Impact of pharmacist participation in the patient care team on value-based health measures

Abstract

Purpose

To evaluate whether pharmacist engagement on the interdisciplinary team leads to improved performance on diabetes-related quality measures.

Methods

This was a retrospective observational study of patients seen in primary care and specialty clinics from October 2014 to October 2020. Patients were included if they had a visit with a physician, nurse practitioner, physician’s assistant, or clinical pharmacist practitioner (CPP) within the study period and had a diagnosis of diabetes. The intervention group included patients with at least one visit with a CPP, while the control group consisted of patients who were exclusively managed by non-CPP providers. The primary outcome of this study was the median change in glycosylated hemoglobin (HbA_1c) from baseline to follow-up at 3, 6, and 12 months. The secondary outcome was the probability of achieving the HbA_1c targets of <7% and <8% at 3, 6, and 12 months.

Results

Patients referred to a CPP had higher HbA_1c levels at baseline and were more likely to have concomitant hypertension (P < 0.01). Patients seen by a CPP had 0.31%, 0.41%, and 0.44% greater reductions in HbA_1c compared to patients in the control group at 3, 6, and 12 months, respectively (P < 0.01). Patients managed by a CPP were also more likely to achieve the identified HbA_1c targets of <7% and <8%.

Conclusion

Patients referred to a CPP were more complex, but had greater reductions in HbA_1c and were more likely to achieve HbA_1c goals included in the organization’s quality measures. This study demonstrates the value of pharmacists in improving patient care and their role in supporting an organization’s achievement of value-based quality measures.

KEY POINTS.

• The shift to value-based practice models provides an opportunity for pharmacists to further demonstrate their value to the healthcare team by improving performance on value-based quality measures.

• This study adds to the growing body of literature demonstrating the value of pharmacist participation in the outpatient interdisciplinary care team.

• This retrospective cohort study found that, regardless of patient complexity, pharmacist involvement in the interdisciplinary care team led to statistically significant reductions in glycosylated hemoglobin and increased probability of achieving value-based health measures associated with diabetes management.

Passage of the Patient Protection and Affordable Care Act in 2010 and the Medicare Access and CHIP Reauthorization Act in 2015 has led to a renewed focus by many payors on quality of care and its utility in determining reimbursement.^1,2 Since these pieces of legislation were enacted, various programs have been implemented by the Centers for Medicare and Medicaid Services aimed at improving outcomes, value, and quality of healthcare. Many programs have begun to shift reimbursement to a pay-for-performance model in which financial compensation is tied directly to meeting clinical quality measures outlined in value-based contracts. Other programs have developed accountable care organizations in which healthcare organizations share the financial risk, incentivizing interventions that decrease the total cost of care. These approaches represent a substantial shift in the economics of healthcare, which historically has been based on a fee-for-service model, and thus have shifted the focus toward achieving quality metrics and decreasing total cost of care.

Many of the quality metrics in value-based contracts are based on the Healthcare Effectiveness Data and Information Set (HEDIS).³ A great number of HEDIS measures are centered around disease states known to require medication management and optimization, including glycosylated hemoglobin (HbA_1c) levels in diabetes and blood pressure targets in various disease states. As medication experts, pharmacists are uniquely positioned to assist healthcare organizations in achieving these quality metrics. Despite the medication expertise of pharmacists, many organizations have not utilized them in the past owing to reimbursement challenges with the fee-for-service model. The ongoing shift to value-based contracts and accountable care organizations presents a new financial model that may support the inclusion of pharmacists as part of the interdisciplinary care team.

Pharmacists have been demonstrated to lead to improvement in several quality metrics, including 30-day readmissions, adherence to medication regimens, and total medication cost, through utilization of more cost-effective therapies.^4-6 While substantial evidence has demonstrated the benefit from pharmacists in the management of diabetes, there is a dearth of literature specifically characterizing the impact of pharmacist participation on achievement of quality measures included in payor contracts beyond traditional comprehensive medication management. Therefore, this study sought to demonstrate the impact of pharmacist involvement in the collaborative patient care team on industry-defined value-based quality metrics.

Our health system comprises 11 hospitals ranging from a 25-bed critical access hospital to an academic medical center with more than 900 beds, with multiple physician practice groups across the state. Pharmacy Services is a system division encompassing the pharmacy departments across the 11 entities, serving all affiliated hospitals and clinics. The University of North Carolina (UNC) Medical Center is home to a level 1 trauma center, a National Cancer Institute–designated comprehensive cancer center, a pediatric center of excellence, a regional hemophilia center, and a regional burn center. The UNC Medical Center department of pharmacy employs over 400 personnel, including over 100 pharmacists and clinical pharmacist practitioners (CPPs). CPPs must be approved by the state board of pharmacy and complete 35 hours of continuing education annually to maintain their licensure. CPPs are found in more than 25 clinics within the health system and are privileged to independently order laboratory tests and adjust medication therapy under a collaborative practice agreement with supervising physicians.

