The overlooked impact of background diet and adherence in nutrition trials.

Importance: Randomized controlled trials in nutrition (RCTN) face unique challenges such as the influence of participants' background diets and varying adherence to the intervention, factors that are difficult to quantify and can mask true intervention effects. However, the exact impact of these factors remains unclear.

Objective: To quantify the impact of background diet and adherence, estimated using validated nutritional biomarkers of flavanol intake, on the outcomes of a large-scale RCTN.

Design Setting And Participants: This study was nested within the COcoa Supplement and Multivitamin Outcomes Study (COSMOS; NCT02422745 ), a randomized, double-blind, placebo-controlled, 2×2 factorial trial conducted among 21,442 older adults in the United States. Analyses focused on participants (n=6,532) in the placebo and cocoa-extract intervention arms who had available biospecimens and outcome data.

Exposures: Daily supplementation with a cocoa flavanol (CF).

Main Outcomes And Measures: We used validated flavanol biomarkers in urine to assess background dietary flavanol intake and adherence to the cocoa extract intervention. Outcomes included total cardiovascular disease (CVD) events, CVD mortality, all-cause mortality, and major CVD events. Hazard ratios (HRs) with 95% CIs were estimated using intention-to-treat, per-protocol, and biomarker-based analyses.

Results: Of the 6,532 participants, 20% in both the placebo and intervention arms had a background flavanol intake similar to that provided by the intervention; only 5% had no background flavanol intake. In the intervention group, 33% of participants were not adherent based on expected biomarker levels, that was larger than the15% estimated based on self-reported pill-taking. Accounting for biomarker-estimated background diet and adherence resulted in a significant impact on effect sizes. For total CVD events, the HR (95% CI) of 0.79 (0.59-1.05) in the per-protocol analysis (PP), was reduced to 0.65 (0.47-0.89) in the biomarker-based analysis. Similar changes were observed for CVD mortality (from 0.51 (0.23; 1.14) to 0.44 (0.20; 0.97)), all-cause mortality (0.69 (0.45; 1.05) to 0.54 (0.37; 0.80)) and major CVD events (from 0.62 (0.43; 0.91) to 0.48 (0.31; 0.74)). Overall, biomarker-based analyses consistently produced stronger effect estimates than ITT and PP analyses that do not consider the impact of background diet or assess adherence using self-reported methods.

Conclusions And Relevance: These results highlight the importance of considering diet and adherence in RCTN using nutritional biomarker-based analyses. This may help to improve consistency and reliability of outcomes of RTC testing nutritional interventions.