Estimating Racial and Gender Disparities in the Use of ...



Estimating Racial and Gender Disparities in the Use of Anti-hypertensive and Lipid-lowering Medications: Results from the Multi-Ethnic Study of Atherosclerosis

Robyn McClelland, Neal Jorgensen, Richard Kronmal

Introduction

The MESA population is heavily treated, with almost 40% of participants on anti-hypertensive medications at baseline, and 16% on lipid lowering medication. These numbers rise to approximately one half and one third of the population by exam 4 respectively. From an epidemiologic perspective, disparities in treatment by gender, age, race/ethnicity or socio-economic position are of great interest. That is, at a given level of cholesterol and cardiovascular risk factors, is there a difference in medication use by gender or race/ethnicity? In a cohort study such as MESA, these types of questions are difficult to answer. Participants not on medications at baseline but who begin taking them later in the study may have been influenced by our reporting of results to participants and physicians, and hence estimated rates may not reflective of the general patterns of use in the population at large. For example, examination of new users would tend to underestimate any disparities that may be due to lack of access to regular health care. Thus, it is more desirable to examine rates of medication use at the time that participants enter the study. However, for participants taking medications at baseline we do not have a measurement of their pre-treatment values in order to control for confounding by indication. For this we can use our previous work on estimating pre-treatment biomarker values to improve our models of medication use.

Research Questions

• At a given level of (estimated) pre-treatment blood pressure, are there racial/ethnic or gender differences in the pattern of use of anti-hypertensive medications? Are these differences explained by CVD risk factors or markers of socio-economic status?

• At a given level of (estimated) pre-treatment cholesterol, are there racial/ethnic or gender differences in the pattern of use of lipid-lowering medications? Are these differences explained by CVD risk factors or markers of socio-economic status?

• Do either gender or race/ethnicity modify the associations of risk factors with medication use?

• How do the inferences obtained using the estimated pre-treatment biomarker values differ from:

▪ Using observed baseline values (even for those on treatment)

▪ Using the population of new-users, excluding those on treatment at baseline

Sample

All MESA participants with non-missing medication use data at baseline and valid blood pressure/cholesterol values will be included for the primary analysis. (n=6808 for blood pressure; n=6762 for cholesterol).

Analysis Plan

For participants taking lipid lowering medication at baseline their pre-treatment cholesterol levels will be estimated using techniques described in McClelland et al [2008]. Similarly, we will also estimate pre-treatment blood pressure for those taking anti-hypertensive medication at baseline. Briefly, pre-treatment values are estimated based on type of drug, observed cholesterol/blood pressure, and demographics. The model used is estimated based on the subset of new drug users throughout the study (that is, using the pre and post treatment values for the new users).

Two sets of regression models for the prevalence of medication use at baseline will be developed: one for use of anti-hypertensive medications, one for the use of lipid lowering medications. A first model for each endpoint will include age, gender, race/ethnicity and estimated pre-treatment blood pressures (or cholesterol for the lipid lowering model). This will tell us whether there are differences in the use of therapies at a given level of blood pressure (cholesterol). A second model will include adjustment for cardiovascular risk factors including diabetes, cholesterol (or hypertension), body mass index, and smoking status. Finally a third model will include income, education, and health insurance to explore the extent to which differences in socioeconomic status may explain the disparities. We will compare the extent to which the ability to control for estimated pre-treatment biomarker values changes the inferences. Multiple imputation using 10 imputations will be compared with using the average of 10 imputations to explore the impact that estimation of pre-treatment values has on the variability of our parameter estimates.

Funding: Analysis for this project will be supported by Dr. McClelland’s R01 on Statistical Methods in CVD Research. The goals of that grant are oriented towards statistical methodology development with applications to modeling blood pressure and cholesterol in the context of medication use.

Possible Overlap: There is no overlap with this study, but there are two MESA manuscripts that used medication use as an endpoint. ML001 looked at time trends in the use of different kinds of anti-hypertensive medications, and whether ALLHAT recommendations had an impact on the types of medications being used. This paper was not concerned with predictors of taking anti-hypertensive medication itself (versus no medication). ML017 looked at whether CAC at baseline influenced initiation of lipid lowering medications. That paper did not consider baseline use of lipid medication, and the focus was exclusively on CAC which we will not consider in this paper. AM035 (Jorgensen) will consider the use of imputed pre-treatment cholesterol values as part of a propensity score model to reduce confounding by indication. That paper will not consider anti-hypertensive medication use, and does not specifically examine the predictors of medication use in a model focused on interpretation.

References

McClelland, R., Kronmal, R., Haessler, J., Blumental, R., and Goff, D. J. (2008). Estimation of risk factor associations when the response is influenced by medication use: an imputation approach. Statistics in Medicine, 27(24):5039–5053.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download