Abstract
BACKGROUND:
Hypertension, hyperlipidemia, and type 2 diabetes (T2D) are 3 of the most common chronic conditions, but related medication adherence rates are far below 80%. Consequences of poor adherence include high health care utilization/costs and increased mortality. There is accumulating evidence in support of the benefits of affording pharmacists the opportunity to practice at the full scope of their licensure by engaging in patients’ clinical care.
OBJECTIVE:
To examine the impact of a large national pharmacy chain’s pharmacist-led interventions to improve medication adherence among older adults with hypertension, hyperlipidemia, or T2D. A secondary objective was to estimate the potential cost savings associated with improved adherence.
METHODS:
Participants were Medicare patients aged 18 years or older who had 2 or more prescription fills in at least 1 of the 3 therapeutic classes. The primary outcome, optimal adherence, was defined as proportion of days covered (PDC) of 80% or higher. A difference-in-differences (DID) design with a generalized linear model analytical approach was applied to examine differences between intervention participants and controls. The study period spanned from 2020 to 2022.
RESULTS:
Intervention participants (n = 317,613, age 70.1 years, female sex 57.0%) had lower baseline optimal adherence than controls (n = 943,389, age 73.3, female sex 56.1%) for diabetes (76.9% vs 79.8%), hypertension (79.0% vs 83.0%), and cholesterol (78.6% vs 82.1%). The DID results showed that between 2020 and 2022, optimal adherence had significant absolute increases for intervention participants (diabetes: +4.0%, hypertension: +6.3%, cholesterol: +6.1%) vs controls who declined in adherence (diabetes: −1.6%, hypertension: −0.4%, cholesterol: −1.4%). All DID models were significant at P < 0.0001. Total cost of care was projected based on improvements in adherence. Based on PDC improvements for the test population, we estimate that the pharmacist consultations were associated with annual total health care cost savings of $10,329,284 ($109 per capita), $31,640,660 ($122 per capita), and $21,589,875 ($75 per capita) for test population patients with diabetes, hypertension, and hyperlipidemia, respectively.
CONCLUSIONS:
The study found that the pharmacist-led interventions were significantly associated with increased optimal adherence over 2 years. These findings demonstrate the potential of pharmacist-led interventions to improve medication adherence among older adults with chronic conditions. Strategies to expand pharmacist-provided care must be further examined.
Plain language summary
Older people can have problems with blood pressure, cholesterol, and diabetes. Sometimes they do not take their medicine as the doctor suggested. We found pharmacists could help them take their medicine correctly and consistently. Health care is less expensive when patients take their medicine.
Implications for managed care pharmacy
New strategies are emerging to reduce the traditional care model for chronic disease management. Pharmacists have training to address many patient care needs and are among the most easily accessible clinicians. Therefore, pharmacists have potential to add further value as care partners. Key industry partners have called for more strategies to expand value-based pharmacist care. The findings from this study demonstrate the potential of pharmacist-led interventions to improve medication adherence among older adults with chronic conditions.
The vast majority (87.6%) of adults aged 65 years or older in the United States have at least 1 chronic disease, and 63.7% have 2 or more chronic diseases.1 Managing one’s own condition is key to preventing or slowing disease progression.2 Chronic disease self-management can be overwhelming to patients, particularly when a patient has 2 or more conditions to manage.3 Barriers and facilitators to chronic disease self-management have been well-documented and largely fall within the areas of personal/lifestyle characteristics, health status, resources, environmental characteristics, and health system care.4 Health care–quality factors most relevant to condition self-management include access to care, continuity of care, and relationships with clinicians.2
Numerous evidence-based programs exist to educate and support patients in disease self-management. Of these programs, medication adherence is consistently cited as a critical component for successful disease management. For instance, and one of the most widely disseminated evidence-based self-management programs—the Chronic Disease Self-Management Program—includes medication management as a key facet of the intervention.5
New strategies are emerging to integrate a diverse group of clinicians using a team-based method toward chronic disease management.6,7 Hypertension, hyperlipidemia, and type 2 diabetes (T2D) are 3 of the 5 most common chronic diseases among older Americans, but related medication adherence rates are far below 80%.8,9 However, since the application of Center for Medicare and Medicaid services (CMS) STARS rating system, medication adherence rates have significantly improved for medications both targeted and not targeted by the STARS system.10 Consequences of poor adherence include high health care utilization/costs, decreased quality of life, and increased mortality.11
Given the complexities of medication adherence, a multidisciplinary approach is needed.12 Pharmacists can add value in care teams because they have training to address many patient care needs and are among the most accessible clinicians. Key industry partners recognize this opportunity and are calling for more strategies to expand value-based pharmacist care.13
Understanding the importance of medication adherence in chronic disease self-management and the role that pharmacists can play, Walgreens Advanced Care pharmacist consultation program was developed and implemented to assist with patients’ medication adherence needs. We conducted a study to measure changes in medication adherence among older adults with hypertension, hyperlipidemia, or T2D after implementation of this pharmacy program. A secondary objective of this study was to estimate the potential cost savings associated with improved adherence.
