Background
Diabetes mellitus can have an impact on behaviour, emotional well-being, and social life. This relationship is bidirectional, in that better psychosocial well-being can contribute to better health outcomes.1 2 Maintaining a high quality of life and promoting physical and psychological well-being are therefore considered to be of equal importance to reaching biomedical treatment goals, such as normal-range blood glucose levels.3 Indeed, the provision of ‘psychosocial care’ is now recognised as one of the pillars of state-of-the-art diabetes care.1 4 However, the implementation of this principle in routine clinical practice has proven difficult. Consequently, clinical diabetes management continues to be based primarily on biomedical outcomes such as HbA1c or time in glycaemic target range as measured by continuous glucose monitoring.5 6
Research studies report high comorbidity of diabetes and depression (10–20%)7, distress (44.6%),7 anxiety (28%)8 and eating disorders (6–50% depending on the criteria used)2 and show a link between these issues and metabolic outcomes.9 For this reason, current diabetes practice guidelines advocate person-centred psychosocial care, including empathic communication, individualised treatment and education, as well as psychological screening with patient-reported outcome measures (PROMs) assessing symptoms of depression, diabetes distress, anxiety, eating disorders and cognitive abilities.1 3 4 However, several barriers to translating these recommendations into practice have been reported, including limited time and resources, tensions with established medical and personal routines, and uncertainty and a lack of skills in assessing and dealing with psychosocial problems10 11 due to a lack of training and viable care pathways.5
There is additionally a lack of clarity about which psychosocial factors are considered relevant, useful, acceptable, and beneficial by people with diabetes.11 12 Indeed, many PROMs were developed for use in clinical research and there is limited evidence of benefits from their routine use in clinical practice.11 13 To address this issue, it is crucial to involve people with diabetes and healthcare professionals in the development of PROMs for clinical practice, their implementation, and the design of health service delivery,6 such as psychosocial care. However, so far, there is little evidence on how these stakeholders perceive and evaluate psychosocial screening procedures as part of diabetes care.14
Methods
Data source and participants
We used the final voting results of a recent international Delphi exercise, which aimed to achieve multi-stakeholder consensus on a set of diabetes-relevant outcomes for use in routine care.15 People with diabetes and healthcare professionals as well as representatives of the industry and authorities participated in a three-round questionnaire-based consensus process in which they rated the importance of including diabetes-relevant outcomes in the outcome set.15 These included eight psychosocial outcomes: psychological well-being, depression, anxiety, eating problems, diabetes distress, diabetes-specific quality of life, social support and sexual health (which included physiological and psychosocial aspects, such as sexual dysfunction, sexuality and family planning). All outcomes had been pre-selected based on a systematic literature review,16 focus groups with people with diabetes and healthcare providers, a national diabetes register17 and the ICHOM diabetes outcome set.18
The subset of participants included in this study were people with type 1 and type 2 diabetes (N=60) and healthcare professionals (N=45) who completed the third survey round. They were from 13 European countries, including Austria, Belgium, Denmark, France, Germany, Italy, the Netherlands, Norway, Romania, Spain, Sweden, Switzerland and the UK; 53% were women and 48% were >60 years old (table 1). Initial data analysis showed associations between stakeholder groups and diabetes type with healthcare professionals being typically experts for both diabetes types (62%) and people with diabetes living predominantly with type 2 (58%). Women who participated in the study were generally younger than men, people with diabetes were older and the proportion of men among them was higher than among healthcare professionals.
Participant characteristics
Data analysis
We used the same consensus threshold as applied in the original study, which was defined as ≥70% of both stakeholder groups rating an outcome ≥7 on a 10-point Likert scale (ranging from 1, meaning ‘not relevant at all’, to 10, meaning ‘of highest importance’).15 Therefore, we assumed that the participants of a stakeholder group considered an outcome important for holistic diabetes care if ≥70% of them rated it ≥7, whereas we assumed that an outcome was considered less important if <70% rated it ≥7. We created a binary variable for each outcome with the values ‘important’ (votes 7–10) and ‘less important’ (votes 1–6) and used cross-tabulation to compare the voting results of the two stakeholder groups for each outcome.
Multivariable linear regression analyses were performed for the outcomes that were considered less important (i.e., those that received <70% of votes ≥7) in one or both stakeholder groups to test the association between age, gender and/or diabetes type and the importance rating they received. The intention was to identify whether these outcomes might be regarded as important by certain subgroups of people with diabetes or healthcare providers, for example, depending on their age, gender, or the diabetes type they are living with and/or an expert for.
