Improvement in Oral Diabetes Medication Adherence after a Comprehensive Wellness Assessment in a Dual Eligible Population

Abstract

Medication adherence is often lower among disadvantaged patients. Drivers of medication adherence may include the quality of communications between patient and medical caregiver.

The research objective is to assess whether an annual Comprehensive Wellness Assessment (CWA) is associated with improved medication adherence. The CWA targeted primarily dual eligible Special Needs Plan (SNP) enrollees in a Medicare Advantage plan. This retrospective panel study used administrative claims data and member-month level data for members who were newly diagnosed with diabetes in 2010, allowing for up to 5 years of follow-up. The intervention of interest is whether the member received a CWA in the past 12 months. Multivariate regression models were estimated using pooled member-month data and a difference-in-difference type approach to assess whether CWA visits improve oral diabetes medication (ODM) adherence among SNP enrollees. Twenty-six percent of pooled member-month observations are from SNP enrollees. Average monthly ODM adherence is 77.5%. Approximately 31% of SNP enrollees had a CWA in the last 12 months, compared to 5% of regular enrollees. Regression results show SNP enrollees with a CWA on average had higher monthly adherence by 3.9 percentage points (P < 0.01), and were 7% more likely to meet the threshold of at least 80% adherence (P < 0.01). Adherence is even higher for the subsample of African American SNP enrollees with CWA. CWA appears to be effective in improving ODM adherence among SNP patients. Care models with components like wellness assessments that include medication review and education may improve medication adherence as well as Medicare Advantage plan star ratings.

Introduction

The issue of medication nonadherence has generated significant interest because of its complexity from both a cost and outcomes perspective. It has been estimated that of the 3.2 billion prescriptions written annually in the United States, half are not taken as prescribed, especially among patients with asymptomatic chronic conditions.¹ By some estimates, the impact of medication nonadherence on health and health care costs account for approximately $100 billion in annual spending associated with hospital admissions that are medication related.^2
–4

Since 2012, the US Centers for Medicare & Medicaid Services (CMS) has included adherence to oral diabetes medications (ODM) among the triple-weighted measures for star ratings (1–5 stars) for Medicare Advantage (MA) plans. The measure is based on what percent of members who are diagnosed with diabetes and not on insulin fill prescriptions often enough to cover at least 80% of the days—also referred to as proportion of days covered (PDC)—they should be taking the medication. Being a triple-weighted measure further incentivizes MA plans to investigate what policies and interventions they can use to improve adherence to ODM.

This study uses data from a large MA plan in Louisiana with approximately 54,000 enrollees, 31.7% of whom are diagnosed as diabetic, and 25% of whom are Medicare-Medicaid eligible (dual eligible). The objective is to assess whether a Comprehensive Wellness Assessment (CWA) program, which the plan launched as part of a patient-centered model of care and that is targeted primarily at dual eligible Special Needs Plan (SNP) enrollees, has helped improve ODM adherence.

Overview

Adherence to medication regimens is critical to the successful management of chronic conditions,⁵ and research on medication adherence has grown in importance as the number of chronic conditions continues to increase and successful treatment becomes more dependent on patient self-management.⁶ Nonadherence to medication regimens is a result of a set of complex interactions involving prescribers, the patient, and the social environment. Scientific theory and previous research indicates that some of the factors that play a role include the provider-patient relationship and communications,⁷ disease burden, patient characteristics including socioeconomic characteristics,⁸ patient beliefs and awareness of what medication adherence will achieve in terms of health outcomes,^9,10 as well as other cost-benefit issues, and the patient's perceptions of social networks such as friends and family.^11
–13

Although medication adherence is a multifactorial phenomenon, there seems to be consensus that improving communication between patients and providers, and tailoring medication regimens with a patient-centered focus, including understanding a patient's beliefs, lifestyle, and relationship to their illness, are of high importance. These are all foundations for improving patient outcomes associated with the patient-centered medical home model.

The MA plan that is studied here has adopted a population health approach to care management. One element of that strategy was to reengineer a traditional medical management/utilization review program into a patient-centered model of care that includes a CWA offered annually to enrollees. The main element of the CWA is an annual 1- to 2-hour evaluation performed by a nurse practitioner (NP), which is targeted primarily toward the SNP enrollees. The CWA is intended to give enrollees an opportunity to participate in an extensive visit with a goal of improving patient self-management and engagement and ultimately improving outcomes. The assessment also affords the enrollee the opportunity to speak with a licensed clinical social worker and a clinical pharmacist. The CWA provides an opportunity for patients to have an unrushed discussion with the NP about their health status, addressing questions/concerns, and includes among its components a medication review and education on the importance of medication adherence. Exams may be completed in a variety of settings based on the enrollees' preference. Locations may include the enrollee's home, the organization's service center locations, or the enrollee's primary care physician (PCP) office. The ultimate goal is to improve patient outcomes and keep the enrollee active and informed about his/her own health management.

