Patient characteristics and intervention effectiveness of a diabetes management intervention program change with recruitment year. Those who were recruited later were less motivated and made fewer improvements in their intervention outcomes.
Keywords: Translational research, Intensive case management, Effectiveness, Recruitment
Abstract
In real-world settings, eligible populations and intervention effectiveness for a translational intervention likely vary across time. To determine the optimal strategies for effective large-scale implementation of evidence-based interventions, it is critical to investigate these potential variabilities. The purpose of this study is to evaluate whether patient characteristics and intervention effectiveness differed by year of enrollment in a multiyear evidence-based translational intervention. The Special Diabetes Program for Indians Healthy Heart (SDPI-HH) Demonstration Project is an intensive case management intervention designed to reduce cardiovascular disease risk among American Indians and Alaska Natives with diabetes. SDPI-HH participants recruited from 2006 through 2008 were included. Baseline characteristics were compared by year of enrollment. We also evaluated the differences in improvements in clinical and behavioral risk factors for cardiovascular disease among participants recruited in different years. The baseline characteristics of the three cohorts significantly differed in demographics, diabetes duration, health behaviors, level of motivation, and clinical measures. Improvements in 13 clinical and behavioral outcomes also differed by enrollment year with the 2006 cohort having the greatest number of significant improvements and the highest rates of participation and retention. Further investigation into the ways to modify the intensive case management model to address differences in levels of motivation and participation is warranted to improve the management of chronic disease in Indian health. Given the evolving nature of translational initiatives of this kind, our analysis results highlight the need to understand and adapt during the natural progression of health behavioral interventions.
Implications
Practice: To optimize participant retention and intervention effectiveness, practitioners of future translational projects of an evidence-based behavioral intervention may need to modify the intervention delivery approach in later enrollment years to address the differences in participants’ motivation, maintenance of the quality of intervention delivery, and other system or staff-level factors that may dilute effectiveness.
Policy: Policymakers who want to decrease the risk of cardiovascular disease among patients with diabetes should consider how their eligible population characteristics might change over time in large-scale implementation initiatives of the intensive case management model and provide advice on how to modify the delivery of care accordingly.
Research: Future research is needed into ways to modify the delivery of intensive case management intervention to address different levels of motivation and participation. Meanwhile, various delivery-setting factors that could impact sustained effectiveness and participant involvement, such as staff turnover, funding instability, and competing priorities should be assessed to sustain intervention effectiveness and participant involvement in future similar translational efforts.
INTRODUCTION
In the USA, the risk of developing diabetes among American Indians and Alaska Natives (AI/ANs) is 2.3 times that of non-Hispanic whites [1]. In addition, complications of diabetes such as heart disease, stroke, and kidney failure are among the leading causes of death among AI/ANs [2], with cardiovascular disease (CVD) now being the leading cause of mortality. Early detection and treatment of risk factors associated with these complications can reduce the risk of CVD development and progression [3].
In 1997, with increasing awareness of the severe diabetes epidemic in AI/AN populations, Congress established the Special Diabetes Program for Indians (SDPI), calling on the Indian Health Service (IHS) to direct funding for diabetes treatment and prevention. In 2003, Congress increased funding levels and directed the IHS to establish a competitive grant program. This program allowed IHS, tribal, and urban Indian health programs to compete for funding to participate in the development and implementation of two translational interventions: one for diabetes prevention among AI/ANs with prediabetes, the other for CVD risk reduction in those with diabetes. In late 2004, the successful grantee sites and key members of the IHS Division of Diabetes Treatment and Prevention began a collaborative process to develop and implement the demonstration projects.
The chronic care model, developed to make routine ambulatory care for patients with chronic diseases proactive [4], was chosen as the framework for the SDPI Healthy Heart Project (SDPI-HH, the CVD risk reduction arm of SDPI). In particular, using case management methods, SDPI-HH grantees decided upon a clinic and team-based approach that consisted of (a) assessment of each patient’s CVD risk and development of an individualized treatment plan; (b) disease management through physical exams and treatment goals for hypertension, dyslipidemia, smoking cessation, and hyperglycemia; and (c) self-management education based on the Honoring the Gift of Heart Health (HGHH) curriculum developed for addressing CVD risk among AI/ANs with diabetes [5]. The clinic-based intensive case management activities were based on what was then current national American Diabetes Association [6] and IHS evidence-based standards of care for patients with type 2 diabetes [7].
Since previous research had proven the efficacy of the intensive case management approaches [8–11], the SDPI-HH effort was considered translational in nature, where current evidence-based treatments were implemented on a large scale in 30 diverse healthcare settings; the effectiveness of the intervention was then assessed. In contrast to rigorously controlled randomized clinical trials in which patient populations are carefully defined, the evaluation of translational initiatives seeks to assess the effectiveness of interventions proven efficacious in diverse patient populations. Indeed, previous analyses of SDPI-HH data have shown significant improvements between baseline and a 1-year follow-up in the primary outcome variables of HbA1c, blood pressure, and lipid control among all participants recruited over a 3-year period, presenting evidence for the overall effectiveness of SDPI-HH [12].
In addition to an evaluation of overall effectiveness, the multiyear longitudinal structure of the SDPI-HH lends itself to evaluating changes over time. In real-world settings, the eligible participant populations likely vary across time; for example, more motivated and/or targeted patients may enroll early. Accordingly, these patients may have a higher level of readiness to change their behaviors and may perceive greater potential program benefit than patients enrolled in later years. Indeed, SDPI site staff reported observing changes in their participants’ motivation and characteristics over the first few years of recruitment. These observations suggest it is important to investigate the association of changes in participant motivation and characteristics with intervention outcomes. To determine the optimal strategies for effective large-scale implementation of evidence-based interventions, it is critical to investigate continued effectiveness as an intervention rolls out to large, diverse populations [13]. Only by understanding whether changes in patient characteristics occur over time, which may lead to changes in intervention effectiveness and suggest the need to consider correspondingly new approaches to motivate patients, we are able to fully address the challenges in scaling up successful interventions to regional, national, or international levels [13].
