Improving Outcomes for Diverse Diabetic Populations
Improving Outcomes for Diverse Diabetic Populations
Project IMPACT: Diabetes was designed to establish a nationwide program that would scale the proven APhA Foundation process model into communities across the United States that are disproportionately affected by diabetes. Consistent with previous APhA Foundation initiatives, the Project IMPACT: Diabetes approach integrates collaborative care with pharmacists, continuous quality improvement, use of patient self-management credentialing, and collection of a minimum data set.
The key objectives of Project IMPACT: Diabetes included the following:
Project IMPACT: Diabetes focused on improving the care and expanding access to evidence-based practices for several heavily burdened populations:
The study evaluated implementation strategies and patient care results in 25 disproportionate share communities across the United States. Through Project IMPACT: Diabetes, each community increased access to patient-centered, team-based care designed to improve clinical, process, and self-management measures related to diabetes care; increased collaboration among physicians, pharmacists, and other members of the health care team; and potentially prevented costly diabetes-related complications, including amputations, blindness, and glucose excursions (hypoglycemia/hyperglycemia) that could lead to hospital visits.
Twenty-five communities were selected through a competitive application process that evaluated communities on their patient population, resources, information accessibility, team motivation and education, plan for incentive alignment, previously demonstrated success, and leadership. The communities engaged local stakeholders, modified existing programs, and used other community resources to implement local models of care for patients with diabetes. Figure 1 displays the distribution of the participating communities overlaid on the Centers for Disease Control and Prevention's (CDC) Behavioral Risk Factor Surveillance System map for people who have been told they have diabetes.Table 1 lists the names of the communities, their locations, and descriptive information about the type of partnering organizations.
(Enlarge Image)
Figure 1.
Distribution of Project IMPACT: Diabetes communities (▾) across states categorized according to mean glycosylated hemoglobin (A1C) values of residents with diagnosed diabetes
Source of A1C data: Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System map for people who have been told they have diabetes.
Project IMPACT: Diabetes is a multisite, observational, pre–post comparison study evaluating the impact of quality-improvement activities on the clinical outcomes of patients with diabetes. The Western Institutional Review Board approved the study and granted a waiver of informed consent.
Patients participating in the study were enrolled from June 1, 2011, to January 31, 2012. Each participant was evaluated for 1 year after enrollment or until January 31, 2013. The corresponding baseline data collection period extended up to 180 days before the start of the study period for clinical measures and 365 days before the start of the study period for process measures. To be eligible for enrollment, patients had to have a diagnosis of diabetes, be newly initiated on diabetes therapy, or be maintained on diabetes therapy but poorly controlled (e.g., A1C >7%).
Each community, led by at least one community champion, adapted APhA Foundation's structure and process model to accommodate the local process of care, which could include physicians, nurse practitioners, physician assistants, specialist providers, organizational administration, health benefits managers, promotoras (lay Hispanic/Latino community members with specialized health education training), patient advocates, and others who affect how care is received in the community.
The common thread through all participating communities was that pharmacists and patients were integrated into the care team and clinical progress was quantified and recorded. As members of the team, pharmacists educated patients on the pathology of diabetes and how medications work to improve health; taught insulin injection techniques and the importance of medication adherence; promoted healthy lifestyles; and reinforced health goals and monitored progress toward those goals.
In addition to involving one-on-one consultations with pharmacists, local care models varied in their inclusion of such offerings as group educational classes, grocery store food tours, exercise programs, joint visits with patients by a combination of providers (e.g., pharmacists, physicians, dietitians, nurse practitioners), and a variety of patient incentives (e.g., bus passes, grocery store gift cards, discounted or free healthy lunches at employer worksites, discounted copayments for antidiabetic medications and supplies). Specific information about the composition of the health care teams and how pharmacists were integrated into the different communities is available at www.projectimpactdiabetes.org and included in a companion article in this issue of JAPhA.
Changes in clinical performance measures were assessed at baseline and then according to practice guidelines for a period of 1 year. For an individual patient's data to be included in the project evaluation, a minimum of two documented postenrollment visits at least 90 days apart was required. Clinical measures included hemoglobin A1C, body mass index (BMI), systolic/diastolic blood pressure, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), total cholesterol, and triglycerides. Each patient was also assessed on key process measures for the care of patients with diabetes, including foot examination status, eye examination status, influenza vaccination status, and smoking status.