In addition, the patient-centered model includes a care management team comprised of a care coordinator registered nurse and social worker for each contracted PCP, adding the capacity needed to address the often significant psychosocial and nonclinical barriers to care. This study is the first empirical evaluation of whether the CWA is associated with improvements in an outcome that CMS gives high weight in its star ratings.

Methods

Population

The data set for this study was derived from a MA plan in southeast Louisiana with 54,000 enrollees. Using the organization's administrative claims data, a “new to diabetes” data set was created that included distinct members whose first definitive diagnosis of diabetes derived from claims data, with at least 1 International Classification of Diseases, Ninth Revision code, occurred as an enrollee in the plan between January of 2010 and December of 2014. Member-months were calculated from first enrollment date to current date, excluding months in which enrollees were not active in the plan.

Study design

A retrospective panel study was conducted employing the organization's administrative claims data, using pooled member-month observations for January 2010 through December 2014. The hypothesis of interest for this study is whether a CWA is associated with improved ODM adherence in the MA plan's SNP population.

The outcome of interest was monthly adherence to ODM. Adherence to ODM was operationalized using 2 different measures. The first was a measure of “ODM monthly adherence,” which is the proportion of days in the month covered by fills of diabetes medication; the second was a binary indicator that was termed “ODM-adherent,” indicating whether the PDC was ≥80% versus <80%. Monthly adherence was calculated using a metric that closely parallels the metric used by CMS to calculate annual adherence. Specifically: (1) All member-months were excluded for years during which a member had any claim for a fill of insulin. (2) Only the member-months after the months on or after the first oral diabetes fill of the year were included. (3) It was assumed that the member received appropriate ODM during acute inpatient or observation days and these days were subtracted from the numerator and denominator; member-months that were completely associated with acute inpatient and/or observation stay days were excluded from the measure. (4) Numerator days were calculated as total ODM fill days—and in cases in which the member had more ODM fill days than denominator days, adherence for the month was set at 100%. One notable aspect in which this metric diverges from the CMS approach is that CMS excludes those member observations whose first ODM fill is within the last 90 days of the year. Because the study team had full access to the data and for completeness, the team included month-level observations after the first fill of the year, even if that occurs in October or later for any year. The main intervention of interest is whether the member who is a SNP enrollee received a CWA in the past 12 months. Other member characteristics included in the statistical analyses are sex (1 if female), race (1 if minority), age in years, binary indicators for specific chronic conditions (ie, end-stage renal disease, chronic obstructive pulmonary disease [COPD], congestive heart failure [CHF] symptoms, cancer, hypertension [HTN], chronic kidney disease [CKD]), the number of consecutive months of program enrollment prior to the diabetes diagnosis, and the number of acute inpatient days and observation days in that current month. Descriptive statistics for the pooled SNP and non-SNP samples are presented in Table 1. The main hypothesis of interest is that receiving a CWA is associated with better adherence among SNP enrollees.

Table 1.

Descriptive Statistics by SNP Status and by CWA Visit

Variables	Non-SNP Mean (N = 118215)	SNP Mean (N = 39107)	Mean Diff
Months on oral medication	0.778	0.769	0.009^***
Months adherent	0.707	0.693	0.013^***
Minority	0.386	0.644	−0.258^***
Female	0.509	0.68	−0.171^***
Age	69.316	66.056	3.261^***
End-stage renal disease	0.001	0.003	−0.002^***
Chronic obstructive pulmonary disease	0.064	0.091	−0.027^***
Congestive heart failure	0.069	0.074	−0.004^***
Cancer	0.034	0.03	0.004^***
Hypertension	0.49	0.553	−0.063^***
Chronic kidney disease	0.136	0.146	−0.010^***
Consecutive months enrolled	17.36	14.549	2.811^***
Acute inpatient days	0.102	0.133	−0.032^***
Observation days	0.019	0.026	−0.007^***

