Diabetes Mellitus
Conditions
Keywords
Incentives,, Diabetes,, Medicaid,, Health behavior,, Cost effectiveness
Brief summary
The purpose of this randomized controlled trial (RCT) study is to examine the extent that financial incentives when combined with diabetes evidence-based practices, improve self-management and biometric measures for adult diabetic Medicaid recipients with an HbA1c of ≥ 6.5 at enrollment. The study will also evaluate the cost-effectiveness of the program. Specific Aims: 1. Evaluate whether financial incentives for completing American Diabetes Association (ADA) recommended tests, exams, health education, biometric outcome goals, and vaccinations will improve primary biometric outcomes. 2. Evaluate whether financial incentives for completing ADA recommended tests, exams, health education, biometric outcome goals, and vaccinations will improve diabetes patients' self-management as assessed by Summary of Diabetes Self-Care Activities Measure (SDSCA) and 36-Item Short Form Health Survey (SF36v2). 3. Evaluate the extent to which financial incentives for healthy behaviors reduce the cost of health care utilization.
Detailed description
Diabetes is the seventh leading cause of death in the United States (OECD 2013). It is also known that certain populations are at greater risk for diabetes. In Hawaii, diabetes disproportionally affects Native Hawaiians and Pacific Islanders as they are three times more likely to be diagnosed with diabetes. In addition, in 2010 the U.S. Department of Health and Human Services Office of Minority Health reported that Native Hawaiians/Pacific Islanders in Hawaii were 5.7 times as likely as Caucasians living in Hawaii to die from diabetes(Office of Minority Health, 2010). In order to address the challenges that chronic diseases impose on individuals and the health care system the Centers for Medicare & Medicaid Services (CMS) is assessing the impact of incentivizing patients to increase self-care and disease management. Previous studies have demonstrated that monetary incentives have been associated with an improvement in behavioral outcomes, most notably when the incentive is received immediately following the targeted behavior (Volpp, K.G., et.al., 2008; Mitchell, M.S., et.al., 2013). Cahill et al. (2008) showed that economic incentives were tied to smoking cessation and led to a decrease in relapse within a year. Our study seeks to build on these findings and determine whether financial incentives may provide a way to improve diabetes self-management. Data: Electronic data (Labs, Outcomes) - January 1st, 2013 through December 31, 2015 Electronic data (Claims) - January 1st, 2011 through December 31, 2015
Interventions
BEHAVIORALFinancial Incentives
This intervention will examine the effects of incentives on improving adult diabetic Medicaid beneficiaries' health outcomes and reducing associated costs through healthy behavior changes in their diabetes self-management. Incentives focus on improving self-management of diabetes, compliance with ADA recommended preventive, treatment and management measures, primary biometric measures of diabetes, and eliminating barriers to a healthy lifestyle
Sponsors
Centers for Medicare and Medicaid Services
Hawaii Department of Human Services (DHS)
Kaiser Permanente
RTI International
IMPAQ International, LLC.
Econometrica, Inc.
University of Hawaii
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
HEALTH_SERVICES_RESEARCH
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
No
Inclusion criteria
1. Clinically diagnosed with diabetes mellitus 2. 18 years of age or older 3. Medicaid recipient 4. At recruitment has an HbA1c level of ≥ 6.5 5. At recruitment is receiving care coordination at Kaiser Permanente Hawaii
Exclusion criteria
1. Current pregnancy - gestational diabetes 2. End-stage Renal Disease 3. Does not speak English
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Changes in HDL From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in HDL from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in Total Cholesterol From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in total cholesterol from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in Triglycerides From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in triglycerides from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in LDL From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in LDL from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in HbA1c From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in Hemoglobin A1c from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in Systolic Blood Pressure From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in systolic blood pressure from baseline to end of study. Change = (End of Intervention score - Baseline score) |