The goal of the current study is to compare the outcomes of intensive case management intervention among SDPI-HH participants recruited in different years (2006–2008). In particular, we examine baseline differences in target population, rates of participation and retention, and intervention effectiveness across different years of enrollment. These data provide a valuable opportunity to evaluate program implementation over time with findings that may inform future, similar large-scale translational intervention.
MATERIALS AND METHODS
Data collection
IHS funded 30 grant programs under SDPI-HH, consisting of 7 IHS, 21 tribal, and 2 urban Indian health programs. Most programs commenced enrollment in January 2006; the few participants (n = 50) who started at the end of 2005 are included in the 2006 cohort. Participants were recruited from each site’s diabetes registry, medical database, and/or community activities. Participants had to be AI/AN (based on eligibility to receive IHS or tribal services), have a previous or new diagnosis of diabetes, and be at least 18 years of age. Exclusion criteria included current pregnancy and end-stage renal disease on dialysis. Based on provider judgment, those with active alcohol or substance abuse, under current cancer treatment, or having unstable CVD were excluded.
Baseline assessments were typically obtained within 1 month before the initiation of intensive case management. Annual assessments were conducted to measure changes in clinical, behavioral, and psychosocial measurements. Each baseline or annual assessment included medical history regarding the date of diabetes diagnosis; physical measurements of height, weight, and waist circumference; clinical measurements of blood pressure and clinical laboratory test results of HbA1c and lipids; smoking status, aspirin use, and minutes of physical activity in the prior month. At each assessment, participants also completed a questionnaire to report psychosocial characteristics and dietary choices. During the annual assessments, participants reported whether they had participated in any HGHH classes in the last year. Throughout the project, case managers recorded the number of case management visits participants had attended. In this study, the baseline and first annual data from SDPI-HH participants enrolled in the first 3 years of the project (i.e., 2006–2008) were compared.
The SDPI-HH protocol was approved by the institutional review board of the University of Colorado Anschutz Medical Center and the National IHS institutional review board. When required, grantees obtained approval from other entities, such as tribal review boards, which oversaw their program. In addition, participants provided written informed consent and Health Insurance Portability and Accountability Act authorization.
Measures
Baseline demographic characteristics contrasted across the 3 years of enrollment included participants’ gender, age, educational attainment, employment status, marital status, and annual household income as reported in the baseline questionnaire. Clinical characteristics included years since diabetes diagnosis, BMI (body mass index), waist circumference, blood pressure, HbA1c, and lipids (high-density lipoprotein [HDL], low-density lipoprotein [LDL], and triglyceride) measured at clinical assessment. At baseline, case managers obtained medical history data on date of diabetes diagnosis. The number of years since diabetes diagnosis was determined by the number of years between baseline assessment date and the date of diagnoses. To evaluate whether a participant was newly diagnosed with diabetes at enrollment, a variable determining whether a participant had been diagnosed for a year or less at baseline was created.
Certain baseline behavioral characteristics were also compared among different cohorts, including smoking status, aspirin use, physical activity, dietary choices, and stages of change. Smoking status was defined as current smoker (yes/no). Aspirin use was defined as using aspirin daily or taking an anticoagulant (yes/no), whether no contraindication for use was reported. To assess the physical activity level for each participant, at baseline and annually, staff members asked participants to report the average minutes of physical activity per week in the previous month.
Details about the dietary choice variables are described elsewhere [14]. Briefly, participants were asked to recall the intake of 18 different types of foods over the last 30 days using a self-administered food frequency questionnaire. The frequency of each type of food was reported as (i) less than once a month, (ii) 1–3 times a month, (iii) about once a week, (iv) 2–3 times per week, (v) about once a day, and (vi) more than once a day. The healthy food score was constructed by averaging the intake frequency of 6 healthy foods, while the unhealthy food score was the mean intake frequency of 12 unhealthy foods.
Stages of change theory suggest that those who move from “precontemplation” to “action” stages will be more likely to make the changes necessary for better health outcomes [15–18]. To evaluate readiness to change behaviors among SDPI-HH participants, baseline stages of change for the following behaviors were compared: (i) Exercise. Asked about their plans for regular exercise (defined as 150 min per week of planned activities to increase physical fitness) [15]; (ii) Diet. Focusing upon the avoidance of high-fat foods [19]; and (iii) Weight Control. Assessed weight loss intent and activities [20].
Intervention outcomes of the SDPI-HH project were improvements in both clinical and behavioral risk factors for CVD between baseline and Year 1 annual assessments. Clinical outcomes included BMI, waist circumference, blood pressure, HbA1c, and lipids. Behavioral outcomes included aspirin use, smoking status, dietary choices, and physical activity. Goals for behavioral outcomes were set as smoking cessation, daily use of aspirin, and ≥150 min/week of physical activity. Treatment to target for clinical outcomes were defined as BMI <30 kg/m2, waist circumference <40 inches for males and <35 inches for females, blood pressure <130/80 mmHg, A1c <7%, HDL >40 mg/dl for males and >50 mg/dl for females, LDL <100 mg/dl, and triglycerides <150 mg/dl.
Program participation was measured by the mean number of case management visits during the first year among all participants who enrolled and by the percentage of participants completed a Year 1 assessment. Additional measures of participation are the mean number of case management visits and percentage of participants attending any HGHH classes among participants with a Year 1 assessment.