Variables	SNP, No CWA ^a Mean (N = 27137)	SNP, CWA ^a Mean (N = 11880)	Mean Diff
Months on oral medication	0.759	0.793	−0.034^***
Months adherent	0.677	0.73	−0.053^***
Minority	0.642	0.65	−0.008
Female	0.673	0.696	−0.023^***
Age	65.634	67.017	−1.383^***
End-stage renal disease	0.003	0.003	0
Chronic obstructive pulmonary disease	0.092	0.09	0.002
Congestive heart failure	0.066	0.091	−0.024^***
Cancer	0.029	0.032	−0.003
Hypertension	0.535	0.595	−0.061^***
Chronic kidney disease	0.123	0.197	−0.074^***
Consecutive months enrolled	14.047	15.697	−1.650^***
Acute inpatient days	0.141	0.116	0.025^**
Observation days	0.026	0.026	0.001

“No CWA” and “CWA” refers to whether a patient had a CWA within the past 12 months or not.

P value for t-test of means ^*** P < 0.01, ^** P < 0.05.

CWA = Comprehensive Wellness Assessment; SNP = Special Needs Plan.

Statistical analysis

The study team estimated multivariate regressions of the following form to test the key hypothesis: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document} \begin{align*}{ \rm ODM - measure}_{ \rm imt} =& \beta_0 + \beta_1 \ { \rm SNP}_{ \rm imt} + \beta_2 \ { \rm CWA}_{ \rm imt} \\ &+ \beta_3 \ ( { \rm SNP}_{ \rm imt} \times \,{ \rm CWA}_{ \rm imt} ) + { \rm X}_{ \rm imt} \lambda \\ &+ { \rm M}_{ \rm m} + { \rm T}_{ \rm t} + { \rm e}_{ \rm ist}\end{align*} \end{document}

where SNP_imt is a binary indicator for whether the member is a SNP enrollee and CWA_imt is a binary indicator for whether the member had a CWA is the past 12 months. Thus, β₁ captures average differences in adherence between SNP and non-SNP enrollees in absence of CWA, β₂ captures the average association of CWA with adherence for all members, and β₃—the coefficient of the interaction between SNP and CWA—captures the information about whether SNP enrollees with a CWA show improved adherence compared to their counterparts. X_imt is a vector of the other member characteristics described earlier. M_m is a vector of “month fixed effects” to capture any possible seasonal variation in adherence, and T_t is a vector of year fixed effects to capture general changes in adherence for all members over time; e_ist is the error term, clustered to account for repeated observations from each member in the data set.

Multivariate linear regression models were estimated for the outcome of ODM monthly adherence, and multivariate Poisson models for the binary outcome of ODM-adherent. The Poisson model was preferred over the more conventional logit model because the study team was interested in how CWA influenced relative risk of adherence for SNP enrollees—and it is well known in the scientific literature that odds ratios from logit models provide a poor approximation of relative risk ratios except for situations in which the outcome of interest is quite rare,¹⁴ and that misinterpretation of odds ratios as risk ratios have sometimes led to substantial distortions of scientific findings.¹⁵ When the outcome of interest is relatively common, the Poisson model provides a more accurate estimate of the relative risk ratio.¹⁴

The study team initially estimated both sets of models without the controls for X_imt, M_m, and T_t, and thereupon estimated models after adding in X_imt, and T_t, and finally also adding in M_m. This allowed the team to see whether main results are sensitive to the controls included. Results for the linear and Poisson models are presented in Tables 2 and 3, respectively. Because extant literature has found evidence of racial differences in adherence, the analyses also were conducted stratified by race. These results are in Table 4.

Table 2.

