Currently Viewing:
The American Journal of Managed Care March 2017
Impact of a Pharmacy-Based Transitional Care Program on Hospital Readmissions
Weiyi Ni, PhD; Danielle Colayco, PharmD, MS; Jonathan Hashimoto, PharmD; Kevin Komoto, PharmD, MBA; Chandrakala Gowda, MD, MBA; Bruce Wearda, RPh; and Jeffrey McCombs, PhD
Applying Organizational Behavior Theory to Primary Care
Samyukta Mullangi, MD, MBA, and Sanjay Saint, MD, MPH
Private Sector Accountable Care Organization Development: A Qualitative Study
Ann Scheck McAlearney, ScD; Brian Hilligoss, PhD; and Paula H. Song, PhD
Scaling Lean in Primary Care: Impacts on System Performance
Dorothy Y. Hung, PhD; Michael I. Harrison, PhD; Meghan C. Martinez, MPH; and Harold S. Luft, PhD
Patient Experience Midway Through a Large Primary Care Practice Transformation Initiative
Kaylyn E. Swankoski, MA; Deborah N. Peikes, PhD, MPA; Stacy B. Dale, MPA; Nancy A. Clusen, MS; Nikkilyn Morrison, MPPA; John J. Holland, BS; Timothy J. Day, MSPH; and Randall S. Brown, PhD
A Better Way: Leveraging a Proven and Utilized System for Improving Current Medication Reconciliation Processes
Ajit A. Dhavle, PharmD, MBA; Seth Joseph, MBA; Yuze Yang, PharmD; Chris DiBlasi, MBA; and Ken Whittemore, RPh, MBA
Currently Reading
Effects of an Enhanced Primary Care Program on Diabetes Outcomes
Sarah L. Goff, MD; Lorna Murphy, MA, MPH; Alexander B. Knee, MS; Haley Guhn-Knight, BA; Audrey Guhn, MD; and Peter K. Lindenauer, MD, MSc
Improvements in Access and Care Through the Affordable Care Act
Julie A. Schmittdiel, PhD; Jennifer C. Barrow, MSPH; Deanne Wiley, BA; Lin Ma, MS; Danny Sam, MD; Christopher V. Chau, MPH; Susan M. Shetterly, MS
Does Paid Versus Unpaid Supplementary Caregiving Matter in Preventable Readmissions?
Hsueh-Fen Chen, PhD; Taiye Oluyomi Popoola, MBBS, MPH; and Sumihiro Suzuki, PhD

Effects of an Enhanced Primary Care Program on Diabetes Outcomes

Sarah L. Goff, MD; Lorna Murphy, MA, MPH; Alexander B. Knee, MS; Haley Guhn-Knight, BA; Audrey Guhn, MD; and Peter K. Lindenauer, MD, MSc
An insurance company—sponsored enhanced primary care program had little effect on selected outcomes for low-income patients with diabetes.

To evaluate the effectiveness of Buena Salud, a multidisciplinary enhanced primary care program for Medicaid Managed Care patients at a community health center serving a low-income Hispanic community.

Study Design: Controlled before-and-after observational study. 

Methods: We extracted data from the electronic health record for patients aged 18 to 64 years with a) type 2 diabetes (T2D) enrolled in the Buena Salud program between August 2011 and January 2012 and b) randomly selected control patients with T2D who had been seen at the study health center during the same time frame. Outcomes included process measures (eg, number of glycosylated hemoglobin measures in a year), target lab and blood pressure values, and utilization measures (eg, emergency department visits). Demographics and other potential confounders were also extracted. We used a difference-in-differences (DID) analysis to estimate the effect of the intervention.

Results: A total of 72 Buena Salud patients with T2D and 247 control patients with T2D were included in the analysis. The Buena Salud group had a greater increase in the percentage of patients with guideline-concordant measurement of microalbumin/creatinine (DID = 22.2%; P = .008), a trend toward fewer hospitalizations than controls, and a mean rise in diastolic blood pressure. We did not find differences in other outcome or utilization measures.

Conclusions: A recently implemented enhanced primary care program had minimal impact on T2D process, outcome, and utilization measures for patients in this study. However, there were some promising trends, and it is possible that the intervention may demonstrate more of an effect as the program matures.