We conducted a residual analysis to test residual distribution and model misspecification. Where model assumptions were violated, we estimated the CI using a robust variance (sandwich estimator). For the regression models, we report global adjusted R2, regression coefficients and their 95% CIs. The importance of variables in the model was assessed by means of standardised regression coefficients and the change in R2. All analyses were done in R.19
Findings
Evaluations of psychosocial diabetes outcomes
For all eight outcomes included in the analysis, either positive (N=3) or negative (N=5) consensus was reached between people with diabetes and healthcare professionals, meaning that either ≥70% of both groups (positive consensus) or <70% of both groups (negative consensus) considered an outcome as ‘important’ (votes ≥7). There was no disagreement between the two stakeholders on any of the outcomes.
People with diabetes and healthcare professionals agreed on the importance of regularly assessing psychological well-being, diabetes distress, and diabetes-specific quality of life to achieve holistic diabetes management. They further agreed that it was less important to regularly assess depression, anxiety, eating problems, social support, and sexual health in this context. For most outcomes, the proportion of votes ≥7 was similar for people with diabetes and healthcare professionals, differing by a few percentage points (pp) only (0.6–4.4 pp). However, the difference was more pronounced for the importance placed on monitoring depression (62% of people with diabetes vs 52% of healthcare providers), distress (85% of people with diabetes vs 71% of healthcare providers) and social support (67% of people with diabetes vs 47% of healthcare providers) (table 2).
Proportion of people with diabetes and healthcare professionals who rated each outcome as ‘important’
Outcomes with relevance for specific subgroups
Multivariable linear regression analyses were performed for the five outcomes that had received <70% of votes ≥7 in both stakeholder groups. The aim was to investigate whether age, gender, and/or diabetes type can predict the importance attached to monitoring these outcomes in each of the stakeholder groups. Table 3 shows the regression coefficients and measures of model fit for each of the analyses. The model predicting the importance that people with diabetes attribute to regular assessment of eating problems was the only one to reach statistical significance (p=.02). It was able to explain 11.4% of the variability in how the importance of the outcome was rated. The variable most substantially associated with importance rating, indicated by a change in R2, was gender (change in R2=24.2%), followed by age (change in R2=8.7%) and diabetes type (change in R2=6.3%) (table 4). The model indicates that being a woman, being younger and living with type 1 diabetes are associated with a higher rating of the importance of regularly assessing eating problems in diabetes care.
Regression coefficients and measures of model fit for all linear regression analyses
Linear regression model for the perceived importance among people with diabetes of regularly assessing eating problems in diabetes care
Discussion
We set out to explore which diabetes-relevant psychosocial outcomes people with diabetes and healthcare professionals consider particularly important for routine diabetes care. We did so to inform a strategy for facilitating the routine provision of psychosocial diabetes care. We found that people with diabetes and healthcare professionals expressed the same opinion about which outcomes they wanted to regularly assess and use for diabetes management, including psychological well-being, quality of life, and diabetes distress. Likewise, they agreed that depression, anxiety, eating problems, social support, and sexual health were aspects that were less important to assess on a regular basis.
Interpretation
Our findings fit well with the positive, affirmative approach to psychosocial diabetes care called for by members of the Psychosocial Aspects of Diabetes Study Group (PSAD) who have criticised the problem-oriented focus that still prevails in diabetes practice.20 In fact, there is evidence that some people with diabetes may reject screening for psychological disorders because of stigma or fear of them.14 20 Diabetes stigma and the experience of discrimination are still commonplace for people with diabetes.21 Being associated with or experiencing other health conditions that are subject to stereotyping and stigmatisation, such as psychological disorders, can lead to experiencing ‘intersecting stigma’.22 The fear of additional (i.e., intersecting) stigma might have influenced the decision of the people with diabetes in our study to (perhaps unconsciously) place less importance on regularly screening for depression, anxiety, and eating disorders.