Models for Monthly ODM Adherence

Variables	Model 1 % of days in month took oral medication (t stat in parenthesis)	Model 2 % of days in month took oral medication (t stat in parenthesis)	Model 3 % of days in month took oral medication (t stat in parenthesis)
Special Needs Plan (SNP)	−0.0197^***	0.00383	0.00251
	(−2.627)	(0.501)	(0.330)
	0.00864	0.616	0.741
Comprehensive Wellness Assessment (CWA) in past 12 months	−0.00907	−0.0173^*	−0.00645
	(−0.920)	(−1.745)	(−0.647)
	0.358	0.0811	0.518
Interaction between SNP and CWA	0.0435^***	0.0395^***	0.0379^***
	(3.338)	(3.093)	(2.974)
	0.000849	0.00199	0.00295
Minority		−0.0584^***	−0.0585^***
		(−9.745)	(−9.807)
		0	0
Female		−0.00949	−0.00934
		(−1.622)	(−1.604)
		0.105	0.109
Age		0.00238^***	0.00237^***
		(6.770)	(6.787)
		0	0
End-stage renal disease		−0.116^*	−0.112^*
		(−1.950)	(−1.898)
		0.0512	0.0578
Chronic obstructive pulmonary disease		−0.0262^**	−0.0254^**
		(−2.228)	(−2.177)
		0.0259	0.0295
Congestive heart failure		−0.0224^*	−0.0226^*
		(−1.900)	(−1.929)
		0.0574	0.0538
Cancer		−0.00434	−0.00422
		(−0.339)	(−0.333)
		0.734	0.739
Hypertension		0.00601	0.00774
		(1.072)	(1.385)
		0.284	0.166
Chronic kidney disease		−0.00323	−0.00103
		(−0.424)	(−0.135)
		0.671	0.892
Number of consecutive months enrolled pre diabetes		−0.000905^***	−0.000909^***
		(−8.649)	(−8.731)
Acute inpatient days		−0.00622^***	−0.00626^***
		(−5.168)	(−5.228)
		2.44e-07	1.76e-07
Observations days		−0.00554	−0.00546
		(−1.349)	(−1.341)
		0.177	0.180
Year Fixed Effects		Yes	Yes
Month Fixed Effects		No	Yes

***

P < 0.01, ^** P < 0.05, ^* P < 0.1.

Table 3.

Poisson Models for Monthly ODM Adherence

Variables	Model 1 IRR (t stat in parenthesis)	Model 2 IRR (t stat in parenthesis)	Model(1) IRR (t stat in parenthesis)
Special Needs Plan (SNP)	0.958^***	1.002	1.001
	(−3.212)	(0.163)	(0.0685)
Comprehensive Wellness Assessment (CWA) in past 12 months	0.995	0.978	0.988
	(−0.329)	(−1.363)	(−0.701)
Interaction between SNP and CWA	1.084^***	1.074^***	1.072^***
	(3.740)	(3.376)	(3.291)
Minority		0.888^***	0.888^***
		(−11.10)	(−11.14)
Female		0.987	0.987
		(−1.325)	(−1.314)
Age		1.004^***	1.004^***
		(6.992)	(7.000)
End-stage renal disease		0.777^*	0.780^*
		(−1.943)	(−1.921)
Chronic obstructive pulmonary disease		0.956^**	0.957^**
		(−2.168)	(−2.138)
Congestive heart failure		0.970	0.970
		(−1.492)	(−1.509)
Cancer		0.990	0.990
		(−0.430)	(−0.427)
Hypertension		1.021^**	1.022^**
		(2.131)	(2.299)
Chronic kidney disease		1.001	1.003
		(0.0693)	(0.240)
Number of consecutive months enrolled pre diabetes		0.998^***	0.998^***
		(−7.919)	(−7.958)
Acute inpatient days		0.990^***	0.990^***
		(−4.429)	(−4.465)
Observation days		0.991	0.991
		(−1.239)	(−1.234)
Year Fixed Effects		Yes	Yes
Month Fixed Effects		No	Yes

***

P < 0.01, ^** P < 0.05, ^* P < 0.1.

IRR = incidence rate ratio.

Table 4.

