Patient Characteristics and Medication Burden Affect Adherence Among Seniors

AJPB® Translating Evidence-Based Research Into Value-Based Decisions®July/August 2014
Volume 6
Issue 4

Adherence among seniors to 3 widely used chronic medication classes differed by medication burden, patient characteristics, and specific comorbid conditions.

By 2030, the number of seniors in the United States is projected to reach approximately 72 million, or 20% of the total population.1 This is a 14% increase from 2012 and is because of longer life spans and the aging of the baby boom generation. Seniors frequently have multiple chronic conditions (MCCs), defined by the Agency for Healthcare Research and Quality (AHRQ) as 2 or more chronic physical or behavioral health problems.2 Compared with individuals aged 45 to 64 years, seniors are significantly more likely to have MCCs (80% vs 49%), and that proportion is expected to increase.2 Patients with MCCs account for 66% of US healthcare expenditures3 and 93 cents of every Medicare dollar spent.2 Seniors with MCCs are at higher risk for hospitalizations and incur more physician office and emergency department visits than seniors with 1 or no chronic condition.4-6 The CMS 2012 Chronic Conditions Chartbook states that Medicare beneficiaries with MCCs accounted for 98% of the 1.9 million hospital readmissions in 2010.4

Management of concurrent conditions often requires complex medication regimens that are associated with suboptimal medication adherence.7,8 Several studies of medication adherence in elderly populations demonstrate an association between higher overall comorbidity levels and lower adherence to medications indicated for the treatment of diabetes, hypertension, and hyperlipidemia.9-11 Poor adherence to these therapies is linked to increased healthcare utilization, increased healthcare costs, and decreased quality of life.12-16 Additional patientrelated risk factors for nonadherence include age, sex, and low socioeconomic status.13,15,16 Disease-related risk factors for nonadherence include cognitive impairment and depression,10,17,18 which are prevalent in the elderly population.19

Because pharmacotherapy is often an essential component of chronic disease treatment, poor adherence to medication compromises the ability to manage chronic diseases and to achieve the best health outcomes, particularly for those with MCCs. We describe how medication adherence for 3 highly utilized chronic medication classes (antidiabetics, renin-angiotensin system [RAS] antagonists, and statins) varied by patient characteristics and comorbidities in a senior population. These medication categories are indicated for the treatment of 3 of the 5 most prevalent chronic conditions in seniors (hypertension [58%], hyperlipidemia [45%], and diabetes [28%])4 and are included in CMS’s Star Ratings adherence performance measures.


We conducted a retrospective analysis using pharmacy fill information from participating MedImpact clients to estimate medication comorbidity and adherence for members 65 years and older. Pharmacy fill information during the calendar year 2013 was used to estimate comorbidity among Medicare beneficiaries with 3 highly prevalent chronic conditions: diabetes, hypertension, and hyperlipidemia. Presence of each condition was determined by using a drug proxy of 1 or more fills for an antidiabetic (including insulin), antihypertensive, or antihyperlipidemic.Combinations of concurrent morbidity among these 3 chronic conditions were calculated (eg, an antihypertensive and an antihyperlipidemic).

Adherence rates for the 3 target medication classes used by CMS in its Part D Star Rating Measures (Medication Adherence for Diabetes, Medication Adherence for Hypertension [RAS antagonists], and Medication Adherence for Cholesterol [Statins]) were estimated for the population subgroup meeting requirements for measurement.20 Adherence was measured as the proportion of days covered (PDC) by medication using technical specifications from the CMS Medicare 2014 Part C and D Technical Notes (updated April 2, 2014) and the Acumen, LLC (CMS contractor) Adherence Measures PDP/MAPD Contract Report User Guide (released April 2014).20,21 Proportion of days covered was measured from first fill to end of the 2013 calendar year or member disenrollment (patient review period). Days of medication coverage were calculated using information on fill dates and days of supply within each patient’s review period.22 Patients with at least 2 fills within the target medication class and a review period of at least 91 days were included. Patients using insulin were excluded from the diabetes measure.