Statistical analyses
The three cohorts for comparison in this study were defined as those recruited in years 2006, 2007, and 2008, respectively. Baseline characteristics of the three cohorts were compared using chi-square tests for categorical variables and ANOVA for normally distributed continuous variables. For non-normally distributed continuous variables (years with diabetes, minutes of physical activity per week in the last month, and triglycerides), a ranked transformed analysis of variance was used [21]. For pairwise comparisons among the three cohorts, a Bonferroni adjustment of multiple comparisons was used to determine significant differences between a pair of means. All analyses were conducted in SAS version 9.3 (SAS Institute Inc., Cary, NC).
Linear mixed models were used to estimate mean changes for each of the normally distributed outcomes between baseline and the Year 1 assessment. Each mixed model included a binary time indicator variable (baseline vs. Year 1), a categorical cohort variable (2006, 2007, 2008), and an interaction between time and cohort as fixed effects. All models also included random intercepts at the individual level to model participant-level heterogeneity [22]. For highly skewed continuous variables (e.g., minutes of physical activity), a generalized linear mixed model with a Poisson distribution and a log link function was employed to estimate the rate ratio of Year 1 versus baseline. To evaluate changes in the proportions of participants obtaining treatment goals, a generalized linear mixed model with a binary distribution and a logit function was used. In all generalized linear mixed models, the same fixed effects were included as well as random residuals to model overdispersion. Estimate and contrast statements were included to estimate changes between baseline and Year 1 assessments in each cohort year and to evaluate differences in these changes between cohorts. Sensitivity analyses were conducted after adding age, gender, time since diabetes diagnosis, and psychological distress into the regression models. The main parameters of interest did not change much in these models. Therefore, we presented the results of the simpler models as the main findings of this study.
RESULTS
Baseline characteristics
By the end of 2008, 2910 diabetes patients had enrolled in the SDPI-HH and completed a baseline assessment. Table 1 shows the baseline differences among the three enrollment year cohorts. The number of enrolled participants decreased over the 3 years. Compared with latter cohorts, the first cohort (2006) had a greater percentage of females, older participants, and had higher incomes. They also had diabetes for a longer period of time, a smaller proportion of newly diagnosed participants, lower diastolic blood pressure, and mean HbA1c but higher triglycerides. Regarding behavioral characteristics, the 2006 cohort consumed healthy foods more frequently, unhealthy foods less often, reported fewer minutes of weekly physical activity, were more likely to be in the maintenance stages for exercise and diet, and were less likely to be current smokers. In addition, Table 1 reveals no baseline differences among cohort years in the percentages of participants with goal attainment for clinical and behavioral outcomes, although the 2008 cohort had a marginally higher percentage meeting the A1c goal at baseline (p = .0523).
Table 1.
Baseline demographics and clinical and behavioral characteristics by enrollment year
| Enrollment year (N = 2910) |
|||||
|---|---|---|---|---|---|
| 2006 n = 1123 |
2007 n = 985 |
2008 n = 802 |
|||
| Demographic characteristics | Category | n (%) | n (%) | n (%) | p Value |
| Gender | Male | 348 (31.0) | 353 (35.8) | 311 (38.8) | .0013 |
| Female | 775 (69.0) | 632 (64.2) | 491 (61.2) | ||
| Age categories (years) | 18 to <40 | 97 (8.6) | 132 (13.4) | 110 (13.7) | <.0001 |
| 40 to <50 | 232 (20.7) | 190 (19.3) | 202 (25.2) | ||
| 50 to <60 | 364 (32.4) | 329 (33.4) | 267 (33.3) | ||
| ≥60 | 430 (38.3) | 334 (33.9) | 223 (27.8) | ||
| Education | Post high school | 558 (55.8) | 495 (57.1) | 354 (54) | .4953 |
| 0–12 years | 442 (44.2) | 372 (42.9) | 301 (46) | ||
| Employment status | Employed | 514 (53.7) | 432 (53) | 306 (50.7) | .0810 |
| Retired | 195 (20.4) | 147 (18) | 103 (17.1) | ||
| Unemployed/student | 249 (26) | 236 (29) | 195 (32.3) | ||
| Marital status | Married or live together | 542 (59.6) | 432 (56.4) | 314 (55.3) | .1036 |
| Separated, divorced, or widowed | 258 (28.4) | 224 (29.2) | 158 (27.8) | ||
| Never married | 109 (12.0) | 110 (14.4) | 96 (16.9) | ||
| Annual household income | <15k | 227 (26.3) | 248 (33.4) | 198 (35.5) | .0058 |
