Comorbidity and lack of education countered participation in a diabetes prevention self-management program


  • Helle Terkildsen Maindal RN, MPH, PhD,

  • Mette Vinther Skriver MSc,

  • Marit Kirkevold RN, DrEd,

  • Torsten Lauritzen MD, DrMed,

  • Annelli Sandbæk MD, PhD

Helle Terkildsen Maindal
Section of General Practice
School of Public Health
Aarhus University
Bartholins Allé 2
8000 Aarhus C
Telephone: +45 89 42 60 44


Maindal HT, Skriver MV, Kirkevold M, Lauritzen T & Sandbæk A (2011) Journal of Nursing and Healthcare of Chronic Illness3, 293–301
Comorbidity and lack of education countered participation in a diabetes prevention self-management program

Aim.  To investigate socio-economic and disease-related predictors for non-participation in the ‘Ready to Act’ program offering self-management support to people with screen-detected dysglycaemia.

Background.  Screening for type 2 diabetes followed by support to people’s self-management strategies is increasingly being offered in primary care. Due to non-participation in self-management programs, healthcare providers often miss the opportunity to provide the basic knowledge and skills resulting in uninformed self-management decisions.

Methods.  A prospective cohort-study was conducted in 2010 within the framework of the Danish part of the Anglo–Danish–Dutch study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care (ADDITION). A total of 322 43- to 75-year-old screen-detected patients, receiving GP-multi-faceted intensive treatment were invited to an additional interdisciplinary 12-week self-management program ‘Ready to Act’. A comparison between participants and non-participants was conducted according to socio-economic outcomes: age, gender, cohabitant and education and disease-related outcomes: diagnosis (diabetes or prediabetes), duration of diagnosis, comorbidity measured by Charlson’s combined 17 medical conditions measure.

Results.  The participant rate was 44%. The most pronounced predictor for participation was vocational education. With no vocational education as the reference group, the OR for participation was 1·91 (95% CI 1·06–3·44) for 1–3 years of vocational education and 2·65 (95% CI 1·31–5·39) for more than 3 years of vocational education. Second, having comorbidity (a score of one or more at Charlson’s index) tended to decrease participation (OR 0·6, 95% CI 0·34–1·06).

Conclusion.  Educational level and comorbidity were the most decisive factors that influenced attendance to a diabetes prevention program. Both factors are available in daily clinical practice for healthcare planning.

Relevance to clinical practice.  The results incite improved awareness of how to develop different recruitment strategies accounting for educational level and comorbidity, involve the target groups in intervention development, and include individual, contextual and health-system factors when shaping self-management interventions.


Type 2 diabetes account for almost 90% of all cases of diabetes worldwide, and pose a significant health problem with increasing prevalence and costs to society (Shaw et al. 2010). The benefits of early detection and intensive treatment of diabetes followed by self-management support are achieving increasing awareness in public health (Deedwania & Fonseca 2005, Gillies et al. 2007, Sandbaek et al. 2008). Non-participation in self-management programs continues to challenge research and general practice, and must be taken into account when planning self-management support (Glasgow et al. 1996, Toobert et al. 2002). A lack of self-management knowledge and skills in the early stages of disease development may have an adverse effect on the patient’s chronic disease trajectory, if their make decisions and actions based on incomplete or wrong assumptions (Adriaanse et al. 2002, Parry et al. 2004).


Diabetes prevention research trials have shown how preventive medication and improved dietary and physical activity reduces the risk of diabetes development by up to 58% (Pan et al. 1997, Tuomilehto et al. 2001, Knowler et al. 2002). Studies undertaken in the USA and Finland had high participation rates (up to 85%), but these studies lack external validity as they are carried out in highly motivated volunteers without comorbidity, and the self-management interventions were comprehensive and resource-demanding, although cost-effective (Tuomilehto et al. 2001, Knowler et al. 2002). Results from ‘ideal’ conditions in research are not always transferable into real-life settings, and they entail a major challenge of further translation and implementation according to components such as incomplete delivery of program and non-participation (Glasgow et al. 1996, Green & Glasgow 2006). A review by Paterson et al. (2010) reported that the factors influencing attendance in chronic disease clinics are complex and multi-factorial. One study recommendation was a greater understanding of missed appointments and non-attendance among health providers. One way of increasing the understanding is by enhanced awareness of the socio-economic, cultural, demographic, and other influences that may affect a person’s decision to enrol in a chronic disease self-management program.

