A codesigned integrated kidney and diabetes model of care improves patient activation among patients from culturally and linguistically diverse backgrounds

Abstract Background Little is known about the relationship between patients' cultural and linguistic backgrounds and patient activation, especially in people with diabetes and chronic kidney disease (CKD). We examined the association between culturally and linguistically diverse (CALD) background and patient activation and evaluated the impact of a codesigned integrated kidney and diabetes model of care on patient activation by CALD status in people with diabetes and CKD. Methods This longitudinal study recruited adults with diabetes and CKD (Stage 3a or worse) who attended a new diabetes and kidney disease service at a tertiary hospital. All completed the patient activation measure at baseline and after 12 months and had demographic and clinical data collected. Patients from CALD backgrounds included individuals who spoke a language other than English at home, while those from non‐CALD backgrounds spoke English only as their primary language. Paired t‐tests compared baseline and 12‐month patient activation scores by CALD status. Results Patients from CALD backgrounds had lower activation scores (52.1 ± 17.6) compared to those from non‐CALD backgrounds (58.5 ± 14.6) at baseline. Within‐group comparisons showed that patient activation scores for patients from CALD backgrounds significantly improved by 7 points from baseline to 12 months follow‐up (52.1 ± 17.6–59.4 ± 14.7), and no significant change was observed for those from non‐CALD backgrounds (58.5 ± 14.6–58.8 ± 13.6). Conclusions Among patients with diabetes and CKD, those from CALD backgrounds report worse activation scores. Interventions that support people from CALD backgrounds with comorbid diabetes and CKD, such as the integrated kidney and diabetes model of care, may address racial and ethnic disparities that exist in patient activation and thus improve clinical outcomes. Patient or Public Contribution Patients, caregivers and national consumer advocacy organisations (Diabetes Australia and Kidney Health Australia) codesigned a new model of care in partnership with healthcare professionals and researchers. The development of the model of care was informed by focus groups of patients and healthcare professionals and semi‐structured interviews of caregivers and healthcare professionals. Patients and caregivers also provided a rigorous evaluation of the new model of care, highlighting its strengths and weaknesses.


