Evaluating the case-positive, control test-negative study design for influenza vaccine effectiveness for the frailty bias.