Thoolen’s study of the Dutch self-management program ‘Beyond Good Intentions’ (Thoolen et al. 2007) conducted by general nurse practitioners and delivered in primary care after screening for type 2 diabetes, had a participating rate of 49%. The most important predictor for attending the program was baseline self-management skills assessed by the summary of Diabetes Self-Care Activities Measure. In a similar American diabetes prevention program, the preintervention level of ‘readiness to change’ measured by a ‘stage of change’ questionnaire was found to be predictive for attendance (Helitzer et al. 2007).

The use of psychosocial measures such as ‘readiness to change’ as screening measurements for healthcare planning presuppose that the target population is well known, but this is seldom the case in prevention studies. The patient’s level of self-management or readiness to change may not be obvious for the healthcare provider before referral to preventive self-management programs, whereas more objective data like age or marital status in most cases is accessible. In Danish healthcare, the general practitioner staff have access to a variety of socio-economic and disease-related routine data about each of the patients on their lists. These data may help to targeted screening interventions, even before seeing the patient. The association between socioeconomic status and health is well known, and many determinants of these health inequalities have been studied. The results of the role of socioeconomic status when using health care and self-management programs are however, conflicting (Mielck et al. 2006).


In this study, we aimed to identify decisive socio-economic and disease-related factors available in a GP setting for signing up for a self-management program after screening for type 2 diabetes in general practice. Our hypothesis is that high socio-economic status and absence of competing diseases increases the probability of participating. We further hypothesise that having fully developed diabetes compared with prediabetic stages increases the probability of participating. Participants are defined as people with dysglycaemia signing up for the ‘Ready to Act’ program. Dysglycaemia is in this study used as the common term for the two prediabetic stages of impaired fasting glucoses (IFG), impaired glucose tolerance (IGT) and type 2 diabetes (T2D). The ‘Ready to Act’ program aims to promote health-related action competence in living with dysglycaemia and involved four learning objectives; motivation, informed decision-making, action experience and social involvement (Maindal et al. 2010, 2011).


A prospective cohort study was conducted in 2010 in a population of 322 persons aged 43–75 years [mean 62·2 SD (6·9)] living in a county of Denmark. They were all recruited from the intervention group of the ‘Ready to Act’ self-management study conducted in 2007 in the Danish part of the ADDITION-trial: ‘Anglo–Danish–Dutch study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care’ (Lauritzen et al. 2000, Sandbaek et al. 2008), which is an international population-based screening and treatment study for type 2 diabetes.

The ‘Ready to Act’ study investigated an optimised intervention consisting of intensive multi-factorial treatment by GPs (medication and lifestyle counselling) supplemented by the 12-week self-management program: ‘Ready to Act’ in a 2:1 prerandomization procedure (Schellings et al. 2005) among 509 persons diagnosed with dysglycaemia.


In the study, 322 participants were allocated to the optimised intervention, and 187 to GP treatment according to the ADDITION protocol (Fig. 1). All GPs in the ADDITION intervention arm were trained to motivate and provide target-driven intensive non-pharmacologic and pharmacologic treatment to people with screen-detected T2D. People with IGT and IFG were to be treated according to the national clinical guidelines for cardiovascular disease prevention [Danish Association of General Practice (DSAM) 2007].

Figure 1.

 Flowchart of study population. *All participants were recruited from GPs in the intensive treatment arm of the ADDITION-study (The Anglo-Danish-Dutch Study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care) in Denmark.

The study cohort (n = 322) received one invitation from the ‘Ready to Act’ program sent from the research unit to their home address followed by a reminder, if they had not answered within 3 weeks. The invitation consisted of an introductory letter to the practical issues, and a 4-page pamphlet with information about diabetes, the program and quotes from former participants about their outcome in a similar program. No further attempts to increase the response rate were made.

The ‘Ready to Act’ program was delivered in primary care settings, e.g. GP clinics or health centres and consisted of two individual counselling interviews and eight group sessions, which totalled 18 hours within a period of 3 months. The first individual session was provided by a nurse and focused on the participant’s motivation, needs and readiness to change. This was followed by group sessions with 8–14 participants led by nurses, dieticians, physiotherapists and GPs and comprised eight themes within the fields of action competence, cardiovascular risk and dysglycaemia. When the program ended, the participants were offered a second individual feedback and planning session with a nurse. We used self-directed action plans to support realistic goals and health actions throughout the program (Maindal et al. 2010).

Data collection

A comparison between those who accepted to participate in the program and those who did not was conducted according to the following socio-economic variables; age, gender, civil status and education and disease-related variables; diagnosis (IFG/IGT or diabetes), duration of diagnosis and comorbidity.

