A patient-centered comparative effectiveness research study of culturally appropriate options for diabetes self-management

doi:10.21203/rs.3.rs-3466636/v1

Download PDF

Research Article

A patient-centered comparative effectiveness research study of culturally appropriate options for diabetes self-management

https://doi.org/10.21203/rs.3.rs-3466636/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

This project compared the effectiveness of two evidence-based models of culturally competent diabetes health promotion: The Diabetes Self-Management Support Empowerment Model (DSMS), and The Chronic Care Model (CCM). Our primary outcome was improvement in patient capacity for diabetes self-management as measured by the Diabetes Knowledge Questionnaire (DKQ) and the Patient Activation Measure (PAM). Our secondary outcome was patient success at diabetes self-management as measured by lower A1c values and depression scores using the PHQ-9. We also gathered data on the cultural competence of the program using the Consumer Assessment of Healthcare Providers and Systems Cultural Competence Set (CAHPS-CC). We compared patient outcomes in two sites in Albuquerque, New Mexico, that serve a large population of Latino diabetes patients from low-income households. Participants were enrolled as dyads—a patient participant (n=226) and a social support participant (n=226). Outcomes over time and by program were analyzed using longitudinal linear mixed modeling, adjusted for patient participant demographic characteristics and other potential confounding covariates. Secondary outcomes were also adjusted for potential confounders. Interactions with both time and program helped to assess outcomes. This study did not find a difference between the two sites with respect to the primary outcome measures and only one of the three secondary outcomes showed differential results. The main difference between programs was that depression decreased more for CCM than for DSMS. An exploratory, subgroup analysis revealed that at CCM, patient participants with a very high A1c (>10) demonstrated a clinically meaningful decrease in A1c. However, given the higher cultural competence rating for the CCM, statistically significant improvement in depression, and the importance of social support to the patients, results suggest that a culturally and contextually situated diabetes self-management and education program design may deliver benefit for patients, especially for those patients with high A1c levels.

Endocrinology & Metabolism

Community-engaged research

patient-engaged research

Comparative effectiveness

Diabetes self-management

Depression

A1c levels

Patient activation

Cultural competence

Hispanic

Latino

LatinX

Social determinants of health

Health disparities

Health equity

Diabetes knowledge

Chronic care model

Social support

Low-income

Overview

Here we report results from a non-randomized, pragmatic, quasi-experimental, patient-engaged, comparative effectiveness research (CER) study [1] conducted by researchers at the University of New Mexico (UNM) comparing two distinct evidence-based models for culturally competent diabetes self-management and education programming. We compared outcomes at two sites in Albuquerque, New Mexico, serving a large Latinx patient population from low-income households. The overall study protocol is described in detail elsewhere [1]. Data collection involved interviews, focus groups, surveys, and testing of patient participants for A1c and depression. Survey responses, blood samples, and height/weight measurements were gathered at four time points (baseline, 3-months, 6-months, and 12-months). The UNM Human Research Review Committee/Institutional Review Board (HRRC/IRB # 16-303) approved all aspects of the research protocol.

Models for comparison

Local context for each model

The two models being compared were the Diabetes Self-Management Support Empowerment Model (DSMS) and [2] the Chronic Care Model (CCM) [3,4]. Each program serves a large population of Latinx patients from low-income households in Albuquerque, New Mexico, and employs a distinct evidence-based approach to create program cultural competence. Patient participants were recruited from both programs. One program, the Center for Diabetes Education at the University of New Mexico Hospital (CDE-UNMH), is based at a university hospital and uses the DSMS [2]. The other, One Hope, is based at a community clinic operated by a faith-based nonprofit and uses the CCM approach [3,4].

The Diabetes Self-Management Support Empowerment Model (DSMS)

The DSMS is a patient-centered, theoretically based educational framework that follows National Standards for Diabetes Self-Management Education [2], is certified by the American Diabetes Association [5], and is accredited by the American Association of Diabetes Educators (AADE) [5]. The DSMS combines a series of clinically informed group didactic sessions that use a patient self-determination approach to empower patients to take control of their own diabetes health with follow-up support to sustain self-management gains achieved during the sessions.

The Chronic Care Model (CCM)

The CCM is “a systematic approach to restructuring medical care to create partnerships between health systems and communities” [3,4] by addressing not only the medical but also the cultural and linguistic needs of patients through the inclusion of cultural competence in the delivery system design [3]. To create a holistic care regime, the CCM focuses on addressing social determinants of health by meeting patients' medical, cultural, and linguistic needs through integrating cultural norms and social relationships from the patient population into program design [6].

Measurements for Outcomes

Improved patient capacity for diabetes self-management

Capacity for diabetes self-management was measured using the Diabetes Knowledge Questionnaire (DKQ) and the Patient Activation Measure (PAM).

Diabetes knowledge was measured using the DKQ summed score. We hypothesized that the Chronic Care Model (CCM) model would result in a larger increase in DKQ summed scores from baseline to 6 months as compared to the Diabetes Self-Management Support Empowerment Model DSMS. Previously published studies evaluating culturally competent diabetes management programs report meaningful changes in DKQ summed scores with Cohen’s f effect sizes of 0.03 to 0.16 in studies ranging in sample sizes per arm from 10 to 189 [7–10].
Patient activation was measured using the PAM-10 raw score. We hypothesized that theCCM model would result in a larger increase in PAM-10 raw scores from baseline to 6 months as compared to the DSMS model. Previously published studies evaluating culturally competent diabetes management programs report changes in PAM-10 raw scores with meaningful Cohen’s f effect sizes of 0.01 to 0.16 in studies ranging in sample size per arm from 26 to 133 (per Shah, Co-I Burge, and colleagues) [11–15].

Improvement in A1c and PHQ-9

To measure physiological responses to out two intervention models, we used A1c and a depression scale score from the patient health questionnaire (PHQ-9) [16–18].

Participants

Dyadic enrollment design

We enrolled participants as dyads, where half were patients diagnosed with diabetes or pre-diabetes (“patient participants”) and half were individuals identified by the patient participant as someone in their lives who provides them with significant social support (“social support participants”). Including patient-social support dyads was a mechanism for incorporating the social dimension of patient’s lives in our research. Patient participants provided data related to their own health or experience. Social support participants primarily provided data on their perspective on their patient partner's health.