Method
PROGRAM DESCRIPTION
The Walgreens Advanced Care program is based on a combination of advanced machine learning models that continuously train and are complemented by a rules engine. The model inputs include demographic, geographic, patient utilization patterns, plan, and other administrative data. The patients are ranked and identified for the appropriate intervention by the optimal channel at the best time. The identified patients were subsequently given the appropriate pharmacist-delivered multichannel adherence interventions. Multiple steps of outreach were introduced intending to build a relationship between the pharmacist and patient, including but not limited to face-to-face, telephonic, and/or digital outreach. Touchpoints included calling patients on their first refill within the calendar year, discussing possible barriers due to their complex therapeutic regimen, providing personalized solutions to improve their adherence and health outcomes, and making follow-up calls to support patients throughout the year. The Advanced Care program had been launched in many Medicare Part D plans since 2019, which would potentially have benefited patient adherence over years.
STUDY DESIGN
This quasi-experimental study used a difference-in-differences (DID) approach to compare 2 groups: a treatment group (patients under selected Medicare Part D plans that partnered with the Advanced Care program at the end of 2020) and a control group (patients under all other Medicare Part D plans, which had never been covered by the Advanced Care program but were exposed to other standard-of-care programs, including walk-in pharmacist consultations and digital and telephonic refill and pick-up reminders) across 3 years (2020-2022). The study period included a pre-intervention period prior to the program (calendar year 2020), whereas calendar years 2021 and 2022 served as the post-intervention periods. Additionally, calendar years 2021 and 2022 were compared to assess the incremental change, given the continuation of the program over 2 years. Eligible patients were enrolled in a Medicare Part D plan, were at least 18 years of age, and had 2 or more prescription fills in at least 1 of 3 therapeutic classes (diabetes, hypertension, and cholesterol) during the observation period (Figure 1).
FIGURE 1.
Study Design
DATA AND METRICS
The primary outcome was the binary variable, optimal medication adherence. Optimal adherence was assessed by calculating the proportion of days covered (PDC) at the patient/therapeutic class level and subsequently categorizing values into 2 groups: those that reached a PDC threshold of 80% during the measurement year and those that did not (PDC < 80%). Certain disease state–specific medications (ie, insulin for diabetes and sacubitril/valsartan for hypertension) were excluded per CMS protocol.14 The demographic characteristics included age, sex, urbanicity, and median income. Adherence data used in this study were obtained from pharmacy claims of eligible Medicare Part D patients in the pharmacy’s internal data warehouse. Missing/incomplete data, although rare, were removed from the analysis. Data were then processed using the year-to-date methodology (measurement period starts from patient’s index date, [first fill date], for assessing performance through the last day of the calendar year) elaborated by EQUIPP (http://www.phamacyquality.com) for year-end performance. The measurement period must have been at least 91 days long for the patient to be included. When multiple prescriptions for the same target medication were dispensed, if there was a days’ supply overlap, the start date for the latter fill was adjusted to be the day after the days’ supply for the previous fill had ended.