At baseline, all enrolled patients completed the Knowledge Assessment from APhA Foundation's PSMP for Diabetes to gauge their knowledge strengths and areas for improvement. Based on their answers to the 36-question assessment, patients earned achievement levels of "beginner," "proficient," or "advanced." Assessment responses were also used to help pharmacists and other providers customize patient education to meet individual needs.
A postproject survey was fielded to communities to collect information about the types of partnering health care workers, formalized collaborative relationships, and sustainability of the clinical services.
The minimum data set was standardized across all 25 communities, and a common data collection tool was provided to streamline data organization and reporting. Community members were responsible for collecting the minimum data set elements through chart extraction, point-of-care testing, or other methods. Data were deidentified. Clinical changes from the baseline period to the 12-month intervention period, process measures, and patient self-management proficiency were assessed. Each patient served as his or her own control for comparative analyses.
Because of the many factors that influenced care delivery within each of the 25 communities, we believed that an analysis controlling for some of the major differences among participants would provide the most conservative and accurate determination of intervention impact. To shape the analysis, we assumed that study participants within a community who experience the same pharmacist interventions are likely more similar than participants from different communities. SAS Proc Mixed (SAS Institute, Inc., Cary, NC) was used to run hierarchical linear mixed models to handle the nesting of participants within communities. Additionally, observations (baseline and most recent) were made of the participants who were cared for by the organizations within those communities in this multilevel model.
We anticipated that health outcomes would change once participants had been exposed to the intervention and that there would be inherent differences among participants of the various knowledge groups as defined by the PSMP. The initial model included a fixed effect for time, a fixed effect for credential group, and an interaction between group and time. Because researchers expected there to be differences among organizations, random effects for time and intercept for each organization were also included in the model. Effect sizes were calculated using least-squares means for the effects of interest (time or credential) and the observed standard deviation of the total sample during baseline for between-group comparisons. Effect sizes within group comparisons were calculated for significant findings using the observed standard deviations and the estimated correlation between time points from the model. The a priori level of significance was set at P <0.05.
Objectives
Project IMPACT: Diabetes was designed to establish a nationwide program that would scale the proven APhA Foundation process model into communities across the United States that are disproportionately affected by diabetes. Consistent with previous APhA Foundation initiatives, the Project IMPACT: Diabetes approach integrates collaborative care with pharmacists, continuous quality improvement, use of patient self-management credentialing, and collection of a minimum data set.
The key objectives of Project IMPACT: Diabetes included the following:
Expand a proven community-based model of care throughout high-risk areas in the United States.
Improve key indicators of diabetes care in these communities.
Scale the existing model nationally by establishing local peer-to-peer network mentoring.
Establish a sustainable platform for permanent change by embedding guiding principles in each community that will drive diabetes care for years to come.
Target Population
Project IMPACT: Diabetes focused on improving the care and expanding access to evidence-based practices for several heavily burdened populations:
Areas with incidences of diabetes higher than that of the state average
Patients with uncontrolled glycosylated hemoglobin (A1C), defined as values greater than 7%, and other indicators of uncontrolled blood pressure, hypercholesterolemia, or weight
Patients with limited access to quality diabetes care because of geographic, financial, or other barriers
Communities that show need, through lack of focused resources or diabetes-related programming, for implementation of enhanced diabetes care
Methods
The study evaluated implementation strategies and patient care results in 25 disproportionate share communities across the United States. Through Project IMPACT: Diabetes, each community increased access to patient-centered, team-based care designed to improve clinical, process, and self-management measures related to diabetes care; increased collaboration among physicians, pharmacists, and other members of the health care team; and potentially prevented costly diabetes-related complications, including amputations, blindness, and glucose excursions (hypoglycemia/hyperglycemia) that could lead to hospital visits.
Setting
Twenty-five communities were selected through a competitive application process that evaluated communities on their patient population, resources, information accessibility, team motivation and education, plan for incentive alignment, previously demonstrated success, and leadership. The communities engaged local stakeholders, modified existing programs, and used other community resources to implement local models of care for patients with diabetes. Figure 1 displays the distribution of the participating communities overlaid on the Centers for Disease Control and Prevention's (CDC) Behavioral Risk Factor Surveillance System map for people who have been told they have diabetes.Table 1 lists the names of the communities, their locations, and descriptive information about the type of partnering organizations.
(Enlarge Image)
Figure 1.
Distribution of Project IMPACT: Diabetes communities (▾) across states categorized according to mean glycosylated hemoglobin (A1C) values of residents with diagnosed diabetes
Source of A1C data: Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System map for people who have been told they have diabetes.
