Gaps in Etiologic Classification of Chronic Kidney Disease in US Veterans: A Retrospective Cohort Study

Many patients living with chronic kidney disease (CKD) may not have a clearly identified cause, which can be a barrier to targeted therapies. We assessed the assignment of CKD etiologies in the national Veterans Health Administration (VHA) electronic health record data, overall and by demographics. We identified regular VHA users aged 18 - 100 years with new-onset CKD based on persistent eGFR <60 ml/min/1.73m2 occurring 2005-2015. CKD etiologies were identified by administrative codes over 3 years before and after the incident CKD date. Etiologies were categorized as diabetic, hypertensive/vascular, obstructive/urologic, glomerular, cystic/congenital, and tubulointerstitial. CKD was considered recognized, but without etiology, if only a general CKD code was present. Logistic regression models assessed disparities by sex and race/ethnicity after adjustment for age, US Census region, and calendar year of incident CKD. Among 452,851 Veterans with CKD, 42.5% were assigned at least one etiology. Hypertensive/vascular (24.5%) was the most common. After adjustment, non-Hispanic Black Veterans and Hispanic Veterans were more likely to be assigned an etiologically-specific diagnosis than non-Hispanic White Veterans (OR 1.61 and 1.23, respectively, each p<0.001). Differences were greater at younger ages. Non-Hispanic Black Veterans were much more likely to be assigned a hypertensive/vascular etiology (OR 2.17; p<0.001), but both non-Hispanic Black and Hispanic Veterans were more likely than White veterans to be assigned diabetic and glomerular etiologies as well (each p<0.001). Women were much less likely to be assigned an etiology overall, and in all categories, compared to men (each p<0.001). Women were less likely to have CKD recognized even if etiology was not assigned. Overall, the assignment of an etiology for CKD is low among U.S. Veterans, particularly among women, possibly due to poor CKD recognition. Types of etiologies assigned differ by race/ethnicity, which may represent true etiologic difference or biased assignment.