A recent trial comparing medical and surgical management of elderly patients with coronary artery disease reported improved quality of life at one year for both treatment arms (Pfisterer, Buser et al. 2003), suggesting that treatment likely reduces disability in the elderly population. However, no studies to date have estimated the effect of increased use of appropriate pharmaceutical treatments over time on disability rates in the elderly population.
Pharmaceutical use is not captured in Medicare claims, since Medicare does not pay more for use of pharmaceuticals. Thus, we do not have time series data on the use of pharmaceuticals by area. We do have a snapshot of data on pharmaceutical use, taken from a survey of medical records in the mid-1990s. The cooperative Cardiovascular Project (CCP) abstracted medical record data on 186,800 Medicare patients hospitalized for an AMI between February 1994 and July 1995 including data on appropriateness for and receipt of guideline recommended treatments (Marciniak, Ellerbeck et al. 1998).
Use among patients most suited for treatment ranged from 51% for beta-blockers in the immediate post-myocardial infarction treatment to 76% for aspirin. Average utilization rates for the mid-1990s are shown in table 4. Researchers at Dartmouth have calculated the average use rate of each of these pharmaceuticals at the HRR level, which we employ in our analysis (O’Connor, Quinton et al. 1999).
While not known at the area level, use of these pharmaceuticals did increase during our study timeframe. Reported use of aspirin for heart attack patients in 1985 and 1995 was 30% and 75%, respectively, and over the same time, use of beta blockers was 48% and 50%, thrombolytics was 9% and 31%, and ace-inhibitors was 0% and 21%, respectively (Heidenreich and McClellan 2001). The change in ace-inhibitors use from this study was based only on changes in the Worcester, Massachusetts area and may not reflect changes in use nationally. birth control Levonorgestrel
For patients with heart failure and stroke we employed published data on state-wide use of evaluation of ejection fraction (heart failure) and use of warfarin for patients with atrial fibrillation and prescription of antithrombotics at discharge for patients with acute stroke or transient ischemic attack (stroke) from 2000-2001 (Jencks, Huff et al. 2003).
The lack of time series data on pharmaceutical use at the area level requires us to make one change in the analysis. Equation assumes that the procedure has the same effectiveness over time. Since use of pharmaceuticals increased over time, we would expect that the coefficient on pharmaceutical use would be greater for the 1994-99 cohort than for the 1984-89 cohort; more patients, after all, would have been treated with the technology. To account for this phenomenon, we interact each treatment variable with the survey year dummy variables.
From our models, we estimate the likelihood of being disabled, dead and alive and nondisabled at follow-up if all respondents lived in HRRs that provided relevant procedures and pharmaceuticals at the 10th and 90th percentiles of care, holding all other covariates constant. We develop separate estimations for relevant procedures and pharmaceuticals to better understand the effect of each treatment type on changes in death and disability over time.