Linear and Poisson Models Adjusted for Race

	Linear		Poisson
Variables	Nonminority Enrollees % of days in month took oral medication	Minority Enrollees % of days in month took oral medication	Nonminority Enrollees % of days in month took oral medication	Minority Enrollees % of days in month took oral medication
Special Needs Plan (SNP)	0.0102	−0.00116	1.010	0.996
	(0.925)	(−0.111)	(0.553)	(−0.224)
	0.355	0.911	0.580	0.823
Comprehensive Wellness Assessment (CWA) in past 12 months	−0.0143	−0.0202	0.981	0.976
	(−1.136)	(−1.302)	(−0.981)	(−0.912)
	0.256	0.193	0.327	0.362
Interaction between SNP and CWA	0.0315^*	0.0474^**	1.059^**	1.087^***
	(1.727)	(2.577)	(1.990)	(2.608)
	0.0842	0.0100	0.0466	0.00912
Female	−0.0163^**	−0.000317	0.975^**	1.004
	(−2.205)	(−0.0338)	(−2.045)	(0.220)
	0.0275	0.973	0.0408	0.826
Age	0.00281^***	0.00194^***	1.005^***	1.004^***
	(6.293)	(3.581)	(6.349)	(3.748)
	3.49e-10	0.000347	2.17e-10	0.000178
End-stage renal disease	−0.0690	−0.138^*	0.729	0.788
	(−0.720)	(−1.912)	(−1.299)	(−1.620)
	0.471	0.0559	0.194	0.105
Chronic obstructive pulmonary disease	−0.00156	−0.0756^***	0.995	0.872^***
	(−0.112)	(−3.523)	(−0.231)	(−3.222)
	0.911	0.000432	0.817	0.00127
Congestive heart failure	−0.0291^*	−0.0145	0.961	0.984
	(−1.920)	(−0.790)	(−1.604)	(−0.490)
	0.0550	0.430	0.109	0.624
Cancer	−0.00588	−0.00278	0.988	0.993
	(−0.363)	(−0.137)	(−0.463)	(−0.168)
	0.717	0.891	0.644	0.867
Hypertension	0.00845	0.00287	1.028^**	1.010
	(1.157)	(0.330)	(2.312)	(0.598)
	0.247	0.741	0.0208	0.550
Chronic kidney disease	−0.0120	0.00715	0.989	1.017
	(−1.232)	(0.595)	(−0.744)	(0.775)
	0.218	0.552	0.457	0.438
Number of consecutive months enrolled pre diabetes	−0.00104^***	−0.000677^***	0.998^***	0.999^***
	(−8.047)	(−3.911)	(−7.597)	(−3.218)
	0	9.38e-05	0	0.00129
Acute inpatient days	−0.00693^***	−0.00507^***	0.989^***	0.993^*
	(−4.544)	(−2.598)	(−4.151)	(−1.943)
	5.71e-06	0.00942	3.31e-05	0.0520
Observation days	−0.00965^*	0.00105	0.986	1.000
	(−1.865)	(0.166)	(−1.557)	(−0.00800)
	0.0623	0.868	0.119	0.994
Year Fixed Effects	Yes	Yes	Yes	Yes
Month Fixed Effects	Yes	Yes	Yes	Yes

***

P < 0.01, ^** P < 0.05, ^* P < 0.1.

Finally, to assess whether any changes in adherence associated with CWA in the past 12 months was confined to the months immediately following CWA, or whether (and to what extent) adherence was higher in later months as well, the study team also ran regressions in which CWA_imt is replaced with a vector of indicators of CWA within the past 30 days, 30–90 days, 90–180 days, and more than 180 days but within 12 months. Those results are in Table 5.

Table 5.

Linear and Poisson Models with Categories for Time Since Comprehensive Wellness Assessment (CWA)

Variables	Linear % of days in month took oral medication (t stats in parenthesis) P	Poisson Met adherence threshold IRR (t stats in parenthesis) P
Special Needs Plan (SNP)	0.003	1.001
	(0.33)	(0.07)
	0.741	0.945
CWA within past 30 days	−0.017	0.951^**
	(−1.57)	(−2.41)
	0.116	0.016
CWA within past 30–60 days	0.010	1.014
	(0.96)	(0.76)
	0.339	0.448
CWA within past 60–180 days	−0.008	0.991
	(−0.60)	(−0.45)
	0.547	0.687
CWA within past 180–365 days	−0.015	0.986
	(−0.93)	(−0.58)
	0.35	0.563
Interactions: SNP x CWA within past 30 days	0.046^***	1.105
	(3.21)	(3.67)
	0.001	0.00
Interactions: SNP x CWA in past 30–60 days	0.017	1.040
	(1.20)	(1.62)
	0.231	0.105
Interactions: SNP x CWA in past 60–180 days	0.044^***	1.078
	(2.73)	(2.75)
	0.006	0.006
Interactions: SNP x CWA in past 180–365 days	0.046^**	1.076
	(2.44)	(2.41)
	0.015	0.016

Notes: Models control for all other variables, year fixed effects, month fixed effects.

***

P < 0.01, ^** P < 0.05, ^* P < 0.1.

IRR = incidence rate ratio.