| INTRODUCTION
Patient activation, defined as an individual's knowledge, skills and confidence in managing chronic diseases, 1 is associated with important health outcomes.Demographic characteristics such as older age, [2][3][4] low education, 2,5 low socioeconomic status 3,5 and smoking 6 are associated with low activation.Individuals with lower levels of activation are more likely to be obese 7 and less likely to achieve cholesterol and glycated haemoglobin targets. 8With respect to healthcare utilisation, individuals who report lower activation levels are more likely to be hospitalised, 9 have a longer length of stay in hospital 10 and have greater healthcare costs. 11 individuals experiencing multimorbidity, especially those with long-term health conditions like depression, type 2 diabetes mellitus, dementia and frailty, 12 activation levels are low.Similarly, activation levels are low among patients with comorbid diabetes and chronic kidney disease (CKD), especially those who have advanced kidney disease, are older and have poor self-reported health status. 13,14ile a number of factors associated with activation in patients with comorbid diabetes and CKD are now known, 13,14 there is a knowledge gap regarding the relationship between patient activation and patients' cultural and linguistic backgrounds.This gap is important, given that people from culturally and linguistically diverse (CALD) backgrounds experience a higher burden of disease and difficulties in accessing health services, 15 which impacts significantly on their health and quality of life.Among patients from CALD backgrounds, access to health services is influenced by challenges at an individual and family level, community and organisational level and systems and policy levels.These challenges include poor health literacy, multimorbidity, diminishing healthy migrants' effect, unhealthy food behaviours and lifestyles, language and communication problems, inadequate interpretation services and inadequate health systems and services to address the needs of CALD populations. 15Additionally, patients from CALD backgrounds are largely underrepresented in research, which limits evidence in how to improve access, utilisation and policies specific to their community needs. 16rthermore, the patient activation measure (PAM), a tool that evaluates the patients' perceived knowledge, skills and confidence to engage in self-management activities, has been shown to perform differently to patients of varied cultural and linguistic backgrounds. 17e risk of bias in the PAM increases given its initial development was conducted through testing with an 88% White population who were predominantly fluent in the English language. 1This highlights the importance of not only assessing variations in PAM across cultural and linguistic backgrounds but also evaluating PAM differences within groups.
Another key knowledge gap is the impact of person-centred integrated models of care on activation levels of people with diabetes and CKD from CALD backgrounds.Person-centred integrated models of care ensure that patients' values and concerns inform the management of long-term conditions instead of focusing on a standard set of disease management processes designed by health professionals. 18This approach encourages patients to actively participate in the selection of their treatment goals and to collaborate with health professionals regarding their specific needs for treatment and support of their chronic diseases. 19Integrated care delivered in this way has been reported to improve patient satisfaction, perceived quality of care and access to services 20 and to lower costs compared with usual care. 21We hypothesise that person-centred integrated models of care may improve activation levels for patients from CALD backgrounds due to several reasons.First, person-centred integrated models of care provide an opportunity for patients to build a certain level of trust in healthcare workers and the healthcare system, allowing for sharing of ideas and decision-making, which is important for improving patient activation. 22cond, given that most individuals from CALD backgrounds are disproportionately impacted by discrimination within the healthcare system, 23 asking them to embrace interventions that promote activation may improve their experience of the healthcare system.An alternative approach is care provided through person-centred care models that focus on addressing implicit attitudes and behaviours among clinicians and the healthcare systems, which have been shown to negatively influence healthcare professionals' willingness to engage in patientcentred care, provide referrals to specialised treatment or even adhere to evidence-based guidelines when serving diverse populations. 24,25e purpose of this study was not only to examine the association between patient activation and CALD background in patients with diabetes and CKD but also to determine the impact of a codesigned integrated kidney and diabetes model of care on patient activation for patients from CALD backgrounds.

| Study design, setting and participants
Adult patients (over 18 years) with diabetes and CKD who were receiving care from the Diabetes Kidney Service (DKS) 26 at Monash Health between January 2015 and August 2017 were recruited and followed up for 12 months.The study design, recruitment and followup of participants have been previously described in detail. 27,28In brief, participants with a diagnosis of diabetes (either type 1 or type 2) and CKD stages 3-5 (estimated glomerular filtration rate [eGFR] < 60 mL/min/1.73m 2 ), including dialysis, were included in the study.This study focused on people with more advanced CKD and diabetes, which was the group we deemed had the greatest need.
The diagnosis of diabetes was written in medical records and/or confirmed by the laboratory.Patients were considered to have CKD if they had a sustained eGFR < 60 mL/min/1.73m 2 (i.e., two or more eGFR readings <60 mL/min/1.73m 2 over a 3-month period) calculated using the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) formula. 29Key exclusion criteria were an eGFR ≥ 60 mL/min/1.73m 2 and a functioning kidney transplant.The study was conducted by following the Strengthening the Reporting of Observational Studies in Epidemiology recommendations. 30We obtained ethics approval from Monash University and Monash Health Human Ethics Review Committees.

| The diabetes and kidney disease model of care
We have previously described the diabetes and kidney disease model of care in detail. 28In brief, this model of care was codesigned by healthcare professionals (general practitioners, endocrinologists, nephrologists and nurse practitioners), patients with diabetes and CKD and patient advocacy groups such as Diabetes Australia and Kidney Health Australia in 2015.Findings from a large multisite formative evaluation of the barriers and enablers of current health services for diabetes and CKD and the needs of patients, carers and their health professionals informed the design of this model of care. 31is study highlighted the presence of guideline-care gaps among patients with diabetes and CKD as well as barriers to health care that included poor continuity of care, inadequate understanding/education about CKD, feeling unwell, inadequate support from family and friends, conflicting advice from specialists and poor communication among specialists, and these varied across CKD stage and hospitals. 31e diabetes and kidney disease model of care aims to provide patient-centred, coordinated multidisciplinary assessment and management of patients with comorbid diabetes and CKD in partnership with primary care.The service focuses on improving patient self-management, communication and better coordination of care between health professionals.General practitioners remain the coordinators of patient care within a team including an endocrinologist and nephrologist, specialist registrars in endocrinology and nephrology, diabetes and renal nurse practitioners, dietitian, administration and a research officer (for evaluation and continual improvement of the service).Patient self-management is achieved through motivational interviewing and tailored education.
Patients also get a diabetes-kidney care plan after each visit to ensure that they are updated with their care goals/management.The service utilises professional interpreting services in preference to family members for non-English speaking patients.Referrals to the integrated service included patients with type 1 or type 2 diabetes with an eGFR < 60 mL/min/1.73m 2 .