| 15 to <30k | 245 (28.4) | 205 (27.6) | 151 (27.1) | ||
| 30 to <50k | 228 (26.5) | 175 (23.6) | 128 (23.0) | ||
| ≥50k | 162 (18.8) | 115 (15.5) | 80 (14.4) | ||
| Clinical characteristics | Mean (SD) | Mean (SD) | Mean (SD) | p Value | |
| Years since diabetes diagnosis Median (5th; 95th percentile) |
6.41 (0.35; 22.51)c | 5.89 (0.22; 20.75) | 5.89 (0.13; 21.28)a | .0028 | |
| Diabetes for less than a year | 120 (11.3%) | 134 (14.1%) | 148 (19.5%) | <.0001 | |
| BMI (kg/m2) | 36.69 (7.83) | 36.65 (8.01) | 36.55 (8.42) | .9270 | |
| Waist circumference (Inches) | 45.38 (6.50) | 45.74 (6.89) | 45.91 (6.97) | .2200 | |
| Systolic blood pressure (mm Hg) | 128.81 (16.49) | 128.74 (16.39) | 128.72 (16.45) | .9912 | |
| Diastolic blood pressure (mmHg) | 75.44 (9.88)c | 76.12 (10.19) | 76.82 (10.37)a | .0130 | |
| HbA1c (%) | 7.57 (1.75)b,c | 7.85 (2.00)a | 8.06 (2.14)a | <.0001 | |
| HDL (mg/dl) | 43.73 (11.88) | 43.92 (11.87) | 43.14 (11.18) | .3526 | |
| LDL (mg/dl) | 98.31 (31.51) | 99.74 (33.07) | 97.63 (32.9) | .3744 | |
| Triglycerides (mg/dl) Median (5th; 95th percentile) |
158 (74; 413)b | 151 (69; 398)a | 158 (71; 467) | .0420 | |
| Healthy diet score (range: 1–6) | 3.62 (0.79)b,c | 3.47 (0.81)a | 3.47 (0.80)a | <.0001 | |
| Unhealthy diet score (range: 1–6) | 2.58 (0.69)b,c | 2.66 (0.75)a | 2.71 (0.76)a | .0005 | |
| Physical activity (minutes per week) Median (5th; 95th percentile) |
52.5 (0; 345)b,c | 60.0 (0; 420)a | 60 (0; 450)a | .0045 | |
| 2006 n = 1123 |
2007 n = 985 |
2008 n = 802 |
p Value | ||
| Behavioral characteristics | n (%) | n (%) | n (%) | ||
| Physical activity goal (≥150 min per week) | 257 (23.2) | 275 (27.9) | 258 (32.3) | <.0001 | |
| Stages of change for exercise | Pre/Contemplation | 194 (19.6) | 171 (20.7) | 129 (20.1) | .0039 |
| Preparation | 364 (36.8) | 301 (36.4) | 250 (38.9) | ||
| Action | 180 (18.2) | 199 (24.1) | 137 (21.3) | ||
| Maintenance | 251 (25.4) | 155 (18.8) | 126 (19.6) | ||
| Stages of change for diet | Pre/Contemplation | 148 (15.7) | 137 (18.0) | 115 (19.2) | .0011 |
| Preparation | 222 (23.6) | 212 (27.8) | 153 (25.5) | ||
| Action | 240 (25.5) | 201 (26.4) | 179 (29.8) | ||
| Maintenance | 332 (35.2) | 212 (27.8) | 153 (25.5) | ||
| Stages of change for weight control | Pre/Contemplation | 235 (23.7) | 223 (26.8) | 165 (25.5) | .4252 |
| Action | 557 (56.2) | 465 (56.0) | 357 (55.3) | ||
| Maintenance | 199 (20.1) | 143 (17.2) | 124 (19.2) | ||
| Current smoker | 199 (17.7) | 203 (20.6) | 198 (24.7) | .0010 | |
| Aspirin use | 841 (78.0) | 724 (76.1) | 583 (75.1) | .3237 | |
| Clinical goal attainment at baseline | n (%) | n (%) | n (%) | p Value | |
| BMI goal (<30 kg/m2) | 204 (18.2) | 180 (18.3) | 163 (20.3) | .4318 | |
| Waist goal (<40 inches male, <35 inches female) | 72 (6.6) | 69 (7.1) | 65 (8.2) | .3800 | |
| Blood pressure goal (<130/80 mmHg) | 486 (43.3) | 420 (42.6) | 334 (41.8) | .7913 | |
| HbA1c goal (<7%) | 499 (44.4) | 408 (41.5) | 312 (39.0) | .0523 | |
| HDL goal (>40 mg/dl male, >50 mg/ dl female) | 367 (32.8) | 320 (32.6) | 273 (34.2) | .7355 | |
| LDL goal (<100 mg/dl) | 594 (54.6) | 509 (52.9) | 443 (57.7) | .1369 | |
| Triglycerides goal (<150 mg/dl) | 512 (45.7) | 481 (48.9) | 367 (45.8) | .2789 | |
Values are estimated differences from a linear mixed model or a generalized mixed model. BMI body mass index; HDL high-density lipoprotein; LDL low-density lipoprotein; SD standard deviation.
aSignificantly different from the mean of 2006 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
bSignificantly different from the mean of 2007 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
cSignificantly different from the mean of 2008 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
Intervention effectiveness
Table 2 presents improvements in clinical and behavioral outcomes by enrollment year for participants completing a Year 1 assessment. In general, the number of improvements in clinical and behavioral indicators decreased with each succeeding year of enrollment. Cohort 2006 achieved the highest rate of improvement with 11 out of 13 intervention outcomes statistically significant. The 2007 cohort obtained improvements in 10 outcomes while the 2008 cohort obtained improvement in only 7 outcomes. In 2006, only HDL and healthy diet choices did not significantly improve, while for the 2007 cohort no significant improvements were seen in systolic blood pressure, HDL, and minutes of physical activity. Unlike the earlier cohorts, the 2008 cohort did show a significant increase in mean HDL levels, but did not achieve significant improvements in BMI, waist circumference, systolic and diastolic blood pressure, triglycerides, or minutes of physical activity.
Table 2.
