Pre-registration v1: Caffeine intake timing and resting heart rate - an age-interaction study

This is the authors' abstract. We don't add key points for this paper.

Pre-registration archive (frozen full text) for an observational study of caffeine intake timing and resting heart rate in US adults (NHANES 2003-2018). Contents: the registrable pre-registration text v1 (Markdown + print PDF), the decision record with the timestamps of each ratified choice, the freeze record (SHA-256 of the criteria text and of the calibration documents, plus the NTP-verified freeze time), and three pre-outcome calibration documents (timing-window calibration, analytic power for the age interaction, medication code scan). Status at archive time: no outcome association had been estimated. The outcome variable (resting heart rate) had been used only for counts and dispersion, in an analytic power calculation; no coefficient, p-value or direction had been computed, and the main analysis script did not exist. Registration anchors, priority fixed in advance: AsPredicted submission time > this Zenodo record > OSF registration (OSF was offline 2026-10-16 09:00 EDT to 2026-10-19 09:00 EDT for a planned cloud migration, so the frozen text was archived here and on AsPredicted during the outage; OSF will hold the primary registration). The AsPredicted answers cite this DOI for every structural specification that does not fit its nine fields (age-spline knots fixed at 40 and 60 years; the three-component decomposition rules; the condition-triggered rules; the frozen medication code lists). Discovery layer: 8 two-year NHANES cycles, adults 18-39 with valid pulse (unweighted n = 16,643); all-adult interaction model n = 38,793. Mechanism layer: 2011-2012 only, n approx. 300-600, single mediator.