Table 1.

Comparison of Primary and Secondary Analysis Methodologies

	Sample used	Period during which influenza vaccinations were measured (i.e., the exposure period)	Exclusion criteria (in addition to those described in Fig. 1)	Model covariates
Primary Analysis: ATT Estimation Using PSM	Propensity-score matched	Follow-up	•Excluded patients if any of the following variables were missing from the data set: geographic region, race, or sex. The number of patients missing each of these variables is provided in Table 2.	The propensity score model included all covariates shown in Table 2.
Secondary Analysis: Time-to-Event Model	Unmatched	Look-back	•Excluded patients if any of the following variables were missing from the data set: geographic region (n = 1,934), race (n = 212,757), or sex (n = 197).	The competing-risks time-to-event model included all covariates shown in Table 2 except for the following:
			•Excluded patients who received more than 6 influenza vaccinations during the 6-year look-back period (n = 31).	•Geographic region •Number of health care visits during the look-back period
			•Excluded patients who received zero influenza vaccinations during the look-back period but≥1 influenza vaccinations during the follow-up period (n = 587,726).	•Number of routine “well visit” examinations during the look-back period

Methods that did not differ between the primary and secondary analyses (i.e., outcome and covariate measurements) are not shown in this table. ATT, average treatment effect in the treated; PSM, propensity score matching.