| Changes in Diastolic Blood Pressure From Baseline to the End of Intervention (December 2015) | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | Changes in diastolic blood pressure from baseline to end of study. Change = (End of Intervention score - Baseline score) |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Change From Baseline to End of Intervention (December 2015) in General Diet Subscale of The Summary of Diabetes Self-Care Activities (SDSCA) Measure | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | SDSCA is a validated, self-reported measure assessing the average # of days the recommended diabetes self-care activities are performed over the past 7 days in the areas of general diet, specific diet, exercise, blood-glucose testing, and foot care at baseline, mid, and end of intervention. Possible scores range from 0 to 7 days. Change = (End of Intervention Score - Baseline Score) |
| Change From Baseline to End of Intervention (December 2015) in Physical Component Summary Measure of the Short Form (SF-36v2) Health Survey | Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment | The SF-36v2 is validated, self reported short-form health survey used to assess changes over time in the well-being of participants. It consists of 2 component summary measures that further summarize 8 health domain scales. The Physical Component Summary (PCS) measure is derived from domain scales of Physical Functioning (10 items), Role-Physical (4 items), Bodily Pain (2 items), and General Health (5 items). Scores of component summary measures and health domain scales range from 0 to 100 with higher scores indicating better outcomes. Norm-based scoring was used so that scores for each health domain scale and component summary measure have a mean of 50 and standard deviation of 10 based on the 2009 U.S. general population. The SF-36v2 was used to assess participants' health and wellbeing at baseline, mid, and endpoint of intervention. Change = (Midpoint Score - Baseline Score) |
| Changes in Health Utilization Cost Before and During Intervention - Amount Paid by Service Providers | Before intervention (3 years prior to baseline) and during intervention (2 years from baseline to end of intervention) | Changes of total cost expenditures including emergency room use and hospitalizations in the intervention and control groups before and during intervention. |
Countries
United States
Participant flow
Recruitment details
The study recruited 320 eligible patients from Kaiser Permanente Hawaii. Participants were randomly assigned into the intervention or control group with 159 and 161 in each group respectively. The randomized control trial (RCT) was conducted from May 2014 to December 31, 2015 with rolling enrollment from May 2014 to January 2015.
Participants by arm
| Arm | Count |
|---|---|
| Financial Incentives Participants (159) in the Incentive Group will: 1) continue to receive usual care; 2) are eligible to receive financial incentives based on completion of recommended ADA benchmarks and achievement of goals that are founded on evidence based guidelines for diabetes; and 3) be compensated for completion of surveys.
Financial Incentives: This intervention will examine the effects of incentives on improving adult diabetic Medicaid beneficiaries' health outcomes and reducing associated costs through healthy behavior changes in their diabetes self-management. Incentives focus on improving self-management of diabetes, compliance with ADA recommended preventive, treatment and management measures, primary biometric measures of diabetes, and eliminating barriers to a healthy lifestyle. | 159 |
| Control Participants (161) in the Control Group will continue to receive usual care and be compensated for the completion of surveys only. They will not receive financial incentives. | 161 |
| Total | 320 |
Withdrawals & dropouts
| Period | Reason | FG000 | FG001 |
|---|---|---|---|
| Overall Study | Death | 1 | 1 |
| Overall Study | Lost Medicaid eligibility | 29 | 22 |
| Overall Study | Switched Managed Care Organization (MCO) | 4 | 4 |
| Overall Study | Withdrawal by Subject | 0 | 3 |
Baseline characteristics
| Characteristic | Financial Incentives | Control | Total |
|---|---|---|---|
| Age, Continuous | 48.5 years STANDARD_DEVIATION 10.7 | 47.8 years STANDARD_DEVIATION 10 | 48.1 years STANDARD_DEVIATION 10.32 |
| Ethnicity (NIH/OMB) Hispanic or Latino | 10 Participants | 14 Participants | 24 Participants |
| Ethnicity (NIH/OMB) Not Hispanic or Latino | 145 Participants | 144 Participants | 289 Participants |
| Ethnicity (NIH/OMB) Unknown or Not Reported | 4 Participants | 3 Participants | 7 Participants |
| History of Heart Disease | 14 Participants | 20 Participants | 34 Participants |
| History of Hypertension | 115 Participants | 106 Participants | 221 Participants |
| History of Smoking or Tobacco Dependence | 47 Participants | 48 Participants | 95 Participants |
| Race (NIH/OMB) American Indian or Alaska Native | 1 Participants | 0 Participants | 1 Participants |
| Race (NIH/OMB) Asian | 32 Participants | 36 Participants | 68 Participants |
| Race (NIH/OMB) Black or African American | 1 Participants | 1 Participants | 2 Participants |
| Race (NIH/OMB) More than one race | 40 Participants | 53 Participants | 93 Participants |
| Race (NIH/OMB) Native Hawaiian or Other Pacific Islander | 54 Participants | 48 Participants | 102 Participants |