Patient participants (PPs)

Patient participants were individuals entering one of the two programs during the period of the study either because they were newly diagnosed with diabetes or pre/diabetes, or because they were having trouble managing their blood sugar levels. Patient participants were adults (≥18 years old), self-identified as Latinx(a), self-reported household income below 250% of the FPL, and were able to identify a social support who was willing to participate in the study.

Social support participants (SSPs)

The only requirements for social support participants were that they had to be adults (≥18 years old) and had to be willing to participate in the research. The social support did not have to be an actual “caregiver” to the patient, nor were they screened for ethnicity or income in order to participate.

Data collection

Each participant was enrolled in the study for 12 months. At baseline, all participants were asked demographic questions. At all four time points, we asked questions from validated survey instruments.

Diabetes Knowledge Questionnaire (DKQ) [7–10].
Patient Activation Measure 10 (PAM-10) [19–27].
Patient Health Questionnaire (PHQ-9) [16–18].
Consumer Assessment of Healthcare Providers and Systems Cultural Competence Set (CAHPS-CC).

At all four time points, we gathered blood samples from patient participants only. We also conducted interviews and focus groups with participants and providers.

Qualitative Analysis

We conducted a rigorous, disciplined, empirical analysis of qualitative data using Hammersley’scriteria for qualitative research based on plausibility, credibility, and relevance [28]. We conducted a theory-driven qualitative content analysis according to standards developed by Gläser and Laudel]. Three members of the research team (two bilingual) read through transcripts to identify conceptual categories and patterns related to specified domains of inquiry, and created a qualitative codebook. We explored interconnections between theme categories and developed a holistic interpretation of the data (“constant comparison”).

Quantitative Analysis

Sample size and Power Calculations

Our goal was to recruit N=240 patient-social support pairs (N=120 per site) in order to obtain complete data on N=96 pairs per site, assuming a 20% attrition rate. Comparing response changes on the DKQ, PAM-10, and PHQ-9 from baseline to 6 months between the CCM to the DSMS, the two-sided Type I error rate was adjusted for the number of comparisons made using a Bonferroni correction (α=0.025). The power analyses for detecting site differences among change scores were based on multiple linear regression models including demographic characteristics, participants’ perceived cultural competence of providers (CAHPS-CC), and social supports’ change scores on the DKQ, PAM-10, and PHQ-9 as covariates. We report Cohen’s f effect sizes based on the regression method, where Cohen’s standards for “small”, “medium”, and “large” effects are 0.02, 0.15, and 0.35, respectively [29,30].

The power for the primary endpoints with n=96 per site and α=0.025 for comparing the CCM to the DSMS was as follows

Change in DKQ summed score: Δ_CCM-DSCS= 2.2 (SD = 3.8), power = 96%, Cohen’s f effect size (ES)=0.09
Change in PAM-10 raw score: Δ_CCM-DSCS= 12.7 (SD = 24.8), power = 85%, Cohen’s f ES=0.07

The power for the secondary endpoints with n=96 per site and α=0.017 for comparing the CCM to the DSMS was as follows

Change in A1c: Δ_CCM-DSCS= -0.5 (SD=1.0), power = 84%, Cohen’s f ES=0.06.
Change in depression scores (PHQ-9): Δ_CCM-DSCS= -3 (SD=6), power = 84% Cohen’s f ES = 0.06.

Statistical analysis

Descriptive statistics were calculated to summarize PP characteristics. Medians and interquartile ranges (IQR) were calculated for continuous variables and were compared across sites by Kruskal-Wallis test. Frequencies and percentages were calculated for categorical variables and were compared with the chi-square test. Significant differences are noted.

Mean outcomes for Aims 1 and 2 over time and by program were analyzed using longitudinal linear mixed modeling [31,32] to account for the repeated measure effects, as well as to adjust for patient participant demographic characteristics. The interaction between time and program model was of primary interest to assess whether each outcome changed to different extents over time by program. Covariates that were excluded because of their strong relationship with site included primary language and type of insurance. The models used an unstructured covariance over time. The full models were fit and then reduced with backward model selection using conditional Akaike information criterion (cAIC)] Model fit assumptions on the residuals were equal variance and normality, which were both assessed visually. However, results were robust to violations of model distributional assumption. Analyses were performed in R 4.1.0. The restricted maximum-likelihood (REML) adjusted least-squares mean difference estimate of the outcome between programs from baseline to 6 months is reported along with its 95% confidence interval. The difference estimate from baseline to 12 months is also reported.

We conducted an analysisto assess patient participant capacity for diabetes self-management, our primary patient-reported outcome, by measuring diabetes knowledge using the DKQ and patient activation using the PAM-10. Descriptive statistics including means, standard deviations, medians, and quartiles were calculated for each outcome measure. Diabetes self-management program models were compared for the CCM vs the DSMS. P-values were compared to a Bonferroni-corrected α = 0.025 to account for multiple comparisons (two primary outcome measures). The primary outcome analyses were:

Improvement in diabetes knowledge: Primary analysis used the DKQ summed score.
Improvement in diabetes-related “patient activation”: Primary analysis used the PAM-10 raw score. We also converted raw scores to scaled scores per a proprietary algorithm [33] and then grouped them into patient activation levels which are displayed descriptively (frequencies and percentages) by diabetes self-management program model.

For our secondary outcome, successful patient participant management of their diabetes, we measured their A1c from blood samples drawn and PHQ-9 depression scores. The modeling used the longitudinal mixed model with the additional covariates of DKQ and PAM-10 for all three secondary outcomes and also PHQ-9 for A1c. Difference estimates are reported along with their 98.3% confidence intervals.

As part of our characterization of programmatic interface with patient participant culture and socio-economic circumstances(Aim #3), we summarized patient participant and social support participant scores on five of the eight subscales of the Consumer Assessment of Healthcare Providers and Systems Cultural Competence Set (CAHPS-CC) (Domains A. and B., F., G., and H.) aggregated over the three follow-up timepoints and treated as a fixed covariate to assess the overall program experience. Medians and quartiles of the subscales by diabetes self-management program model were reported. Social support participant scores were combined with the patient participant scores and used as covariates in the analysis for the quantitative primary and secondary outcome measures.

