Abstract P130: A Map of SPRINT’s Data-free Zone

Abstract

120 mmHg is an optimal SBP goal according to the SPRINT trial. However, certain inclusion and exclusion criteria cloud its broad applicability. It is critical to understand which patients are well represented and reasonable candidates for intensive BP goals. Using only trial inclusion and exclusion criteria diminishes the fact that subjects are unevenly distributed across these criteria. A patient may fit study constraints, yet be poorly represented. Conversely, a patient may be excluded based on a parameter, and declared an inhabitant of a "data-free zone," yet in other respects resemble the trial population. We defined and mapped the "data-rich, data-limited, and data -free zones" of SPRINT based on subjects’ baseline characteristics and not on inclusion and exclusion criteria. For each participant (n=9245), a z-score was computed for 6 variables: age, SBP, glucose, non-HDL-C, creatinine, and BMI. Standardized coefficients from multivariable logistic regression, based on SPRINT’s primary end-point, were used to weigh variables. Summary Scores (SS) were generated for each subject to scale with the Euclidean distance of participants from the theoretical "average patient" in six dimensional space. A SS of 0.56 represents the 90th percentile and 0.74 represents the 97.5th. These were chosen as borders between the data-rich, data-limited, and data-free zones. SS were then calculated for 2007-14 NHANES participants with age >35, SBP≥130, and HbA1c<7. The NHANES population mapped onto SPRINT data zones shows a landscape of applicability by race and sex (Figure). Defining data zones based on patient characteristics holds promise to refine the applicability of trial results.