Results

Table 1 shows that adherence is lower among SNP enrollees than non-SNP enrollees, both in terms of ODM monthly adherence (by 11%, P value for t test <0.01) and the proportion who are ODM-adherent (by 1.4 months, P < 0.01). Moreover, significant differences exist for SNP enrollees with a CWA in the past 12 months versus not, both for ODM monthly adherence (79.3 versus 75.9, P < 0.01) and the proportion who are ODM-adherent (73.0 versus 67.7, P < 0.01). Significant differences also are seen between SNP and non-SNP enrollees in other patient characteristics. The SNP group has substantially higher proportions of minorities and females, tends to be somewhat younger with fewer months of consecutive enrollment prior to the diabetes diagnosis, but has higher rates of all the chronic conditions except for cancer, as well as more acute inpatient days and observation days per month. Certain differences also are seen for SNP enrollees with a CWA versus those without. The group with CWA is slightly more likely to be female, older, have more months of consecutive enrollment prior to the diabetes diagnosis, and to have higher rates of CHF, HTN and CKD, but to have fewer acute inpatient days per month.

Linear regression results (Table 2) for the unadjusted model show that SNP enrollees with a CWA visit, on average, have ODM monthly adherence that is higher by more than 4 percentage points (P < 0.01). In the adjusted model with individual characteristics and year fixed effects, the corresponding number is almost 4 percentage points (P < 0.01). In the model that additionally includes month fixed effects, the corresponding number is close to 4 percentage points (P < 0.01), which translates into an increase of approximately 5% using the average ODM adherence of SNP enrollees (76.9%) as a baseline. Corresponding Poisson results (Table 3) indicate that SNP enrollees with a CWA visit are over 8% more likely to be ODM-adherent (incidence rate ratio [IRR]: 1.084, P < 0.01) in the unadjusted model, and over 7% more likely in the adjusted model without month-fixed effects (IRR: 1.074, P < 0.01) and in the adjusted model with month fixed effects (IRR: 1.072, P < 0.01). With respect to other member characteristics, in general, age is positively associated with adherence, while being minority is negatively associated with adherence. COPD, acute inpatient days, and higher consecutive months of enrollment before diagnosis also are negatively associated with adherence.

For robustness checks, the study team also estimated logistic models with the binary outcome and found results for SNP enrollees with CWA on ODM adherence to be virtually identical with respect to direction and statistical significance, although—as predicted in the scientific literature—the odds ratio provided an overestimation of the relative risk. These results are available on request.

Results for regressions that are stratified by minority and nonminority status (Table 4) indicate that the associations between CWA and adherence tend to be larger and more significant statistically for SNP enrollees who are minorities than nonminorities. Nonminority SNP enrollees with a CWA visit, on average, have ODM monthly adherence that is higher by approximately 3 percentage points and only weakly significant (P < 0.10), while the corresponding figure for minority SNP enrollees is more than 4.5 percentage points (P < 0.01). Nonminority SNP enrollees with a CWA visit are about 6% more likely to be ODM-adherent (IRR: 1.059, P < 0.05), whereas corresponding minority enrollees are more than 8.5% more likely to be ODM-adherent (IRR: 1.087, P < 0.01).

Finally, results for regressions in which a CWA in the past 12 months is replaced by a CWA within the past 30 days, 30–90 days, 90–180 days, and 180–365 days (Table 5) indicate that the highest adherence occurs within the 30 days after CWA, that there is less evidence of improved adherence in the 30–90 day period, but thereafter higher adherence resumes and stays fairly stable, although it does not reach the level of the first 30 days. Unadjusted ODM percentage and probability of adherence for SNP enrollees in each 30-day period after CWA for the 12-month period (presented in online Appendix A, available at www.liebertpub.com/pop) show a similar pattern, with a fall off in adherence occurring toward the very end of the 12-month period.

Discussion

The aim of this study was to evaluate the effectiveness of the CWA on an important area for the organization—medication adherence among dual eligible enrollees. Overall adherence to medications is an important measure for MA plans and their overall star rating results because these measures are categorized as outcome measures and are triple weighted.

The results indicate that a CWA is associated with significant improvements in ODM adherence among SNP enrollees, with the improvements being more marked for minority SNP enrollees than nonminority SNP enrollees. The study also finds that there is little difference between SNP and non-SNP enrollees in ODM adherence at baseline (ie, no CWA) once minority status is adjusted for, although being a minority per se compared to a nonminority is associated with significantly worse adherence. These results are in accordance with some previous findings in the literature.^16,17 In addition to analyzing the ODM adherence rate in accordance with the CMS star program technical specifications, an additional question of interest relates to the sustainability of adherence over time. The results of this analysis demonstrate improved adherence rates in the month after the CWA, then a brief decline followed by sustained improvement in adherence rates for much of the remainder of the 12 months, although not to the level of the first month after the CWA was completed. This finding tentatively suggests that the initial improvement in ODM adherence after CWA may lead to some habit forming such that high adherence persists for several months after the CWA. This suggests that there may be added benefits to assessing patients' habits related to taking medication, which may give practitioners insight into the likelihood of adherence to long-term medications. Providing habit-forming advice to patients may be a way to reinforce potential benefits associated with CWA in terms of improving medication adherence and, eventually, health outcomes.^18,19