Changes in outcomes between baseline and Year 1 by enrollment year
| Enrollment year | ||||
|---|---|---|---|---|
| 2006 (n = 818) |
2007 (n = 701) |
2008 (n = 508) |
||
| Changes in clinical outcomes | ∆ Mean (SE) | ∆ Mean (SE) | ∆ Mean (SE) | Overall p value |
| BMI (kg/m2) | –0.21 (0.08)* | –0.18 (0.09)* | –0.16 (0.10) | .93 |
| Waist circumference (inches) | –0.29 (0.10)* | –0.44 (0.11)* | –0.22 (0.13) | .40 |
| HbA1c (%) | –0.14 (0.06)* | –0.18 (0.06)* | –0.31 (0.07)* | .17 |
| Systolic BP (mmHg) | –1.94 (0.62)* | –0.82 (0.67) | 0.23 (0.78) | .09 |
| Diastolic BP (mmHg) | –0.94 (0.38)* | –1.24 (0.41)* | –0.47 (0.47) | .46 |
| HDL (mg/dl) | 0.21 (0.29)c | –0.03 (0.31)c | 1.88 (0.36)*,a,b | .0001 |
| LDL (mg/dl) | –6.65 (1.06)* | –7.69 (1.14)* ,c | –3.32 (1.34)*,b | .04 |
| Triglycerides (mg/dl) (geometric mean and SE) | –1.08 (1.02)* | –1.05 (1.02)* | –1.03 (1.02) | .23 |
| Changes in behavioral outcomes | ||||
| Healthy diet score | 0.05 (0.03) | 0.11 (0.03)* | 0.10 (0.04)* | .32 |
| Unhealthy Diet Score | –0.15 (0.02)* | –0.17 (0.03)* | –0.13 (0.03)* | .75 |
| Physical activity (log estimate) | 0.43 (0.07)b,c | 0.09 (0.07)a | 0.07 (0.8)a | .0007 |
|
(rate ratio of Yr 1:baseline estimate ± 95%CI) |
1.54 (1.34, 1.77)* |
1.10 (0.95, 1.27) |
1.07 (0.91, 1.26) |
|
| ∆ % | ∆ % | ∆ % | ||
|---|---|---|---|---|
| Current smoker | –3.9%* | –2.9%* | –3.1%* | .37 |
| Aspirin use | 11.1%*,c | 12.3%*c | 7.6%*,a,b | .01 |
| Changes in goal attainment | ∆ % | ∆ % | ∆ % | |
| BMI goal (<30 kg/m2) | 1.2% | 1.9% | 1.8% | .88 |
| Waist goal (<40 inches men, <35 inches female) | 0.8% | 1.9% | 1.3% | .70 |
| Blood pressure goal (<130/80 mmHg) | 9.8%*,c | 4.0% | 1.1%a | .03 |
| HbA1c goal (<7%) | 5.3%* | 1.9% | 6.4%* | .21 |
| HDL goal (>40 mg/dl male, >50 mg/dl female) | –0.4% | 0.4% | 5.0%* | .08 |
| LDL goal (<100 mg/dl) | 10.6%* | 10.8%* | 4.2% | .05 |
| Triglycerides goal (<150 mg/dl) | 4.7%* | 5.5%* | 5.1%* | .95 |
| Physical activity goal (≥150 min/week) | 16.1%*,b,c | 6.6%*,a | 2.2%a | <.0001 |
Values are estimated differences from a linear mixed model or a generalized mixed model. BMI body mass index; HDL high-density lipoprotein; LDL low-density lipoprotein; SE standard error.
aSignificantly different from the mean of 2006 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
bSignificantly different from the mean of 2007 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
cSignificantly different from the mean of 2008 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
*p < .05 for test with H0: mean difference is equal to 0 (within-cohort changes).
The changes in achieving intervention goals also varied by enrollment year and, in many cases, were consistent with the mean improvements (Table 2). The 2006 cohort had the greatest number of significant improvements in targeted goals among the three cohorts. These participants achieved significant changes in the proportions of attaining intervention goals for five outcomes: blood pressure, HbA1c, LDL, triglycerides, and physical activity. The 2007 cohort only achieved significant changes in the percentages of goal attainment for three goals: LDL, triglycerides, and physical activity. The 2008 cohort also had significant changes in the proportions of obtaining three goals: HbA1c, HDL, and triglycerides.
Table 2 statistically contrasts differences in intervention outcomes by cohort as well. For most of these changes, the differences when compared across cohorts were not statistically significant. However, the magnitude of the 2008 cohort’s improvement in their mean HDL was greater than that of the other two cohorts (p = .0001). For LDL, the magnitude of change was similar in both 2006 and 2007 cohorts, but the reduction in 2007 cohort was greater than that in the 2008 cohort. Increase in physical activity in the 2006 cohort was significantly greater than those in the 2 succeeding years. In addition, 2006 cohort demonstrated greater success at Year 1 in attaining targeted goals than the latter two cohorts for physical activity and blood pressure, although there was no difference in the magnitude of change between 2006 and 2007.
Program participation
Table 3 presents data on participation over the 3 years of enrollment. The mean number of case management visits during the participant’s first year differed based on enrollment year. Cohort 2008 had significantly fewer case management visits than cohorts 2006 and 2007 (p < .0001). The percentage of participants completing their Year 1 assessment decreased from the 2006 to 2007 to 2008 cohorts. Among the participants who completed their Year 1 assessment, significant differences were found among the three cohorts, with the 2007 cohort receiving the most case management visits and the 2008 cohort the fewest. Further, the percentages of participants attending an HGHH class during the first year of participation significantly differed by enrollment year (p < .0001), with cohort 2006 showing the highest rates.
Table 3.
Measurements of participation by enrollment year
| Enrollment year | ||||
|---|---|---|---|---|
| Measurement of participation | 2006 | 2007 | 2008 | p Value |
| Total participants | 1123 | 985 | 802 | |
| Mean number of case management visits during 1st year, all participants (SD) | 6.98(3.29)c | 7.12(3.38)c | 6.03(3.29)a,b | <.0001 |
| Participants with Year 1 assessment n (%) | 818(72.8) | 701(71.2) | 508(63.3) | <.0001 |
| Mean number of case management visits during 1st year among participants with Year 1 assessment (SD) | 7.77(2.95)b,c | 8.20(2.96)a,c | 7.21(3.02)a,b | <.0001 |
| Participants attending any HGHH classes during 1st year among participants with Year 1 assessment n (%) | 613(75.0) | 403(57.5) | 268(53.5) | <.0001 |
p value for differences in means was determined by ANOVA; p value for differences in frequencies was determined by a chi-square test. SD standard deviation; HGHH Honoring the Gift of Heart Health.
aSignificantly different from the mean of 2006 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
bSignificantly different from the mean of 2007 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
cSignificantly different from the mean of 2008 cohort at α = 0.017 (based on Bonferroni correction for multiple comparisons).
