Blood Pressure Reduction in Multiomic Digital Food-as-Medicine: a single-arm cohort study with a regression-to-the-mean-adjusted estimand

Elevated systolic blood pressure (SBP) is a major modifiable driver of cardiovascular morbidity. Digital food-as-medicine and home-blood-pressure-monitoring programs increasingly support its management outside the clinic, but most published evaluations are single-arm, pre/post designs; because enrolment is conditioned on an elevated baseline reading, regression to the mean (RTM) is expected to inflate any observed improvement, and this is rarely checked against an internal comparator. We conducted a retrospective, single-arm cohort study of home blood-pressure monitoring within a commercial digital food-as-medicine program serving a US employer-sponsored population with obesity. Of 329 members included, 136 were at-goal (<130/80 mmHg), 74 at Stage 1 (130-134/80-84 mmHg) and 119 at Stage 2 ([&ge;]135/85 mmHg) at baseline; follow-up SBP was evaluable for 128, 68 and 112 members in these strata at 3 months and for 62, 34 and 53 at 6 months, respectively. The primary estimand was the unadjusted within-member change in home SBP at 6 months among members with Stage-1 and Stage-2 blood pressure at baseline; a secondary estimand corrected for RTM using an ANCOVA linear mixed model with the at-goal stratum as an internal negative-control population. Among Stage-1 members (n = 34), unadjusted SBP change was -4.36 mmHg (95% CI -7.43 to -1.28; p = 0.0069), paired with an RTM-corrected secondary change of -3.33 mmHg (-6.40 to -0.26; p = 0.033). Among Stage-2 members (n = 53), unadjusted change was -10.17 mmHg (-12.94 to -7.40; p < 0.001), paired with an RTM-corrected secondary change of -5.50 mmHg (95% CI -8.50 to -2.50; p = 0.00037); RTM accounted for ~24% and ~46% of the unadjusted decline in Stage-1 and Stage-2, respectively. RTM-corrected estimates were robust to sensitivity analyses on measurement modality, seasonality, covariates and model specification, with no evidence of attrition bias. Program participation was associated with a decline in home SBP among Stage-1 and Stage-2 members that remained significant, though smaller, after RTM correction, supporting RTM-adjusted SBP change as the appropriate metric for evaluating food-as-medicine interventions.