Diabetes Distress, Medication Taking, Glycaemic Control and Self-Management: Comparing a Minority Migrant Group With Mainstream Society

Diabetes distress (DD) has broad-ranging effects on type 2 diabetes (T2DM) management and outcomes. DD research is scarce among ethnic minority groups, particularly Arabic-speaking immigrant communities. To improve outcomes for these vulnerable groups, healthcare providers, including pharmacists, need to understand modiable predictors of DD. To assess and compare DD and its association with medication-taking behaviours, glycaemic control, self-management, and psychosocial factors among rst-generation Arabic-speaking immigrants and English-speaking patients of Anglo-Celtic background with diabetes, and determine DD predictors. Setting: Various healthcare settings in Australia. Method: A multicentre cross-sectional study was conducted. Adults with T2DM completed a survey comprised of validated tools. Glycated haemoglobin, blood pressure, and lipid prole were gathered from medical records. Multiple linear regression models were computed to assess the DD predictors. Main outcome measure: Diabetes distress level.


Introduction
Globally, diabetes is associated with signi cant morbidity and burden. Its prevalence continues to increase and is projected to reach 10.4% by 2040, placing increasing demands on healthcare systems [1].
Urgent actions are needed to combat this global burden, including better support for patients to achieve optimal self-management [2,3].
Successful integration of diabetes care into daily life is challenging. Most individuals are expected to comprehend large amounts of information about diabetes and its treatment, make many selfmanagement decisions, spend substantial time often performing complex self-care activities, and access multiple healthcare services [4,5]. Therefore, it is unsurprising that diabetes distress (DD) is prevalent among people with type 2 diabetes (T2DM) [6,7]. The prevalence of moderate and severe DD among adults with T2DM varies globally; it has been reported to be as high as 45.4% in community settings [8].
DD is rooted in how patients experience diabetes management within their socio-cultural environment, and it refers to the emotional burdens, worries, and concerns among individuals living with this progressive, challenging disease [9]. Previous studies demonstrated signi cant linkages between DD and poor glycaemic control, worse self-management activities, low quality of life and self-e cacy, and increased morbidity levels and healthcare costs [9][10][11][12][13][14][15].
Multidisciplinary integrated team-based approach, structured patient-centred support, and appropriate follow-up mechanisms have led to improvements in diabetes control [16,17]. Pharmacist-led diabetes management program improved the care and outcomes of people with diabetes including ethnic minorities [18,19]. Despite being important for the comprehensive management of T2DM, healthcare providers including pharmacists often fail to screen for or assess DD. Delaying early detection of DD negatively impacts patients' quality of life and increases healthcare costs [14]. Psycho-education interventions designed to reduce DD have demonstrated improved health outcomes, particularly among those who are not achieving clinical goals or have di culty adhering to their prescribed treatment and lifestyle modi cations [20,21].
In recent years, the increasing ethnic diversity among patients with diabetes in upper-middle and highincome western countries has received growing attention. DD levels are higher among migrant communities compared with the general population [22]. These minority groups often have poorer health outcomes, higher morbidity rates, lower health literacy, face more barriers to access and use of healthcare services, and often receive less medical information from their doctors compared with mainstream society [23,24].
There is a paucity of data about DD and contributing factors among Arabic-speaking immigrant communities in western countries [22]. Our previous research among this rapidly growing minority identi ed that they have poorer glycaemic control, reduced adherence to self-management activities and prescribed medications, experience more challenges in accessing health services, and negative health beliefs when compared with English-speaking people of Anglo-Celtic origin [25][26][27]. To date, no previous study assessed all these factors that are postulated to be associated with DD among Arabic-speaking immigrant communities.

Aim Of The Study
The study aimed to assess and compare DD level and understand its relationship with key clinical, behavioural, psychosocial, and sociodemographic factors among Arabic-speaking immigrants and English-speaking patients of Anglo-Celtic background with T2DM (comparison group) and determine the extent to which these factors predict DD levels. Such assessment would be used to guide clinicians and pharmacists to improve medication adherence, glycaemic control and self-management among Arabicspeaking immigrant communities.

Ethics approval
Ethics approval was obtained from the Monash University Human Research Ethics Committee (CF09/0956: 2009000462) and ethics committees at participating hospitals.

Study design and setting
A multicentre cross-sectional study was conducted in various healthcare settings in metropolitan and rural Victoria, Australia. Participants were recruited through diabetes outpatient clinics at three major hospitals, ten primary care practices, and ve community support groups. Rural participants were recruited through diabetes outpatient clinics at a major rural hospital and its associated clinics, three general medical practices, and various community support groups in Shepparton.