DISCUSSION
SDPI-HH, as a translational intervention recruiting participants over several years, lends itself to the evaluation of baseline characteristics and continued effectiveness by cohort, defined here as year of enrollment. The results of this study demonstrate differences in cohort composition, degree of participation, and intervention effectiveness over time in this specific effort and suggest broader lessons for deploying translational initiatives of this nature.
Baseline participant characteristics differed across cohorts. Those enrolled in 2006 were older and had a longer duration of diabetes. A post hoc analysis, to aid in understanding these differences, showed that the first cohort had higher mean baseline diabetes knowledge and possibly a better understanding of the risk of CVD for diabetes patients. Therefore, they may be more motivated to make changes than those enrolled in the 2 latter years. This is demonstrated by the reduction, for subsequent cohorts, in the effectiveness on clinical and behavioral outcomes, and lower retention rates.
The 2006 cohort did not improve on two intervention outcomes: HDL and frequency of healthy food choices. The latter is due, perhaps, to the longer diabetes duration of this cohort, and, therefore, many had already changed to healthier food choices. This hypothesis is supported by the 2006 cohort reporting the highest value of baseline healthy diet score among the three cohorts. Cohorts 2007 and 2008, meanwhile, had greater numbers of newly diagnosed diabetes participants with lower baseline values for healthy food choices, with a resulting greater potential for improvement. Cohort 2006 also had the highest proportion of participants in the maintenance stage for stages of change in diet [20], a stage where the participants were expected to maintain their established healthy diet habits instead of achieving further improvements in their diet behaviors.
The lack of improvement in HDL was seen in cohort 2006 as well as 2007 even though both 2006 and 2007 cohorts had significant improvements in LDL and triglycerides levels. Cohort 2008 was the only cohort that obtained a significant mean improvement in HDL levels, along with a significant improvement in LDL levels but not triglycerides. The magnitude of LDL improvement in cohort 2008 was less than those achieved by the other two cohorts, however. The significant increase in HDL among 2008 participants could be due to HDL dysfunction in patients with poorly controlled or newly diagnosed diabetes [23]. Supporting this, the 2008 cohort also had the greatest level of HbA1c improvement; HbA1c reduction has been reported to improve HDL, but not LDL [24].
Participants in the 2008 cohort were recruited during the last year of the demonstration project funding and achieved the fewest improvements in outcomes, even while their baseline characteristics indicated considerable need. This lack of performance correlates with reduction in program participation. This may indicate a decrease in enthusiasm for the program, as funding came to an end, from not only the participants but, perhaps, also the staff. Given the insecurity of further funding in 2008, fewer participants may have been enticed to enroll. Additional factors could have affected the number of case management sessions attended, participation in the HGHH classes, enrollment, and retention rates. Programs could have initially targeted participants who were highly motivated during the implementation of the program hoping to achieve initial results that were valuable to stakeholders, such as IHS, and their healthcare providers, tribes, and communities, in part to garner support and to maximize the probability of continued funding. On grantee annual recruitment and retention reports, staff members from SDPI-HH grantee sites indicated staffing issues, such as staff turnover and insufficient staffing levels, were endemic. With the increase in the number of participants annually requiring case management (participants were encouraged to remain in the program while funding lasted), in addition to recruiting and retention efforts, staffing issues could have led to a reduction in participation over time. Moreover, the primary reasons for lack of patient participation in the program, as reported by staff, were lack of time for the visits, transportation, and other competing commitments. Although the differences may be most apparent in the 2008 cohort, the trends across time are clear and suggest a complex interaction of system and funding issues in the face of higher participant needs concomitant with lower effectiveness. A model incorporating in-person case management with current electronic methods of communication and a more permanent and self-sustained form of funding may help maintain participation and adequate staffing.
All these findings have critical implications for future implementation policy decision making for large-scale translation of evidence-based interventions. For example, making case management reimbursable by health insurance could potentially increase staff stability, patient participation, and continued intervention effectiveness of a multiyear program. Furthermore, by demonstrating the importance of assessing changes over time in baseline characteristics, intervention effectiveness, and program participation, this study informs future revisions of the intervention to maximize continued effectiveness of an evidence-based intervention as it rolls out to real-world settings. Unlike randomized controlled trials where concerns about internal validity are paramount, external validity comes to the fore in translational interventions. Thus, as suggested by the results of the current study, a critical step for translational initiatives should be to develop data-driven approaches for intervention adaptation to address changes in patient, staff, and system characteristics over time to maintain similar levels of effectiveness and program participation. Such modifications must be thoughtful, of course, and attend to both internal and external validity concerns. Further, as always, such adaptations should also be designed in concert with the programs and patients.
Limitations
Important limitations deserve acknowledgment. Voluntary participation in the intervention may have resulted in a self-selection bias that limits the generalizability of these findings. Given that about 30% of the participants did not complete the first annual assessment, mixed models were used to provide unbiased estimates of the intervention effectiveness under the assumption of missing-at-random. This assumption cannot be easily assessed; thus, the lack of bias in our estimates may not be guaranteed, which could further limit the generalizability of our results [25]. Since the intervention activities in this initiative were based on previously demonstrated efficacy and current standards of care, an evaluation of the effectiveness of the intervention in real-world settings was deemed more appropriate than a design to document effects by comparison to a control group. Yet, without a control group, we cannot exclude the possibility that the improvements achieved by SDPI-HH participants were due to secular trends in usual care beyond the project instead of the SDPI-HH intervention. Lastly, provider and staffing issues over the 3 years may have contributed to the decline in participant recruitment, retention, and intervention outcomes. However, the lack of quality data on provider and staff characteristics precludes us from rigorously evaluating this association which needs to be taken into consideration when designing future translational initiatives.