STATISTICAL APPROACH
Descriptive statistics for age (means) and sex (proportions) were analyzed at the patient level. Demographic variables for urbanicity (proportions) and median income (median) were measured at the store level. The comparison between groups was performed by Student’s t-test (age and median income) or chi-square test (sex and urbanicity) in SAS EG 9.4 (SAS Institute Inc.). A DID ([treatment post-intervention − control post-intervention] − [treatment pre-intervention − control pre-intervention]) approach was applied to compare between groups and between pre-intervention and post-intervention periods; P values were assessed using the generalized linear model. The assumption was that the treatment group would have a similar trend of adherence over time to the control group in the absence of treatment. The model was fit to the data using Maximum Likelihood Estimation, after which the DID estimate was obtained by taking the estimated coefficients for the interaction between treatment and time using least squares means estimation, which represented the treatment effect. The hypothesis test was conducted to assess the statistical significance of the DID estimate. All statistical tests were evaluated at a 2-sided significance level of P ≤ 0.0001.
TOTAL COST-OF-CARE SAVINGS
Results from the DID estimate for each medication group were used to approximate the incremental total cost-of-care savings. Total cost of care, defined as the sum of all medical and pharmacy claim allowable amounts (plan, member, and third-party payer), was derived from an adherence-based cost-savings model published by Prime Therapeutics.15 The totals were represented as per patient per year (PPPY) based on a study population of Medicare patients. To fit the model, our PDC results were categorized as follows: below 50 (reference), 50-79, 80-94, 95-98, and 99 and higher. Compared with the reference category of below 50 PDC, cost savings for increased adherence varied from $888 to $3,365 PPPY for diabetes, from $421 to $2,901 for hypertension, and from $262 to $1,563 for cholesterol.15 The total cost-of-care savings was applied to each PDC category for 2021 and 2022. The DID between test and control was calculated to estimate the incremental savings associated with the Advanced Care program. All dollar amounts were adjusted for health care inflation.16
Results
In 2020 (pre-intervention), the treatment group had 317,613 unique patients who enrolled in selected Medicare Part D plans before the Advanced Care program was launched. The control group had 943,389 unique patients who enrolled in other Medicare Part D plans and were never exposed to the Advanced Care program but received the standard pharmacy care. In 2021 (the first post-intervention year), there were 405,794 unique patients in the treatment group and 940,761 unique patients in the control group. In 2022 (the second post-intervention year), 547,852 unique patients were included in the treatment group; 845,537 unique patients were included in the control group. The treatment group continued to grow after the Advanced Care program was launched, whereas the control group maintained a similar size over 3 years (Table 1).
TABLE 1.
Sample Attrition Table
| Attrition criteria | n |
|---|---|
| Total number of patients from 2020 to 2022 | 2,611,290 |
| Patients with exclusion criteria | 475,333 |
| Missing or undefined sex value | 15 |
| Eligible patients for this study | 2,135,942 |
| 2020 Cohort | 1,261,002 |
| Treatment group | 317,613 |
| Control group | 943,389 |
| 2021 Cohort | 1,346,555 |
| Treatment group | 405,794 |
| Control group | 940,761 |
| 2022 Cohort | 1,393,389 |
| Treatment group | 547,852 |
| Control group | 845,537 |
Before the launch of the Advanced Care program in 2020, the treatment group had lower baseline performance compared with the control group in optimal adherence (P < 0.0001). Within the same period, treatment group patients had a lower frequency of suburban residence and lower median income compared with those in the control group (Table 2). The initial patient characteristic distributions suggested that these two groups are comparable. Our initial treatment group consisted of nearly half of participants with more than 1 chronic condition (a population at higher risk for nonadherence because of the complexity of their regimen) (Supplementary Table 1 (159.8KB, pdf) , available in online article).
TABLE 2.
