Despite the fact that the effectiveness of RAAS inhibitors in the primary prevention of AF has been evaluated by several population-based observational studies [23-25], it is not clear whether the RAAS inhibitors remain effective in the reduction of new-onset AF in the older adult hypertensive population when competing risks (e.g. all-cause death) exist. All of these studies simply ignored the existence of the competing risks in their analyses. However, in the presence of the competing risks, it is flawed to examine the effect of the antihypertensive medications on reducing AF risk using the traditional survival analysis approaches such as the Kaplan-Meier methods and the Cox proportional …show more content…
The chronic disease algorithms include specific criteria for reference time periods (searching one to up to three years), diagnosis and procedure codes, number and type of qualifying claims (i.e., must have two Carrier claims during reference time period) and coverage (must have Part A [in-patient] and Part B [out-patient] coverage). These predefined chronic conditions include atrial fibrillation and hypertension, which allows the present study to identify the study cohorts and determine the outcome events. The Part D prescription drug event (PDE) data include pharmacy claims submitted by pharmacies to the Part D health plans for beneficiaries enrolled in Medicare Part D, which record prescription drugs identified by the National Drug Code, the date of dispensing, and the length of supply (also known as “days’ supply”). The use of the PDE data allows the present study to ascertain the starting time and duration of drug …show more content…
The study included Medicare beneficiaries with evidence of HTN who filled prescriptions of one of three study medications (i.e., RAAS inhibitors, BBs, and CCBs) between January 1, 2007, and September 30, 2011 (the “index prescription”). The resulting dataset of eligible beneficiaries was used to study Aim 1. In order to maintain the study power, the present study allowed the inclusion of prevalent drug users (i.e., allowed hypertensive beneficiaries to receive their respective index prescription drugs during the twelve-month baseline period). For the prevalent drug users in 2006 whose index drug therapies were ended after January 1, 2007, the index dates were mandatorily adjusted to be January 1, 2007. Otherwise, the present study did not include those whose prevalent drug uses were ended before January 1, 2007. For this aim, two measures were used to assess utilization of the study drugs including the duration of therapy (in days) and the proportion of days covered defined in the following section “3.4 Measurement of Key Variables”. In addition, all available research questions relevant to time-invariant baseline covariates were examined to characterize the study participants in each comparison group before and after the propensity-score (PS)