(P28-009-23) Nutri©, Clinical Decision Support To Incorporate Precision Behavioral Nutrition in Primary Care: Preliminary Results From a Pilot Trial

Objectives: This study assesses whether Nutri©, an interactive software for data-driven collaborative diet goal setting, could enable precision behavioral nutrition interventions in primary care. Specifically, we hypothesized that 1) Nutri’s inference engine would differentiate diet goal recommendations across patients, and 2) Nutri’s user interface would encourage primary care providers (PCPs) to select a recommended diet goal.

Methods: We tested Nutri in a clinic-based cluster randomized trial, wherein PCPs were randomized to receive Nutri or standard of care control. Nutri is an expert system that 1) computationally processes patient data delivered via API from ASA24 automated diet recall system and a REDCap survey through a set of rules based on a synthesis of expert knowledge and the Dietary Guidelines for Americans (i.e., inference engine, IE); and 2) presents IE results in a 3-step workflow to promote goal setting between patients and PCPs (i.e., user interface, UI). Nutri’s IE prioritizes a set of 9 diet goals (e.g., "choose lean proteins," "reduce portion sizes") according to their likely impact on the patient’s diet and, when needed, provides a warning if a patient’s data triggers system guardrails (e.g., a patient under-consumes kcals for their age and sex). Nutri’s UI guides collaborative diet goal setting with features we identified in previously published user-centered design studies (e.g., color coding, prompts, food lists). Here we describe the distribution of goal recommendations from Nutri’s output for all patient participants and the frequency with which PCPs who used Nutri selected one of the recommended or warning goals.

Results: Nutri’s IE classified a different combination of the 9 goals as needing improvement for each patient who completed ASA24 (n=58; goal recommendation frequency 9%-69%). Among PCP-patient dyads (n=18) who were presented with Nutri during an appointment, 100% set a goal with Nutri; 16/18 were a Nutri-recommended or warning-related goal.

Conclusions: Nutri recommended the full breadth of the available goals based on a patient’s data and PCPs who received Nutri used it to select an appropriate goal in 89% of appointments. These results suggest the feasibility and potential value of a clinical decision support tool for precision behavioral nutrition in primary care.

Funding Sources: Agency for Healthcare Research & Quality