Statistical Analysis
The statistical adjustments used in the analysis were prespecified and
have been used for all comparative analyses of the Micra CED Study
cohort. Propensity score overlap weights9,10 were used
to account for differences in baseline and encounter characteristics
between the leadless-VVI and transvenous-VVI cohorts. Unadjusted and
overlap-weight adjusted 3-year complication, reintervention, and heart
failure hospitalization rates were estimated using the cumulative
incidence function. Fine-Gray competing risk models were used to compare
the unadjusted and adjusted risk for 3-year chronic complications,
device reinterventions, and heart failure hospitalizations between study
groups, and Cox proportional hazards models were used to compare
all-cause mortality and the composite of heart failure hospitalization
and all-cause mortality through 3 years. A sub-analysis of the heart
failure hospitalization endpoints was also conducted among patients in
the study cohort without a history of heart failure at baseline. Because
heart failure hospitalizations are not a primary or pre-specified
secondary endpoint of the Micra CED Study, the inclusion of this
endpoint is designed to be hypothesis-generating. Thus, no statistical
correction for the additional endpoint was made and all endpoints were
evaluated at a significance level of P<.05 and all P values
were 2-tailed. Events occurring in between one and 10 patients were
suppressed to protect beneficiary privacy as required by
CMS.11 All statistical analyses were conducted in SAS
version 9.4 (SAS Institute).