Study participants
People attending participating clinics were invited to complete study questionnaire while waiting for their appointments. Those who expressed interest were screened for eligibility via a brief researcheradministered checklist. Inclusion criteria were a diagnosis of T2DM and aged ≥ 18 years. Participants were recruited into two groups based on self-reported ethnic background. First, Arabic-speaking rstgeneration immigrants whose rst language was Arabic and born in any of the 22 countries of the Arab League were selected. Second, Australian English-speaking participants of Anglo-Celtic identity and born in North America, Australia, United Kingdom, Ireland, or New Zealand were selected. Eligible consenting patients completed a paper-based questionnaire immediately after consenting. On average, participants took 45 minutes to complete the survey.
The rst section of the questionnaire measured sociodemographic and health characteristics. The second section assessed psychosocial and behavioural factors using validated and reliable tools and items, including:

Medication-taking behaviours
Medication adherence was measured using the modi ed Morisky scale [28]. This scale consists of four items, scored as 'Yes' (0) or 'No' (1). Based on the total score, patients were categorised as either adherent (total score = 4) or non-adherent (total score < 4). Participants' medication underuse in the past 12 months was measured by two items used in previous studies [29], and data was presented in four categories: no underuse, cost-related underuse, non-cost related underuse, and both cost-and non-cost-related underuse.

Diabetes self-management
The Summary of Diabetes Self-Care Activities (SDSCA) was used to assess adherence to dietary behaviours, exercise, blood glucose testing, foot care, and smoking status [30]. The SDSCA is a brief, reliable, and valid measure that is widely used globally and it has 11 items measuring the number of days in the previous week that patients have engaged in self-care activities on a scale of 0-7 [30]. Data is reported as the mean days of activity in the prior week.
Health literacy and self-e cacy Functional health literacy was assessed using three validated items measuring di culties in reading medical forms or learning about medical conditions [31]. This instrument provides a total score that can be dichotomized into adequate functional health literacy (scores ≥ 10) and inadequate functional health literacy (scores < 10). Self-e cacy was de ned as the con dence a participant feels about conducting two speci c tasks: taking diabetes medicines as prescribed and performing self-care activities successfully. Participants recorded their response on a 5-point Likert scale, categorized as not con dent (responses 1-3) or con dent (responses 4-5). These two items have been validated among patients with diabetes [32,33].

Diabetes distress (main outcome measure)
Diabetes distress level was assessed using the Diabetes Distress Scale (DDS-17) [34]. This brief validated and reliable scale has 17 items in four subscales: emotional burden, physician-related distress, regimenrelated distress, and interpersonal distress. Participants' responses were recorded on a scale of 0 to 5; they were asked to indicate the degree to which each 'issue' might have distressed them in their daily lives. A total score (range: 0-85) was calculated for each patient; higher scores represent higher DD levels.
The last section collected the most recently documented clinical tests within the last 12 months of HbA1c level, blood pressure and lipid panel. Data was retrieved from medical records with participant consent.
The questionnaire was pilot-tested among three academics, two diabetes educators, and nine patients for clarity and the time to complete the questionnaire and to establish face and content validities. It was translated into Arabic by a professional translation service. The Arabic version was back-translated into English. The two versions were compared, and no differences were detected.

Sample size
If signi cant difference in the mean DD levels exists between the Arabic-speaking immigrants and mainstream, the obtained sample size of 701 enabled us to determine the signi cance of a mean absolute minimum difference of 2.7 (standard deviation 10.0) in DD scores between Arabic-speaking immigrants the mainstream group with a test power of 95%. Only participants with data on the outcome variable, which is DD, were included in the analysis.

Data analysis
Data were analysed using both SPSS Statistics for Windows, version 27.0, Armonk, NY, USA, and Minitab version 17.3. Data are summarized in percentages for categorical variables, mean ± standard deviation (SD) for continuous variables, and mean differences [95% con dence interval (CI)] for comparing two continuous variables. Bivariate associations between the two ethnic groups and other categorical variables were analysed using the Pearson Chi-square test, and the results were reported as numbers and percentages. Bivariate associations between continuous normally distributed variables were tested with Pearson's Correlation (r). Mean differences in the DD levels were tested through Independent-samples T-Test or one-way ANOVA, when the variables had more than two categories. All tests are two-tailed, and P values < 0.05 are considered statistically signi cant.
Multiple linear regression was used to model the extent to which the covariates in question could predict DD levels. Signi cant covariates for predicting diabetes distress were declared at a 5% signi cance level. All analyses were strati ed by the two groups.

