Screening for CKD To Improve Processes of Care among Nondiabetic Veterans with Hypertension: A Pragmatic Cluster-Randomized Trial

Background and objectives We conducted a pilot, pragmatic, cluster-randomized trial to evaluate feasibility and preliminary effectiveness of screening for CKD using a triple-marker approach (creatinine, cystatin C, and albumin/creatinine ratio), followed by education and guidance, to improve care of hypertensive veterans in primary care. We used the electronic health record for identification, enrollment, intervention delivery, and outcome ascertainment. Design, setting, participants, & measurements We randomized 1819 veterans without diabetes but with hypertension (41 clusters) into three arms: (1) CKD screening followed by patient and provider education; (2) screening, education, plus pharmacist comanagement; or (3) usual care. The primary clinical outcome was BP change over 1 year. Implementation and process measures included proportion screened; CKD detection rate; and total and new use of renin-angiotensin system inhibitors, nonsteroidal anti-inflammatory drugs, and diuretics. Results Median age was 68 years, 55% were white, 1658 (91%) had a prior creatinine measure, but only 172 (9%) had prior urine albumin/creatinine ratio, and 83 (5%) had a prior cystatin C measure. Among those in the intervention, 527 of 1215 (43%) were identified with upcoming appointments to have CKD screening. Of these, 367 (69%) completed testing. Among those tested, 77 (21%) persons had newly diagnosed CKD. After 1 year, change in systolic BP was −1 mm Hg (interquartile range, −11 to 11) in usual care, −2 mm Hg (−11 to 11) in the screen-educate arm, and −2 mm Hg (−13 to 10) in the screen-educate plus pharmacist arm; P=0.49. There were no significant differences in secondary outcomes in intention-to-treat analyses. In as-treated analyses, higher proportions of participants in the intervention arms initiated a renin-angiotensin system inhibitor (15% and 12% versus 7% in usual care, P=0.01) or diuretic (9% and 12% versus 4%, P=0.03). Conclusions The pragmatic design made identification, enrollment, and intervention delivery highly efficient. The limited ability to identify appointments resulted in inadequate between-arm differences in CKD testing rates to determine whether screening improves clinical outcomes.