Associations between knowledge, illness perceptions, self-management and metabolic control of type 2 diabetes among African and European-origin patients


Dr Abdul-Razak Abubakari
Department of Health and Applied Social Sciences
Liverpool Hope University
Hope Park
L16 9JD
Telephone: 0044-151-291-3795


abubakari a-r, jones mc, lauder w, kirk a, anderson j & devendra d (2011) Journal of Nursing and Healthcare of Chronic Illness3, 245–256
Associations between knowledge, illness perceptions, self-management and metabolic control of type 2 diabetes among African and European-origin patients

Aims.  Using Leventhal’s Common-sense self-regulation model, this study investigated associations between illness perceptions, self-management and metabolic-control outcomes for diabetes among European and African-origin patients with type 2 diabetes.

Background.  Compared to the general populations of their host countries, African-origin populations in the diaspora are disproportionately affected by diabetes and its microvascular complications. However, comparatively little is known about how African-origin patients with type 2 diabetes perceive about their condition and whether such perceptions influence their self-management behaviours.

Design.  Cross-sectional design.

Methods.  Participants were recruited from diabetes clinics in London. Illness perceptions and adherence to self-management recommendations were assessed using questionnaires and data for metabolic-control outcomes were obtained from patient records. Associations between variables were investigated using correlations and multiple-regression techniques. Data collection for the study was conducted between February and June 2008.

Results.  In all, 359 patients participated in the study. The results indicated that perceiving diabetes with severe consequences was associated with poor self-management in both African-origin (black-African and black-Caribbean) and European-origin (white-British) patients with type 2 diabetes. However, personal control perceptions were associated with better self-management in African-origin patients. In multivariate analysis, illness perceptions explained significant proportions of the variations in self-management behaviours in both European-origin and African-origin patients. Perceived personal control over diabetes was a main determinant of the variations in self-management among European and African-origin patients. Other dimensions of illness perceptions, including consequences and emotional representations contributed to variations in self-management among African-origin patients. After controlling for demographic/disease characteristics, self-management of diabetes did not predict any metabolic-control outcome for diabetes in any ethnic group.

Conclusion.  Illness perceptions are important determinants of self-management among African-origin patients.

Relevance to clinical practice.  Illness perception-based interventions may be used to promote self-management among African-origin patients. Healthcare professionals should consider using patients’ illness perceptions to inform service delivery.


Type 2 diabetes is a chronic condition with significant medical and psychological impact on patients and their families. Patients with diabetes are at increased risk of both microvascular and macrovascular complications, including eye, nerve, kidney and heart disease (Fowler 2008). Due to the burden and the sequelae associated with diabetes, patients are encouraged to perform daily self-management activities to minimise, or at least, delay the onset of long-term complications (Stratton et al. 2000). The recommended daily self-management activities for diabetes patients include dietary regulation, physical exercise, self-monitoring of blood glucose (SMBG) and foot care [American Diabetes Association (ADA) 2008; National Institute for Health and Clinical Excellence (NICE) 2008]. Considerable lifestyle changes are often required if patients are to meet the recommended self-management activities.

A range of psychological frameworks exist to explain patient health-related actions in response to an illness (Ajzen 1991, Prochaska & Velicer 1997, Bandura 2001, Leventhal et al. 2003). One such important framework is the Leventhal’s Common-sense self-regulation model (CS-SRM) (Leventhal et al. 2003). According to this framework, individuals’ faced with a health threat, such as the experience of symptoms or diagnoses, tend to form emotional and cognitive representations about their condition. These representations are important determinants of the coping strategies and behaviours individuals adopt in response to the perceived health problem. Illness perceptions may be unique to the individual, but they are significantly influenced by patients’ cultural and social contexts, as well as by their past experiences and beliefs about health in general, and the disease condition in particular (Diefenbach & Leventhal 1996). As a common-sense scientist, the individual also appraises the effectiveness of the coping and health-seeking strategies that are subsequently adopted, which may lead to adjustments or changes in representations, health-seeking behaviours, or both (Leventhal et al. 2003). The CS-SRM has been used widely as a guide for understanding and predicting patients’ adherence to treatment and disease management recommendations (Horne & Weinman 2002, Horne 2003). Most of the studies guided by this framework are, however, conducted in European-origin populations. The few studies conducted with multi-cultural patient populations show different patterns of correlations in different groups, suggesting that illness perceptions may explain self-management behaviour differently for different ethnic groups (Barnes et al. 2004, Bean et al. 2007). Therefore, findings from the largely European-origin patient populations may not be applicable to those from minority ethnic groups, including African-origin patients.