CONCLUSIONS
The results from this study demonstrate that the changes in baseline characteristics of SDPI-HH participants as time progressed, along with changes in staff composition, staff enthusiasm, and funding status, might have altered the effectiveness of the intervention. The changes in participant characteristics could be due to different participant availability for recruitment, with those at the start of the intervention being more motivated given their longer diabetes duration. Thus, it is critical to remain alert to the need for different strategies in subsequent years as well as being open to modifying the health behavioral models that underlie the intervention to accommodate potentially shifting characteristics of the target population. For instance, efforts to meaningfully target participant readiness to change should be considered for latter cohorts. Varying the method of delivering the intervention may also warrant further attention. Intensive case management has been reported to improve response to intervention, but may be impeded by high costs and the lack of availability of interested participants due to the level of commitment [26, 27]. An intervention that combines in-person case management and internet-based, peer-led programs or telephone technology to motivate, deliver, and monitor participants could reduce the burden on participants and staff, allowing more patients to benefit from the program. In addition, future studies investigating the impact of delivery-setting factors such as staff turnover, funding stability, and participants’ competing priorities are critically needed to ensure continued provider and staff enthusiasm for the project and sustained fidelity of the intervention delivery. Lastly, investigations into how to modify a program to improve reach, motivation, and maintenance of a larger number of patients with diabetes are warranted to maximize the success of similar interventions for diabetes management among minority populations and to inform policy.
Acknowledgments
We would like to express our gratitude to the Indian Health Service as well as tribal and urban Indian health programs and participants involved in the Special Diabetes Program for Indians Healthy Heart (SDPI-HH) Project. Funding for the SDPI-HH project was provided by the Indian Health Service (HHSI242200400049C, S.M. Manson). Manuscript preparation was supported in part by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (1P30DK092923, S.M. Manson and R21-DK-108187, L. Jiang).
Footnotes
Grantees participating in the Special Diabetes Program for Indians Healthy Heart Project: Absentee Shawnee Tribe of Oklahoma; Albuquerque Service Unit; Bad River Band of Lake Superior Chippewa; Blackfeet Tribe; Choctaw Nation of Oklahoma; Confederated Salish and Kootenai Tribes; Montana/Wyoming Tribal Consortium with Assiniboine & Gros-Ventre, Chippewa Cree Tribe, and Crow Nation; Hualapai Tribe; Indian Health Care Resource Center of Tulsa, Inc. in consortium with Northeastern Tribal Health System Miami Service Unit; Indian Health Council, Inc.; Leech Lake Reservation Tribal Council; Mille Lacs Band of Ojibwe in consortium with St. Croix Chippewa Indians of Wisconsin; Muscogee Creek Nation Health System; Navajo Area Indian Health Service with Northern Navajo Medical Center and Inscription House Clinic; Northwest Washington Indian Health Board with Lummi Indian Nation, Nooksack Tribe of Indians, Swinomish Tribal Community, and Upper Skagit Indian Tribe; Ramah Navajo School Board, Inc.; Redding Rancheria Indian Health Clinic; Hoopa Valley Tribe; Riverside-San Bernardino County Indian Health, Inc.; Santo Domingo Tribe; Sault Ste Marie Tribe Chippewa; Seattle Indian Health Board; St. Regis Mohawk Health Services; Taos-Picuris Service Unit; Tohono O’Odham Healthy Heart Project; Toiyabe Indian Health Project, Inc.; Uintah & Ouray Indian Health Service (IHS) Clinic; Wagner Health Care Center IHS; Whiteriver IHS Unit; Yakama Indian Health Center—IHS/DHHS; Yukon-Kuskokwim Health Corporation.
Compliance with Ethical Standards
Primary Data: The authors have full control of all primary data, and they agree to allow the journal to review their data if requested.
Conflict of Interest: No authors have any competing interests or conflicts of interest.
Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The SDPI-HH protocol was approved by the institutional review board of the University of Colorado Anschutz Medical Center and the National IHS institutional review board. When required, grantees obtained approval from other entities, such as tribal review boards, which oversaw their program. This article does not contain any study with animals performed by any of the authors.
Authors’ Contributions: All authors have approved this manuscript for publication, including all grantees participating in the Special Diabetes Program for Indians Healthy Heart Project. KAP analyzed, interpreted the data, drafted the manuscript, and revised the manuscript for further revisions and publication. JB participated in the design of the SDPI-HH project, contributed to the discussion of the analyses, reviewed, and edited the manuscript. AJ participated in the interpretation of analysis results and revised the manuscript critically for important intellectual content. AB participated in the conceptualization and design of the SDPI-HH project, contributed to discussion, reviewed, and edited the manuscript. SMM conceptualized and designed the SDPI-HH project, contributed to the discussion, reviewed, and edited the manuscript. LJ designed the study, researched the data, contributed to the discussion, wrote, reviewed, and edited the manuscript. All authors read and approved the final manuscript.
Informed Consent: Informed consent was obtained from all individual participant included in this study. In addition, participants signed a Health Insurance Portability and Accountability Act authorization.