The more significant long-term question, and area for future research, relates to improved glycemic control as measured by improvement in glycated hemoglobin (HbA1c). Several research studies have indicated an inverse relationship between ODM adherence and HbA1c levels. The mean value of HbA1c in these studies was 8.0 and the research demonstrated that each 10% increase in ODM adherence was associated with a range of 0.1%–0.16% decrease in HbA1c levels.^20

–26 Additional research analyzing nonadherence demonstrated that a corresponding 10% increase in nonadherence was associated with a 14% increase in HbA1c levels, also with a mean value of 8.0.²⁶ Further research in this area is needed to assess long-term declines in HbA1c and reductions in other complications from diabetes related to changes in adherence levels.

Implementing a patient-centered model of care requires a clinical transformation that does not rely on a “one size fits all” model. In addition to the general assessment by the NP, goals of the program include discussions about the importance of completing preventive screenings, the identification of psychosocial and/or community resource needs, development of a personal health goal, and medication review and education on the importance of taking prescribed medications as directed by the enrollee's PCP.

The CWA has been targeted primarily toward SNP enrollees, because dual eligible members typically have worse results for CMS star measures than non-dual members.²⁷ Because the CWA is designed to mitigate several of the barriers to adherence that have been identified in prior literature (eg, an ineffective communication between the provider and patient on the benefits as well as side effects of medications, lack of consideration for the patient's lifestyle, complex medication regimens that are not well explained or reviewed, poor communication and relationships between patient and caregiver¹⁰), it is anticipated that medication adherence is among the outcomes that can be improved by completing a CWA. There are 9 triple-weighted measures in the current star rating program—6 of which are directly and indirectly related to medication therapy and adherence. This study is the first of a series of evaluations of whether the CWA is associated with improvements in the triple-weighted measure outcomes.

Limitations

This study has limitations. The study used the organization's claims data, which relies on provider completeness of submitted claims. The sample was confined to enrollees who have been with the plan for at least 5 years post diabetes diagnosis, and may not be generalized to members who stay for shorter periods of time. It is likely that there is some element of self-selection involved in which SNP enrollees choose to actually go to the CWA—and such enrollees may be engaged in healthier behaviors and also may participate in other opportunities and benefits provided by the patient-centered model of care. Therefore, caution must be exercised against causally attributing better adherence solely to the CWA. Furthermore, these results pertain to 1 MA plan in the southeast, and its results may not be generalized across all MA plans. Finally, this study only considers the association of CWA with 1 of the triple-weighted CMS star measures, and cannot inform on the overall effectiveness of CWA. However, the study team will build on this study in the next research project and explore associations between the CWA and other CMS star measures.

Conclusion

In summary, the completion of a CWA appears to be an effective method to improve medication adherence to ODM in the study sample. Further studies are planned to evaluate the impact on other triple-weighted star program outcome measures as well as cost-benefit analyses that compare the costs incurred to conduct the CWA versus the expected benefits from any predicted improvements in the plan's star rating. Thus, this study provides important information on increasing ODM adherence and a potential intervention that can improve MA plan star ratings, which future studies can build on.

Footnotes

Author Disclosure Statement

Drs. Guerard, Omachonu, and Sen, and Mr. Perez declared no conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors received the following financial support: This study was partially supported by Peoples Health, a Medicare Advantage plan in southeast Louisiana. The data used in this work are the property of the Peoples Health organization. The organization had no role in the design or conduct of the study or approval of the manuscript. The content is solely the responsibility of the authors and does not represent the official views of the organization.

References

Zolnierek

, Dimatteo

. Physician communication and patient adherence to treatment: a meta-analysis. Med Care. 2009; 8:826–834.

McDonnell

, Jacobs

. Hospital admissions resulting from preventable adverse drug reactions. Ann Pharmacother. 2002; 36:1331–1336.

Senst

, Achusim

, Genest

, et al. Practical approach to determining costs and frequency of adverse drug events in a health care network. Am J Health Syst Pharm. 2001; 58:1126–1132.

Shalansky

, Levy

. Effect of number of medications on cardiovascular therapy adherence. Ann Pharmacother. 2002; 36:1532–1539.