| Demographic and clinical variables
Age, gender, CALD status, stage of kidney disease and duration of diabetes were obtained from a questionnaire (Supporting Information: Appendix S1), which was prospectively completed by site study staff or the clinician, using standardised procedures from the doctor's notes and laboratory results from the clinic.

| CALD background
The definition of CALD background has evolved from the one proposed by the Australian Bureau of Statistics that utilised the country of birth, language spoken at home, English proficiency or other characteristics (including year of arrival in Australia) to determine CALD status. 32The most recent literature suggests that CALD status would best be defined as people born in non-English speaking countries and/or who do not speak English at home. 33For the purpose of our study, we used the latter definition mainly because we collected data on country of birth as well as the primary language spoken at home.We used the term CALD over 'Non-English-Speaking Background' for it is inclusive of individuals based on more than simply language and more welcoming and reflective of the diversity of the entire population. 34

| Patient activation
We evaluated patients' level of involvement in their health care by administering the PAM-13 at baseline and after 12 months. 35This 13-item measure has a similar reliability and validity to the 22-item version across different ages, genders and health condition status (Cronbach's α of .91 and a Rasch person statistic of 0.81 for the real and 0.85 for the model on which it was based). 1 The PAM scale has 13 questions, with four alternative responses to each of the 13 items, namely, 'disagree strongly, disagree, agree and agree strongly', and the fifth response option, 'not applicable' (N/A), was available for all items. 35A standardised spreadsheet provided by Insignia Health ® was used to calculate the PAM score. 36Participants who responded to less than seven items or who answered all questions with 'disagree strongly' or 'agree strongly' were excluded as per the developer's recommendations.The mean PAM score was then calculated on all items leaving out the ones thought to be nonapplicable by the participants.The raw mean score was converted into a standardised activation score ranging from 0 to 100, creating the PAM scores which were classified into the four levels of activation: Level 1 (score: <47.0),Level 2 (score: 47.1-55.1),Level 3 (score: 55.2-67.0)and Level 4 (score: >67.0) as per Insignia Health ® scoring rules. 36

| Statistical analysis
Participants who completed the PAM questionnaire were included in this analysis.First, the χ 2 test or the independent t-test examined differences in covariates by CALD status at baseline.Second, using the PAM score as a continuous variable, a multivariate regression analysis was performed to assess the relationship between patient activation and CALD background, adjusting for factors that have been known to influence patient activation at baseline.These factors include gender, age, duration of diabetes, stage of CKD and patientreported barriers to health care.Given the importance of these covariates to patient activation, they were all included in the model.
The process of determining patient-reported barriers to health care among patients with comorbid diabetes and CKD has been reported previously. 37Third, paired t-tests examined changes in patient activation within groups.A sensitivity analysis compared baseline activation scores for patients from CALD background who did not complete 12 months follow-up and those who did to determine the effect of baseline patient activation scores on participation and retention of these patients in the study.Confidence intervals (CIs) were reported at the 95% level and results were considered significant at conventional p < .05level.All analyses were performed using Stata version 16.0 (StataCorp).

