Documentations of Advanced Heath Care Directives Where Are They TAI_SEALE
Disparities in Antihypertensive Medication Adherence ADAMS
1. Antihypertensive Medication Adherence
among Newly Treated Patients:
Opportunities for Disparities Reduction?
Alyce S. Adams, PhD
Connie Uratsu, RN 18th Annual HMO
Wendy Dyer, MS Research Network
David Magid, MD, MPH Conference
Patrick O’Connor, MD, MA, MPH April 29-May 2, 2012
Arne Beck, PhD Seattle, WA
Melissa Butler, PhD
P. Michael Ho, MD, PhD
Julie A. Schmittdiel, PhD
2. Acknowledgements
INSTITUTIONS
Kaiser Permanente Division of Research, Oakland, CA; Institute for Research,
Kaiser Permanente, Denver, CO; Kaiser Permanente Center for Health
Research Southeast, Atlanta, GA; HealthPartners Research Foundation,
Minneapolis, MN; Denver VA Medical Center, Denver, CO
FUNDERS
National Heart, Lung, and Blood Institute and the National Institute for
Mental Health as a supplement to the HMO Research Network Cardiovascular
Disease Network [3U19HL091179-04S1].
National Institute for Diabetes, Digestive and Kidney Diseases Health Delivery
Systems Center for Diabetes Translational Research [P30DK092924]
(Adams, Schmittdiel, O’Connor)
OTHER
Dr. Alan Go (critical edits), Ms. Karen R. Hansen (manuscript preparation)
4. Conceptual Framework
Predisposing Factors
•Beliefs about risks and
Mediators
benefits of medicines
Primary
Health Status
Income •Medication Coverage Non-Adherence
•Patient-Provider Relationship
Race/Ethnicity Education
Geography •Perceived affordability
Whites
•Blacks Rural/Urbanicity
Social Support Enabling Factors
•Hispanics Early
Culture •Health Literacy/Education
•Asians Preferences •Patient self-care skills Non-Persistence
Racism •Medication Affordability
Stress •Medication Tolerability
Perceived Barriers
•Affordability/Ease of Access
•Competing Demands Non-Adherence
•Cognitive Issues/Complexity
5. Research Questions
1. Are racial and ethnic differences in antihypertensive
medication taking behavior consistent over time?
2. What factors contribute to differences in mediation taking
Behavior at different stage of adherence by race and
Ethnicity?
6. Methods
Setting: Kaiser Permanente Northern California
Patients: Adults (≥18 years) with hypertension who were new users of
antihypertensive therapy in 2008
Outcome Measures
Primary non-adherence: failing to fill a prescribed antihypertensive agent within
60 days after it was ordered by physician
Early non-persistence: failing to refill within 90 days of running out of the
first prescription
Non-adherence: not having medication available for 20% or more of days
during the 12 months following initiation of therapy
Modeling: Multivariate logistic regression analysis, with sensitivity analyses
using proc genmod and multiple imputation
8. Stages of Non-Adherence by Race/Ethnicity
45
40
35
30
25
20
15
10
5
0
White (non- Black (non- Asian Hispanic
Hisp) Hisp)
Primary Non-Adherent Early Non-Persistent
Non-Adherent
10. Logistic Regression Model Estimating Non-
Adherence with Antihypertensive Agents
Black (non- Asian (non- Hispanic
Hispanic) Hispanic)
Model 1: Age, Gender 1.73 (1.53-1.96) 1.20 (1.07-1.35) 1.68 (1.51-1.87)
+ smoking status, BMI, SBP 1.71 (1.51-1.94) 1.22 (1.08-1.37) 1.67 (1.51-1.86)
+ household income 1.67 (1.47-1.89) 1.22 (1.09-1.38) 1.65 (1.48-1.83)
+physical comorbidity 1.67 (1.47-1.90) 1.23 (1.09-1.38) 1.65 (1.48-1.84)
+mental health comorbidity 1.67 (1.47-1.90) 1.23 (1.09-1.39) 1.65 (1.48-1.84)
+ physician visits 1.68 (1.48-1.90) 1.23 (1.09-1.39) 1.65 (1.48-1.84)
+medication copay & mail order 1.54 (1.35-1.75) 1.13 (1.00-1.28) 1.48 (1.33-1.65)
pharmacy use
11. Key Findings
• In this setting where patients have more or less equal
access to care, non-white race was associated with
both early non-persistence & non-adherence
• These relationships were robust to the inclusion of
sociodemographic and clinical factors.
• However, the relationship between race/ethnicity and
non-adherence was appreciably attenuated by the
inclusion of medication copay and mail order
pharmacy use.
12. Limitations
• Unmeasured confounders
• beliefs and preferences unlikely to change over time
• limits our understanding of differences and why they
occur
• Logistic regression
• OR may overestimate effects, additional sensitivity
analyses planned
• Missing Data
• Results robust to multiple imputation
• Racial/Ethnic misclassification
• may bias results if the misclassification is correlated
with both race/ethnicity and adherence
13. Conclusions
• Racial and ethnic differences in medication
taking behavior occur early in the course of
treatment.
• System level changes that ease access to
medications may have the potential to
attenuate persistent gaps in the use of
these and other clinically effective therapies.
Sensitivity Analyses: No or very small differences between results from Proc Genmod and when multiple imputation used for missing BMI, systolic, HHincome, medvisits and copay for either model.