Patient Demographic Characteristics of Study Cohorts in the Pre-Intervention and Post-Intervention Periods (2020-2022)
| Year | 2020 | 2021 | 2022 | |||
|---|---|---|---|---|---|---|
| Group | Treatment | Control | Treatment | Control | Treatment | Control |
| Unique patients, n | 317,613 | 943,389 | 405,794 | 940,761 | 547,852 | 845,537 |
| Diabetes | 74,681 | 207,365 | 95,102 | 211,612 | 129,379 | 196,563 |
| Hypertension | 204,452 | 610,589 | 259,180 | 602,184 | 348,037 | 535,690 |
| Cholesterol | 218,962 | 666,615 | 289,792 | 678,650 | 399,487 | 613,452 |
| Age (mean) | 70.1 | 73.3 | 71.0 | 73.5 | 71.7 | 73.6 |
| Sex, % | ||||||
| Female | 57.0 | 56.1 | 56.3 | 56.2 | 55.3 | 56.2 |
| Male | 43.0 | 43.9 | 43.7 | 43.8 | 44.7 | 43.8 |
| Urbanicity, % | ||||||
| Urban | 10.0 | 10.6 | 8.9 | 10.4 | 8.2 | 11.6 |
| Rural | 35.6 | 33.6 | 39.5 | 34.0 | 42.8 | 32.5 |
| Suburban | 54.4 | 55.8 | 51.6 | 55.6 | 49.0 | 55.9 |
| Median income, $ | 60,795 | 64,327 | 62,003 | 64,361 | 62,880 | 64,323 |
One year after the Advanced Care program launch (2021), the treatment group increased, at which point optimal adherence was quite similar between the treatment and control groups: 79.2% ± 0.1% vs 79.4% ± 0.1% for diabetes, 83.0% ± 0.1% vs 83.2% ± 0.1% for hypertension, and 81.8% ± 0.1% vs 81.5% ± 0.1% for cholesterol (Figure 2). The DID model showed that from 2020 to 2021, compared with the control group, the treatment group had an absolute increase of 2.8% ± 0.2% in PDC80 for diabetes, an increase of 3.7% ± 0.1% in PDC80 for hypertension, and an increase of 3.8% ± 0.1% in PDC80 for cholesterol (all with P < 0.0001) (Table 3).
FIGURE 2.
Impact of Advanced Care Program on Optimal Adherence
TABLE 3.
Statistical Results Summarized from Differences-in-Differences Model
| Least squares means estimate, % | SE, % | z value | Pr > |z| | |
|---|---|---|---|---|
| 2020 vs 2021 | ||||
| Diabetes | 2.8 | 0.2 | 12.02 | <0.0001 |
| Hypertension | 3.7 | 0.1 | 26.55 | <0.0001 |
| Cholesterol | 3.8 | 0.1 | 27.92 | <0.0001 |
| 2021 vs 2022 | ||||
| Diabetes | 2.9 | 0.2 | 14 | <0.0001 |
| Hypertension | 3.0 | 0.1 | 23.79 | <0.0001 |
| Cholesterol | 3.6 | 0.1 | 29.96 | <0.0001 |
| 2020 vs 2022 | ||||
| Diabetes | 5.7 | 0.2 | 25.75 | <0.0001 |
| Hypertension | 6.6 | 0.1 | 45.08 | <0.0001 |
| Cholesterol | 7.5 | 0.1 | 52.82 | <0.0001 |
By the end of 2022, after the Advanced Care program had been operational for these patients for 2 years, the optimal adherence for the treatment group continued to climb and surpassed the control group: 80.9% ± 0.1% vs 78.2% ± 0.1% for diabetes, 85.3% ± 0.1% vs 82.6% ± 0.1% for hypertension, and 84.7% ± 0.1% vs 80.7% ± 0.1% for cholesterol (Figure 2). From 2021 to 2022, compared with the control group, the treatment group had a percentage point increase of 2.9% ± 0.2% in PDC80 for diabetes, a percentage point increase of 3.0% ± 0.1% in PDC80 for hypertension, and a percentage point increase of 3.6% ± 0.1% in PDC80 for cholesterol (all with P < 0.0001) (Table 3). These findings suggested that the program was associated with sustainable improvements in medication adherence.
The treatment group’s improvements in optimal adherence increased across the 2-year study period for the 3 disease states. The overall gain for the treatment group compared with the control group was 5.7% ± 0.2% in PDC80 for diabetes, 6.6% ± 0.1% in PDC80 for hypertension, and 7.5% ± 0.1% in PDC80 for cholesterol (all with P < 0.0001) (Table 3). The consecutive increase in performance for the treatment group demonstrates the positive impact of the Advanced Care program on medication adherence over time.