Am J Manag Care. 2017;23(3):e75-e81
Takeaway Points

As risk for population health is increasingly being shared, insurance companies are exploring the use of interdisciplinary care teams in primary care to improve chronic disease and population health management. In this study, we evaluated the outcomes of a new insurance company–sponsored enhanced primary care program, Buena Salud, in a controlled before-and-after study. Buena Salud targeted Hispanic patients with diabetes at a community center serving a mostly low-income population. We found limited effects on diabetes outcomes, including process, utilization, and patient outcome measures, in the program’s first year of existence. 
  • Implementation studies of team-based primary care programs may enhance program effectiveness outside of an experimental setting. 
  • Policy makers may want to allow more time for programs implemented in a natural setting to mature before determining effectiveness.
  • Policies that address the social determinants of health may be necessary for enhanced primary care programs to have sustained substantial impact.
Case management, peer health coaching, and other team-based systems of care, such as the patient-centered medical home, have been implemented throughout the United States in an effort to improve care coordination and patient outcomes in the ambulatory setting.1-9 These types of programs are often created and implemented by healthcare delivery systems, but insurance companies have played an increasing role in developing and funding strategies to improve chronic disease management as well.10 The term "enhanced primary care” refers to a team-based care model that aims to improve care processes and outcomes.11 This multi-disciplinary team model includes the use of clinical tools such as practice guidelines, patient monitoring and tracking systems, and measures of resource use. 

Team-based care models often focus on management of chronic diseases, such as type 2 diabetes (T2D), because of the higher morbidity, mortality, and costs associated with chronic disease.12 T2D has proven particularly challenging to manage in the primary care setting, with less than 20% of patients with T2D achieving targets for glycated hemoglobin (A1C), low-density lipoprotein cholesterol (LDL-C), and blood pressure.13 Although team-based care coordination programs for patients with T2D have shown promise,8,9 disease management remains suboptimal, particularly for vulnerable populations.12

Racial, ethnic, and economic disparities in T2D prevalence, care, and outcomes in the United States cause a disproportionate burden of disease in vulnerable populations.14,15 Arguably, patients with the highest risk for poor outcomes may benefit most from the additional resources provided by an enhanced primary care team. In this study, we aimed to determine the impact of a newly implemented insurance company–sponsored team-based enhanced primary care program (Buena Salud) on process measures, patient outcomes, and healthcare utilization for a low-income racial and ethnic minority population of patients with T2D. 


We conducted a controlled before-and-after study to assess the effect of the Buena Salud program on T2D process, outcome, and utilization measures. Although the Buena Salud team provided support for patients with other chronic diseases and promoted preventive care, this analysis is limited to patients with T2D.

Design, Setting, and Participants 

Buena Salud is a bilingual enhanced primary care program for Medicaid Managed Care patients. The program was financed by the Health New England insurance company and implemented by Brightwood Health Center (BHC), an urban community health center with a largely Hispanic population (88%) insured primarily by either Medicaid (59%) or Medicare (28%). More than 50% of BHC patients prefer Spanish as their spoken language, and 17% of adult BHC patients have T2D. Patients were generally enrolled in Buena Salud through a referral from their primary care provider, but they also could have been enrolled through a periodic auto-enrollment process or through self-referral. 

The Buena Salud enhanced primary care team consisted of 2 registered nurse care managers, 2 medical assistants trained as community health workers, and a caseworker. The care team was unusual compared with many managed care teams in that each team member was bilingual (Spanish/English) and from the same racial/ethnic group as the majority of BHC patients. The Buena Salud intervention was based on the Chronic Care Model, which focuses on 6 areas that can optimize chronic disease care: self-management support, healthcare delivery system design, decision support, use of clinical information systems, organization of healthcare, and community support.16,17 

The Buena Salud team offered patients education and coaching for disease self-management, provided additional contact with patients in the clinic and at home, interfaced with diabetes specialists for decision support, used electronic health registries to identify patients in need of care and services, and utilized linkage to community-based support services (eAppendix [eAppendices available at]). A total of 450 patients were enrolled in the Buena Salud program between August 2011 and January 2012. Each Buena Salud nurse was expected to actively manage up to 50 of these patients at any given time, not all of whom had T2D. Care intensity varied depending on individual need, but was not formally documented by the team. 