| Participants
Of a total of 393 patients screened, 290 met the inclusion criteria for the model of care evaluation (Figure 1).Of these, 77 (27%) were from CALD backgrounds.During follow-up, 11 died (3.8%) before the 12-month visit and 170 (57%) completed the 12-month patient activation questionnaires.The baseline demographic and clinical characteristics of the study population are shown in Table 1.At baseline, the mean (SD) age for all patients was 67 ± 12 years, with a predominance of men (64%), and patients were evenly distributed across all stages of kidney disease.There were no differences in age, gender, stage of kidney disease, duration of diabetes, body mass index, barriers to health care and access to diabetes and kidney care by CALD status (all p > .05).

| Patient activation
At baseline, individuals from CALD backgrounds had significantly lower patient activation scores (52.1 ± 17.6) compared to those from non-CALD backgrounds (58.5 ± 14.6) (mean difference: −6.4,95% CI: −10.8 to −2.5, p = .002)(Figure 2 and Table 1).Among patients with low activation (Levels 1 and 2), 40% were from non-CALD backgrounds and 50% were from CALD backgrounds (Figure 3).In multivariate analysis adjusted for age, gender, duration of diabetes, eGFR and presence of barriers to health care at baseline, participants from CALD backgrounds scored on average 7 points lower than those from non-CALD backgrounds (Table 2).At 12 months, no difference in activation scores was observed between individuals from non-CALD and CALD backgrounds (Figure 2).Forty percent of patients from CALD background and non-CALD backgrounds had low levels of activation (Figure 3).Within-group comparisons showed that patient activation scores for ZIMBUDZI ET AL.
| 2587 patients from CALD backgrounds significantly improved by 7 points from baseline to 12 months follow-up (52.1 ± 17.6-59.4± 14.7), and no significant change was observed for those from non-CALD backgrounds (58.5 ± 14.6-58.8± 13.6) (Figure 2).To determine whether baseline patient activation scores influenced participation and retention of participants in this study, a sensitivity analysis was performed, which showed that baseline activation scores for patients from CALD backgrounds who did not complete 12 months follow-up (51.0 ± 13.8) and those who did (52.9 ± 20.1) were not different (Supporting Information: Table S1).Among patients from non-CALD backgrounds, no difference was observed as well between those who did not complete 12 months follow-up (58.4 ± 14.2) and those who did (58.6 ± 14.9).There were no differences with respect to other characteristics, except that patients from CALD backgrounds who did not complete 12 months follow-up were older (71.8 ± 11.1) than those who did (66.0 ± 12.4) (Supporting Information: Table S1).

| DISCUSSION
We studied the relationship between patient activation and CALD status among patients with diabetes and CKD and showed that patients from CALD backgrounds had lower activation scores compared to those from non-CALD backgrounds at baseline.Additionally, we have shown that a codesigned integrated kidney and diabetes model of care significantly improved activation levels of patients from CALD backgrounds, but not those from non-CALD backgrounds at 12 months follow-up.
Patient activation is low among patients with diabetes and CKD as described previously, 13,14 but in the current study, we report even much lower activation scores among the subgroup of patients from CALD backgrounds.In our sample, patients from CALD backgrounds scored 7 points less on the PAM compared to those from non-CALD backgrounds.Considering that a 1-point incremental change in activation equates to an improvement in health outcomes of about 3% and a reduction in health costs of about 3%, 38 it is clear that patients from CALD backgrounds may experience disparity in the quality and safety of health care.

