Optimal Diabetes Care Outcomes Following Face-to-Face Medication Therapy Management Services

Abstract

Pharmacists play an integral role in influencing resolution of drug-related problems. This study examines the relationship between a pharmacist-led and delivered medication therapy management (MTM) program and achievement of Optimal Diabetes Care benchmarks. Data within Fairview Pharmacy Services were used to identify a group of patients with diabetes who received MTM services during a 2007 demonstration project (n=121) and a control group who were invited to receive MTM services but opted out (n=103). Rates of achieving optimal diabetes clinical management for both groups were compared using the D5 diabetes measure for years 2006, 2007, and 2008. The D5 components are: glycosolated hemoglobin (HbA1c<7%); low-density lipoprotein (<100 mg/dl); blood pressure (<130/80 mmHg); tobacco free; and daily aspirin use. Multivariate difference-in-differences (DID) estimation was used to determine the impact of 1 year of MTM services on each care component. Patients who opted in for MTM had higher Charlson scores, more complex medication regimens, and a higher percentage of diabetes with complications (P<0.05). In 2007, the percentage of diabetes patients optimally managed was significantly higher for MTM patients compared to 2006 values (21.49% vs. 45.45%, P<0.01). Nonlinear DID models showed that MTM patients were more likely to meet the HbA1c criterion in 2007 (odds ratio: 2.48, 95% confidence interval [CI]: 1.04–5.85, P=0.038). Linear DID models for HbA1c showed a mean reduction of 0.54% (95% CI: 0.091%–0.98%, P=0.018) for MTM patients. An MTM program contributed to improved optimal diabetes management in a population of patients with complex diabetes clinical profiles. (Population Health Management 2013;16:28–34)

Background

A s a professional practice, pharmacy has witnessed enormous change over the last 40 years, becoming increasingly involved in patient care, and evolving into a more patient-oriented practice. This evolution has been accompanied by greater recognition of clinical services as an integral part of professional practice. Striking statistics showing the relationship between mortality and morbidity and improper drug therapy¹ called for a new paradigm of pharmacy practice that placed drug therapy optimization at its core. That was the motive behind coining the term “pharmaceutical care.”

Pharmaceutical care has been defined as “the responsible provision of drug therapy for the purpose of achieving definite outcomes that improve a patient's quality of life.”² These outcomes can include curing disease, eliminating or reducing a patient's symptoms, slowing disease progression, or disease prevention.² The proponents of pharmaceutical care stressed the need for this practice in the community setting where most patients usually acquire their medications. The philosophy was embraced by pharmacists and pharmacy organizations and culminated in the official recognition of pharmaceutical care practices in the Prescription Drug, Improvement, and Modernization Act (MMA) approved by Congress in 2003. The MMA added a prescription drug benefit via Medicare Part D that is administered by insurers.³

Small projects and pilot trials have generated evidence of economic, clinical, and humanistic (quality of life) benefits from pharmacist's provision of patient-centered medication management to populations with chronic diseases. Examples include programs in Iowa,⁴ Minnesota,⁵ and North Carolina's “Ashville project.”⁶ With the passage of Medicare Part D, cognitive services offered by a pharmacist with the goal of optimizing medication use became officially recognized as medication therapy management (MTM) services.

With the passage of the MMA legislation, insurers are required to offer MTM as a quality improvement program to a defined subset of beneficiaries with the goal of optimizing therapeutic outcomes by improving medication use and reducing adverse drug events.⁷ This subset of patients usually is expected to be more complex in terms of disease state, co-morbidities, or medication regimens. Evidence has shown that pharmacists' involvement in managing therapeutic regimens for populations of patients with complex chronic diseases (eg, diabetes,⁸ hypertension,⁸ heart failure,^9,10 hypercholesterolemia⁵) results in better clinical outcomes. However, less attention has been given to the pharmacist's potential to align medication therapy goals to overall patient management and the role this may play in achieving quality measurement benchmarks for population health management.