To determine the degree of influence of various risk factors and comorbidities on medication adherence, the proportion of total patients adherent (PDC >80%, the threshold used by CMS) to each of the 3 medication classes was calculated and stratified by several patient characteristics: age, sex, receiving low-income subsidy (LIS), number of target chronic conditions as determined by drug proxy, and select conditions as determined by drug proxy that have high pill burden (eg, human immunodeficiency virus infection), have high costs (eg, cancer), or are known to negatively impact medication adherence (eg, depression, Alzheimer’s disease). We used a proprietary drug proxy developed for identification of 30 chronic disease categories. Disease categories were based on the MCC list used by CMS and AHRQ.2,4 Additionally, counts of unique prescribers and pharmacies visited were measured for each patient to identify regimen complexity and potential challenges associated with coordination of care. All analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC).


Among the total number of patients with at least 1 prescription fill for any of the 3 chronic medication classes, antihypertensive use was the most frequent (88.6%), followed by use of an antihyperlipidemic (64.6%) and an antidiabetic (26.8%). Approximately 39% filled a prescription for only 1 medication class (29.0% antihypertensive, 8.8% antihyperlipidemic, and 1.3% antidiabetic;

Figure 1

). The most common dyad was antihypertensive and antihyperlipidemic use (35.5%), and 19.0% of patients filled a prescription for all 3 medication classes.

Among the subgroup of patients meeting criteria for adherence measurement, incremental increases in adherence were observed for patients utilizing additional target classes (

Figure 2

). The proportion of patients adherent (PDC >80%) to medications for diabetes among those taking all 3 classes (85.4%) was 12.3 percentage points greater than the proportion of adherent patients utilizing antidiabetics only (73.1%). The proportion of adherent patients increased with additional target medication class utilization at a lesser rate for the other 2 measures: 81.0% to 84.8% for RAS antagonists and 77.5% to 80.2% for statins.

Overall, the percentage of adherent patients by class for antidiabetics, RAS antagonists, and statins was 82.5%, 83.3%, and 79.2%, respectively (


). Approximately half (49.7%) of patients utilized more than 1 target medication class and 86.2% of patients were nonadherent to 1 or more target medication classes: 48.4% were nonadherent to 1 class, 28.8% to 2 classes, and 9.0% to all 3 classes (data not shown). The proportion of adherent patients varied slightly by age group for all 3 target classes; however, adherence was lower among females and patients receiving medication subsidies based on low-income status. Decreased adherence was similarly observed among patients who had 4 or more prescribers and were utilizing 4 or more pharmacies. Adherence trends by total comorbidity were consistent for all 3 classes; rates were lower for patients with 1 condition or 10 or more conditions compared with patients who had between 2 and 9 conditions. Adherence to target medication classes was consistently lower for patients treated with concomitant antidepressants, antimigraine agents, or medications for asthma or chronic obstructive pulmonary disease (COPD). Adherence ranged from 77.9% for statins to 81.1% for RAS antagonists among patients treated for depression and 78.2% to 80.7% among patients treated for asthma or COPD. Patients treated for Alzheimer’s disease demonstrated lower adherence (4 to 5 percentage points) to target medication classes than patients without treatment.


This study used a large Medicare population to assess the influence of medication burden and patient characteristics on adherence to 3 frequently used chronic medication classes. The largest percentage of patients utilized antihypertensives, followed by antihyperlipidemics, and lastly, antidiabetics, which is consistent with national prevalence of these 3 conditions. The majority of patients were utilizing medications for more than 1 condition, which likely indicates considerable MCC burden among Medicare beneficiaries.

The proportion of adherent patients was 83%, 83%, and 79% for the target medication classes (antidiabetics, RAS antagonists, and statins, respectively). These rates, although higher than national rates (77%, 79%, and 75%) reported by CMS for the same 2013 calendar year, followed a similar relative trend across the medications measured (lowest rates for statins and highest rates for RAS antagonists.