The healthcare professionals’ voting results may have been influenced by their professional identity, which is largely formed by what is expected of them in their field and by the competencies (e.g., knowledge and skills) they acquire to fulfil this role.23 Identifying and treating certain psychological issues, a task traditionally undertaken by psychologists and psychiatrists, may feel out of scope for healthcare providers specialising in the provision of diabetes care,14 possibly explaining why the healthcare professionals in our study did not vote for regularly assessing these outcomes in diabetes care. Similar reasoning may explain the limited importance attributed to routinely asking about social support. Indeed, it was shown that both healthcare professionals and people with diabetes are reluctant to initiate conversations about social support in consultations, as this topic is often seen as outside the remit of diabetes care professionals.11 Moreover, reluctance or embarrassment may be reasons why the participants did not speak out in favour of regularly addressing sexuality in routine diabetes care.24
For diabetes care to be effective and safe it is crucial that people with diabetes and healthcare providers are comfortable with the required processes, although it is also important that (potentially unpleasant) scientific evidence is translated into practice. Indeed, the risk of experiencing depression, anxiety, and problematic eating behaviour has been shown to be elevated in people living with diabetes compared with those living without.2 Yet, these outcomes were considered less important to regularly assess by people with diabetes and healthcare professionals in this analysis. Still, evidence suggests that assessing the symptoms of these conditions should be part of psychosocial diabetes management.1 25
We believe that well-coordinated interdisciplinary care and clear care pathways can enable holistic psychosocial diabetes care that is neither perceived as stigmatising or uncomfortable by people with diabetes nor as overwhelming or out of scope by healthcare providers.
Practical implications
Based on our findings and interpretation, we propose a two-staged approach for facilitating the implementation of psychosocial diabetes care in routine clinical practice, including (1) priority outcomes assessed regularly and universally for every adult with type 1 and type 2 diabetes, and (2) outcomes assessed because of specific preconditions, for example, when irregularities in the priority outcomes are detected. Following the conceptual categorisation of outcomes into health and quality of life outcomes,26 we arrive at the following two psychosocial care pathways: (1) indications of reduced psychological well-being and/or increased levels of diabetes distress should prompt an assessment of potential psychological disorders, such as depression, anxiety and disordered eating; (2) reports of reduced quality of life should trigger inquiries about social support and/or sexual health (figure 1).
Psychosocial care pathways for person-centred diabetes care. *The measures proposed here are suggestions, not exhaustive and need to be adapted to different settings, circumstances and available resources.
Psychological well-being pathway
Based on our findings, we suggest that psychosocial assessment in routine diabetes care begins with an overview of the person’s psychological well-being and level of diabetes distress. This approach can ensure that the experiences of people with diabetes are not ‘over-pathologised’,20 as emotional well-being is important for everyone’s health, and diabetes distress is considered a normal, expected emotional response to living with diabetes.27 A similar approach that used the WHO Five Well-Being Index (WHO-5) and Problem Areas in Diabetes (PAID) questionnaires to monitor the mental status of people with diabetes has been shown to be feasible and acceptable in a multicentre prospective observational study.13 The WHO-5 has been valued by people with diabetes as a mood screening tool and proved to be suitable for screening depression in this group.28 The PAID is suitable for identifying clinically relevant diabetes distress and supporting meaningful conversations in diabetes consultations.29 Depressive symptoms as measured by PROMs can often be partially explained by diabetes distress causing false-positive depression screening results, but these two presentations can be distinguished in professional psychological interviews.27 30 Thus, starting with an appraisal of psychological well-being and diabetes distress before deciding whether psychological disorders should be investigated by appropriate professionals can avoid false-positive findings, which can be alarming for the individual. Indeed, a step-by-step approach based on verbal screening questions, followed by a validated screening tool and finally a clinical interview has already been recommended in some places.30
Quality of life pathway
Based on the responses in this study, we suggest that quality of life should be regularly assessed in all adults with diabetes and that inquiry about social and sexual health may be considered thereafter. Both are recognised, along with others, as important domains of diabetes-specific quality of life.31 For example, sexual dysfunction (including vaginal dryness and pain, inability to orgasm and erectile dysfunction), which is more common in men and women with diabetes than in those without, and impaired sex life due to body image or fear of hypoglycaemia can affect the quality of life.24 Moreover, it was shown that social support can mitigate the effects of stress and distress on the quality of life of people with diabetes.32
Both social and sex life are intimate topics, and people with diabetes and healthcare providers may feel reluctant or uncomfortable talking (or asking) about them. This barrier may be overcome by starting with a less intrusive inquiry about the quality of life and then progressing further. Using a PROM for this purpose may help overcome inhibitions and facilitate conversation about these topics.11 For instance, a short questionnaire on diabetes-specific quality of life, such as the DAWN2 Impact of Diabetes Profile (DIDP),7 could be used in a way that the subscales of physical health and/or relationships elicit questions on social support and/or sexual health. Such an approach may be realised using computerised adaptive testing, which offers more flexibility by adapting the questions to the respective situation and reduces the burden on people with diabetes and medical staff when dealing with these issues.26 Addressing these issues can significantly improve the lives of people with diabetes as it can help them and their caregivers understand their situation more holistically and take the necessary steps to improve it.24 32
Paying attention to specific subgroups
We found that the importance of assessing eating problems was rated higher by women and younger participants with type 1 diabetes. This is consistent with the fact that their prevalence is particularly high in this group, that is, young people with either diabetes type and specifically young women with type 1 diabetes.33 Diabetes care providers should therefore pay particular attention to symptoms of problematic eating behaviour in people with these characteristics and consider actively addressing these issues with them.
