UCLA

Progression to end stage kidney disease (ESKD) is often sudden due to few symptoms during earlier stages of chronic kidney disease (CKD), and health care access barriers preventing recommended CKD screening. Due to nonrandom processes dictating selection into pre-ESKD nephrology care, permanent access versus catheter, and arteriovenous graft (AVG) versus arteriovenous fistula (AVF), measurement of outcomes subsequent to these processes must consider remedies to potential selection bias. In paper one, a recursive bivariate probit estimated factors associated with permanent access creation prior to hemodialysis after accounting for selection into pre-ESKD nephrology care. Nearly all patient factors had small effects, whereas pre-ESKD nephrology care increased the likelihood of improved care (permanent access) more than fourfold. Polices related to health insurance access and pre-ESKD nephrology care are therefore essential for improving rates of early permanent access creation. Paper two examines determinates of vascular access type and infection, with some factors (e.g., pre-ESKD nephrology care) relating to infection risk indirectly, through selection into an access type. An endogenous Poisson was used to model variability in hospitalization for vascular access infection after adjusting for selection into permanent access. The results suggest ESKD patients under 30, patients with a history of intravenous drug use, and residents of nursing homes should be a focus of interventions to reduce vascular access infection among ESKD patients, as adjustment for selection into an access type substantially increased estimates of these groups’ infection risks, relative to estimates from a Cox proportional hazards model that does not correct for selection bias. In paper three, rates of filled opioid prescription did not differ by access type, and a modeling approach accounting for selection into AVF versus AVG was found to be inappropriate. However, among patients filling an opioid prescription, AVG recipients had, on average, 2 additional 5 milligram hydrocodone equivalents prescribed to them relative to AVF recipients. Federal, state, and health system policymakers, as well as members of surgical societies and research scientists, may wish to use these findings to inform methodological approaches and decisions related to the identification and care of late-stage CKD and early-stage ESKD patients.