Socio-economic measures

Data on age and gender were derived from the Danish Civil Registration System, which supplies an unambiguous identification number to all residents in Denmark. Civil status data and education were obtained from self-administrated questionnaires. Civil status was defined as either cohabitant or single. Education was defined as the highest formal vocational education attained and was categorised into three groups: no education, 1–3 years of education and more than 3 years of education.

Disease-related measures

Data on diagnosis and body mass index (BMI) were available from our own research database and supplied by GPs case records. All disease-related data were collected at the time of study entrance from the Hospital Discharge Registry. Comorbidity was assessed using Charlson’s combined comorbidity index, which encompasses 19 medical conditions such as congestive heart failure, chronic obstructive pulmonary disease and cancer weighted 1–6 based on severity (Charlson et al. 1987). We excluded type 1 and type 2 diabetes, leaving 17 conditions in the index for the analysis.

Outcome measures

Participation was defined as a positive answer and non-participation defined as a negative or no answer to the invitation of attending the ‘Ready to Act’ program.

Ethical considerations

Ethics approval was sought and gained from the Regional Ethics Committee. Participants were informed of their right to withdraw from the study at any time without consequence and assured that confidentiality would be maintained at all times. Participants were asked to sign a consent form prior to participation. The Danish Data Protection Agency approved the study and the maintenance of the database. The ADDITION study is registered as a clinical trial at, registration no. NCT00237549.

Statistical analysis

Characteristics of the population were reported as counts and percentage for categorical data and as means and standard deviation for continuous variables. The association between participating in the self-management program and socio-economic and disease-related factors was analyzed by use of odds ratio with 95% confidence intervals. Both bivariate and multivariate analyses were performed. Crude and adjusted odds ratios were calculated using logistic regression. All variables were included in the multivariate model and subsequently adjusted. Individuals with missing values for one or more of the included variables were excluded from the multiple analyses. Binomial probability tests using large-sample statistic were used to evaluate whether any of the 17 medical diseases included in the Charlson index had a specific impact on participation. Statistical analysis of data was conducted in stata version 10.0 (Statscorp, Texas, USA).


The baseline characteristic of the study population is shown in Table 1. The study population had an average age at 62·3 years, 53% was male and 54% had diabetes. 35·6% had a Charlson’s comorbidity index score of 1 or more (Tables 2 and 3). Totalling 142 persons (44%) signed up for the ‘Ready to Act’ program, whereas 180 (56%) were categorised as non-participants. Among the participants, 123 of 142 (87%) completed the 12-week program. Those who did not complete the program, despite signing up (n = 19) gave reasons marginal to the program, e.g. vacation or other appointments at the proposed time schedule. There were no differences in baseline characteristics between those who completed the program and those who signed up for the program without completing it.

Table 1.   Baseline characteristics of participants and non-participants in the ‘Ready to Act’ program (n = 322)
Age [mean (SD)], year62·36·918062·07·0142
Gender, female (%)46·1 18048·6 142
Cohabit, not living alone (%)73·3 17677·9 140
Vocational education
 No (%)39·5 15726·8 123
 1–3 years (%)43·3 46·3 
 More than 3 years (%)17·2 26·8 
Diagnosis, IFG & IGT (%)47·8 18043·7 142
Duration of diagnosis [mean (SD)], year1·91·81801·71·7142
Body mass index [mean (SD)], kg/m229·24·717630·75·4139
Charlson comorbidity index (%)
 064·4 18077·5 142
 1+35·6 22·5 
Table 2.   Socio-economic and disease-related factors associated to participation
 nOR crude95% CIOR adjusted* (n = 272)95% CI
  1. *All variables in the table were included in the multivariate model and subsequently adjusted.

 Male322Ref Ref 
Civil status
 Alone316Ref Ref 
 No vocational education280Ref Ref 
 Vocational education 1–3 years0·630·36–1·080·520·29–0·94
 Vocational education 3+ years0·440·22–0·840·380·19–0·77
 IFG or IGT322Ref Ref 
 Type 20·850·54–1·320·740·44–1·27
Duration (years)
 <1322Ref Ref 
Body mass index
 <25315Ref Ref 
Charlson score
 0322Ref Ref 
Table 3.   Diseases in the Charlson’s comorbidity index among participants and non-participants
Diagnostic categoryIndex pointNon-participants (n = 180)Participants (n = 142)%%p-value
  1. The bold p-value 0·001 is the significant value.

