On Being Better Kidney Doctors: Understanding Trajectories, Probabilities, Predictability, and People

Abstract

Listen

Translate article

catalyst for this increase in awareness, 3 giving some sensible structure and nomenclature to the condition. Since 2002, the number of publications and research studies describing improved understanding of the biology of the condition and the outcomes of people with CKD has increased exponentially. In this issue of theAmerican Journal of Kidney Diseases, 3 important articles are presented that help inform both research and clinical care in this area. 4-6 These reports focus on movement among CKD states: speed of the travel, shape of the path, and implications of uncertainty (lack of our understanding) of this journey. The main theme is the uncertainty and diversity evident in pathways to clinical outcomes of those with CKD, particularly with respect to progression of kidney disease. The authors all emphasize the need to better characterize and understand that heterogeneity. By so doing, we will not only be able to care better for patients, but will also be better able to design and execute clinical trials and generate evidence on which to base that clinical care. The report by O’Hare et al 4 describes the trajectories of more than 5,000 persons within the Veterans Affairs system who started dialysis therapy between 2001 and 2003 and had at least 2 years of data prior to that start. Using the integrated laboratory and administrative database available through the Veterans Affairs system, they were able to capture patient characteristics and care practices to explore not only the trajectories, but also potentialinfluencesonthosetrajectories.Whilerecognizing that this study examines only those who ultimately ended up on dialysis therapy (ie, the results are limited by survivor bias), there are important lessons here. First, although most people who initiated dialysis therapy within the 2-year follow-up were those with an estimated glomerular filtration rate (eGFR)!30 mL/min/ 1.73 m 2 (63%) and had relatively slower slopes of decrease, a significant proportion were patients with higher eGFRs who had faster rates of decrease (34%), whereas 3% of patients with eGFRs"60 mL/min/1.73 m 2 had catastrophic rates of decrease. Anchoring the outcome as time to dialysis therapy within 2 years, these findings are not unexpected given that to get to dialysis therapy within 2 years from a higher GFR, one would have to have a steep trajectory. Hospitalizations and episodes of acute kidney injury (AKI) were important associations seen in those with steeper eGFR trajectories. Those with the highest levels of kidney function were least likely to receive predialysis care, likely due to nonappreciation of their potential to progress. The report reminds us that those who arrive at dialysis therapy do so in various ways: the authors define 4 different patterns of eGFR decrease prior to dialysis therapy and note the need for more flexible approaches to preparation for renal replacement therapy initiation based on the knowledge of the heterogeneity of progression within populations of identified individuals. Li et al, 5 using theAfricanAmerican Study of Kidney Disease (AASK) cohort, describe the trajectories of 846 African Americans who participated in both the clinical trial and subsequent longitudinal cohort follow-up study for a period of up to 12 years. They describe a 41% nonprogression rate, variability in trajectories with both acceleration and deceleration of slopes over time, and a significant deviation from linearity in GFR slopes within individualsduringthisextensivefollow-upperiod.There are a number of important observations from this study that are pertinent to clinical care: nonlinearity is related to duration of follow-up. The longer one is followed up, the greater the opportunity to have events occur that change the trajectory. Although this is obvious to most clinicians, implications for the design of clinical studies are profound, as is the impact on planning for resource allocation. In clinical practice, we often predict trajectories based on the most recent values of kidney function and other clinical parameters, but given these results, clinicians would need to constantly readjust assessments based on the totality of data from a given patient and consider most recent values in the context of prior values. Seasoned clinicians may already do this implicitly, but those earlier in their career or less well-versed in