Sample

We enrolled 452 (226 dyads of patient participants and social support participants) in the study: 120 dyads from the DSMS and 106 from the CCM). The CONSORT diagram shows the study flow from eligibility through the 12-month assessment (Fig. S 1).

Descriptive Baseline Statistics

Physiological characteristics

Compared to patients in the DSMS, patients in the CCM were less likely to have diabetes (48.0% vs 66.1%, p = 0.011) compared to pre-diabetes, had a sigficantly shorter median time since diagnosis of their current diabetes category (1.0 vs 2.0, p = 0.028), roughly the same median A1c (6.5 vs 6.7, p = 0.347) with 33.3% of those with diabetes having A1c over 10, roughly the same likelihood of the A1c category “diabetes” (53.8% vs 60.0%, p = 0.636).

Capacity for diabetes self-management

All patient participants (PP) generally scored low on diabetes knowledge (DKQ), with an average score of 14/23, and compared to DSMS, PPs in CCM scored lower by two points (13 vs 15, p < 0.0005) (Table S 2b). However, these values were consistent with baseline assessments of similar populations. The average PAM-10 score was 77.3, with 96% scoring at Level 3 (taking action) or 4 (maintaining behaviors/pushing further) (Table S 2b). Compared to DSMS, PPs in CCM scored higher by four points (p = 0.018) with more PPs at Level 4 (72.6% vs 55.0%) and with fewer at Level 1 (disengaged/overwhelmed) and Level 2 (becoming aware, but struggling) (1.8% vs 5.8%).

Depression

More than half (52.2%) of the patient participants ranked as having symptoms of depression, with 22.2% in the “Moderate,” “Moderately Severe,” or “Severe” categories (Table S 3). More than half in either program scored as having mild to severe depression although a higher percentage of subjects in DSMS tested for moderately severe to severe depression (11.7%) compared to CCM (3.8%, p = 0.057).

Qualitative Analysis

Themes from interviews and focus groups

During Year 1, we conducted 11 provider interviews—five at CCM, six at CDE. During Years 1 and 2, we conducted eight focus groups (2/site/year) with a mix of 36 patient participants and social support participants, and 44 interviews with 23 unique patient participants (12 CCM, 11 CDE) and 21 unique social support participants (12 CCM, 9 CDE). For Aim #1, participant interviews and focus groups informed us about participant knowledge, activation, and motivation, and the barriers to both. For Aim #2, we learned about challenges to reducing A1c, making lifestyle changes, and mental health challenges related to reducing stress and depression.

Participants at both sites discussed the importance of social support in helping them deal with and manage their diabetes. Social support provided by the programs was identified as very important.

Quantitative Analysis

Aim #1: Primary outcomes

The primary analysis for Aim #1 was to measure and compare improvement in patient participant capacity for diabetes self-management as indicated by improvement at 6 months in diabetes knowledge and patient activation. For each outcome variable the resulting model is interpreted for model fit, time-by-program interaction, main effects not associated with interactions, and interactions (Fig S 2A). Model fit assumptions were not violated.

Changes in diabetes knowledge

Scores on the PP DKQ-23 were used to measure diabetes knowledge. Scores increased for patient participants in both programs. Both sites increased their knowledge from baseline to 3 months (approximately 0.75 to 1 point) and sustained their improvement through the end of the study. There was no Time-by-Program interaction (Fig S 2C, p-value = 0.265), so both programs changed to similar extents over time; in particular, the Baseline-to-6-month change for CCM DKQ-23 (a 23-point scale) compared to the change for DSMS was 0.483, 95% CI (-0.119, 1.086). DSMS scored approximately 1.25 points higher than CCM (Fig S 2B, p-value = 0.0009). The multivariate analysis (Table S 10, contrasts not shown) revealed that the PP DKQ-23 Score was positively related to SSP DKQ-23 Score (regression coefficient (coef) = 0.198, 95% CI: (0.137, 0.258)).

Patient activation

We used the PAM-10 as a proxy for measuring the patient’s capacity for self-managing their diabetes, conceptualized as patient activation. The average PAM-10 score was 77.3, with the vast majority scoring at Level 3 (taking action) or 4 (maintaining behaviors and pushing further). Patients in the CCM compared to patients in the DSMS scored higher by four points with more patients at Level 4 (72.6% vs 55.0%) and with fewer at Levels 1 and 2 (1.8% vs 5.8%) (Table S 2).

We used the PAM-10 numeric score because it is more sensitive to detecting differences than rounding the scores into four activation categories. PAM-10 scores increased for PPs in both programs (Fig S 3A). Model fit assumptions were not violated. There was no Time-by-Program interaction (Fig S 3C, p-value = 0.612), so both programs changed to similar extents over time; in particular, the Baseline-to-6-month change for CCM compared to the change for DSMS was − 1.278, 95% CI (-4.56, 2.01). DSMS PAM scores (scale from 0 to 100) were close to 77 at baseline and CCM was close to 78 at baseline, but both increased to about 82 at 3 months and beyond (Fig S 3B). Both sites increased their activation scores by 6 months (p-value < 0.0001). Patient activation (PAM-10) was positively related to social support for DSMS (coef = 0.172, 95% CI: (0.105, 0.238)) but only slightly positive for CCM (coef = 0.0526, 95% CI: -0.0205, 0.126). Lastly, patient activation was positively related to CAHPS G Trust for DSMS (coef = 3.67, 95% CI: 2.18, 5.15) but not related for CCM (coef = -0.70, 95% CI: -2.63, 1.23).

Aim #2: Secondary outcomes

Our second aim was to measure and compare PP success at diabetes self-management as indicated by improvement at 6 months in A1c and depression index score.

Changes in A1c

A1c on the log₂ scale (log₂ (A1c), here called “A1c”) had an initial slight decrease for patient participants in both programs (Fig S 4A), and the A1c distribution was hyper-exponential showing right skewness even after log transformation. Model fit assumptions were not violated. There was no Time-by-Program interaction (Fig S 4C, p-value = 0.616), so both programs changed to similar extents over time; in particular, the Baseline-to-6-month change for CCM compared to the change for DSMS was − 0.0165, 95% CI (-0.0786, 0.0456). DSMS was slightly higher than CCM for the study period. Both sites decreased their A1c from baseline to 3 months and sustained their improvement throughout the end of the study (Fig S 4B).