Myocardial infarction11176·114·930·323
Congestive heart failure11417·780·700·001
Peripheral vascular disease1412·220·700·137
Cerebrovascular disease11588·335·630·175
Chronic pulmonary disease11467·784·230·094
Rheumatologic disease1432·222·110·473
Peptic ulcer disease1763·894·230·561
Mild liver disease1311·670·700·219
Hemiplegia or paraplegia2000·000·00
Renal disease2100·560·000·187
Moderate to severe liver disease3000·000·00
Metastatic solid tumour6301·670·000·061

Table 2 presents socio-economic and disease-related variables in relation to participation. The crude, unadjusted estimates show that the probability of participating in the self-management program were statistically significantly higher among people with a vocational education lasting more than 3 years (OR 2·30; 95% CI 1·19–4·45) and statistically significantly lower among people with other medical conditions, represented with Charlson’s comorbidity index of 1 or more (OR 0·53; 95% CI 0·32–0·87).

The adjusted estimates in Table 2 confirm the impact of education stating that the higher level of education increased the participation rate; namely with no vocational education as reference group, we found an OR for participation of 1·91 (95% CI 1·06–3·44) for 1–3 years of vocational education and 2·65 (95% CI 1·31–5·39) for more than 3 years of vocational education. Comorbidity still tends to be the number two slot for predicting participation (OR 0·60; 95% CI 0·34–1·06). Furthermore, there seems to be an association between BMI and participation, i.e. that the higher the BMI the higher probability of participating, but this is not statistically significant. There was no difference between participants and non-participants in this sample in terms of age, gender, cohabitant, diagnose and duration of diagnosis.

Of the 17 medical conditions included in the Charlson’s comorbidity index, the patients suffering from congestive heart failure varied mostly between the participants (0·7%) and non-participants (7·8%), p < 0·001 (Table 3).


A total of 44% of 322 people screen-detected with a diagnosis of IGT, IFG or type 2 diabetes signed up for the 12-week self-management ‘Ready to Act’ program and 123 (87%) among these completed the program. The hypothesis in this study stating that high socio-economic status and absence of competing diseases increased the probability of participating was confirmed.

Vocational education compared with no education was found to be the most important factor predicting participation. This is in line with other studies in the field, some of the studies found that each social and educational stratum differs according to use of health care, health and mortality (Anderson et al. 2005, Lakerveld et al. 2008). Comorbidity was present for more than one third of the study population and proved to be another important predictor of participation. Persons with one or more medical conditions were more likely not to participate. According to comorbidity barriers, it is notable how persons suffering from Congestive Heart Failure participate less. This is from a patient’s perspective very understandable as the condition may be severe and life-threatening. Patients with Congestive Heart Failure may have achieved health information and self-management support in early stages of their disease trajectory, but the complex understanding of how dysglycaemia interacts with their heart condition may be an entirely new agenda. So, from a healthcare perspective, this specific population requires exactly the same information and skills as others with a dysglycaemic condition, e.g. physical activity is crucial for this specific group, preferably with guidance to prevent uncertainty and overload.

Participation or lack of participation are however not necessarily associated with disease severity. The second hypothesis stating that people having fully developed diabetes compared with prediabetic stages would be more likely to participate than those in a prediabetic state was not confirmed. The almost uniform participation rate among people with diabetes and prediabetes demonstrated that factors other than the diagnosis are important to take into account in further development of strategies for recruitment. Engagement, perceived degree of control, illness perception, coherence to symptoms and treatment effects may influence participating, but also circumstances in the health systems such as staff involvement, being inpatient or outpatient or being rewarded for participation (Roberts & Bailey 2011).

Strengths and limitations

A strength of this study was that the intervention was offered in a ‘real-life’ environment in primary care, either in healthcare centres or GP clinics. Participants were recruited by GPs. All GPs in Denmark are independent contractors with the public scheme (the regional health authorities). They act as gatekeepers as 98% of the Danish population is registered with a particular GP with whom they consult for medical advice and referral to other medical specialties or community programs. Approximately 90% of the population contacts their general practice annually, and the average number of contacts is eight, so as such the general practice is the most obvious healthcare site in Denmark to recruit from. It is however possible, that this particular study population were more motivated than a ‘realistic’ population due to being participants in the ADDITION screening research study (Sandbaek et al. 2008). On the other hand, the population could have been less motivated compared with people in conventional treatment in a realistic clinical setting, due to the fact that they were already being offered optimal medical and behavioural treatment initiated by their GP.