| Race (NIH/OMB) Unknown or Not Reported | 4 Participants | 1 Participants | 5 Participants |
| Race (NIH/OMB) White | 27 Participants | 22 Participants | 49 Participants |
| Region of Enrollment United States | 159 participants | 161 participants | 320 participants |
| Sex: Female, Male Female | 88 Participants | 86 Participants | 174 Participants |
| Sex: Female, Male Male | 71 Participants | 75 Participants | 146 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk |
|---|---|---|
| deaths Total, all-cause mortality | 0 / 159 | 0 / 161 |
| other Total, other adverse events | 0 / 159 | 0 / 161 |
| serious Total, serious adverse events | 0 / 159 | 0 / 161 |
Outcome results
Primary
Changes in Diastolic Blood Pressure From Baseline to the End of Intervention (December 2015)
Changes in diastolic blood pressure from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in Diastolic Blood Pressure From Baseline to the End of Intervention (December 2015) | -2.1895 mmHg |
| Control | Changes in Diastolic Blood Pressure From Baseline to the End of Intervention (December 2015) | -1.1002 mmHg |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.388195% CI: [-3.5632, 1.3846]Generalized Estimating Equation Model
Primary
Changes in HbA1c From Baseline to the End of Intervention (December 2015)
Changes in Hemoglobin A1c from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in HbA1c From Baseline to the End of Intervention (December 2015) | 0.0660 percentage of glycosylated hemoglobin |
| Control | Changes in HbA1c From Baseline to the End of Intervention (December 2015) | -0.3440 percentage of glycosylated hemoglobin |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.055395% CI: [-0.0092, 0.8293]Generalized Estimating Equation Model
Primary
Changes in HDL From Baseline to the End of Intervention (December 2015)
Changes in HDL from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in HDL From Baseline to the End of Intervention (December 2015) | 0.6705 mg/dL |
| Control | Changes in HDL From Baseline to the End of Intervention (December 2015) | -0.0472 mg/dL |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.491595% CI: [-1.3272, 2.7626]Generalized Estimating Equation Model
Primary
Changes in LDL From Baseline to the End of Intervention (December 2015)
Changes in LDL from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in LDL From Baseline to the End of Intervention (December 2015) | 0.4523 mg/dL |
| Control | Changes in LDL From Baseline to the End of Intervention (December 2015) | 1.3620 mg/dL |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.86295% CI: [-9.3509, 11.1702]Generalized Estimating Equation Model
Primary
Changes in Systolic Blood Pressure From Baseline to the End of Intervention (December 2015)
Changes in systolic blood pressure from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in Systolic Blood Pressure From Baseline to the End of Intervention (December 2015) | -1.3579 mmHg |
| Control | Changes in Systolic Blood Pressure From Baseline to the End of Intervention (December 2015) | 0.9337 mmHg |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.241695% CI: [-6.1274, 1.5442]Generalized Estimating Equation Model
Primary
Changes in Total Cholesterol From Baseline to the End of Intervention (December 2015)
Changes in total cholesterol from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in Total Cholesterol From Baseline to the End of Intervention (December 2015) | -3.9141 mg/dL |
| Control | Changes in Total Cholesterol From Baseline to the End of Intervention (December 2015) | 2.5709 mg/dL |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.861795% CI: [-16.4517, 13.7653]Generalized Estimating Equation Model
Primary
Changes in Triglycerides From Baseline to the End of Intervention (December 2015)
Changes in triglycerides from baseline to end of study. Change = (End of Intervention score - Baseline score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model. The mixed effects model approach can provide unbiased results when the type of missing data is missing at random, which is common for longitudinal studies.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Changes in Triglycerides From Baseline to the End of Intervention (December 2015) | 0.4843 mg/dL |
| Control | Changes in Triglycerides From Baseline to the End of Intervention (December 2015) | 18.2713 mg/dL |
Comparison: Generalized estimating equation (GEE) modeling was used to examine the pre- and post-intervention changes between groups due to interventions in biometric indicators. The average change in scores over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no differences in the biometric measure from baseline to endpoint between groups.p-value: 0.533395% CI: [-73.7478, 38.1737]Generalized Estimating Equation Model
Secondary
Change From Baseline to End of Intervention (December 2015) in General Diet Subscale of The Summary of Diabetes Self-Care Activities (SDSCA) Measure