There is also evidence of a link between age and gender of people with diabetes and their quality of life and level of distress, with women on average experiencing a lower quality of life and a higher level of distress.32 However, the fact that this is not reflected in the outcome preferences found in our analysis has led us to believe that their importance for person-centred psychosocial care is nevertheless universal.
Limitations
This is a retrospective and exploratory study, and the analyses performed were not adjusted for multiplicity. Our results should therefore be regarded as a source of new hypotheses. Based on our findings, we suggest a practice-oriented framework for psychosocial diabetes care that consists of two clear care pathways and might help facilitate its implementation in routine practice. Its feasibility, efficacy, and safety should now be investigated in future prospective studies.
Our findings may be limited by the rather small (N=105) and homogeneous (high-income to middle-income European countries) sample they are based on.15 Only limited information was available about the participants (e.g., self-reported gender, age group, diabetes type). It might be fruitful to consider the influence of other factors such as diabetes duration, treatment regimen or blood glucose control on outcome preferences. For example, long diabetes duration and poorly controlled blood glucose levels are associated with increased diabetes distress and potential psychological disorders.2 27 If it had been possible to include these variables in the regression models, specific outcome preferences might have been identified for these groups, which would have had an influence on the care pathways developed. Moreover, participants’ responses may have been influenced by the fact that they were asked to consider clinical practicability when rating outcome importance.15 It is possible that lower ratings reflect the belief that a regular assessment is more difficult to achieve for the respective outcome, which is why we believe that the step-by-step and interdisciplinary approach we propose can be helpful. Indeed, both people with diabetes and healthcare providers have expressed that it is crucial for psychosocial diabetes care that teams are interdisciplinary and that appropriate referral options are available.14
Clinical implications
We have developed two psychosocial care pathways for diabetes care based on the outcome preferences of people with diabetes and healthcare providers. We recognise that implementing new routines in healthcare takes time and effort.11 However, one can find positive examples that show that integrating psychosocial diabetes care into routine practice is both valuable and feasible. For example, a psychosocial screening and care pathway was developed and successfully integrated into routine diabetes care in Australia.34 The model is well received by patients and healthcare providers and consistent with our proposed model: it includes screening for diabetes distress and emotional well-being and provides a decision tree for following up on responses, suggesting usual care, diabetes education, or referral to a psychologist.34 This shows that previously reported implementation barriers, such as tensions with established routines and uncertainty and lack of guidance on how to respond to psychosocial screening results,10 11 35 can be overcome.
Other efforts to integrate the assessment of psychosocial outcomes into diabetes care have been hampered by a predominant focus on the use of lengthy, multidimensional PROMs,6 35 a lack of consideration of quality of life or psychological outcomes other than depression and diabetes distress,35 and limited evidence of sustainability and long-term impact.6 We believe that structuring outcome assessments according to the preferences of people with diabetes and healthcare providers in the way we have suggested can facilitate the routine and timely implementation of person-centred psychosocial care for people with diabetes. This will also require changes and support at the management and healthcare system level.
Future studies are required to test whether this is feasible, safe, and effective. It will also be important to investigate the applicability of the proposed care pathways in different health settings and in relation to available resources and refine them accordingly.
The insights we have derived can also assist in adapting the diabetes outcome set on the development on which our analyses were based to ensure its clinical feasibility, acceptability, and meaningfulness in the long term. An upcoming study on the implementation of the outcome set may contribute to testing the pathways proposed here.