Study participants
Data was analysed for 696 participants, data for ve further participants were removed as they did not complete the DD section: 392 were rst-generation Arabic-speaking immigrants and 304 English-speaking participants. Sociodemographic, clinical, behavioural and psychosocial characteristics are summarized in Table 1. Arabic-speaking immigrants were signi cantly younger than the English-speaking (p=0.001), more frequently had an education level below high school (p=0.005), had an inadequate functional health literacy (p<0.001), and a shorter duration of diabetes (p<0.001) than the English-speaking. Signi cantly more Arabic-speaking immigrants were non-adherent to their prescribed medication compared with English-speaking patients (p<0.001) and they also reported more cost-related and non-cost-related medication underuse (p<0.001). Comparing to English-speaking group, Arabic-speaking immigrants had a higher use of oral hypoglycaemic agents (OHAs) (p<0.001) and insulin (p<0.001). Documentation of clinical parameters was incomplete for a considerably higher proportion of Arabic-speaking immigrants. Among those with documented clinical values, Arabic-speaking immigrants more frequently had an inadequate glycaemic control [HbA1c > 7% (>53 mmol/mol)] (p=0.047), high blood pressure (p=0.020) and elevated levels of total cholesterol (p<0.001) and HDL (p=0.001) compared to English-speaking patients.

Diabetes distress levels
As shown in Figure 2, Arabic-speaking immigrants had signi cantly higher total DD levels (p<0.001) than English-speaking patients. This trend was also observed in three domains: emotional burden, regimenrelated distress, and diabetes-related interpersonal distress (p<0.001 for all).
Diabetes distress and sociodemographic characteristics, by ethnicity A negative weak correlation was found between DD levels and age in both Arabic-speaking (p=0.009) and English-speaking patients (p<0.001), meaning that older patients had lower levels of DD (Table 2). Among Arabic-speaking immigrants, higher DD levels were reported among those with lower levels of education (p=0.004), those with inadequate functional health literacy (p<0.001), and patients who were not employed (p<0.001).

Diabetes distress and clinical characteristics, by ethnicity
Mean DD level was signi cantly different according to perceived health status among Arabic-speaking and English-speaking patients: those with poor health status presented higher DD levels (p<0.001 and p=0.002, respectively). Among Arabic-speaking immigrants, those with a longer duration of diabetes had signi cantly lower distress levels (p=0.007). Those with co-morbidities and with inadequate glycaemic control [HbA1c > 7% (>53 mmol/mol)] had higher levels of DD (p=0.003 and p<0.001, respectively). Among English-speaking patients, weak positive correlations were found between DD levels and HDL levels (p=0.007).
Diabetes distress and behavioural and psychosocial characteristics, by ethnicity For both groups, greater adherence to dietary recommendations was weakly correlated with lower levels of DD (p<0.001 in both groups). In the Arabic-speaking group, exercise, self-monitoring of blood glucose, and foot care were moderately negatively correlated with DD (p<0.001 for all the correlations). Also in this group, higher levels of DD were found among those who are non-adherent to medication (p<0.001).
Those reporting cost-related medication underuse had signi cantly higher mean DD levels for both Arabic-speaking immigrants and English-speaking patients than those reporting no underuse (p=0.017 and p<0.001, respectively). Also, among English-speaking, those reporting both cost-and non-cost-related underuse had signi cantly higher DD levels compared with participants who reported non-cost-related underuse only (p<0.001).
In both groups, there were signi cant differences in DD levels according to self-e cacy. DD level was lower among participants who were con dent about taking diabetes medication precisely as prescribed (p<0.001 among Arabic-speaking and p=0.015 among English-speaking) and who were con dent in their ability to carry out self-care activities (p<0.001 among Arabic-speaking and p=0.001 among Englishspeaking).

Prediction of diabetes distress
The total DD score of both groups was regressed onto all signi cant covariates (Table 3). Overall, our regression models were able to explain 52% of the variation in DD scores for Arabic-speaking immigrants and 35% for English-speaking participants.
In the Arabic-speaking group, younger patients (p<0.001), those with lower education levels (less than high school) compared to those with a high school level (p=0.026), and those who were not employed (p=0.009) were more likely to present higher DD levels. Moreover, higher levels of DD among Arabicspeaking patients were predicted by inadequate glycaemic control [HbA1c > 7% (>53 mmol/mol)] (p=0.006), poor health status (p=0.028), lower adherence to exercise (p=0.006), cost-related underuse of medication (p<0.001), and lack of con dence in carrying out self-care activities (p=0.009).
In the English-speaking group, being younger (p<0.001), having lower adherence to dietary behaviour recommendations (p=0.004) and reporting both cost-and non-cost-related underuse of medication (p=0.002) predicted higher levels of DD.