The multivariate analysis (Table S 12, contrasts not shown) revealed that A1c was positively related to depression (coef = 0.0204, 95% CI: (-0.00129, 0.0422)). A1c was positively related to PP DKQ-23 Score at baseline, but much less so at follow-up, and negatively related to SSP DKQ-23 score at baseline, with slightly positive relationships at 3 months and beyond.

Changes in Depression

As we previously reported [34], overall, depression as measured by the PHQ-9 and transformed as log2(PHQ-9 + 2), showed an initial decrease from baseline to 3 months which sustained over time. Model fit assumptions were not violated. There was a Time-by-Program interaction (p-value = 0.006) because the degree of change over time differed between programs, in particular, the Baseline-to-6-month change for CCM compared to the change for DSMS was − 0.266, 95% CI (-0.464, -0.0669). CCM started lower than DSMS and showed a large initial decrease and continued downward throughout the study while DSMS showed a small initial decrease through 6 months, then started increasing again by 12 months nearly rebounding to their baseline levels.

Our hypothesis was that diabetes self-management programs are most successful if their design is culturally and contextually “situated” [35–37] by positively leveraging the cultural values and accommodating the socio-economic circumstances of a patient population in a way that creates synergy with patients’ everyday lives [6, 38–46]. To test this hypothesis, we compared two program models–one based in an academic medical center using an approach based on group educational sessions taught by a trained diabetes educator and the other, a community-run program based on a wrap-around services model. Although they are both described as “culturally competent,” they differ strongly in the extent to which they embody characteristics of cultural and contextual situatedness.

Primary outcomes: Diabetes knowledge and patient activation

For diabetes knowledge and patient activation (capacity for change), we found no statistical differences between programs.

Diabetes knowledge increased slightly in patient participants in both programs. Most patient participants at both sites reported high capacity for diabetes self-management and scored highly on our patient activation measure, which made it difficult to show any differences (ceiling effect). Both programs were generally similarly effective at increasing a patient participant’s activation from baseline to three months and tended to stay at the higher level for the rest of the study period.

However, there were differences between the sites, with CCM patient participants generally being more activated than DSMS patient participants. We believe that the higher-level activation at CCM is related, in part, to the need for CCM participants to rely on themselves rather than medical treatment for managing their diabetes or their health in general because, as indicated above, few CCM patients have health insurance. Moreover, in the interviews and focus groups, we gathered data suggesting that few CCM patient participants qualify for safety net benefit programs that provide supports for unemployment, disability, food insecurity, or rental assistance. Most patient participants at DSMS were using these safety net programs, including many on Social Security Disability Insurance (SSI). Therefore, we found that CCM patient participants are generally much more concerned than DSMS patient participants about the potential of diabetes to negatively impact their ability to work and to provide support for a family—and more patient participants at CCM were married or living with a partner than at DSMS.

In addition, patient participants in the CCM program who scored very low on the PAM at baseline made more and sustained change, ending up as high or higher than those with higher scores at both time points (please interpret this subgroup observation with caution). Those with low activation at DSMS were not able to become activated to the level of the vast majority of other participants in the study.

Secondary outcomes: changes in A1c and depression

For our secondary outcomes of change in A1c, we did not find statistically significant differences between programs. We found that blood sugar levels at both sites decreased and that the difference in the decrease between the sites was not statistically significant. At both sites, after initial decrease at three months, the change was sustained for the rest of the study—although the change was not statistically significant at either site. The decrease seems slightly better sustained at DSMS, but the actual A1c values began lower and stayed lower at CCM.

Baseline values at both sites were similar in relation to the number of patient participant A1cs indicating prediabetes (6 vs 8), low (39 vs 41), medium (36 vs 31), and very high (13 vs 15) A1c values, but DSMS had notably more patient participants in the high range (26 vs 11).

Although neither program lowered participant A1c into a non-diabetes range, patient participants with prediabetes in both programs had A1c scores that remained below 6.4 and lower diabetes range scores (6.4-9.0) did not elevate higher. While stable A1c levels could be due to a variety of factors, and prediabetes does not inevitably get worse, especially over a short period of observation, this is a desired clinical outcome with long-term health implications. The literature demonstrates [47–51] that every 1% decrease in A1c is associated with improved health outcomes and reduced risk related to developing complications such as retinopathy/blindness, kidney disease, neuropathy, diabetes-related hospitalization, cardiovascular disease (CVD), and diabetes-related mortality. Skyler and colleagues [47] found that for each 1% decrease in A1c, there was a statistically significant 18% reduction in CVD events—the primary risk of death for people with diabetes. Similarly, Huang and colleagues [52] found that risk of complications and death became significantly higher above 9.0. The risk threshold for diabetic ketoacidosis (“diabetic coma”) [53] is also an A1c of 9. Therefore, of greater interest, although the difference in the change in A1c over time was not found to be statistically significant between programs, nor was the change dramatic, there was a key clinically meaningful difference, especially true for patient participants with higher baseline A1c values. At CCM, average “very high” A1cs dropped below 10 and the median was below 9. While on average CCM PPs, even those entering at the higher levels, dropped below 10—and the median was below 9, the higher A1c group in the DSMS program did not achieve that degree of diabetes control. The results for patient participants who had high A1c at baseline and dropped on average below 10 with the median below 9 have positive health implications.

Patient participants at CCM, perhaps in part because they tended to start with lower A1c values overall (including fewer with A1c over 10), were more activated [54], and were less depressed (discussed below) at baseline, were able to bring their A1c down to approximately 9.0, the risk threshold for diabetic coma [53]. At DSMS, in part because they started with higher A1cs, were less activated, and more depressed (discussed below), patient participants tended not to lower their A1c near 9.0. Because of the clinically important impact of each 1% increase/decrease in A1c on risk for diabetes-related health complications and the increased risk of complications over time, projected long-term clinical outcomes for the two sites would appear to be meaningfully different. Because more change and a sustained lower level was obtained by CCM patient participants, CCM came much closer to achieving the goal of managing their diabetes than patient participants at DSMS.