Depending on the country, purpose and time of classification, terminologies such as black-African, black-Caribbean, African-Caribbean or African-American, have been used to describe African-origin populations (Agyemang et al. 2005). Regardless of their route of migration or present location, these populations of African ancestry seem to share common health problems including diabetes (Rotimi et al. 1995, Agyemang et al. 2009). For example, compared to the general populations of the UK, the USA, or even Australia African-origin populations are disproportionately affected by diabetes and its microvascular complications (Harris et al. 1998, Whitty et al. 1999, Saleh et al. 2002). For example, the prevalence of diabetes among the black-Caribbean population in England is more than twice (9·0% vs. 3·8%) that in the general population (Department of Health 2005). Diabetes patients of African-origin also have higher incidence of microvascular complications and treatment outcomes (Karter et al. 2002, Baskar et al. 2006). For instance, the risk of renal failure as a result of diabetes is 6·5 times higher among patients of African-origin compared to their European-origin counterparts (Roderick et al. 1996, Raleigh 1997). Similar to the onset of diabetes itself, the incidence of renal replacement therapy and renal failure also occurs at a much younger age in African-origin populations than their European counterparts (Lambie et al. 2008). Three decades ago diabetes was uncommon among indigenous African populations. However, recent reviews suggest that prevalence of diabetes and its risk factors in sub-Saharan Africa may soon mirror figures in the developed world (Abubakari et al. 2008, 2009, Mbanya et al. 2010).

Despite their high susceptibility to diabetes and its complications, comparatively little is known regarding how African-origin patients represent their diabetes. It is also not clear if illness perceptions about diabetes in African-origin patients, as captured by existing measures (Moss-Morris et al. 2002), predict the extent of their adherence to self-management recommendations and glycaemic control.


The aim of this study was to investigate the relationship between diabetes-specific knowledge, illness perceptions and self-management behaviours, and the relationship between self-management behaviours and metabolic control among African-origin patients in the UK.


For the purpose of validating our findings and for reasons of comparison as reported in other studies, it was deemed appropriate to also investigate these relationships in European-origin patients (Barnes et al. 2004, Bean et al. 2007). In line with the conceptual framework in Fig. 1, our investigations tested the hypotheses that among European and African-origin populations with type 2 diabetes:

Figure 1.

 Conceptual model of the hypothesised relationships between patients’ diabetes knowledge, illness perceptions, levels of self-management and metabolic control.

  • 1 Diabetes-specific knowledge will be associated with adherence to self-management recommendations.
  • 2 Diabetes-specific knowledge will significantly correlate with dimensions of the Revised Illness Perceptions Questionnaire (Moss-Morris et al. 2002).
  • 3 Illness representations by patients will be significantly associated with their adherence to self-management recommendations and
  • 4 Adherence to self-management recommendations will be associated with metabolic-control outcomes for diabetes.

To avoid confusion in terminologies the African-origin populations used in this study refer to black-African and black-Caribbean participants combined whilst European-origin refers to white-British patients (see Agyemang et al. 2009 for a note on ethnicity).


Participants and procedures

Participants in this study were recruited consecutively from diabetes and retinal screening clinics in the London boroughs of Brent and Hackney, UK. Hackney is in Inner London and Brent is in Outer London. Both boroughs have diverse ethnic populations, with higher proportions of African-origin populations compared to other parts of the UK. Twenty-five per cent of the population of Hackney and 20% of Brent are of African ancestry (Office for National Statistics 2001).

Patients waiting for their appointments at the above clinics were approached with an invitation letter, participant information sheet, and informed consent forms of the study. Patients who agreed to take part in the study were asked to sign the consent form and subsequently complete the study questionnaires. For patients who could understand and speak English but were unable or had difficulty reading, the consent form and the questionnaires were administered to them. Participants were asked to either complete the questionnaires at the study sites or at home (whichever was deemed convenient and acceptable by the individual).

Participants eligible were adults (≥18 years) with type 2 diabetes, who identified their ethnicity as European (white-British) or African-origin (black-Caribbean or black-African) and also understood and spoke English. Patients who could not speak English, showed evidence of psychotic disorder, had dementia or were suffering from severe pain during their visit to the clinic were excluded and were not approached to take part. Seven patients were excluded at various stages of the interview, due to non-eligibility. Specific reasons for their exclusion were inability to speak English (two patients); uncertainty about type of diabetes (one patient); uncertainty about ethnic classification (three patients); and suspected psychotic disorder (one patient). Where there were doubts about eligibility, the researcher consulted with screeners and clinic staff on the appropriateness of including a patient. Figure 2 shows a diagrammatic representation of the numbers of participants recruited from the two boroughs by ethnicity. Data collection for the study was conducted between February–June 2008.

Figure 2.

 Flow chart showing numbers of patients recruited from Brent and Hackney. BA, black-Africans; BC, black-Caribbean; WB, white-British.


