Stenosis surveillance for hemodialysis patients: A pre-post treatment analysis

Zhenqiang Lu, Purdue University

Abstract

Access stenosis is a major problem for hemodialysis patients. The access is a vessel, a vein or a graft where blood comes in and out of the patient for cleaning. Stenosis is a narrowing of the access caused by the build-up of a deposit. Stenosis causes thrombosis, the loss of a site (only a few sites are possible), expensive hospital procedures, and death. Two measurement methods for access surveillance are (VascAlert, VA)—electronically monitored, continually trended, venous access pressure, and (Transonic, TS)—ultrasound dilution access flow. We studied the ability of VA and TS to classify venous access stenosis. Standard approaches of study such cannot be used directly because the stenosis status of patients not sent for intervention is not known with sufficient accuracy. We use pre-post treatment analysis: study VA and TS data before and after a successful intervention treatment in which the present of stenosis is confirmed by clinical means. Data before intervention are measurements with stenosis and data after are measurement without. We study TS and VA data for nearly one thousand patients from three dialysis centers at Cornell Medical School, Bakersfield Hospital, and The University of Chicago Medical Center. Statistical models are built and validated; the models are the bases of statistical method used to assess VA and TS. The bootstrap is used to characterize statistical uncertainty. Our conclusions are (1) VA performs well, (2) TS performs poorly, and (3) TS might be improved by establishing a clinical TS baseline for each patient.

Degree

Ph.D.

Advisors

Cleveland, Purdue University.

Subject Area

Statistics

COinS