We found that adherence increased as patients utilized more than 1 of the target medication classes. These results are similar to findings of a large longitudinal study that demonstrated increased adherence with higher comorbidity in patients with hypertension and hypercholesterolemia.9 This trend is possibly driven by the increase in adherence interventions used by health plans to increase performance to these CMS Star Ratings measures.

We did find decreased adherence to the target medications among patients with the highest medication burden. Generally, adherence to the target medication measures was lower in patients using antidepressants, antineoplastics, and medications to treat Alzheimer’s disease, migraines, asthma, or COPD. Our finding of lower adherence in members with depression but higher adherence among members with hypertension and diabetes is consistent with previous research. A large observational study of Medicaid patients newly initiating statins found greater odds (19%) of suboptimal persistence in patients treated for depression but decreased odds of suboptimal persistence in patients with hypertension, stroke, congestive heart failure, or diabetes.11

Regimen complexity is a possible reason for decreased adherence with greater comorbidity. Several studies have demonstrated lower statin and antihypertensive adherence with greater medication use, higher number of prescribers, and higher number of pharmacy visits.6,7 Low health literacy is also a likely driver of lower medication adherence.23,24 Research demonstrates that adults 65 years or older have the poorest health literacy skills,25 which places them at risk for not comprehending provider instructions.24,26 Health literacy and the study of the progression of concurrent conditions are 2 research areas emphasized by the AHRQ MCC Research Network.2

Our findings of lower adherence among patients receiving LIS are congruent with CMS’, published adherence rates for the same 2013 measurement period. CMS Patient Safety Reports released in May 2014 demonstrate lower adherence rates, ranging between 3% and 7% lower, for LIS beneficiaries compared with non-LIS beneficiaries in Medicare Part D populations.21 With rising medication costs, greater attention should be given to helping seniors under financial constraints.

When stratifying adherence by multiple patient characteristics and conditions, we observed several patient groups with poor adherence rates. Most notable was the 9.0% of patients who were nonadherent to all 3 target classes. More detailed stratification found that patients who used an antidepressant and had a history of using 4 or more prescribers and visiting 4 or more pharmacies had meaningfully lower adherence (12 percentage points) to diabetes medications than a patient group without depression who used 1 pharmacy and prescriber. These rates were found to be even lower in patients receiving LIS. We also observed lower rates of adherence (9%) to statins in females 85 years or older who received LIS and had more than 6 chronic conditions compared with younger females with lower total comorbidity counts.

Payers, health systems, and accountable care organizations are now more likely to segment and target populations based on their adherence risk profile. Key contributors to poor medication adherence identified in this analysis provide some clues on what factors should be included in predictive models and targeted interventions. Tailored strategies that target segments of the senior population with interventions that address medication complexity, disease burden, coordination of care, and so forth will be critical to improving outcomes among seniors. As the proportion of older adults with MCCs increases, health plans that are able to improve care coordination with various touch points in a patient continuum of care will have an advantage. A more patient-centered approach versus a disease-specific model of care, as outlined in the Affordable Care Act, may address the growing complexities among the senior population.

Our analysis based on pharmacy fill information may underestimate true prevalence of the conditions we measured; however, use of pharmacy fill information as a surrogate for medical conditions has been demonstrated to be effective at predicting medical expenditures and adjusting for patient mix in large populations.27-29 Our study may overestimate actual patient adherence, but the use of refills has been shown to yield valid estimates of adherence to maintenance medications within integrated health systems.30,31 Although this analysis was conducted on a portion of the total Medicare population, our findings are based on a large, diverse Medicare study population.


Adherence to 3 widely used chronic medication classes differed by medication burden, patient characteristics, andspecific comorbid conditions. Poor adherence was more common in seniors receiving LIS, patients with cognitive impairment, patients with behavioral health conditions, and patients treated by multiple providers. Variation in adherence rates reveals subgroups of patients who are particularly at risk for suboptimal adherence. Medication adherence improvement strategies should consider interventions that identify and focus attention on groups with multiple concurrent conditions and coordination-of-care opportunities.

Related Videos
Practice Pearl #1 Active Surveillance vs Treatment in Patients with NETs
© 2024 MJH Life Sciences

All rights reserved.