Horne

. Adherence to medication: a review of existing research. In: Midence

, Myers

, eds. Adherence to Treatment in Medical Conditions. London: Harwood Academic; 1998:285–310.

DiMatteo

. Variations in patients' adherence to medical recommendations: a quantitative review of 50 years of research. Med Care. 2004; 42:200–209.

Ley

. Satisfaction, compliance and communication. Br J Clin Psychol. 1982; 21(pt 4):241–254.

De Geest

, Sabaté

. Adherence to long-term therapies: evidence for action. Eur J Cardiovasc Nurs. 2003; 2:323–323.

Blaschke

, Osterberg

, Vrijens

, Urquhart

. Adherence to medications: insights arising from studies on the unreliable link between prescribed and actual drug dosing histories. Annu Rev Pharmacol Toxicol. 2012; 52:275–301.

10.

Osterberg

, Blaschke

. Adherence to medication. N Engl J Med. 2005; 353:487–497.

11.

Jette

, Cummings

, Brock

, Phelps

, Naessens

. The structure and reliability of health belief indices. Health Serv Res. 1981; 16:81–98.

12.

Kelly

, Mamon

, Scott

. Utility of the health belief model in examining medication compliance among psychiatric outpatients. Soc Sci Med. 1987; 25:1205–1211.

13.

Conn

, Taylor

, Kelley

. Medication regimen complexity and adherence among older adults. Image J Nurs Sch. 1991; 23:231–236.

14.

McNutt

L-A

, Wu

, Xue

, Hafner

. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol. 2003; 157:940–943.

15.

Schwartz

, Woloshin

, Welch

. Misunderstandings about the effects of race and sex on physicians' referrals for cardiac catheterization. N Engl J Med. 1999; 341:279–287.

16.

Davis

, Schoenbaum

, Audet

. A 2020 vision of patient-centered primary care. J Gen Intern Med. 2005; 20:953–957.

17.

Egger

, Day

, Scammon

, Li

, Wilson

, Magill

. Correlation of the care by design primary care practice redesign model and the principles of the patient-centered medical home. J Am Board Fam Med. 2012; 25:216–223.

18.

Gardner

. A review and analysis of the use of ‘habit’in understanding, predicting and influencing health-related behaviour. Health Psychol Rev. 2014; 21:1–19.

19.

Alison Phillips

, Leventhal

. Assessing theoretical predictors of long-term medication adherence: patients' treatment-related beliefs, experiential feedback and habit development. Psychol Health. 2013; 28:1135–1151.

20.

Rozenfeld

, Hunt

, Plauschinat

, Wong

. Oral antidiabetic medication adherence and glycemic control in managed care. Am J Manag Care. 2008; 14:71–75.

21.

Lawrence

, Ragucci

, Long

, Parris

, Helfer

. Relationship of oral antihyperglycemic (sulfonylurea or metformin) medication adherence and hemoglobin A1c goal attainment for HMO patients enrolled in a diabetes disease management program. J Manag Care Pharm. 2006; 12:466–471.

22.

Schectman

, Nadkarni

, Voss

. The association between diabetes metabolic control and drug adherence in an indigent population. Diabetes Care. 2002; 25:1015–1021.

23.

Krapek

, King

, Warren

, et al. Medication adherence and associated hemoglobin A1c in type 2 diabetes. Ann Pharmacother. 2004; 38:1357–1362.

24.

Venturini

, Nichol

, Sung

, Bailey

, Cody

, McCombs

. Compliance with sulfonylureas in a health maintenance organization: a pharmacy record–based study. Ann Pharmacother. 1999; 33:281–288.

25.

Boccuzzi

, Wogen

, Fox

, Sung

, Shah

, Kim

. Utilization of oral hypoglycemic agents in a drug-insured US population. Diabetes Care. 2001; 24:1411–1415.

26.

Pladevall

, Williams

, Potts

, Divine

, Xi

, Lafata

. Clinical outcomes and adherence to medications measured by claims data in patients with diabetes. Diabetes Care. 2004; 27:2800–2805.

27.

Inovalon i. An Investigation of Medicare Advantage Dual Eligible Members Level of Performance on CMS 5- Star Quality Ratings. 2015. http://resources.inovalon.com/h/i/60106948-an-investigation-of-medicare-advantage-dual-eligible-member-level-performance-on-cms-five-star-quality-measures. Accessed May 8, 2015 .

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.09 MB

0.00 MB