Abstract: TH-PO1210
Concordance Between AKI and Blood Pressure (BP) Assessment in Routine Clinical Practice and Research: Results from the SPRINT Electronic Health Record (EHR) Ancillary Study
Session Information
- Late-Breaking Clinical Trials Posters
November 07, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Hypertension and CVD
- No subcategory defined
Authors
- Drawz, Paul E., University of Minnesota, Minneapolis, Minnesota, United States
- Rahman, Mahboob, Case Western Reserve University, Cleveland, Ohio, United States
- Tuot, Delphine S., University of California, San Francisco, California, United States
- Rocco, Michael V., Wake Forest School of Medicine, Winston-Salem, North Carolina, United States
- Pajewski, Nicholas M., Wake Forest School of Medicine, Winston-Salem, North Carolina, United States
Background
In SPRINT, intensive BP lowering was associated with increased risk for AKI. However, AKI ascertainment was based in part on billing codes and was limited to ER visits and hospital admissions. Additionally, it is unknown how BPs obtained using a standardized protocol in SPRINT compared to BPs in routine clinical practice.
Methods
Participant level EHR data were obtained from 15 SPRINT clinical sites. We merged these data with trial data to compare creatinines and BPs obtained in routine clinical practice with data obtained in SPRINT. AKI was defined by a 1.5 fold or greater increase in creatinine from the most recent SPRINT value. The primary outcome was inpatient or outpatient AKI. We used Cox models to assess the effect of intensive BP lowering on AKI events. We also report the difference between SPRINT and EHR BPs.
Results
Out of 9361 randomized participants, at least one EHR creatinine or BP value was available for 3827 participants. Among these, 87 in the intensive and 54 in the standard arm had an AKI event as adjudicated in SPRINT (HR 1.58, 95% CI 1.1 to 2.2). Using EHR-based creatinines, 147 in the intensive and 127 in the standard arm had an inpatient/ER AKI event (HR 1.15, 95% CI 0.9 to 1.5). For our primary outcome, 246 participants in the intensive and 170 in the standard arm had either an inpatient or outpatient AKI event (HR 1.49, 95% CI 1.2 to 1.8). Mean systolic BP in the EHR from 6 to 18 months after randomization was 7.2 (± 11.1) mm Hg higher than BPs obtained in SPRINT in the intensive arm and 4.7 (± 11.7) mm Hg higher than SPRINT BPs in the standard arm. Bland-Altman limits of agreement were wide.
Conclusion
Using linked EHR and SPRINT data, we demonstrate:
a) a higher rate of AKI using data from routine clinical practice compared to clinical trial data,
b) the increased risk for AKI with intensive BP lowering shown in SPRINT was attenuated using data obtained in routine clinical practice; this is the first report of the effect of an intervention on outpatient AKI, and
c) BPs obtained in routine clinical practice were higher than readings obtained in SPRINT; though, the concordance between these BPs varied widely.
These findings should inform the translation of SPRINT results into clinical practice.
Funding
- NIDDK Support