Based on PDC improvements for the test population, we estimate that the Advanced Care pharmacist consultations were associated with $108.61 PPPY in annual total health care cost savings (Table 4) for diabetes ($10,329,284 for the 2021 test population), $122.08 PPPY in annual total health care cost savings for hypertension ($31,640,660 for the test population), and $74.50 PPPY in annual total health care cost savings for cholesterol ($21,589,875 for the test population).15,16
TABLE 4.
Total Cost-of-Care Savings (2021-2022)
| Diabetes | Savings based on Prime Therapeutics model10 ($) | Inflation-adjusted savingsa ($) |
|---|---|---|
| Total savings per patient difference: test | 61.88 | 72.21 |
| Total savings per patient difference: control | (31.19) | (36.40) |
| Savings per patient (difference) | 93.07 | 108.61 |
| Hypertension | ||
| Total savings per patient difference: test | 75.83 | 88.49 |
| Total savings per patient difference: control | (28.78) | (33.59) |
| Savings per patient (difference) | 104.61 | 122.08 |
| Cholesterol | ||
| Total savings per patient difference: test | 48.74 | 56.88 |
| Total savings per patient difference: control | (15.10) | (17.62) |
| Savings per patient (difference) | 63.84 | 74.50 |
a Prime Therapeutics model savings estimates, adjusted for inflation by $1.167.
Discussion
Evidence supports the value of engaging pharmacists in proactive chronic condition management. The study presented here found that, compared with participants in the control group, patients who received the Advanced Care pharmacist outreach experienced significant improvements in optimal adherence over the 2-year intervention period. These results strongly agreed with our hypothesis that patients with multiple chronic conditions would receive benefit from personalized care. Using health care expenditure data from published data,14 our model shows that the improvements in adherence equate to a PPPY cost savings of $108.61, $122.08, and $74.50 for diabetes, hypertension, and cholesterol, respectively.
Collaborative partnership models between pharmacists, payers, and other clinicians have led to successful patient outcomes across disease states such as cardiovascular, diabetes, chronic pain management, and HIV, among others.17,18 A 2023 research review found that pharmacist-provided care in community pharmacies reduced drug-related problems and benefited patient education, clinical outcomes, and economic outcomes.18 Another recent review found that pharmacists-led chronic condition management is associated with improvements in outcomes across disease states, including diabetes, hyperlipidemia, HIV/AIDS, cardiovascular conditions, and respiratory diseases.17
LIMITATIONS
Our study had several limitations. First, this was an observational study wherein patients in only select Medicare Part D plans, which had opted to partner with the Advanced Care program, were assigned to the treatment group. To address this potential for bias, we examined key demographics between test and control groups. Although the groups appeared similar, it is possible other unmeasured demographics differed between the groups. Second, the program evaluated in the study specifically targets patients who were at risk for nonadherence; therefore, not all patients in the treatment group received the intervention. Finally, since the treatment group started with lower medication adherence in the pre-intervention period, these patients naturally had more potential for change over time compared with the control group, which started with higher adherence.
Conclusions
As suggested by the current study, pharmacists are playing an active role in chronic condition management by ensuring better adherence to prescribed medication. Key industry partners recognize this opportunity and are calling for more strategies to expand value-based pharmacist care.9 Leveraging pharmacists to deliver a broader range of patient services can facilitate continuity of care across providers, facilitate stronger patient-pharmacist relationships, and allow the patient to benefit from receiving comprehensive care delivered through multiple lenses of expertise. However, more research is needed on clinical and economic outcomes to support value-based care strategies and pathways to pay pharmacists for such services.
ACKNOWLEDGMENTS
The authors acknowledge Peggy Wonders, Mouna Zamran, and Bhooma Shivakumar for their critical review and editing of the description of Walgreens Advanced Care program.
Funding Statement
This study was funded internally by Walgreen Co.