An important area where we found a statistically and clinically significant difference between the two sites and also revealed important dynamics related to social support was in relation to depression symptoms [34]. CCM started with baseline depression scores that were generally lower than those at DSMS, though at both sites a majority of participants had either “none-minimal” or “mild” severity scores (CCM 84.9%, DSMS 71.7%). Both sites reduced their level of depression symptoms after baseline, but despite starting with less depression (which one could hypothesize should make it more difficult to change), depression scores at CCM improved more than DSMS, and while CCM improvement continued over the period of the study, at DSMS, the improvement was not sustained. Again, the biggest change was in the first three months with some continuing improvement at 6 months, except for the most highly scoring patient participant in the CCM program, where improvement appeared only at 6 months. The least depressed group did not improve further.

As anticipated from research on diabetes and depression, we found that depression symptoms were related to A1c. Patient participants with higher depression scores tended to have higher A1cs. We also found that the diabetes knowledge of the social support was important—higher diabetes knowledge of the social support was associated with lower depression symptoms and better A1c values. The design of the CCM program that involves ongoing peer support together with the demonstrated high level of trust and cultural competence at CCM can be hypothesized to provide a more “supportive” environment involving more in-person contact and the creation of a larger social network. We hypothesize that more trust in the site as indicated in scores for the Consumer Assessment of Healthcare Providers and Systems Cultural Competence Set (CAHPS-CC) (Table S 16) may lead participants to be more interested in participating and develop a feeling of commitment to others related to “showing up” to ongoing group and individual meetings. These meetings become a type of peer support, with patient participants who attend sharing recipes and strategies for diabetes management. Our interpretation is that this sharing and empathy likely contribute to the lower and more sustained levels of depression symptoms. Participants at both sites indicated the importance of these factors in qualitative data, and our broader research with Latina and Mexican immigrant women and CHWs in Albuquerque strongly supports this finding of the deep importance of social support in the lives of Latinxs [55–60]. In fact, this is why our patient advisors were adamant that we find a way to include social support and social context in the design of our study, and why we enrolled participants as patient participant-social support participant dyads rather than merely recruiting patient participants to participate as individuals.

Furthermore, although patient participants at both sites rated their program high in cultural competence and expressed that they really like the program and the staff, when posed with a question about what was missing from the program design, DSMS patient participants tended to indicate that the DSMS model of six discrete group sessions was not really sufficient and was less attractive than a program like the CCM that provides ongoing classes and support. DSMS patient participants and their social support participants reported that patient participants often feel socially isolated and alone in relation to the issue of diabetes—and CCM patient participants discussed the importance of having the social support of the program in an ongoing way. Greater social support and interaction and less social isolation at CCM might help to explain the lower levels of depression symptoms overall, as well as the greater and more sustained decrease in depression after joining the program.

Data modeling and visualization revealed that changes were not uniform, thus the exploratory, post-hoc subgroup analysis helped provide additional understanding of who had bigger changes than others. A subgroup analysis by A1c baseline category showed that depression symptoms at DSMS is higher than at CCM overall, decreases at CCM were greater at all A1c baseline categories with median values ending below 5 for all categories, while DSMS ends below 5 in only the lowest A1c category and all categories showing increases in depression by 12 months.

Outcomes and Comparative Effectiveness of the Two Programs

Findings from this study demonstrate comparative effectiveness of two culturally and contextually situated models of diabetes self-management and education. The two models were similar in many ways: both were deemed culturally competent by PPs and SSPs, and we did not find statistically different results between the two sites with respect to the primary outcome measures and only one of the three secondary outcomes showed differential results. Moreover, the design of this study was observational, and given major differences in site context and patients enrolled, residual confounding is likely. However, given the higher cultural competence rating (0.87 vs 0.83), statistically significant improvement in depression (larger decrease by -0.266) [34], and the importance of social support to the patients (84.9% vs 71.4% strongly agree), the results suggest that culturally and contextually situating a diabetes intervention may deliver benefit for patients, especially for some subgroups of patients.

While the results appear to align with findings in the broader literature that cultural competence plays a role as beneficial for health promotion programming [44, 61], especially for some populations, we recognize that the CCM model requires a substantial commitment to creating wrap-around services that put the patient and the patient’s needs at the center of care and include taking values, customs, beliefs, and language from Latinx culture into account. This is not a model currently in wide use, and it could be challenging for some sites to implement. While it was not within the scope of this study to speak further to the generalizability of the model, further study may show the more nuanced ways that each of these models operates to produce cultural competence for sub-populations.

Our results suggest that a holistic, culturally and contextually situated approach to diabetes prevention and self-management can be beneficial to patients and to their families, caregivers and social supports. The data suggest that integrated, ongoing support is potentially helpful for patients with diabetes. We learned the important role that social support plays in the lives of diabetes patients and care givers and that more attention needs to be paid to integrating these dimensions of a patient’s reality into care.

This study compared diabetes self-management over a 12-month period for 226 Latinx diabetes patients from two different diabetes self-management programs. Although they are both described as “culturally competent,” they differ strongly in the extent to which cultural characteristics are embedded in their structure and function. The primary outcome, improved capacity for diabetes self-management measured through improvements in diabetes knowledge and diabetes-related patient activation, was equivalent for both programs. Both programs resulted in improved A1c levels, although no change in BMI. The only statistically significant difference between programs was the effect on one of the secondary outcomes of depression scores, which were more improved in the CCM than the DSMS program [34], and were more significant in patients with severe diabetes (A1c > 10). In this A1c catagory, improvement in A1c levels followed decreasing depression scores, indicating a subtle but powerful impact of that treatment model. Decreases of A1c in this range are associated with fewer secondary effects of the hyperglycemia clinically.

Using exploratory, post-hoc subgroup analyses, we found differences between small subgroups of patients based on lower baseline PAM and higher baseline A1c and depression scores, and based on scores on the Consumer Assessment of Healthcare Providers and Systems Cultural Competence Set (CAHPS-CC) that were supported by qualitative data and clinically meaningful differences in A1c outcomes, all of which pointed toward CCM as the more culturally and contextually situated model. In future a study focusing on patients with high A1c levels in a large enough cohort for more statistical power would be useful and may reveal treatment and prevention modalities for the life-threatening sequelae of diabetes. As such, this study demonstrates that culturally and contextually situated approaches can successfully deliver effective benefits for diabetes self-care, and that even more graded interventions may be required to enhance self-management and eliminate the long-term lethal complications of diabetes in this population.