The current emphasis on MTM services comes at a time when quality improvement and pay-for-performance initiatives also have increased in prominence. In Minnesota, an independent, nonprofit community organization, Minnesota Community Measurement (MNCM),¹¹ has been publicly reporting medical group performance measures since 2004. Included in these measures is a comprehensive, all-or-none, 5-component optimal diabetes care measure (D5). The 5 components are: glycated hemoglobin (HbA1c;<7%); low-density lipoprotein cholesterol (LDL;<100 mg/dl); blood pressure (<130/80 mmHg); tobacco free; daily aspirin use as appropriate. The purpose of this study is to examine the relationship between MTM and achievement of optimal diabetes care components.

Methods

Setting

The MTM program evaluated in the present study is a service of Fairview Pharmacy Services, a subsidiary of Fairview Health Services, a Minnesota nonprofit corporation and one of the largest health care provider organizations in the state. Fairview Health Services consists of a “network of 7 hospitals, 48 primary care clinics, 55 specialty clinics, and 28 retail pharmacies that serves Minneapolis-St. Paul, as well as communities throughout greater Minnesota and the Upper Midwest.”¹² A standardized patient care process is used by all MTM pharmacists within the system.¹² The MTM program enrolls patients in an “opt-in” procedure through direct referral, mailed letters, and telephonic outreach. Patients with diabetes who attended one of the Fairview Health System clinics staffed by MTM pharmacists were eligible for the study and invited to visit with an MTM pharmacist about their therapeutic regimens. Patients who scheduled and completed an MTM visit were considered to have opted into the program.

MTM is provided primarily to patients through face-to-face consultations. A validated standardized process designed to identify and resolve drug therapy problems and promote optimal patient outcomes is followed during each patient's visit. MTM pharmacists' responsibilities include assessing the patient's medications, identification of patient's drug-related needs, resolution and prevention of drug-related problems, determining appropriate follow-up measures, and documentation of the intervention outcomes. Collaborative practice agreements that allow the MTM pharmacist to initiate, modify, or discontinue drug therapy and order laboratory tests related to diabetes, hypertension, and hyperlipidemia are in effect.¹² All MTM activities are documented using a commercially available software package.

Sample definition, data collection, and data analysis plan

Patients with diabetes (n=127) who participated in an MTM demonstration project and had MTM visits to any Fairview clinic offering MTM services between January 1, 2007, and December 31, 2007 were identified in the Fairview electronic medical record system. A random selection of 121 patients with diabetes who were eligible for the demonstration project but who did not actively participate in MTM services served as the control group. The final analysis included data on patients for whom all information on medications was available at baseline and all D5 quality measure components were available for 2006, 2007, and 2008 (121 cases in the MTM group and 103 cases in the non-MTM group). Additional baseline variables including age, sex, and percent of patients having diabetes complications and co-morbidities to calculate a Charlson comorbidity index score¹³ were extracted from the electronic medical records. Rates of achieving the optimal management benchmark (satisfying the 5 criteria simultaneously or not) and the individual rates for the 5 components (satisfying an individual component benchmark or not) were available from files prepared for submission to MNCM for the year preceding the demonstration project (2006), the year of MTM services (2007), and the year following the end of the project (2008).

Additional control variables were constructed from electronic medical records to reflect varying levels of medication regimen complexity. These included a binary measure of insulin status (whether the patient had insulin in addition to other oral hypoglycemic), a 3-level categorical measure defining hypertensive medication complexity (whether the patient had angiotensin-converting enzyme inhibitors/angiotensin receptor blockers in addition to other hypertensive medications, had 1 hypertensive medication only, or did not have any hypertensive medication), and a 3-level categorical measure defining hypercholesterol medication complexity (whether the patient had statins in addition to other hypercholesterolemia medications, had 1 hypercholesterolemia medication only, or did not have any medication from that therapeutic class). A 2-level categorical variable, intensity of MTM visits, also was defined to reflect varying MTM services exposure levels among the treatment group patients (Intensity 1=1–4 visits, Intensity 2=5 or more visits).