References
- 1. Indian Health Service. Diabetes in American Indians and Alaska natives: facts at-a-glance Department of Health and Human Services, Indian Health Service Division of Diabetes Treatment and Prevention; 2012; Available at https://www.ihs.gov/MedicalPrograms/Diabetes/HomeDocs/Resources/FactSheets/Fact_sheet_AIAN_508c.pdf. Accessibility verified June 17, 2018. [Google Scholar]
- 2. Espey DK, Jim MA, Cobb N, et al. . Leading causes of death and all-cause mortality in American Indians and Alaska Natives. Am J Public Health. 2014;104(suppl 3):S303–S311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Sakao S, Voelkel NF, Tatsumi K. The vascular bed in COPD: Pulmonary hypertension and pulmonary vascular alterations. Eur Respir Rev. 2014;23(133):350–355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Coleman K, Austin BT, Brach C, Wagner EH. Evidence on the chronic care model in the new millennium. Health Aff (Millwood). 2009;28(1):75–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. National Heart, Lung, and Blood Institute (NHLBI). Honoring the Gift of Heart Health. A Heart Health Educator’s Manual. Department of Health and Human Services, National Heart, Lung and Blood Institute, National Institutes of Health and Indian Health Service, NIH Publication Number 03-5218; 2003. Available at https://www.nhlbi.nih.gov/files/docs/resources/heart/ai_manual.pdf. Accessibility verified June 17, 2018. [Google Scholar]
- 6. American standards of medical care in diabetes. Diabetes Care. 2005;28(suppl 1):s4–s36. [PubMed] [Google Scholar]
- 7. Indian Health service. Standards of Care for Adults with Type 2 Diabetes. Indian Health Service, Division of Diabetes Treatment and Prevention, Department of Health and Human Services; 2006. Available at https://www.ihs.gov/diabetes/includes/themes/responsive2017/display_objects/documents/clinicaldocs/Recom_AtaGlance_508c.pdf. Accessibility verified June 17, 2018. [Google Scholar]
- 8. California Medi-Cal Type 2 Diabetes Study Group. Closing the gap: Effect of diabetes case management on glycemic control among low-income ethnic minority populations: The California Medi-Cal type 2 diabetes study. Diabetes Care. 2004;27(1):95–103. [DOI] [PubMed] [Google Scholar]
- 9. Gilmer TP, Philis-Tsimikas A, Walker C. Outcomes of project Dulce: A culturally specific diabetes management program. Ann Pharmacother. 2005;39(5):817–822. [DOI] [PubMed] [Google Scholar]
- 10. Norris SL, Nichols PJ, Caspersen CJ, et al. . The effectiveness of disease and case management for people with diabetes. A systematic review. Am J Prev Med. 2002;22(suppl 4):15–38. [DOI] [PubMed] [Google Scholar]
- 11. Cook CB, Ziemer DC, El-Kebbi IM, et al. . Diabetes in urban African-Americans. XVI. Overcoming clinical inertia improves glycemic control in patients with type 2 diabetes. Diabetes Care. 1999;22(9):1494–1500. [DOI] [PubMed] [Google Scholar]
- 12. Moore K, Jiang L, Manson SM, et al. . Case management to reduce cardiovascular disease risk in American Indians and Alaska natives with diabetes: Results from the special diabetes program for Indians healthy heart demonstration project. Am J Public Health. 2014;104(11):e158–e164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Glasgow RE, Vinson C, Chambers D, Khoury MJ, Kaplan RM, Hunter C. National Institutes of Health approaches to dissemination and implementation science: Current and future directions. Am J Public Health. 2012;102(7):1274–1281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Teufel-Shone NI, Jiang L, Beals J, et al. . Demographic characteristics and food choices of participants in the special diabetes program for American Indians diabetes prevention demonstration project. Ethn Health. 2015;20(4):327–340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Marcus BH, Banspach SW, Lefebvre RC, Rossi JS, Carleton RA, Abrams DB. Using the stages of change model to increase the adoption of physical activity among community participants. Am J Health Promot. 1992;6(6):424–429. [DOI] [PubMed] [Google Scholar]
- 16. Marcus BH, Selby VC, Niaura RS, Rossi JS. Self-efficacy and the stages of exercise behavior change. Res Q Exerc Sport. 1992;63(1):60–66. [DOI] [PubMed] [Google Scholar]
- 17. Velicer WF, Fava JL, Prochaska JO, Abrams DB, Emmons KM, Pierce JP. Distribution of smokers by stage in three representative samples. Prev Med. 1995;24(4):401–411. [DOI] [PubMed] [Google Scholar]
- 18. Velicer WF, Hughes SL, Fava JL, Prochaska JO, DiClemente CC. An empirical typology of subjects within stage of change. Addict Behav. 1995;20(3):299–320. [DOI] [PubMed] [Google Scholar]
- 19. Greene GW, Rossi SR. Stages of change for reducing dietary fat intake over 18 months. J Am Diet Assoc. 1998;98(5):529–534. [DOI] [PubMed] [Google Scholar]
- 20. The University of Rhode Island CPRC. Weight control: stages of change (short form). Available at https://web.uri.edu/cprc/weight-control-stages-of-change-short-form/. Accessibility verified June 17, 2018.
- 21. Conover WJ, Iman RL. Analysis of covariance using the rank transformation. Biometrics. 1982;38(3):715–724. [PubMed] [Google Scholar]
- 22. Verbeke G, Molenberghs G.. Linear Mixed Models for Longitudinal Data. New York, NY: Springer; 2000. [Google Scholar]
- 23. Farbstein D, Levy AP. HDL dysfunction in diabetes: Causes and possible treatments. Expert Rev Cardiovasc Ther. 2012;10(3):353–361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Barbarossa G, Renzi A, D’Erasmo L, et al. . The relation between glycemic control and HDL-C in type 2 diabetes: A preliminary step forward?Diabetes Res Clin Pract. 2014;104(1):e26–e28. [DOI] [PubMed] [Google Scholar]
- 25. Fairclough DL. Design and Analysis of Quality of Life Studies in Clinical Trials: Interdisciplinary Statistics. Boca Raton, FL: Chapman & Hall/CRC; 2002. [Google Scholar]
- 26. Joo JY, Huber DL. An integrative review of case management for diabetes. Prof Case Manag. 2012;17(2):72–85. [DOI] [PubMed] [Google Scholar]
- 27. Gary TL, Hill-Briggs F, Batts-Turner M, Brancati FL. Translational research principles of an effectiveness trial for diabetes care in an urban African American population. Diabetes Educ. 2005;31(6):880–889. [DOI] [PubMed] [Google Scholar]