Our finding of an association between low activation and CALD
background is important in several ways.First, our study opens up an opportunity to explore potential causal and mediating factors for this relationship.A previous study among elderly minority patients 39 reported that the relationship between African American race and lower patient activation was fully mediated by health literacy.
Another study among patients who attended primary care clinics reported lower activation scores among Black participants but suggested that the relationship between race and activation could be mediated by income. 40Unfortunately, in our study, we did not collect health literacy and income data to confirm this finding.
Second, when interpreting our results, it is important to consider that even though the PAM has been used across multiple racial, ethnic and language groups, 41 it is unknown if the contextual components of PAM are race or culturally specific. 17Participants who speak English have been reported to score significantly higher than non-English speakers 42 and this may explain why patients from non-CALD backgrounds scored higher in our study.Third, these results can inform the design of interventions that target people from CALD backgrounds among patients with comorbid diabetes and CKD to ensure disparities that exist in patient activation are identified and addressed.
One such intervention is our codesigned integrated kidney and diabetes model of care that significantly improved activation levels of patients from CALD backgrounds after 12 months.We observed an improvement of 8 points in patient activation, which doubled the minimal clinically important difference of 4 points 35,43 on the PAM scale.Components of this model of care have been described previously, 26 but key features that may have led to improvements in patient activation among patients from CALD backgrounds include the focus on person-centred care, interdisciplinary management of each patient and effective communication with the patient's primary care home team.With this approach, our model of care effectively improved access to health care by reducing barriers associated with navigating the healthcare system, for example, some primary care professionals who refer patients to the DKS felt that referrals for new patients were triaged and processed in a timely way compared to other individual specialist services. 44Additionally, the model of care allows for a strong and trusting relationship between healthcare professionals and patients to be built and this was facilitated by the ability of patients to see the same cohort of healthcare providers consistently. 45Supportive and trustworthy interactions enable patients to take a more active role in their health as they remove the inherent imbalance in power between patients and healthcare professionals. 46The improvement in patient activation is consistent with an Australian study 3 that incorporated core principles of the Patient-Centred Medical Home model among primary care patients presenting with chronic diseases.
We observed a significant improvement in activation among patients from CALD backgrounds even though both groups received the same intervention due to several reasons.8][49][50] Second, patients from non-CALD background who reported better activation scores at baseline may not have responded to the intervention because they had already attained optimal activation levels.In support of this, a prospective longitudinal study among patients with type 2 diabetes reported that patients who were at activation Level 3 at baseline were most likely to remain in Level 3 at follow-up. 51is study has several strengths and limitations.The study did not only demonstrate that patient activation is low among patients with diabetes and CKD from CALD backgrounds, but that a codesigned integrated kidney and diabetes model of care can effectively improve patient activation in this group of patients who are largely underrepresented in research.In terms of limitations, our study was not originally designed to examine small subgroups of patients; however, at baseline, there were similarities with respect to demographic and clinical characteristics between patients from CALD and those from non-CALD backgrounds.This meant that differences in activation scores at 12 months follow-up could be associated with the intervention.Additionally, we had a good representation of patients from CALD backgrounds based on similarities of the proportion of patients who spoke a language other than English reported by the 2021 Australian census (21.1%) 52 and in our study (27%).While it would have been interesting to determine the mediating effect of health literacy, we did not collect data to support these analyses.Future studies should examine whether health literacy has any mediating role in the relationship between patient  T A B L E 2 Adjusted multivariate regression analysis of the relationship between patient activation and CALD status at baseline.| 2591

F I G U R E 1
Strengthening the Reporting of Observational Studies in Epidemiology flow diagram of included patients.CALD, culturally and linguistically diverse; DKS, diabetes kidney service; eGFR, estimated glomerular filtration rate.

| 2589 F
I G U R E 2 Patient activation scores for patients from CALD and non-CALD backgrounds at baseline and 12 months follow-up.CALD, culturally and linguistically diverse background.F I G U R E 3 Sankey plot showing changes in patient activation levels from baseline to 12 months follow-up.Level 1 (score: <47.0),Level 2 (score: 47.1-55.1),Level 3 (score: 55.2-67.0)and Level 4 (score: >67.0).CALD, culturally and linguistically diverse background.activation and CALD status in patients with diabetes and CKD, and if this is the case, interventions that include elements targeted at improving health literacy and removing complexity from health systems need to be prioritised.In conclusion, among patients with diabetes and CKD, those from CALD backgrounds report worse activation scores.Interventions that target people from CALD backgrounds among patients with comorbid diabetes and CKD, such as the integrated kidney and diabetes model of care, are needed to address disparities that exist in patient activation.
T A B L E 1 Patient characteristics by CALD status at baseline.