For the baseline variables, pairwise comparisons were carried out between the 2 groups using t tests for continuous variables and the chi-square test for categorical variables. The change in rate of individual components and the composite optimal management rate were tracked and compared for the MTM group separately over time using the McNemar test for correlated proportions and in comparison to the control group using the chi-square test for differences in proportions.

To assess the effects of MTM on the 5 components of the D5 measure, difference in differences (DID) estimation in both linear and nonlinear regression frameworks was used while controlling for baseline clinical characteristics of the patients' groups.^14
–16 The Karaca-Mandic notation was followed to specify the DID model in both linear and nonlinear frameworks.¹⁵

For the linear framework, the DID model utilizes a standard linear regression equation form. “Y” is a continuous outcome variable (either HbA1c or LDL levels) with a set of time-independent covariates (X) that include age, Charlson index score, a binary measure of diabetes status (diabetes with complications or not), diabetes medication regimen complexity, cholesterol medication regimen complexity, and hypertensive medication regimen complexity. If the observation was from the MTM group, a “Trt” variable was coded as 1; if the observation was from the non-MTM group, the “Trt” variable was coded as zero. To denote time of observation, the variable “After” that is equal to 1 if the observation is from the posttreatment period (2007) and zero if from the pretreatment period (2006) was constructed. In addition, the model included an interaction effect of “Trt” and “After” variables that represents the DID measure of the treatment effect-primary coefficient of interest.

The nonlinear logistic DID framework was used to assess the impact of MTM on each of the 5 components separately where the dependent variable “Y” was coded 1 if the patient met the optimal criteria for the specific component and 0 if not. The nonlinear DID model was set as follows¹⁵: \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 Pr}(y = 1) = F(\beta_0 + \beta_1 X_i + \beta_2 Trt_i + \beta_3 After_i + \beta_4(Trt_i \times After_i))\end{align*} \end{document}

where Pr(y=1) represents the probability of meeting the diabetes care component benchmark. This probability is represented as a function (F) of time-independent explanatory variables (X), the “Trt” variable, the “After” variable, and their interaction as explained previously. The cumulative logistic function (F) was used to model the nonlinear relationship between the binary dependent variables and the explanatory covariates.

The DID interaction coefficient is still a measure of the treatment effect but is interpreted as a likelihood of meeting the diabetes care component benchmark criteria in the MTM group after being exposed to MTM in 2007 as compared to the pre-MTM period 2006 and as compared to the Non-MTM group in 2007 (ratio of an odds ratio). Generalized estimating equations were used to account for repeated measurements of the same patient over the 2 years. Statistical analysis was performed using SAS 9.2 (SAS Institute Inc., Cary, NC).

Lastly, the authors examined the effect of intensity level of MTM visits on meeting optimal diabetes care goals by running the DID model comparing each intensity level outcomes to the non-MTM group separately.

Results

Comparison of baseline data for participants and nonparticipants

The percentage of patients with diabetes complications in 2006 was significantly higher in the MTM group (P<0.001) (Table 1). At baseline, the proportion of patients using insulin (an indicator of diabetes medication complexity) as a part of their regimen or using additional drugs in addition to statins for their hypercholesterolemia (an indicator of hypercholesterolemia medication complexity) was significantly higher in the MTM group as well (P<0.05). Finally, patients in the MTM group had significantly higher Charlson scores reflecting higher coexisting comorbid conditions in this group of patients.

Table 1.

Baseline Characteristics

Characteristic	MTM group	Non-MTM group	P value
Number of patients	121	103	NA
Age	58.44	58.19	NS
% women	52.07%	51.46%	NS
% with diabetic complications	95.04%	15.53%	<0.001
% with insulin therapy	53.33%	34.95%	0.005
Charlson index score^*	3.7	2.72	<0.001
% Medicare	12.73%	20.59%
% Medicaid	5.45%	1.96%
% on statins +others	36.36%	23.3%	0.03
% on ACE/ARB+others	56.20%	58.25%	NS

^* Charlson Index Score: A measure of the level of comorbidity in which each medical condition (17 different conditions) is assigned a score depending on the risk of mortality from that specific condition. For this study, the minimum value assigned to any patient was 1 because all patients had to have diabetes.