Funding Statement

The production of this manuscript was partially funded through Patient-Centered Outcomes Research Institute (PCORI) Awards Tier I-#7738704, Tier II-#7738704, & #CER-1511-32910, and UNM Clinical and Translational Science Center (CTSC) Award #CTSC006-2. The views presented in this work are solely the responsibility of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute (PCORI), its Board of Governors or Methodology Committee, or those of the UNM CTSC.

Conflict of Interest Statement

The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Institutional Review Board Statement

The study was approved by the Human Research Protections Office of the University of New Mexico on October 6, 2016, #16-303.

Clinical Trials Registration

This study was registered with ClinicalTrials.gov on May 2, 2018, #NCT03004664.

Informed Consent Statement

Written informed consent was obtained from all participants involved in the study.

Data Statement

UNM’s Digital Repository is an open access institutional repository created to preserve, gather and disseminate scholarship, research and cultural works created by UNM faculty, staff and students within UNM departments, centers and programs. The repository supports teaching and lifelong learning within the UNM campus and broader New Mexico community by making deposited materials permanently and freely accessible. The supplemental data consisting of codebooks can be freely downloaded at https://doi.org/10.25827/MTPV-KG27.

Page-Reeves J, Regino L, Murray-Krezan C, Bleecker M, Erhardt E, Burge M, et al. A comparative effectiveness study of two culturally competent models of diabetes self-management programming for Latinos from low-income households. BMC Endocr Disord. 2017 Dec;17(1):1–8.
Funnell MM, Brown TL, Childs BP, Haas LB, Hosey GM, Jensen B, et al. National Standards for diabetes self-management education. Diabetes Care. 2011 Jan;34(Supplement 1):S89-96.
Dauvrin M, Lorant V, d’Hoore W. Is the chronic care model integrated into research examining culturally competent interventions for ethnically diverse adults with Type 2 diabetes mellitus? A review. Evaluation & the Health Professions. 2015 Feb 4;38(4):435–63.
Stellefson M, Dipnarine K, Stopka C. The chronic care model and diabetes management in US primary care settings: a systematic review. Preventing Chronic Disease. 2013;10:E26.
American Diabetes Association. Recognized Education Programs– DiabetesPro [Internet]. [cited 2015 Nov 10]. Available from: https://professional.diabetes.org/diabetes-education
Nam S, Janson SL, Stotts NA, Chesla C, Kroon L. Effect of culturally tailored diabetes education in ethnic minorities with type 2 diabetes: a meta-analysis. J Cardiovasc Nurs. 2012 Nov 1;27(6):505–18.
Sixta CS, Ostwald S. Texas-Mexico border intervention by promotores for patients with type 2 diabetes. The Diabetes Educator. 2008 Mar;34(2):299–309.
Garcia AA, Villagomez ET, Brown SA, Kouzekanani K, Hanis CL. The Starr County Diabetes Education Study: development of the Spanish-language diabetes knowledge questionnaire. Diabetes Care. 2001 Jan 1;24(1):16–21.
Vincent D, Pasvogel A, Barrera L. A feasibility study of a culturally tailored diabetes intervention for Mexican Americans. Biological Research for Nursing. 2007 Oct;9(2):130–41.
Kim S, Love F, Quistberg DA, Shea JA. Association of health literacy with self-management behavior in patients with diabetes. Diabetes Care. 2004 Dec 1;27(12):2980–2.
Shah VO, Carroll C, Mals R, Ghahate D, Bobelu J, Sandy P, et al. A Home-Based Educational Intervention Improves Patient Activation Measures and Diabetes Health Indicators among Zuni Indians. PLoS ONE. 2015 May 8;10(5):e0125820.
Wolever RQ, Dreusicke M, Fikkan J, Hawkins TV, Yeung S, Wakefield J, et al. Integrative health coaching for patients with type 2 diabetes: a randomized clinical trial. The Diabetes Educator. 2010 Jul;36(4):629–39.
Druss BG, Zhao L, von Esenwein SA, Bona JR, Fricks L, Jenkins-Tucker S, et al. The Health and Recovery Peer (HARP) Program: A peer-led intervention to improve medical self-management for persons with serious mental illness. Schizophrenia Research. 2010 May 1;118:264–70.
Grønning K, Rannestad T, Skomsvoll JF, Rygg LØ, Steinsbekk A. Long-term effects of a nurse-led group and individual patient education programme for patients with chronic inflammatory polyarthritis - a randomised controlled trial. Journal of Clinical Nursing. 2014 Apr;23(7–8):1005–17.
Lorig K, Ritter PL, Villa FJ, Armas J. Community-based peer-led diabetes self-management: a randomized trial. The Diabetes Educator. 2009 Jul;35(4):641–51.
Huang FY, Chung H, Kroenke K, Delucchi KL, Spitzer RL. Using the Patient Health Questionnaire-9 to measure depression among racially and ethnically diverse primary care patients. Journal of General Internal Medicine. 2006 Jun;21(6):547–52.
Gilbody S, Richards D, Brealey S, Hewitt C. Screening for depression in medical settings with the Patient Health Questionnaire (PHQ): a diagnostic meta-analysis. Journal of General Internal Medicine. 2007 Nov;22(11):1596–602.
Reuland DS, Cherrington A, Watkins GS, Bradford DW, Blanco RA, Gaynes BN. Diagnostic accuracy of Spanish language depression-screening instruments. Annals of Family Medicine. 2009 Sep 1;7(5):455–62.
Parchman ML, Zeber JE, Palmer RF. Participatory decision making, patient activation, medication adherence, and intermediate clinical outcomes in type 2 diabetes: a STARNet study. Annals of Family Medicine. 2010 Sep 1;8(5):410–7.
Schiøtz ML, Bøgelund M, Almdal T, Jensen BB, Willaing I. Social support and self-management behaviour among patients with Type 2 diabetes. Diabetic Medicine: A Journal of the British Diabetic Association. 2012 May;29(5):654–61.
Rygg LØ, Rise MB, Grønning K, Steinsbekk A. Efficacy of ongoing group based diabetes self-management education for patients with type 2 diabetes mellitus. A randomised controlled trial. Patient Education and Counseling. 2012 Jan 1;86(1):98–105.
Hibbard JH, Mahoney ER, Stock R, Tusler M. Do increases in patient activation result in improved self-management behaviors? Health Services Research. 2007 Aug;42(4):1443–63.
Hibbard JH, Mahoney E. Toward a theory of patient and consumer activation. Patient Education and Counseling. 2010 Mar;78(3):377–81.
Hibbard JH. Community-based participation approaches and individual health activation. The Journal of Ambulatory Care Management. 2009 Oct 1;32(4):275–7.
Hendriks M, Rademakers J. Relationships between patient activation, disease-specific knowledge and health outcomes among people with diabetes; a survey study. Health Services Research. 2014 Dec;14:1–9.
Dixon A, Hibbard J, Tusler M. How do People with Different Levels of Activation Self-Manage their Chronic Conditions? The Patient: Patient-Centered Outcomes Research. 2009 Dec;2(4):257–68.
Hibbard JH, Mahoney ER, Stockard J, Tusler M. Development and testing of a short form of the patient activation measure. Health Services Research. 2005 Dec;40(6p1):1918–30.
Hammersley M. Questioning Qualitative Inquiry: Critical Essays. London: Sage; 2008.
Kroenke K, Spitzer RL, Williams JBW. The PHQ-9: Validity of a Brief Depression Severity Measure. Journal of General Internal Medicine. 2001 Sep 1;16(9):606–13.