Methods

Practice setting

This study evaluated patients treated at 4 clinics affiliated with a large academic medical center: family medicine, internal medicine, heart and vascular, and endocrinology. Each practice site has a mixture of interdisciplinary providers, including attending and resident physicians, nurse practitioners (NPs), physician’s assistants (PAs), and CPPs.

Intervention

The collaborative care team at each of the 4 practice sites includes at least one CPP providing management of chronic diseases, including diabetes. Each CPP maintains their own panel of patients who have previously established care with a provider in the clinic. Patients are referred to a CPP on the basis of a number of potential criteria, including disease severity, comorbid conditions, polypharmacy, or lack of non-CPP availability for clinic visits. CPP visits focus on optimization of medication and lifestyle factors contributing to poor control of specific chronic diseases. There was no shared treatment algorithm among the clinics, but each CPP utilized current American Diabetes Association targets when setting goals for their respective patients.

Study design

This was a retrospective observational study approved by the UNC institutional review board. Patients were included if they were seen between October 1, 2014, and October 1, 2020; were 18 years of age or older; had at least one visit with a qualifying provider defined as a physician, NP, PA, or CPP; and had a diagnosis of diabetes at their first visit. The dates were selected to include the earliest implementation of the Epic electronic medical record system at our institution through the date of the data pull. All patients who had at least one visit with a CPP were included in the intervention group. Patients who had visits with only qualifying non-CPP providers during the study period were included in the control group. Patients were excluded if they had gestational or steroid-induced diabetes, end-stage renal disease, chronic kidney disease requiring hemodialysis, or HbA_1c of <4.5%; were pregnant; or were incarcerated at the time of the data pull.

Outcomes

The primary clinical outcome was median change in HbA_1c from baseline to follow-up at 3, 6, and 12 months. Baseline HbA_1c was defined as the most recently documented HbA_1c level within 6 weeks of the index encounter. Follow-up HbA_1c values were identified at 3, 6, and 12 months and included values documented within 6 weeks of the designated follow-up time points. Secondary outcomes were the percentages of patients achieving HbA_1c targets of <7% and <8%, as these targets are featured in HEDIS measures for diabetes control.

Data collection

Demographics, encounter dates and providers, and laboratory values of interest were queried from the organization’s data warehouse, the Carolina Data Warehouse for Health, and extracted from the health system’s electronic medical record. Relevant diagnoses for both inclusion and exclusion were identified by International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10) code.

Data analysis

Differences in baseline characteristics between the cohorts and all clinical outcomes were analyzed using either a 2-sided Wilcoxon rank-sum test for continuous variables or χ² analysis for categorical variables, with α set at 0.05, using Stata/SE 15.1 (StatCorp, College Station, TX). To address missing data with the minimum amount of bias, individuals who did not have an index HbA_1c level were dropped from the study cohort. Any individuals who did not have a 90-day, 180-day, and/or 360-day HbA_1c level had their most recent HbA_1c value carried forward (which would indicate no change in HbA_1c level over the time period). Inverse probability of treatment weight (IPTW) values were calculated from propensity scores that were generated using age, concurrent hypertension diagnosis, race, and biological sex. Evaluation of the primary outcome was performed in a pre-post fashion utilizing a difference-in-differences evaluation of the treatment effect, leveraging the IPTW values to help moderate selection bias. The pre-post method assumes that, in the absence of the intervention, both arms would experience the same change in the selected outcome over time with the difference in difference representing the difference in these groups across a period of time. This method strives to control for the effect of time and pretreatment differences between the treatment and control groups. To maintain the validity of the comparison, it was necessary to control for covariates that might impact a patient’s assignment to the intervention group.^7,8

Results

A total of 5,285 patients met the inclusion criteria and had a diagnosis of diabetes at baseline. Of these, 1,078 patients had a documented visit with a CPP and were included in the intervention group while 4,207 patients represented the control group, as shown in Figure 1. Baseline characteristics and demographics of the patient population are summarized in Table 1. The majority of patients were female (55.0%) and had a concomitant diagnosis of hypertension (69.5%), and the median age was 55 years. The intervention cohort had a median baseline HbA_1c level of 8.6% compared to a baseline level of 7.4% in the control group (P < 0.01). Other statistically significant differences were observed in baseline characteristics between the cohorts, with the patients assigned to the intervention cohort being more likely to be black, have a concomitant diagnosis of hypertension, and be of greater median age.

Figure 1.

Subject disposition flow chart. A1c indicates glycosylated hemoglobin; CKDS, chronic kidney disease; CPP, clinical pharmacy practitioner; DM, diabetes melitus; ESRD, end-stage renal disease.

Table 1.