The Danish registers provide complete data on almost all variables, making the analyses of participation rates possible postintervention. Achieving the comorbidity data was especially a unique opportunity (Tables 2 and 3). Comorbidity was an important predictor of participation and in this study, comorbidity data may even be underestimated, as they are based on hospitalisation. Some of the diseases (e.g. dementia, peptic ulcer disease) in the index are more likely to be diagnosed and treated in general practice in Denmark. The direction of the bias that these potential information problems impact on the study conclusion remain unknown.

Only data on education was included in the socio-economic analyses, as former studies have shown a correlation between education, income and occupation (Dalsgaard et al. 2009). Furthermore, education is known to be a strong predictor for health and health behaviour, but the causal pathways and possible adjustable factors to the associations between education and health are still preliminary investigated (Ross & Wu 1996). A British study (Singh-Manoux et al. 2002) stated that a comparison of the relative importance of the different measures of social position in predicting health is meaningless if the causal relationships among these measures are not accounted for. Whether data on income and occupation would have changed the conclusions in this study remains unknown as these data were not available.

Relevance for clinical practice

Implication of this study could be to differentiate the invitation strategies according to education and comorbidity and to target the content of preventive self management support after screening to the participants needs. The impact of education on the recruitment of patients for self-management programs may be reduced by revising the invitation material. Analyses of how the material appeals to people with a low educational level and low health literacy are needed. Blanch et al. (2008) aimed to eliminate health literacy-related barriers to participation in an arthritis self-management study by having only oral interaction and by using short, everyday terms, and they found no evidence of refusal to participate due to educational attainment or gender. The possibility of having an oral communication in the diabetes prevention recruitment phase is present in the Danish healthcare system due to the mentioned regular contact between the GP and patients.

Another way of complying with non-participation is to differentiate the content of the self-management programs offered. Basic and advanced levels may appeal to people with different educational levels, as well as programs based on different pedagogic approaches such as skills training, psychological education or dissemination of information may appeal to different target groups. A step closer to shape the interventions right could be to involve the target groups more actively in the early program development stage by using for example participatory action research (Reason & Bradbury 2009).

Furthermore, the lack of impact of prediabetes and diabetes diagnoses’ duration is important information for healthcare planners. A common recommendation seems to be that a self-management program should be offered at the time of diagnosis as many people are highly motivated at that point (Helitzer et al. 2007, Thoolen et al. 2007). As a consequence, a future recommendation could be that people with chronic illnesses should have the opportunity to enter a self-management program at different stages of their disease trajectory.

According to the ‘Ready to Act’ program an implication of this study is to target the interventions offered to a larger degree, by for example making the first individual interview a separate independent part of the preventive program. An extended individual interview made by the GP or the clinic nurse could be the first step of a tailor-made intervention, where people are given the opportunity to join the 12-week program. One possibility would be to investigate the ‘readiness of change’ level or self-management level, as recommended in former studies (Toobert et al. 2002, Thoolen et al. 2007). Still, a study of participation in an outpatient preventive cardiovascular disease lifestyle intervention (Lakerveld et al. 2008) found no difference in such psychosocial measures between the participants and the non-participants. The study reported participants to be younger, single and with a higher level of education than non-participants.

These implications underline the urgent need for innovative, systematic approaches to increase the participation in self-management programs. The healthcare planners should systematically consider the predictors for inequality in participation in self-management programs, and give them high priority when translating from efficacy studies to effectiveness studies and dissemination studies (Green & Glasgow 2006, Sussman et al. 2006). Paterson et al. (2010) stated that most non-participation research has focused on individual predictors, whereas much less attention has been paid to the contextual or health-system factors. This study points out components of importance for contextual and health-system factors and shows a new direction for research in strategies in the field of self-management support.


Low educational level and comorbidity were the most decisive factors that hindered participation in the early self-management program offered to people with dysglycaemia. Neither vocational level nor comorbidity are changeable factors, but data on both factors are easily accessible to healthcare providers from GPs’ case records and can as such be used as components in a revitalised approach to recruitment strategies in primary care. This is in relation to both screening and interventions following screening procedures, such as self-management programs.


Our appreciation goes out to all the participants who generously volunteered to enter the study, as well as all the healthcare providers who delivered the ‘Ready to Act’ program. We thank the ADDITION-secretariat for administering the logistic procedures in the study.

Funding source

Our work was funded by the Center for Innovation in Nursing Education in Aarhus, the Danish Nurses’ Organization, Novo Nordic (Clinical Nursing) and the Danish Diabetes Association.


Study design: HT, MV, MK; data collection and analysis: HT, MV and manuscript preparation: HT, MV, MK, AS, TL.

Conflicts of interest