Agency for Healthcare Research and Quality. About the CAHPS Cultural Competence Item Set [Internet]. 2012 [cited 2021 May 16]. Available from: https://archive.ahrq.gov/cahps/surveys-guidance/item-sets/cultural/2312_about_cultural_comp.pdf
Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. 2015;67(1):1–48.
Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest Package: Tests in Linear Mixed Effects Models. Journal of Statistical Software. 2017 Dec 6;82(13):1–26.
Insignia Health. Patient Activation Measure (PAM)® [Internet]. Products. 2021 [cited 2016 Feb 2]. Available from: http://www.insigniahealth.com/products/pam-survey
Erhardt E, Murray-Krezan C, Regino L, Perez D, Bearer EL, Page-Reeves, Janet. Associations Between Depression and Diabetes Among Latinx Patients from Low-Income Households in New Mexico. Social Science & Medicine. 2023 Jan;320:115713.
Clark L, Vincent D, Zimmer L, Sanchez J. Cultural values and political economic contexts of diabetes among low-income Mexican Americans. Journal of Transcultural Nursing. 2009 Oct;20(4):382–94.
Trickett EJ, Beehler S, Deutsch C, Green LW, Hawe P, McLeroy K, et al. Advancing the science of community-level interventions. American Journal of Public Health. 2011 Aug;101(8):1410–9.
Trickett EJ. Multilevel community-based culturally situated interventions and community impact: an ecological perspective. American Journal of Community Psychology. 2009 Jun;43:257–66.
Barrera Jr M, Castro FG, Strycker LA, Toobert DJ. Cultural adaptations of behavioral health interventions: A progress report. Journal of Consulting and Clinical Psychology. 2013 Apr;81(2):196–205.
Hawthorne K, Robles Y, Cannings-John R, Edwards AGK. Culturally appropriate health education for Type 2 diabetes in ethnic minority groups: a systematic and narrative review of randomized controlled trials. Diabet Med. 2010 Jun;27(6):613–23.
Kong A, Tussing-Humphreys LM, Odoms-Young AM, Stolley MR, Fitzgibbon ML. Systematic review of behavioural interventions with culturally adapted strategies to improve diet and weight outcomes in African American women. Obes Rev. 2014 Oct;15(Supplement 4):62–92.
Lie DA, Lee-Rey E, Gomez A, Bereknyei S, Braddock CH. Does cultural competency training of health professionals improve patient outcomes? A systematic review and proposed algorithm for future research. J Gen Intern Med. 2011 Mar;26(3):317–25.
Pottie K, Hadi A, Chen J, Welch V, Hawthorne K. Realist review to understand the efficacy of culturally appropriate diabetes education programmes. Diabetic Medicine. 2013 Sep;30(9):1017–25.
Ricci-Cabello I, Ruiz-Pérez I, Rojas-García A, Pastor G, Rodríguez-Barranco M, Gonçalves DC. Characteristics and effectiveness of diabetes self-management educational programs targeted to racial/ethnic minority groups: a systematic review, meta-analysis and meta-regression. BMC Endocr Disord. 2014 Dec;14(1):1–3.
Whittemore R. Culturally competent interventions for Hispanic adults with type 2 diabetes: a systematic review. Journal of Transcultural Nursing. 2007 Apr;18(2):157–66.
Page-Reeves J, Niforatos J, Mishra S, Regino L, Gingrich A, Bulten R. Health Disparity and Structural Violence: How Fear Undermines Health Among Immigrants at Risk for Diabetes. Journal of Health Disparities Research and Practice. 2013 Jan 1;6(2):30–47.
Page-Reeves J, Mishra SI, Niforatos J, Regino L, Bulten R. An Integrated Approach to Diabetes Prevention: Anthropology, Public Health, and Community Engagement. Qual Rep. 2013;18:1–22.
Skyler JS, Bergenstal R, Bonow RO, Buse J, Deedwania P, Gale EA, et al. Intensive glycemic control and the prevention of cardiovascular events: implications of the ACCORD, ADVANCE, and VA diabetes trials: a position statement of the American Diabetes Association and a scientific statement of the American College of Cardiology Foundation and the American Heart Association. Journal of the American College of Cardiology. 2009 Jan 20;53(3):298–304.
Arnold LW, Wang Z. The HbA1c and all-cause mortality relationship in patients with type 2 diabetes is J-shaped: a meta-analysis of observational studies. The review of diabetic studies. 2014;11(2).
Diabetes Control and Complications Trial Research Group. The relationship of glycemic exposure (HbA1c) to the risk of development and progression of retinopathy in the diabetes control and complications trial. Diabetes. 1995 Aug 1;44(8):968–83.
Sherwani SI, Khan HA, Ekhzaimy A, Masood A, Sakharkar MK. Significance of HbA1c test in diagnosis and prognosis of diabetic patients. Biomarker insights. 2016 Jan;11:95–104.
Pankratz VS, Choi EE, Qeadan F, Ghahate D, Bobelu J, Nelson RG, et al. Diabetes status modifies the efficacy of home-based kidney care for Zuni Indians in a randomized controlled trial. Journal of Diabetes and its Complications. 2021 Feb 1;35(2):1–6.
Huang ES, Liu JY, Moffet HH, John PM, Karter AJ. Glycemic control, complications, and death in older diabetic patients: the diabetes and aging study. Diabetes care. 2011 Jun 1;34(6):1329–36.
Alameri M, Wafa W, AlTikriti A, Saadane I, Alkaf B, Lessan N. Incidence and Rate of Progression of Prediabetes to Type 2 Diabetes: A 10-Year Retrospective Study From UAE. Endocrine Practice. 2019 Apr 1;25:141–2.
Nelson RG, Pankratz VS, Ghahate DM, Bobelu J, Faber T, Shah VO. Home-based kidney care, patient activation, and risk factors for CKD progression in Zuni Indians: a randomized, controlled clinical trial. Clinical Journal of the American Society of Nephrology. 2018 Dec 7;13(12):1801–9.
Page-Reeves J, Murray-Krezan C, Regino L, Perez J, Bleecker M, Perez D, et al. A randomized control trial to test a peer support group approach for reducing social isolation and depression among female Mexican immigrants. BMC public health. 2021 Dec;21(1):1–18.
Page-Reeves J, Shrum S, Rohan-Minjares F, Thiedeman T, Perez J, Murrietta A, et al. Addressing Syndemic Health Disparities Among Latin Immigrants Using Peer Support. Journal of Racial and Ethnic Health Disparities. 2019 Apr;6(2):380–92.
Page-Reeves J, Moffett ML, Steimel L, Smith DT. The evolution of an innovative community-engaged health navigator program to address social determinants of health. Progress in Community Health Partnerships: Research, Education, and Action. 2016;10(4):603–10.
Page-Reeves J editor. Women Redefining the Experience of Food Insecurity: Life Off the Edge of the Table. Lexington Books; 2014.
Page-Reeves J. Conceptualizing Food Insecurity and Women’s Agency: A Synthetic Introduction. In: Page-Reeves J, editor. Women Redefining the Experience of Food Insecurity: Life Off the Edge of the Table. Lanham, Maryland: Lexington Books; 2014. p. 3–44.
Janet Page-Reeves, Amy Anixter Scott, Maurice Moffett, Veronica Apodaca, Vanessa Apodaca. “Is always that sense of wanting … never really being satisfied”: Women’s Quotidian Struggles With Food Insecurity in a Hispanic Community in New Mexico. Journal of Hunger & Environmental Nutrition. 2014 Apr 3;9(2):183–209.
Zeh P, Sandhu HK, Cannaby AM, Sturt JA. The impact of culturally competent diabetes care interventions for improving diabetes-related outcomes in ethnic minority groups: a systematic review. Diabet Med. 2012 Oct;29(10):1237–52.