Baseline Characteristics for Patients With Diabetes Mellitus

Characteristic	Intervention group (≥1 CPP visit; N = 1,078)	Control group (no CPP visit; N = 4,207)	P value
Age, median (IQR), years	59 (49-67)	54 (43-64)	<0.01
Baseline HbA1c, median (IQR), %	8.6 (7.4-10.0)	7.4 (6.5-8.6)	<0.01
Baseline SBP, median (IQR), mm Hg	122 (111-133)	127 (116-139)	<0.01
Baseline DBP, median (IQR), mm Hg	70 (62-78)	76 (68-83)	<0.01
Male sex, No. (%)	489 (45.4)	1,890 (44.9)	0.80
Race, No. (%)			<0.01
White	489 (45.4)	2,387 (56.7)
Black	440 (40.8)	1,206 (28.7)
Asian	17 (1.6)	108 (2.6)
Other/unknown	132 (12.2)	506 (12.0)
Hypertension diagnosis, No. (%)	885 (82.1)	2,789 (66.3)	<0.01

Abbreviations: CPP, clinical pharmacist practitioner; DBP, diastolic blood pressure; HbA_1c, glycosylated hemoglobin; IQR, interquartile range; SBP, systolic blood pressure.

As shown in Table 2, for the primary clinical outcome at 12 months, patients in the intervention group had a 0.44% greater decrease in HbA_1c when compared to patients managed by non-CPP providers (P < 0.01). A similar benefit was observed at 3 months and 6 months, with patients in the CPP group having a 0.31% and 0.41% greater decrease in HbA_1c than patients in the control group, respectively (P < 0.01). As noted in Table 3, for the secondary outcome of achieving an HbA_1c level of <7%, at 12 months, an 8.9% greater proportion of patients in the CPP group were able to meet this goal (P < 0.01). At 6 months, 8.1% more patients in the CPP group achieved the target HbA_1c of <7% (P < 0.01) when compared to the non-CPP group. A similar result was observed for the secondary outcome of achieving an HbA_1c level of <8%, with 7.4%, 10.7%, and 10.9% more patients in the CPP group meeting this goal at 3, 6, and 12 months, respectively (P < 0.01) compared to patients in the non-CPP group.

Table 2.

Primary Outcome: Difference in Glycosylated Hemoglobin Between CPP and Non-CPP Patient Groups With IPTW Adjustment^a

Time point	Parameter	Non-CPP	CPP	Difference	P value
90 days	Pre	7.7	8.9	1.2	<0.01
	Post	7.6	8.5	0.9	<0.01
	Difference in difference			–0.31%	0.002
180 days	Pre	7.7	8.9	1.2	<0.01
	Post	7.6	8.3	0.8	<0.01
	Difference in difference			–0.41	<0.01
360 days	Pre	7.7	8.9	1.2	<0.01
	Post	7.6	8.3	0.7	<0.01
	Difference in difference			–0.44	<0.01

Abbreviations: CPP, clinical pharmacist practitioner; HbA_1c, glycosylated hemoglobin; IPTW, inverse probability of treatment weight.

^aData shown as % unless indicated otherwise.

Table 3.

Secondary Outcome: Achievement of Glycosylated Hemoglobin Targets with IPTW Adjustment^a

Time point	HbA1c goal	Parameter	Non-CPP	CPP	Difference in difference	P value
90 days	<7	Pre	38.9	16.2	–22.6	<0.01
		Post	41.2	22.1	–19.1	<0.01
		Difference in difference	3.5	0.096
	<8	Pre	64.4	35.7	–28.7	<0.01
		Post	67.9	46.6	–21.3	<0.01
		Difference in difference	7.4	0.003
180 days	<7	Pre	38.9	16.2	–22.6	<0.01
		Post	41.1	26.5	–14.6	<0.01
		Difference in difference	8.1	<0.01
	<8	Pre	64.4	35.7	–28.7	<0.01
		Post	68.0	50.0	–18.0	<0.01
		Difference in difference	10.7	<0.01
360 days	<7	Pre	38.9	16.2	–22.6	<0.01
		Post	40.6	26.8	–13.8	<0.01
		Difference in difference	8.9	<0.01
	<8	Pre	64.4	35.7	–28.7	<0.01
		Post	67.4	49.6	–17.8	<0.01
		Difference in difference	10.9	<0.01

Abbreviations: CPP, clinical pharmacist practitioner; HbA_1c, glycosylated hemoglobin; IPTW, inverse probability of treatment weight.

^aData shown as % unless indicated otherwise.

