable assay, available values may be outside of our pre-specified time window (7 to 365 days before the index hospitalization), or values obtained during the eligible time window may not actually reflect steady-state renal function. All studies of AKI are affected by this challenge given that currently available consensus definitions of AKI are based on the magnitude of change of serum creatinine concentration. ASSESS-AKI will have strong internal validity given that both participants with and without AKI will have a consistently measured baseline measure of renal function that will serve as an anchor for assessment of longitudinal changes in kidney function.

As patients who suffer from AKI also tend to systematically differ from those without AKI on relevant demographic and clinical characteristics, accurate assessment of these variables is paramount to delineating the risk attributable to AKI for poor outcomes. Toward that end, existing studies based on their retrospective design and limited data quality have been susceptible to notable residual confounding and biases that limit the ability to delineate the independent contribution of AKI, especially less severe episodes, on adverse clinical outcomes. In particular, the use of administrative codes to capture the primary exposure (AKI) and critically important modifiers such as underlying CKD can lead to misclassification bias reflected in a substantial exaggeration or underestimation of the effect observed. Most studies also did not control for the presence and severity of CKD before the AKI episode, which is problematic given that CKD and CKD severity are both potent predictors of experiencing AKI as well as CKD progression, ESRD and other adverse outcomes. In addition to using standardized measurements of serum creatinine in detailing both AKI and CKD, ASSESS-AKI will also implement a 1:1 AKI:non-AKI participant matching algorithm along with advanced analytic methods that should help to mitigate major confounding for evaluating the independent impact of AKI on targeted outcomes of interest during long-term follow-up.

Current efforts to develop and validate diagnostic biomarkers that can define and predict both AKI and CKD progression based on absolute levels rather than on changes from baseline--as is necessary when utilizing serum creatinine concentrations--may mitigate the challenges of requiring a standard definition. One of the major goals of ASSESS-AKI is to evaluate the incremental utility of blood and urine biomarkers to refine the diagnosis and prognosis of AKI above current definitions using a prospective serial biospecimen collection protocol; standardized collection, processing, storage and testing methodology; and a diverse set of clinical settings and patients. Most of the recently discovered putative biomarkers of AKI are known to demonstrate time-dependent fluctuations around the time of AKI.[18] Some biomarkers are most elevated within a few hours after a known insult such as cardiac surgery (e.g., NGAL, L-FABP), while other prominent biomarkers have a delay in their expression, with peak values occurring 24–48 hours after the time of presumed injury (e.g., IL-18, KIM-1).[18] We anticipate that the timing of the presumed renal insult will vary across the three Clinical Research Center populations that comprise ASSESS-AKI. The TRIBE-AKI cohort will, in general, experience injury occurring during or after the time of cardiopulmonary bypass; the VALID cohort may have multiple episodes of injury in the setting of critical illness, and the Kaiser Permanente cohort will have variation in timing of injury depending on the clinical setting (e.g., medical or surgical wards, oncology or cardiology service) within a general hospitalized population. Thus, our approach is tailored to promote generalizability and pragmatism for biospecimen collection, as both blood and urine will be collected on the day of or as soon as possible after the identification of clinical AKI based on serum creatinine concentration change. Based on this approach, it remains questionable whether biomarkers of current interest will be at or near their peak concentration. However, if biomarkers of AKI are to be widely used by practicing physicians, they are most likely to be initially used when there is evidence of AKI by standard diagnostic criteria to improve the diagnosis and/or prognosis of AKI. Therefore, the results from ASSESS-AKI should be broadly generalizable to clinical practice.

In addition to the challenges mentioned above, characterizing the type of AKI, which could influence short- and long-term outcomes, has been relatively understudied in epidemiologic studies of AKI. Both AKIN and RIFLE consensus criteria do not encourage discrimination of the type of AKI (i.e., ATN, pre-renal azotemia, other), implying that the etiology and type of AKI do not affect subsequent outcomes after accounting for the magnitude of change in serum creatinine concentration and/or urine output. However, ASSESS-AKI will prospectively evaluate the contribution of type of AKI on long-term clinical outcomes -above and beyond conventional measures of AKI severity. We believe that attempting to differentiate AKI "phenotype" (i.e., ATN vs. pre-renal azotemia vs. other) is important to address this knowledge gap. We also recognize that there currently exists no "gold standard" method to distinguish between phenotypes and that the same patient could have multiple contributing etiologies for AKI. Thus, we will implement several strategies, including targeted enrollment of more severe AKI based on greater change in serum creatinine concentration and/or longer duration of serum creatinine elevation that is more likely to represent ATN; urine analysis and microscopy at the time of AKI for identification of urine sediment consistent with ATN (e.g., granular casts); and standardized adjudication of cases via consultation notes, discharge summaries, and other clinical variables. Overall, the ASSESS-AKI study will provide one of the most comprehensive efforts to characterize AKI phenotype and whether it independently alters the risk of long-term clinical outcomes.

While ASSESS-AKI has numerous strengths, it also has several limitations. The cohort will be enriched with patients suffering AKI in common at-risk settings such as surrounding cardiac surgery or within the ICU as well as in more general hospitalized settings, but we will have limited power to examine whether the clinical setting modifies the association between AKI and clinical outcomes. The inclusion of children expands the age range being evaluated but is only applicable to those requiring cardiopulmonary bypass. The relatively small number of children targeted for enrollment means we will be able to detect only large effect sizes. The multi-center cohort includes various U.S. and Canadian sites and health care delivery systems, but the results may not be generalizable to all practice settings and populations. Despite various design and analytic approaches as well as standardized data collection and quality control efforts, as an observational study, we cannot rule out the impact of residual confounding and bias. However, since a randomized comparison is not possible, our prospective cohort design and systematic follow-up will provide a rigorous assessment among eligible participants. There are ongoing plans to address some of these limitations by augmenting ASSESS-AKI with possible ancillary studies (e.g., recruitment of a larger number of pediatric patients).


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.