Glycemic Outcome Not Predicted by Baseline Psychological Measures in a Diabetes Management Program

Introduction

T ype 2 diabetes is now recognized as the fastest growing chronic disease in Australia and other western countries. The Australian prevalence of diabetes is 7.4%, with an additional 16.4% of the population having either impaired fasting glucose or impaired glucose tolerance. ¹ There are significant personal and societal costs associated with type 2 diabetes. ² In developed countries, diabetes is a leading cause of cardiovascular disease and renal failure, and it is a leading cause of blindness and nontraumatic lower limb amputations in those who are older than 60 years of age. ^3,4

Interventions to improve the health outcomes of people with type 2 diabetes have increased steadily over the last 10 years. Despite this, the burden of type 2 diabetes continues to increase in Australia and internationally. Given the magnitude of people with diabetes and finite health resources for treating chronic diseases, there is a pressing need to identify factors that compromise the effectiveness of current best-practice diabetes care models.

The Northern Health Diabetes Hospital Admission Risk Program (HARP) service is a multidisciplinary disease management service for patients with poorly controlled diabetes (defined as glycated hemoglobin [HbA1c]>8.0% or the presence of diabetes-related complications). HARP services are funded by the Victorian Department of Health; the Northern Health Diabetes Service is based in outer metropolitan Melbourne. The service offers multidisciplinary care comprising endocrinology, diabetes education, dietetics, physiotherapy, and podiatry services. Patients undergo a comprehensive multidisciplinary assessment and an individualized care plan is developed in consultation with the patient. The service uses an integrated self-management coaching approach to assist clients in making the behavioral changes necessary to improve the management of their diabetes.

Recently, there has been an increasing focus on patient psychological characteristics and their influence on diabetes interventions. Following discussions with clinicians from the HARP service and a review of the literature, it was hypothesized that patients who did not achieve adequate blood glucose control were more likely to have poor cognition, an external locus of control, anxiety and/or depression, and poor self-efficacy to manage their condition. ^{5 –15}

A previous program evaluation of the Northern Health Diabetes HARP service demonstrated that more than 20% of patients did not achieve target glycemic control (HbA1c<8%) 12 months after program enrolment. ¹⁶ Identifying those patients who are less likely to respond early after enrollment is the first step to offering them an alternative, more effective management pathway. The aim of this study was to determine which of these patient characteristics were associated with glycemic outcome (change in HbA1c) after 12 months.

Methods

A prospective cohort of patients attending the Northern Health Diabetes HARP service was recruited to the study at enrollment into the service (April to December 2008) and followed for 12 months. All new patients were screened for eligibility for the study. Exclusion criteria were: <18 years of age, significant intellectual disability, major psychiatric diagnosis, or documented dementia or dysphasia. This study was approved by the Northern Health Human Research Ethics Committee and informed consent was obtained from all participants prior to study enrollment.

At enrollment, the research nurse administered the following questionnaires to participants: Assessment of Quality of Life (AQoL); ^17,18 Diabetes Empowerment Scale-Short Form (DES-SF); ^19,20 Hospital Anxiety and Depression Scale (HADS); ^21,22 Diabetes-Specific Locus of Control Questionnaire; ¹² Rowland Universal Dementia Assessment Scale (RUDAS); ²³ and the Brief Diabetes Readiness to Change questionnaire. ¹¹ If the patient was not fluent in English, a professional interpreter was used to administer questionnaires. In addition to the questionnaires, the following routinely collected data were also obtained: demographic information, HbA1c, diabetes medications and adherence (upon admission to the service), and the presence of diabetes-related complications (following the first endocrinology assessment). At 12 months, HbA1c was obtained from the service medical record or the participant's general practitioner, and the number of contacts with the service was recorded.

Change in HbA1c was assessed using paired t tests. Univariate and multivariate linear regression analyses were performed to investigate associations between baseline demographic, cognitive, personality, self-efficacy, and mental health characteristics and change in HbA1c at 12 months. Only variables with more than 10 positive cases were included in the regression analysis to ensure stability of the models. Co-linearity and linearity assumptions were tested prior to multivariate analysis. Variables with a P value of ≤0.25 in univariate analysis were entered into the multivariate models.

Results

A total of 196 patients were screened; 5 were ineligible, 48 were unable to be contacted, and 17 declined to participate in the study. The final study group had 126 participants. Change in HbA1c after 12 months was obtained for 86 (68%) of the participants. Table 1 summarizes baseline statistics for participants with complete HbA1c data (N=86). Glycemic control was poor on admission to the service with a mean HbA1c of 8.9%. The mean HbA1c at 12 months was 8.2 (standard deviation [SD] 1.4). Mean reduction in HbA1c was 0.7% (SD 1.6), (P<0.0001), and 35% of participants achieved an improvement in HbA1c of 1% or more. Participants attended the service an average of 8 times; 10% attended fewer than 4 times.

When compared to the recruited sample of participants (N=126), those with follow-up HbA1c measures (N=86) were more likely to be female (42% vs. 49%, P=0.03) and more likely to have neuropathy (20% vs. 26%, P=0.03). There were no other significant differences in any of the baseline variables measured.