ACE, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; MTM, medication therapy management; NA, not applicable; NS, not significant.

Composite score changes over time

Comparing the change in the optimal composite score shows that there was a statistically significant improvement in achieving the optimal level for the MTM group between 2006 (pre MTM) and 2007 (post MTM) (21.49% to 45.45%, P<0.001). Upon discontinuation of MTM at the end of 2007, there was a statistically significant drop in the optimal rate for the MTM group observed at the end of 2008 (45.45% to 25.62%, P=0.0002). For the non-MTM group, a significant improvement also was noticed between 2006 and 2007 (24.27% to 39.81%, P=0.0002) but the decline from 2007 to 2008 was not statistically significant (39.81% to 30.10%, P=0.077).

The change in achievement of each individual component of D5 was examined for the MTM group (Table 2). By the end of 2007, achievement of the optimal levels of LDL, tobacco abstinence, and HbA1c% were significantly improved among MTM participants relative to the pre-MTM period (2006). Upon discontinuation of MTM visits, the optimal rates declined significantly for the HbA1c level only. Other components did not change significantly.

Table 2.

Proportions of Patients Meeting Diabetes Care Components Defined Goals (MTM group, n=121)

Component	2006	2007	P values
HbA1c	43.80%	73.55%	<0.001
LDL	63.64%	83.47%	0.0002
BP	66.12%	71.07%	0.33
Tobacco	85.95%	91.74%	0.01
Aspirin	97.52%	100%	NS

	2007	2008	P values
HbA1c	73.55%	42.15%	<0.001
LDL	83.47%	79.34%	0.35
BP	71.07%	76.03%	0.37
Tobacco	91.74%	91.74%	NS
Aspirin	100%	99.17%	NS

BP, blood pressure; HbA1c, glycated hemoglobin; LDL, low-density lipoprotein; MTM, medication therapy management.

Comparison of participants and nonparticipants

Results from the MTM patients were compared to the control group of non-MTM patients (Table 3). Univariate analysis showed that rates of achievement of optimal management were not significantly different between the 2 groups in 2006 even though the optimal management rate was higher in the non-MTM group. By the end of 2007, however, the MTM group achieved a higher, though not statistically significant, rate of optimal management compared to the non-MTM group.

Table 3.

Proportions of Patients Meeting Diabetes Care Components Defined Goals in Study Groups (MTM vs. non-MTM)

Component (%pass)	MTM group (n=121)	Non-MTM group (n=103)	P value
2006
Optimal Score 2006	21.49%	24.27%	0.62
HbA1c	43.80%	63.11%	0.003
LDL	63.64%	65.05%	0.82
BP	66.12%	61.17%	0.44
Tobacco	85.95%	79.61%	0.20
Aspirin	97.52%	93.29%	0.11
2007
Optimal Score 2007	45.45%	39.81%	0.39
HbA1c	73.55%	72.82%	0.9
LDL	83.47%	73.79%	0.07
BP	71.07%	72.82%	0.77
Tobacco	91.74%	81.55%	0.02
Aspirin	100%	98.06%	0.12
2008
Optimal Score 2008	25.62%	30.10%	0.45
HbA1c	42.15%	59.22%	0.01
LDL	79.34%	73.79%	0.32
BP	76.03%	69.9%	0.3
Tobacco	91.74%	82.52%	0.03
Aspirin	99.17%	99.03%	0.9

BP, blood pressure; HbA1c, glycated hemoglobin; LDL, low-density lipoprotein; MTM, medication therapy management.

After discontinuation of MTM services, an assessment of the optimal rate at the end of 2008 showed the non-MTM group had higher, though not statistically significant, rates of achieving optimal management compared to the MTM group.

Component scores change over time

Analysis of the 5 individual components of the D5 showed that the rate of meeting the HbA1c goal (<7%) was significantly higher in the non-MTM group compared to the MTM group prior to receiving MTM services in 2006. There were no significant differences between the 2 groups of patients for the 4 remaining components. One year after receiving MTM services, difference between the 2 groups in the HbA1c vanished. The MTM group also had a significantly higher rate of achieving tobacco abstinence and meeting the LDL component goal of <100 mg/dl. One year after stopping the service, the rate of achieving the HbA1c goal for MTM patients became significantly lower than that of the non-MTM group. The percent of tobacco-free patients was significantly higher in the MTM group (Table 3).