FigureS01consortdiagram.tif
Consort Diagram Fig S 1. Diagram of enrollments (CONSORT diagram). DSMS, diabetes self-management support empowerment model; CCM, chronic care model. This figure was previously published in [34].
FigureS02compositedkq.tif
DKQ Composit Scores Fig S 2. Comparison of results from the Diabetes knowledge questionnaire (DKQ). Scores are taubluated and primary contrast over time between programs. (See Fig 2for panel descriptions.) Since the 95% CI crosses zero in all three contrasts, there is no difference between the programs for DKQ. DSMS, diabetes self-management support empowerment model; CCM, chronic care model.
FigureS03compositepam.tif
PAM Composit Scores Fig S 3. Patient Activation Measure (PAM-10) primary contrast over time.. (See Fig 2for panel descriptions.) All three contrasts overlap 0; thus there is no statistically significant difference between them. DSMS, diabetes self-management support empowerment model; CCM, chronic care model, PAM-10, Patient Activation Measure.
FigureS04compositea1c.tif
A1c Composit Scores Fig S 4. Analysis of A1c measures. A1c longitudinal data and primary contrast over time between programs. (See Fig 2for panel descriptions.) DSMS, diabetes self-management support empowerment model; CCM, chronic care model; A1c, meaure of glycoylated hemoglobin proportional to chronic glucose levels. This figure was previously published in [34].
FigureS05compositebmi.tif
BMI Composit Scores Fig S 5. Analysis of body mass index (BMI) over time and comparison between cohorts. BMI outcome longitudinal model data and primary contrast over time between programs. (See Fig 2for panel descriptions.) DSMS, diabetes self-management support empowerment model; CCM, chronic care model; BMI, body mass index.
FigureS06bya1cpam.tif
A1c and PAM Scores Fig S 6. PAM-10 outcome by A1c Baseline category. Small improvement in most A1c catagories occurs in both groups in the first 3 months and in the lowest category over the year. Numbers of subjects in each A1c category for each cohort appears above the trajectory. (See Fig 3 for categories.). DSMS, diabetes self-management support empowerment model; CCM, chronic care model; PAM-10, Patient activation measure.
FigureS07bya1cbmi.tif
A1c and BMI Scores Fig S 7. BMI outcome by A1c Baseline category. BMI trajectories are similar between programs. (See Fig 3 for categories.) DSMS, diabetes self-management support empowerment model; CCM, chronic care model; BMI, body mass index.
SupplementoryTables.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

A patient-centered comparative effectiveness research study of culturally appropriate options for diabetes self-management

Status:

Version 1

Abstract

Figures

Introduction

Overview

Models for comparison

Local context for each model

The Diabetes Self-Management Support Empowerment Model (DSMS)

The Chronic Care Model (CCM)

Measurements for Outcomes

Improved patient capacity for diabetes self-management

Improvement in A1c and PHQ-9

Participants

Dyadic enrollment design

Patient participants (PPs)

Social support participants (SSPs)

Data collection

Qualitative Analysis

Quantitative Analysis

Sample size and Power Calculations

Statistical analysis

Results

Sample

Descriptive Baseline Statistics

Physiological characteristics

Capacity for diabetes self-management

Depression

Qualitative Analysis

Themes from interviews and focus groups

Quantitative Analysis

Aim #1: Primary outcomes

Changes in diabetes knowledge

Patient activation

Aim #2: Secondary outcomes

Changes in A1c

Changes in Depression

Discussion

Primary outcomes: Diabetes knowledge and patient activation

Secondary outcomes: changes in A1c and depression

Outcomes and Comparative Effectiveness of the Two Programs

Conclusions

Declarations

References

Supplementary Files

Status:

Version 1