REFERENCES
- 1.Boersma P, Black LI, Ward BW.. Prevalence of multiple chronic conditions among US adults, 2018. Prev Chronic Dis. 2020;17:E106. doi:10.5888/pcd17.200130 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Grady PA, Gough LL.. Self-management: A comprehensive approach to management of chronic conditions. Am J Public Health. 2014;104(8):e25-31. doi:10.2105/AJPH.2014.302041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Koch G, Wakefield BJ, Wakefield DS.. Barriers and facilitators to managing multiple chronic conditions: A systematic literature review. West J Nurs Res. 2015;37(4):498-516. doi:10.1177/0193945914549058 [DOI] [PubMed] [Google Scholar]
- 4.Schulman-Green D, Jaser SS, Park C, Whittemore R. A. metasynthesis of factors affecting self-management of chronic illness. J Adv Nurs. 2016;72(7):1469-89. doi:10.1111/jan.12902 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Allegrante JP, Wells MT, Peterson JC.. Interventions to support behavioral self-management of chronic diseases. Annu Rev Public Health. 2019;40(1):127-46. doi:10.1146/annurev-publhealth-040218-044008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Smith SM, Wallace E, O’Dowd T, Fortin M.. Interventions for improving outcomes in patients with multimorbidity in primary care and community settings. Cochrane Database Syst Rev. 2021;1(1):CD006560. doi:10.1186/s13643-021-01817-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hogg W, Lemelin J, Dahrouge S, et al. Randomized controlled trial of anticipatory and preventive multidisciplinary team care: For complex patients in a community-based primary care setting. Can Fam Physician. 2009;55(12):e76-85. [PMC free article] [PubMed] [Google Scholar]
- 8.Kenreigh CA, Wagner LT.. Medication adherence: A literature review. MedScape. 2005. Accessed April 24, 2023. https://www.medscape.com/viewarticle/514164
- 9.Kleinsinger F. The unmet challenge of medication nonadherence. Perm J. 2018;22(3):18-033. doi:10.7812/TPP/18-033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Parekh N, Munshi KD, Hernandez I, Gellad WF, Henderson R, Shrank WH.. Impact of star rating medication adherence measures on adherence for targeted and nontargeted medications. Value Health. 2019;22(11):1266-74. doi:10.1016/j.jval.2019.06.009 [DOI] [PubMed] [Google Scholar]
- 11.While A. Medication adherence: Understanding the issues and finding solutions. Br J Community Nurs. 2020;25(10):474-9. doi:10.12968/bjcn.2020.25.10.474 [DOI] [PubMed] [Google Scholar]
- 12.Milosavljevic A, Aspden T, Harrison J.. Community pharmacist-led interventions and their impact on patients’ medication adherence and other health outcomes: A systematic review. Int J Pharm Pract. 2018;26(5):387-97. doi:10.1111/ijpp.12462 [DOI] [PubMed] [Google Scholar]
- 13.Pharmacy Quality Alliance. Strategies to expand value-based primary care: Action guide for community pharmacists, healthcare payers, and other stakeholders. 2019. Accessed April 24, 2023. https://pqa.memberclicks.net/assets/S2S/Pharmacist-Provided%20Care%20Action%20Guide.pdf
- 14.Centers for Medicare and Medicaid Services 2023 Medicare-Medicaid Plan Performance Data Technical Notes. Centers for Medicare and Medicaid Services. 2023. Accessed August 23, 2023. https://www.cms.gov/files/document/mmpperformancedatatechnotes.pdf [PubMed]
- 15.Qiu Y, Karaca-Mandic P, Gleason; P. Prime Therapeutics, LLC. Accessed August 23, 2023. https://www.primetherapeutics.com/wp-content/uploads/2021/09/SP_AMCP_-F20_Adherence.pdf
- 16.US Bureau of Labor Statistics. Databases, tables & calculators by subject. 2023. Accessed August 23, 2023. https://data.bls.gov/timeseries/CUUR0000SAM?output_view=pct_12mths
- 17.Newman TV, San-Juan-Rodriguez A, Parekh N, et al. Impact of community pharmacist-led interventions in chronic disease management on clinical, utilization, and economic outcomes: An umbrella review. Res Social Adm Pharm. 2020;16(9):1155-65. doi:10.1016/j.sapharm.2019.12.016 [DOI] [PubMed] [Google Scholar]
- 18.Eldooma I, Maatoug M, Yousif M.. Outcomes of pharmacist-led pharmaceutical care interventions within community pharmacies: Narrative review. Integr Pharm Res Pract. 2023;12:113-26. doi:10.2147/IPRP.S408340 [DOI] [PMC free article] [PubMed] [Google Scholar]