To control for potential confounders simultaneously, a multivariate DID analysis was carried out. Table 4 shows the MTM treatment effects from the analysis with the 5 optimal score components as dependent variables while controlling for possible confounders. Exposure to MTM services had the most significant impact on the levels of HbA1c; odds of achieving the HbA1c<7% goal for patients in the MTM group were 2.48 times higher in 2007 compared to 2006. Treating HbA1c as a continuous variable in the model showed that patients in the MTM group had a significant reduction of 0.54% in HbA1c levels in 2007. There also was a higher trend for MTM patients to achieve the therapeutic goal for the LDL component (<100 mg/dl) but the results were not statistically significant. All other components of the D5 measure (blood pressure, smoking, and aspirin intake) were not statistically significant between the MTM and non-MTM groups.

Table 4.

Difference-in-Differences Models (Whole Sample, n=224)

Outcome Variable	Difference in Difference Estimator (C.I)	P Value
Non-Linear Models
HbA1c (binary)	2.48 (1.04,5.85)	0.038^**
	^*GEE estimate:	0.01^**
	2.44 (1.22,4.86)
LDL (binary)	1.95 (0.81,4.84)	0.13
BP (binary)	0.73 (0.32,1.65)	0.45
Tobacco (binary)	1.62 (0.54,4.84)	0.38
Aspirin (binary)	2.32 (0.08,65)	0.62
Linear Models
HbA1c (continuous)	−0.54 (−0.98,−0.091)	0.018^**
	^*GEE estimate:	0.0009^**
	−0.54 (−0.22, −0.86)
LDL (continuous)	−11.12 (−45.36, 23.12)	0.52

Results shown represent coefficient estimates from models that did not account for repeated observations of the same individual (non-GEE models); GEE estimates are only shown for models in which the coefficient on the interaction term was statistically significant. ^**Significant at p<0.05.

BP, blood pressure; CI, confidence interval; GEE, generalized estimating equation; HbA1c, glycated hemoglobin; LDL, low-density lipoprotein.

Examining outcomes for patients who were exposed to 4 or fewer MTM visits per year (Intensity=1) within the nonlinear framework showed that MTM patients at this visit intensity had a higher likelihood of meeting the HBA1c and LDL goals, but the results were not statistically different from the non-MTM group (Table 5). The linear DID model treating HbA1c and LDL as continuous measures showed that patients exposed to 4 or fewer MTM visits annually had a mean reduction of 0.42% in their HbA1c levels as compared to the non-MTM group (P<0.05).

Table 5.

Difference-in-Differences Models (by Intensity Levels)

Outcome Variable	Difference in Difference Estimates using GEE estimation (CI)	P Value
Nonlinear Models
Intensity=1 (4 MTM visits or less annually) vs. non-MTM group (n=183)
HbA1c (binary)	1.79 (0.84,3.84)	0.12
LDL (binary)	1.29 (0.54,3.08)	0.56
Intensity=2 (5 or more MTM visits annually) vs. non-MTM group (n=144)
HbA1c (binary)	5.11 (1.78,14.68)	0.0025^*
LDL (binary)	7.4 (1.67,33.18)	0.0085^*
Linear Models
Intensity=1 (4 MTM visits or less annually) vs. non-MTM group (n=183)
HbA1c (continuous)	−0.42 (−0.07, −0.78)	0.01^*
LDL (continuous)	−2.41 (−37.79, 42.62)	0.90
Intensity=2 (5 or more MTM visits annually) vs. non-MTM group (n=144)
HbA1c (continuous)	−0.76 (−0.26, −1.27)	0.0026^*
LDL (continuous)	−28.97 (0.48, −58.43)	0.053

Significant at P<0.05.

