RESULTS
We identified 32,591 MIs in SynPUF and assembled a cohort of 1,589
patients experiencing first MI with no history of aspirin or clopidogrel
use. Of those patients, 234 (15%) initiated aspirin or clopidogrel
within 30 days and 411 (26%) additional patients initiated within 90
days. Only 5 initiated treatment on day 0. There were 593 outcomes
within the 180-day risk window.
The “0-30 day” regimen censored 1,355 patients for failing to start by
day 30, while the “0-90 day” and “30-90 day” regimens censored 944
patients for failing to start by day 90. The “30-90 day” regimen
censored 229 patients for starting treatment too early.
After IPCW, 180-day outcome risk was 40.2% for the clones in the “0-30
day” regimen, 35.7% in the “0-90 day” regimen, and 35.2% in the
“30-90 day” regimen. This resulted in risk differences of -4.6% and
-5.0% comparing “0-90 day” and “30-90 day” regimens to the “0-30
day” regimen, respectively.
Figure 3 provides cumulative probabilities of initiation
between day 0 and 90 within each treatment regimen after IPCW.
Initiation rates were relatively consistent throughout the interval, as
expected given the synthetic data. The figure highlights that exposure
patterns during the first 30 days are identical for the “0-30 day” and
“0-90 day” regimens and that the rate of initiation (i.e., the slope
of the line) is higher for the “30-90 day” regimen than the “0-90
day” regimen from day 30 to 90.
Table 1 shows SMDs between individuals censored due to failing
to initiate treatment on day 30 or 90 and recent initiators before and
after applying inverse probability of censoring weights. Absolute SMDs
are less than 0.100 after IPCW and covariates are better balanced after
applying the weights except for the renal covariate in the “0-30 day”
regimen. The total number of initiators after weighting (“0-30 day”
recent initiator weighted N: 1254.2; “0-90 day” weighted N: 766.3;
“30-90 day” weighted N: 896.4) and the pre-weight cohort the weights
were constructed in (“0-30 day” total N: 1254, “0-90 day” censored
N: 765, “30-90 day” censored N: 893.9) were also almost identical,
suggesting minimal positivity violations.