We reviewed eligible patients’ electronic health records (EHRs) to extract data pertinent to the study aims. Eligibility was defined as follows: 1) all BHC patients aged 18 to 64 years with T2D who were newly enrolled in the Buena Salud program between August 1, 2011, and January 31, 2012 (intervention); and 2) all BHC patients aged 18 to 64 years who had T2D, had been seen for a clinical encounter at BHC between August 1, 2011, and January 31, 2012, and were not enrolled in Buena Salud (potential controls). All intervention and control patients were cared for in the same clinical setting. From the list of potential controls, we randomly selected 3 patients with T2D who had been seen in the clinic in the same month that a Buena Salud patient had been enrolled. We selected controls in this manner rather than matching on variables such as age, race/ethnicity, or gender for 2 reasons: 1) the population attending the clinic is relatively homogeneous for these measurable variables and 2) selecting patients seen in a similar time frame reduces the potential for differences in unmeasured confounders related to temporal changes in practice at the health center. Buena Salud patients and controls also must have had at least 1 visit to BHC in the 12 months prior to the enrollment/index visit so that baseline data could be extracted. Because there were fewer patients with T2D enrolled in Buena Salud than anticipated, we had slightly higher than a 3:1 ratio. This study was approved by the Baystate Institutional Review Board, which waived informed consent.

Time Period Studied and Outcome Variables 

We identified target clinical outcomes, care processes, and healthcare utilization measures for patients with T2D using the national American Diabetes Association diabetes care guidelines that were in place in 2011 when the study was designed and systematic reviews of studies that included diabetes process and outcome measures.18-22 Clinical outcome measures included values for A1C, LDL-C, systolic blood pressure, and diastolic blood pressure (DBP). Care processes included the number of times A1C, microalbumin/creatinine ratio, and lipids were measured. Utilization measures included emergency department (ED) visits and unplanned hospitalizations, defined as any hospitalization other than for a nonemergent procedure (Figure). Baseline data were collected from the year prior to enrollment/index visit, beginning with the first day of the month in which they were enrolled, and the intervention period was defined as a window ranging from 12 to 15 months following the enrollment/index month. Timing of this window varied slightly for different outcomes in order to account for the earliest time after enrollment that the intervention would have been expected to affect the outcome measured and to provide grace periods for guidelines requiring a certain frequency of a measure (Figure). The overall duration of follow-up was chosen so that we could capture the maximum number of Buena Salud patients in the study. In addition to T2D process and outcomes data, we also extracted demographic information and data regarding potential confounders (eg, comorbidities and the number of years a patient received care at BHC) (Table 1). 

Data Extraction

We first oriented data extractors to the study’s data dictionary and extraction protocol, which included where to locate pertinent data in the EHR. After establishing baseline consistency using standardized extraction forms, 2 extractors also independently reviewed 20 randomly selected EHRs to assess inter-extractor consistency at 6 and 18 months into the course of data extraction to test whether consistency was maintained. Study data were collected and managed using REDCap electronic data capture tools.23


Participant characteristics are presented as means and standard deviations (SDs) for continuous variables and as frequencies and percentages for categorical variables. To estimate differences between the groups studied, we used a difference-in-differences (DID) approach (the difference between the pre-post change in the Buena Salud group compared with the control group). Study outcomes were modeled using generalized estimating equations with exchangeable correlations and robust standard errors (clustering on patient). Continuous outcomes were modeled using the identity link, Gaussian family binary outcomes were modeled using the logit link and binomial family, and count outcomes were modeled using the log link and negative binomial family. Models were estimated with main effects for the intervention group and time period, with an interaction term between these 2 representing the DID. Because Buena Salud participants were frequency-matched to controls based on enrollment month, we used enrollment month as an indicator variable in the model. This term was not significant in the models, so it was removed. 

Predicted outcomes and 95% confidence intervals (CIs) are presented in their original metrics (eg, mm/Hg for blood pressure) using Stata’s “-margins-” postestimation command. Statistical significance was set at an alpha of <0.05. Multivariable models included potential confounders (eg, demographic data, comorbidities, and number of years receiving care at BHC). Using Wald tests, models were reduced to include variables that were significant at the <0.05 level. To control for possible residual confounding and for face validity, we retained age, mental health, and substance use in all models. Original power calculations estimated that if at least 100 Buena Salud patients and 175 controls were included, this sample size would provide >85% power to detect a medium effect size (Cohen’s d = 0.40) at an alpha of <0.05. Clinically, this would be the equivalent of a DID estimate for A1C of 0.75, assuming a pooled SD of change of 1.70. Our achieved sample size was less than the estimated 100, but there was still 85% power to detect the same effect size. The analysis was conducted using Stata version 13.1 (StataCorp LP, College Station, Texas). 


Copyright AJMC 2006-2020 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up