CI, confidence interval; GEE, generalized estimating equation; HbA1c, glycated hemoglobin; LDL, low-density lipoprotein; MTM, medication therapy management.

Table 5 shows that patients who were exposed to 5 or more MTM visits annually (Intensity=2) were more likely to meet the HbA1c goal in 2007 than they were in 2006 and more likely than the non-MTM group to meet the HbA1c goal in 2007. Patients exposed to 5 or more MTM visits annually also were more likely to meet the LDL goals in 2007—either compared to their 2006 performance or compared to the non-MTM group in 2007. When comparing the outcomes on the continuous level, the MTM group with 5 or more visits annually had a mean reduction of 0.76% in their HbA1c levels between 2007 and 2006 as compared to the non-MTM group. In addition, the MTM group had a trend to have a mean reduction of 28.97 mg/dl in their LDL levels as compared to the non-MTM group.

Discussion

This study adds to the growing literature exploring the potential benefits of MTM services in achieving clinical goals of therapy for patient populations with chronic diseases. It is among the first research endeavors to explore the ability of MTM to meet composite measures of optimal care for diabetes. Most previously published work focused on improvement of individual clinical components of care such as HbA1c for patients with diabetes and blood pressure and LDL levels for patients with vascular diseases.^17,18 This study has the advantage of showing the pharmacist role in coordinating multiple dimensions of care for diabetes through MTM services.

In an attempt to overcome less rigorous inference associated with pre–post designs and analyses for patients who received MTM services only (treatment groups), a design that has been used extensively for evaluation in this setting, we used a comparator group. The addition of this comparison group highlights the need to examine factors associated with motivation to opt in to receive MTM services. This was a demonstration project that provided MTM services free of charge to participants; therefore, cost did not contribute to the opt-in decision. However, the findings related to differences between the 2 groups at baseline would suggest that patients who did opt in for MTM services had a higher perceived need for assistance with their medication regimens. The differences between the 2 groups identified at baseline (age, sex, Charlson index score, diabetes status, diabetes medication regimen complexity, cholesterol medication regimen complexity, and hypertensive medication regimen complexity) were controlled for in the multivariate DID analysis.

We found that MTM services provided to patients with diabetes had a positive impact on the key quality measures of diabetes control. MTM pharmacists succeeded in bringing the more complex patients into a level of blood glucose control that was similar to the less complex Non-MTM patients. Patients in the MTM group had higher rates of optimal diabetes management while they had face-to-face encounters with MTM pharmacists. However, when MTM services were discontinued, patients in this group returned to rates that were not significantly different from their baseline measures. Patients with chronic diseases like diabetes may achieve better outcomes from regular follow-up with an MTM pharmacist to sustain the beneficial effect.

Data from this study showed that MTM services were delivered to a distinct population of patients with diabetes. Patients who received MTM services were patients with a higher number of comorbid conditions and more complex medication regimens for diabetes. This finding suggests that a patient channeling effect is taking place in MTM provision. Patients with more challenging diabetes problems are more likely to be referred to and to participate in MTM programs. Differences in baseline characteristics of participants and nonparticipants need to be controlled for carefully in evaluations of MTM programs. Thus, multivariate statistical models that control for baseline differences between participants and nonparticipants and account for both baseline and posttreatment values of the clinical outcomes were used to evaluate the impact of being exposed to MTM. Our study design had the unique advantage of recording the patients' outcomes before exposure to MTM started, after exposure, and 1 year after the exposure had ended. This constitutes a huge improvement that enabled a more defensible robust estimation of the treatment effects. Among the individual components of the MNCM optimal diabetes measure (D5), DID multivariate models showed that MTM services received over a period of 1 year resulted in significant reduction of HbA1c levels compared to the year prior to receiving MTM (2006). Lowering HbA1c levels to meet a goal of <7% is a recommended target for all patients with diabetes. It is important to note, however, that HbA1c achievement seems to be very sensitive to receiving MTM service. Assessment of the proportion of patients who achieved optimal HbA1c levels 1 year after discontinuing MTM in 2008 showed a significant drop to nearly the baseline values of 2006. The pharmacist role in managing and optimizing medications in terms of appropriateness and dosage can lead to this sensitivity phenomenon. This finding highlights the need to receive MTM services on a regular basis in order to maintain beneficial clinical control of hyperglycemia.

