known_biases: Single VA biobank of predominantly older male US veterans (low external validity to women/general population); diet by one-time self-administered FFQ (exposure misclassification) and outcomes by electronic health records (ascertainment tied to VA care access); short mean follow-up. Any finding resting on MVP moves with these together.