Discussion

This study evaluated a large and diverse population of patients managed collaboratively by pharmacists across multiple primary care and specialist clinics, helping to better characterize the benefit of utilizing pharmacists in a value-based care model. The baseline HbA_1c level of patients referred to a pharmacist for management of diabetes was significantly higher (by >1 percentage point), and these patients were of older age and more frequently had comorbid hypertension. The increased baseline complexity of these patients represents additional risk in value-based care models as it may influence the likelihood that these patients achieve goals set by quality measures. If the greater baseline complexity is not accounted for, then it is possible that the true benefit of pharmacist involvement in achieving value-based quality measures will be underestimated. The results of this study demonstrate the importance of understanding patient complexity as a key consideration when discussing position justification based on revenue that is derived from achievement of value-based quality measures.

Although representing a more complex population, the patients managed by a CPP saw statistically significant reductions in HbA_1c levels within 3 months of being referred for care compared to the difference observed in patients managed by a non-CPP provider. The observed 0.4% difference in HbA_1c reduction is clinically significant, as research has shown that even small reductions in HbA_1c are associated with a large reduction in microvascular complications.⁹ Additionally, the CPP-managed patients had a greater probability of meeting the HEDIS definition of HbA_1c control, which directly contributes to increased performance on value-based quality measures. Therefore, despite the increased complexity in patients referred to a pharmacist, this study demonstrates that pharmacists operating within a collaborative practice agreement can successfully serve as not only extenders of care but also as providers of high-quality care.

The relatively large sample size allowed us to perform IPTW adjustment to control for the significant difference in baseline characteristics, which helped to mitigate the risk of selection bias, an important consideration in cohort study designs.¹⁰ The difference-in-differences method has been shown in other literature to be a useful method of controlling for pretreatment differences when evaluating HbA_1c.¹¹ This statistical method permitted us to adjust for the differences in baseline HbA_1c that are seen when a pharmacist is managing the most uncontrolled and treatment-resistant patients in a clinic. As a result, we can suggest with some confidence that the difference in observed outcomes between these 2 cohorts is due to the involvement of a pharmacist on the care team. The fact that the treatment benefit was present across multiple clinics indicates that the addition of a pharmacist to the interdisciplinary care team can improve the quality of care provided across a variety of practice settings.

Previous literature has shown that pharmacists can have a positive impact on diabetes-related quality measures.^12-14 Joseph and colleagues¹² reported that intervention by a pharmacist can lead to improvement in diabetes-related quality measures, specifically dilated eye exams, but were unable to evaluate HbA_1c-associated measure performance owing to limitations of the available data. A previous study describing a randomized controlled trial for a pharmacist-led diabetes intervention demonstrated statistically significant reductions in HbA_1c; however, the patient population was small and restricted to patients enrolled in a single value-based care contract, limiting the generalizability of the findings.¹³ This study is the first, to our knowledge, to demonstrate the positive impact of pharmacists on value-based quality measures in a large patient population as well as the first of its kind to control for baseline differences in patients assigned to pharmacist-based treatment groups, leveraging inverse probability of treatment weighting.

Limitations

As with other retrospective cohort studies, there were limitations that should be considered.¹⁰ Although we controlled for multiple potential covariates to account for differences in baseline characteristics, there is always the potential that a relevant covariate was not included, which could lead to differences in observed treatment effect. The number of variables available to control for baseline differences also poses a limitation. In the future, it may be helpful to control for financial class, baseline laboratory values, hyperlipidemia, and other comorbidities. Additionally, it was not possible to control for the differences in starting HbA_1c, so it is unclear whether the lower baseline HbA_1c in the nonintervention arm contributed to the reduced pre-post difference. The higher baseline HbA_1c in the intervention arm may have led to more aggressive treatment decisions being made compared to the nonintervention arm. The reliance on ICD-10 coding within patients’ electronic medical records to determine eligibility could have impacted the appropriate selection of our patient population if the diagnoses were coded incorrectly. Lastly, while patients were assigned to each cohort on the basis of provider visits, it is possible that a pharmacist could have been involved in the care of a patient seen by a non-CPP provider through informal consults or staff messages, which could have impacted the observed differences in treatment outcomes between the cohorts.

Conclusion

Patients managed by a pharmacist had a significant improvement in HbA_1c as well as the likelihood that they achieved disease control as defined by diabetes-associated value-based quality measures. These improvements were noticed within 3 months of the first CPP visit and persisted at 6- and 12-month evaluations. Therefore, pharmacists have the potential to have a vital role in influencing medication-related value-based care measures in more complex and difficult-to-manage patient populations. Building on this study, future research could focus on other HEDIS measures to demonstrate the benefit of pharmacist management for other chronic conditions. Additionally, future research is needed to characterize the revenue generated and costs avoided as a result of clinical pharmacist involvement in outpatient disease management.

Disclosures

This publication was supported by grant number UL1TR002489 from the National Center for Advancing Translational Sciences at the National Institutes of Health. The authors have declared no potential conflicts of interest.

Additional information

The findings of this study were presented as a platform presentation at the UNC Eshelman School of Pharmacy’s Research in Education and Practice Symposium in May 2021.