In addition, analyses stratified on number of MTM visits showed that patients who had 5 or more visits annually had significant improvements in the 2 main care components: HbA1c and LDL in 2007, both on the linear and nonlinear scales. The magnitude of improvement in patients who had 5 or more visits was significantly higher than for patients who had 4 visits or fewer. This strengthens the importance of establishing an adequate number of MTM encounters for patients with complex chronic diseases (eg, diabetes) to meet the optimal clinical goals of therapy.

An interesting observation was that the rate of tobacco abstinence in the MTM group was higher than the non-MTM group 1 year after discontinuation of the service. This might allude to a sustained effect of the pharmacist education on that part of the patient's lifestyle management.¹⁹ Smoking is known to be a risk factor for many diabetes and vascular diseases complications.^20
–22 Thus, the ability of the pharmacist to modify that part positively underlines the importance of pharmacist involvement in lifestyle management recommendations for chronic disease patients.

One last point of note is that the rate of smoking abstinence within the MTM group persisted from 2007 to 2008 after discontinuing MTM services. This is a very powerful result that is in contrast to high rates of smoking recidivism within 1 year that have been reported in the literature.²³ Thus, pharmacists' efforts related to smoking cessation seem to result in longer term abstinence from tobacco. MTM programs may be a viable option for smoking cessation efforts led by pharmacists.

An unexplored but potentially important aspect of MTM services that is closely related to glycemic control is medication adherence. Given the increasing complexity of therapeutic regimens in terms of numbers of drugs and dosing schedules, patients may face difficulties in adhering to the therapies as prescribed. Pharmacist involvement in educating patients about the importance of adherence and following up with the patients in regular MTM consultations has the potential to better control HbA1c levels.²⁴ Previous studies have reported an association between improvement in medication adherence and enrollment in MTM programs.

Limitations

Given the nature of the observational data and lack of randomization in our design, there is a chance for residual confounding by unmeasured variables. This could have been manifested in the unmeasured patients' characteristics that possibly made them more likely to opt into MTM services and to be more responsive to the MTM pharmacists' recommendations.

Moreover, we cannot simply rule out the possible role that physician and nurse education could have played in helping the patient achieve the goals of therapy. Not only would physician and nurse education have this beneficial effect, but exposure to non-MTM pharmacists in normal community pharmacy visits could have had the same impact. Pharmacists are required by law to provide medication information and education to patients when dispensing medications on a regular basis. Thus, a possible spillover effect in the control group could not be ruled out completely.

We also assumed that the medications regimens were time invariant in the multivariate DID models. That assumption is based on the short period of observation and the unlikelihood of a major change in the classes of medication prescribed for persistently ill patients represented in our analysis sample.

Finally, the number of patients included in our study was relatively small. Budget and resource constraints were the main considerations in the a priori determination of sample size. Nonetheless, this relatively small sample size did show statistically significant improvement in the HbA1c levels in MTM patients. A larger sample size could have had the potential to elucidate additional significant relationships between exposure to MTM services and the remaining optimal diabetes care components.

Conclusion

In summary, this study suggests that a pharmacist-led and delivered MTM program has the potential to improve optimal diabetes management rates in a population of complex diabetes patients. Therefore, policies at the federal and local levels should be designed to increase patient access to MTM services.

Footnotes

Acknowledgments

The authors would like to thank Bryan Dowd, Ph.D., for providing valuable input on the statistical analyses and for reviewing earlier versions of this manuscript.

Author Disclosure Statement

Drs. Brummel and de Oliveira are employees of Fairview Pharmacy Services. Mr Soliman and Dr. Carlson disclosed no conflicts of interest in the research, authorship, and/or publication of this article. This work was supported by an unrestricted grant from Novartis.

References

Ernst

, Grizzle

. Drug-related morbidity and mortality: Updating the cost-of-illness model. J Am Pharm Assoc (Wash), 2001; 41:192–199.