SDSCA is a validated, self-reported measure assessing the average # of days the recommended diabetes self-care activities are performed over the past 7 days in the areas of general diet, specific diet, exercise, blood-glucose testing, and foot care at baseline, mid, and end of intervention. Possible scores range from 0 to 7 days. Change = (End of Intervention Score - Baseline Score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model; this provides unbiased results when missing data is random. Not all participants submitted surveys at each data point.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Change From Baseline to End of Intervention (December 2015) in General Diet Subscale of The Summary of Diabetes Self-Care Activities (SDSCA) Measure | 0.7796 units on a scale |
| Control | Change From Baseline to End of Intervention (December 2015) in General Diet Subscale of The Summary of Diabetes Self-Care Activities (SDSCA) Measure | 0.1393 units on a scale |
Comparison: The average change in scores over time for the intervention group was compared to that for the control group. The difference in these average changes is referred to as the difference-in-differences values which was assessed for significance at p = 0.05. Null hypothesis assumed no difference in the General Diet subscale from baseline to endpoint between groups. Generalized estimating equation modeling was used to examine changes in SDSCA between groups at different time points.p-value: 0.04295% CI: [0.0233, 1.2572]Generalized Estimating Equation Model
Secondary
Change From Baseline to End of Intervention (December 2015) in Physical Component Summary Measure of the Short Form (SF-36v2) Health Survey
The SF-36v2 is validated, self reported short-form health survey used to assess changes over time in the well-being of participants. It consists of 2 component summary measures that further summarize 8 health domain scales. The Physical Component Summary (PCS) measure is derived from domain scales of Physical Functioning (10 items), Role-Physical (4 items), Bodily Pain (2 items), and General Health (5 items). Scores of component summary measures and health domain scales range from 0 to 100 with higher scores indicating better outcomes. Norm-based scoring was used so that scores for each health domain scale and component summary measure have a mean of 50 and standard deviation of 10 based on the 2009 U.S. general population. The SF-36v2 was used to assess participants' health and wellbeing at baseline, mid, and endpoint of intervention. Change = (Midpoint Score - Baseline Score)
Time frame: Baseline to end of intervention - 12 months minimum to 19 months maximum due to rolling enrollment
Population: Repeated measurements for each subject were used. The analysis was per protocol with all available observations for each subject included to estimate parameters in the mixed effects model; this provides unbiased results when the type of missing data is random. Not all participants submitted surveys at each data point.
| Arm | Measure | Value (LEAST_SQUARES_MEAN) |
|---|---|---|
| Financial Incentives | Change From Baseline to End of Intervention (December 2015) in Physical Component Summary Measure of the Short Form (SF-36v2) Health Survey | 0.5162 units on a scale |
| Control | Change From Baseline to End of Intervention (December 2015) in Physical Component Summary Measure of the Short Form (SF-36v2) Health Survey | -1.6027 units on a scale |
Comparison: The average change in scores over time for the intervention group was compared to that of the control group. The difference in the average changes is referred to as the difference-in-differences values which was assessed for significance at p=0.05. Null assumed no group difference in the Physical Component Summary measure of SF-36v2 from baseline to midpoint. Generalized estimating equation (GEE) modeling was used to examine changes in health measures between groups at different time points.p-value: 0.039995% CI: [0.0981, 4.1395]Generalized Estimating Equation Model
Secondary
Changes in Health Utilization Cost Before and During Intervention - Amount Paid by Service Providers
Changes of total cost expenditures including emergency room use and hospitalizations in the intervention and control groups before and during intervention.
Time frame: Before intervention (3 years prior to baseline) and during intervention (2 years from baseline to end of intervention)
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Financial Incentives | Changes in Health Utilization Cost Before and During Intervention - Amount Paid by Service Providers | -0.185 dollars/day | Standard Error 0.674 |
| Control | Changes in Health Utilization Cost Before and During Intervention - Amount Paid by Service Providers | -0.203 dollars/day | Standard Error 0.484 |
Comparison: A standard diffs-in-diffs model was utilized to estimate the causal effect of the intervention on medical costs per patient/day. The average change in cost over time for the intervention group was compared to that for the control group. The difference-in-differences values was assessed for significance at p = 0.05. Null hypothesis assumed no difference in the cost per patient per day from baseline to endpoint between groups.p-value: >0.05standard diffs-in-diffs model