Study Design
Project IMPACT: Diabetes is a multisite, observational, pre–post comparison study evaluating the impact of quality-improvement activities on the clinical outcomes of patients with diabetes. The Western Institutional Review Board approved the study and granted a waiver of informed consent.
Patients participating in the study were enrolled from June 1, 2011, to January 31, 2012. Each participant was evaluated for 1 year after enrollment or until January 31, 2013. The corresponding baseline data collection period extended up to 180 days before the start of the study period for clinical measures and 365 days before the start of the study period for process measures. To be eligible for enrollment, patients had to have a diagnosis of diabetes, be newly initiated on diabetes therapy, or be maintained on diabetes therapy but poorly controlled (e.g., A1C >7%).
Each community, led by at least one community champion, adapted APhA Foundation's structure and process model to accommodate the local process of care, which could include physicians, nurse practitioners, physician assistants, specialist providers, organizational administration, health benefits managers, promotoras (lay Hispanic/Latino community members with specialized health education training), patient advocates, and others who affect how care is received in the community.
The common thread through all participating communities was that pharmacists and patients were integrated into the care team and clinical progress was quantified and recorded. As members of the team, pharmacists educated patients on the pathology of diabetes and how medications work to improve health; taught insulin injection techniques and the importance of medication adherence; promoted healthy lifestyles; and reinforced health goals and monitored progress toward those goals.
In addition to involving one-on-one consultations with pharmacists, local care models varied in their inclusion of such offerings as group educational classes, grocery store food tours, exercise programs, joint visits with patients by a combination of providers (e.g., pharmacists, physicians, dietitians, nurse practitioners), and a variety of patient incentives (e.g., bus passes, grocery store gift cards, discounted or free healthy lunches at employer worksites, discounted copayments for antidiabetic medications and supplies). Specific information about the composition of the health care teams and how pharmacists were integrated into the different communities is available at www.projectimpactdiabetes.org and included in a companion article in this issue of JAPhA.
Changes in clinical performance measures were assessed at baseline and then according to practice guidelines for a period of 1 year. For an individual patient's data to be included in the project evaluation, a minimum of two documented postenrollment visits at least 90 days apart was required. Clinical measures included hemoglobin A1C, body mass index (BMI), systolic/diastolic blood pressure, low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), total cholesterol, and triglycerides. Each patient was also assessed on key process measures for the care of patients with diabetes, including foot examination status, eye examination status, influenza vaccination status, and smoking status.
At baseline, all enrolled patients completed the Knowledge Assessment from APhA Foundation's PSMP for Diabetes to gauge their knowledge strengths and areas for improvement. Based on their answers to the 36-question assessment, patients earned achievement levels of "beginner," "proficient," or "advanced." Assessment responses were also used to help pharmacists and other providers customize patient education to meet individual needs.
A postproject survey was fielded to communities to collect information about the types of partnering health care workers, formalized collaborative relationships, and sustainability of the clinical services.
Data Collection
The minimum data set was standardized across all 25 communities, and a common data collection tool was provided to streamline data organization and reporting. Community members were responsible for collecting the minimum data set elements through chart extraction, point-of-care testing, or other methods. Data were deidentified. Clinical changes from the baseline period to the 12-month intervention period, process measures, and patient self-management proficiency were assessed. Each patient served as his or her own control for comparative analyses.
Data Analysis
Because of the many factors that influenced care delivery within each of the 25 communities, we believed that an analysis controlling for some of the major differences among participants would provide the most conservative and accurate determination of intervention impact. To shape the analysis, we assumed that study participants within a community who experience the same pharmacist interventions are likely more similar than participants from different communities. SAS Proc Mixed (SAS Institute, Inc., Cary, NC) was used to run hierarchical linear mixed models to handle the nesting of participants within communities. Additionally, observations (baseline and most recent) were made of the participants who were cared for by the organizations within those communities in this multilevel model.
We anticipated that health outcomes would change once participants had been exposed to the intervention and that there would be inherent differences among participants of the various knowledge groups as defined by the PSMP. The initial model included a fixed effect for time, a fixed effect for credential group, and an interaction between group and time. Because researchers expected there to be differences among organizations, random effects for time and intercept for each organization were also included in the model. Effect sizes were calculated using least-squares means for the effects of interest (time or credential) and the observed standard deviation of the total sample during baseline for between-group comparisons. Effect sizes within group comparisons were calculated for significant findings using the observed standard deviations and the estimated correlation between time points from the model. The a priori level of significance was set at P <0.05.