Discussion
This study is the rst to assess DD and its correlation with medication-taking behaviours, glycaemic control, self-management practices, and self-e cacy among Arabic-speaking immigrants and Englishspeaking patients with T2DM in an upper-income western country. Similar to the recent ndings from the Dutch Diabetes Pearl Cohort, our Arabic-speaking immigrants had higher DD levels, lower medication adherence, and worse self-management practices and health outcomes compared with Caucasian participants [35]. Consistent with previous research with diabetic patients [10], a strong correlation between DD and HbA1c was found in our study, but only among Arabic-speaking immigrants.
Pharmacists have contributed signi cantly to improve diabetes management [36,37]. There are several modalities for pharmacist's involvement in diabetes care, including disease management, education, medication counseling, screening for CVD risk factors, referral, and follow-up [18,38,39]. However, the active screening for DD has not yet been integrated into existing pharmacist-led diabetes interventions. Our results demonstrated a signi cant association between medication adherence and underuse and DD levels. This calls for pharmacists to focus their future efforts on integrating DD assessment into their encounters with patients with T2DM, especially those from ethnic minority groups who typically have higher DD levels. Pharmacists could use brief and simple self-report measures to assess DD levels, facilitating such assessment into routine care [40].
In multivariate analysis, age was strongly associated with DD in both groups after controlling for the effect of other factors. The overall ability of the regression model to explain 52% of the variation in DD among Arabic-speaking immigrants was impressive, compared with just 35% for English-speaking patients. Self-e cacy was a substantial contributor to DD among both groups. The apparent excess burden of DD among Arabic-speaking immigrants may indicate less appropriate healthcare models, access inequalities, and/or greater challenges with self-management [41].
Consistent with previous studies, our results demonstrated that younger age was associated with higher DD levels in both groups [13,42]. Successful diabetes management requires patients to engage and maintain complex self-management behaviours. Within this context, it is postulated that those older individuals who typically have longer diabetes duration have a better opportunity to adopt and improve these self-management activities and integrate them into a daily routine. Moreover, compared to younger individuals, older adults are expected to have more time to accept disease diagnosis and build resilience and emotion regulation strategies to respond to diabetes demands [43]. Among Arabic-speaking patients, a cost-related underuse of medication, a lower educational level, and being unemployed predicted higher DD levels. This constellation of factors is revealing of how health is in uenced by interactions between environment, social and economic factors [44], which are unequally distributed, often penalizing individuals belonging to ethnical minorities.
It has been demonstrated that DD levels can be reduced with appropriate intervention [8]. Previous strategies to reduce DD relied on educational, behavioural/skill-based interventions (aiming at improving diabetes self-management knowledge, problem solving, goal setting, action planning) while the psychological intervention focused on the remediation of diabetes worries that underlie DD [8, 21,45]. Our results suggest that DD-inclusive interventions might have a particularly large impact on T2DM outcomes for Arabic-speaking migrant especially for those with high initial DD levels, and the focus of such intervention should be on promoting adherence to medication and self-management activities, and improving self-e cacy.

Strengths and weaknesses
Some limitations must be acknowledged. The cross-sectional design precluded cause-effect inferences of the observed correlations. The negative relationship between DD levels and self-management can be interpreted in two directions: increased DD can negatively impact self-management behaviours [46], and also the di culty in performing self-management could contribute to develop or increase DD [10]. Thus, to avoid reverse causality, future studies must include DD and self-management in longitudinal analyses to establish the direction and causality. Caution must be exercised in generalizing study ndings to all Arabic-speaking immigrants due to the non-random sampling and heterogeneity of this group. However, this was mitigated by recruiting a large and diverse sample from various primary, secondary, and tertiary healthcare settings in metropolitan and rural areas. The comprehensive assessment of factors known to be associated with DD and the inclusion of these in the regression model to identify modi able predictors to be used as starting point to develop targeted interventions is another major strength.

Conclusion
Arabic-speaking immigrants had higher DD levels, lower medication adherence and worse selfmanagement compared with English-speaking patients. Healthcare providers must routinely screen for and assess DD as part of comprehensive diabetes care, particularly among Arabic-speaking immigrants. The identi ed modi able factors that predicted higher DD levels among Arabic-speaking immigrants were low adherence rates to medication and exercise, and low-self-e cacy.

Declarations
Funding: This research was supported by an operational grant from the University of Sharjah and an internal grant from the Centre for Medicine Use and Safety, Monash University.
Con icts of interest: The authors declare that they have no con icts of interest.
Author contributions: HA designed the study, collected and partially analysed data, wrote, reviewed and edited the manuscript. HS conducted bivariate and multivariate analysis and edited the manuscript. KM contributed signi cantly to the discussion, reviewed and edited the manuscript. CB reviewed and edited the manuscript. CS contributed signi cantly to data analysis and interpretation, created tables, and edited the manuscript.
Availability of data and material (data transparency): Data is available from the author by request.
Code availability (software application or custom code): Not applicable.
Additional declarations for articles in life science journals that report the results of studies involving humans and/or animals: Not applicable.   underuse, 'Not con dent' for each self-e cacy item. Figure 1