Measures of personality factors, cognition, mental health, and quality of life also are summarized in Table 1. Study participants had fair self-reported health status, which represents a significantly worse status than the general age-adjusted Australian population based on normative data (AQoL utility score=0.83, SD 0.20). ¹⁸ Anxiety was common, but few participants were classified as having probable depression (HADS score > 10).

The mean score for the DES-SF indicates that study participants had a moderate level of diabetes-related psychosocial self-efficacy. The locus of control questionnaire showed that study participants saw their diabetes control as something that they could individually influence, but that they also regarded health professionals and family as being able to strongly influence their diabetes control. Chance was viewed as a less important influence on their diabetes control. The readiness-to-change questionnaire indicates that 25% of clients were in the precontemplation and contemplation stages and therefore not considered to be “ready” to change their diabetes management behaviors, while 56% were in the preparation and action stage and were motivated to change their behavior to improve their diabetes control. There were no significant differences between the recruited sample (N=126) and those with follow-up HbA1c data (N=86) in any baseline questionnaire results.

Discussion

Evaluation of a multidisciplinary diabetes service demonstrated statistically and clinically significant improvements in HbA1c at 12 months. Patients with longer duration of diabetes were less likely to experience improvement in glycemic control and those with a higher number of contacts with the service had greater improvement in glycemic control. Patients who attended a minimum of 4 or more sessions were more likely to experience improvement in glycemic control while those who attended 6 or more times achieved the greatest improvement in glycemic control.

The HARP service was designed for individuals with poor glycemic control and diabetes-related vascular complications; therefore, it targeted individuals with a longer duration of disease. Aging is known to cause both a decrease in insulin production and an increase in insulin resistance. The progressive nature of diabetes means that these individuals may require more intensive medical management and complex medication regimens to achieve optimal glycemic control. ²⁴ This may explain why the number of service contacts and duration of disease were significant predictors of achieving target HbA1c levels, and suggests that such diabetes programs may be more effective earlier in the disease course, along with multiple (at least 4) clinician sessions.

The psychological measures in this study were not associated with glycemic control in contrast to findings of other studies. ^{5,7 –11} There are several potential explanations for these findings.

The prevalence of depression in this sample was relatively low at 7%. The prevalence of depression in adults with diabetes is reported in the literature to be between 11% and 31%, and to be higher in hospital/clinical setting than in community settings. ²⁵ We previously reported the prevalence of depression in the outpatient diabetes clinic population to be 16%. ²⁶ The lower prevalence of depression in this study cohort may suggest that the cohort is not entirely representative of a typical diabetes population and the low prevalence of depression may have made it difficult to demonstrate a relationship with glycemic outcome.

Some of the questionnaires used may not have been sensitive enough to measure the characteristics of participants in this study accurately because they were developed and validated in highly educated English-speaking populations. For example, both the DES-SF and Diabetes-Specific Locus of Control questionnaires were developed and validated in predominantly college-educated populations. ^12,19,20 In contrast, half of this population did not receive any schooling beyond primary school, and over one third preferred to speak a language other than English.

The finding that locus of control was not associated with improvements in glycemic control is consistent with a recently published meta-analysis of 17 studies, which also concluded that there is no correlation between locus of control and glycemic outcome. ²⁷

This study utilized a series of questionnaires that focus on evaluation of baseline beliefs (self-efficacy, stages of change, locus of control) and impairments/health status (depression and anxiety, cognitive function, quality of life). It did not assess or capture information relating to behaviors. The premise of the study was that beliefs and impairments could influence behaviors, and therefore outcomes. Clearly, there may have been aspects of behavior change (that were not measured) that may have influenced outcome (eg, carer support or access to additional primary health care services).

Importantly, this sample may have been too small to identify variations in the personality characteristics measured and the glycemic control outcome. Although there were no major significant differences in the baseline characteristics of the overall sample (n=126) and the analysis sample (n=86), a 32% loss to follow-up highlights the difficulty with adherence in this population. In addition, 70 patients who were screened for participation in the study (n=196) were not recruited. Although only minimal data were collected on these participants, we know that there were no significant differences in age and sex between those screened and those who consented. There may have been systematic differences in other baseline characteristics that could have affected the outcomes of the study.

This study sought to explore associations between personality characteristics and glycemic outcomes for patients attending a diabetes service. Further research is required to develop and validate instruments that measure self-efficacy in culturally and linguistically diverse populations with poor health literacy. Longitudinal studies of trends in stages of change throughout a patient's progression through treatment should be undertaken to provide more detailed information on their capacity to move between stages of change while undertaking chronic disease self-management programs. Larger longitudinal cohort studies also are required to study in detail the personal and environmental factors that influence an individual's capacity to modify behavior and improve glycemic and health outcomes.

The measures of cognition, diabetes self-efficacy, locus of control, mental health, and quality of life used in this study were not associated with improvements in glycemic control at 12 months. Creation of more sensitive psychometric questionnaires is necessary to profile patients with diabetes, given the great educational and cultural heterogeneity that exists in such populations. Importantly, this study identified that those with longer duration of disease and fewer contacts with the service were significantly less likely achieve optimal glycemic control, suggesting that chronic disease management programs may be more effective if they target participants earlier in their disease and provide a more intensive intervention.