Hepler

, Strand

. Opportunities and responsibilities in pharmaceutical care. Am J Hosp Pharm, 1990; 47:533–543.

Medicare Prescription Drug Improvement, Modernization Act of 2003. http://www.gpo.gov/fdsys/pkg/PLAW-108publ173/html/PLAW-108publ173.htm. 2012 March 1.

Chrischilles

, Carter

, Lund

et al. Evaluation of the Iowa Medicaid pharmaceutical case management program. J Am Pharm Assoc (2003), 2004; 44:337–349.

Isetts

, Schondelmeyer

, Artz

et al. Clinical and economic outcomes of medication therapy management services: The Minnesota experience. J Am Pharm Assoc (2003), 2008; 48:203–211; 3 p following 211.

Bunting

, Cranor

. The Asheville project: Long-term clinical, humanistic, and economic outcomes of a community-based medication therapy management program for asthma. J Am Pharm Assoc (2003), 2006; 46:133–147.

Smith

, Clancy

. Medication therapy management programs: Forming a new cornerstone for quality and safety in Medicare. Am J Med Qual, 2006; 21:276–279.

Cranor

, Bunting

, Christensen

. The Asheville project: Long-term clinical and economic outcomes of a community pharmacy diabetes care program. J Am Pharm Assoc (Wash), 2003; 43:173–184.

Murray

, Ritchey

, Wu

, Tu

. Effect of a pharmacist on adverse drug events and medication errors in outpatients with cardiovascular disease. Arch Intern Med, 2009; 169:757–763.

10.

Murray

, Young

, Hoke

et al. Pharmacist intervention to improve medication adherence in heart failure: A randomized trial. Ann Intern Med, 2007; 146:714–725.

11.

MN Community Measurement. Measuring and reporting health care quality in Minnesota. http://www.mncm.org/site/. 2012 February 28.

12.

Ramalho de Oliveira

, Brummel

, Miller

. Medication therapy management: 10 years of experience in a large integrated health care system. J Manag Care Pharm, 2010; 16:185–195.

13.

Deyo

, Cherkin

, Ciol

. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol, 1992; 45:613–619.

14.

, Norton

. Interaction terms in logit and probit models. Econ Lett, 2003; 80:123–129.

15.

Karaca-Mandic

, Norton

, Dowd

. Interaction terms in nonlinear models. Health Serv Res, 2012; 47:255–274.

16.

Bertrand

, Duflo

, Mullainathan

. How much should we trust differences-in-differences estimates? Quart J Econ, 2004; 119:249–275.

17.

Toscani

, O'Connor

, Nash

. A decentralized, patient-centered approach to diabetes disease management in the primary care setting. Dis Manag, 2001; 4:163–171.

18.

Johnson

, Chen

, Cheng

et al. The impact of clinical pharmacy services integrated into medical homes on diabetes-related clinical outcomes. Ann Pharmacother, 2010; 44:1877–1886.

19.

Branham

, Moose

, Ferreri

. Retrospective analysis of medication adherence and cost following medication therapy management. (Article 12) Innov Pharm, 2010; 1:1–8.

20.

Al-Delaimy

, Manson

, Solomon

et al. Smoking and risk of coronary heart disease among women with type 2 diabetes mellitus. Arch Intern Med, 2002; 162:273–279.

21.

Haire-Joshu

, Glasgow

, Tibbs

. Smoking and diabetes. Diabetes Care, 1999; 22:1887–1898.

22.

Al-Delaimy

, Willett

, Manson

, Speizer

, Hu

. Smoking and mortality among women with type 2 diabetes: The nurses' health study cohort. Diabetes Care, 2001; 24:2043–2048.

23.

Serraino

, Franceschi

, Tassan

, Baron

, Talamini

. Cigarette smoking recidivism after participation in the "five-day plan to stop smoking" in northeastern Italy. Prev Med, 1993; 22:272–283.

24.

Rhee

, Slocum

, Ziemer

et al. Patient adherence improves glycemic control. Diabetes Educ, 2005; 31:240–250.