Demographic/clinical characteristics

A demographic questionnaire was used to collect data on patients’ age, gender, country of origin, educational and employment status, presence of other long-term conditions and type of treatment for their diabetes. Data on glycaemic control levels (HbA1c) and history of microvascular complication was obtained from hospital and General Practitioner electronic databases. The most recent HbA1c data (within the last six months) were used. HbA1c is a measure of average blood glucose level for the past two to three months. Its values are expressed in percentages. Generally, patients with HbA1c values of ≤7·0% are said to have good glycaemic control, and those with values >7·0% have poor glycaemic control (NICE 2008). Microvascular complication status was defined as the presence or absence of retinopathy, neuropathy or kidney disease (nephropathy).

Diabetes-specific knowledge

Patients’ knowledge about basic principles of diabetes and its management was assessed using the brief Diabetes Knowledge Test (DKT) (Fitzgerald et al. 1998). The brief DKT is a 23-item tool with two subscales. The general knowledge subscale is made up of the first 14 questions and is suitable for use with either type 1 or type 2 diabetes patients. The second component, the insulin use subscale, is made up of the last nine items and is only appropriate for use among patients treated with insulin. We therefore used only the general knowledge subscale as this study involved all patients with type 2 diabetes regardless of their treatment regime. The brief DKT has good reliability coefficient (α = 0·70) and was developed in a heterogeneous ethnic sample of patients including people of African-origin (Fitzgerald et al. 1998).

Recommended self-management activities

Performance of the recommended self-management activities by individual patients was measured using the revised Summary of Diabetes Self-Care Activities scale (SDSCA). The SDSCA is a 25 item self-report instrument. Patients are required to indicate how many of the past seven days they have performed various diabetes self-management activities. The self-management activities covered by the SDSCA include diet, exercise, SMBG, foot care and smoking. However, we report only diet, exercise, SMBG and foot care in this study. Levels of overall self-management performance by individual patients in this study were derived by adding their scores for diet, exercise, SMBG and foot care. Psychometric evaluation conducted on the revised SDSCA reported acceptable mean inter-item correlation coefficient (r = 0·47) (Toobert et al. 2000). The revised SDSCA has 11 core items carefully selected from the previous versions and additional 14 newly expanded items. As reliability and validity are not known for the recently added items, only the 11 core items were used in this study.

Illness perceptions about diabetes

The Revised Illness Perception Questionnaire (IPQ-R) was used to collect data on participants’ own representations (or perceptions) about their diabetes (Moss-Morris et al. 2002). The IPQ-R updated and improved the psychometric properties of the original IPQ, which was developed to measure the theoretical constructs of Leventhal’s CS-SRM (Weinman et al. 1996). One major step in the revision of the IPQ was the addition of the emotional representation subscale, enabling the simultaneous capture of the cognitive and emotional representations of the individual, a key feature of the CS-SRM.

In line with the general dimensions of illness perceptions, the IPQ-R consists of the Identity, Timeline (Timeline-acute/chronic and Timeline-cyclical), Consequences, Personal control, Treatment control and the Causal representation subscales. Additionally, the IPQ-R also has Illness coherence and Emotional representation subscales. Identity refers to the label an individual associates with the illness condition. It includes the symptom(s) the individual recognises as part of the illness. In the IPQ-R for diabetes, 14 symptoms are listed. Patients are requested to indicate which symptoms they have experienced and those that they think are experienced as a result of their diabetes. The Timeline-acute/chronic subscale asks whether patients perceive their illness to be long-term (or permanent) or short-term (or temporary), whilst Timeline-cyclical describes individuals’ perceptions regarding the stability or changeability of their illness. Individuals’ perceptions about the ‘seriousness’ of their illness, and the impact it has on their lives is determined by the Consequences subscale. In the IPQ-R, the impact of the illness on individuals’ own life, their close associates and financial consequences are explored. Whilst Personal control examines individuals’ perceptions about their own personal effectiveness or their perceived control over the condition, Treatment control explores their perceptions regarding the effectiveness of their treatment in controlling and/or curing their illness. The Illness coherence subscale is unique to the IPQ-R and measures the extent to which individuals’ representations about the illness reflects their ‘coherence’ and understanding about the condition. Thus, it evaluates how the entire illness ‘makes sense’ to the individual. The Emotional representation subscale, investigates the emotional arm of illness representations. It assesses individuals’ emotional feelings about their illness condition. For instance, to what extent are individuals emotionally hurt by the condition? The cause component of illness representations seeks to explore patients’ own perceptions about the possible causes of their illness. The developers of the IPQ-R generated four preliminary causal components (psychological, risk factor, immunity and chance/accident) for this scale. However, they recommended that, if appropriate, individual researchers should conduct factor analysis on the causal items to identify groups of causal beliefs.

The IPQ-R has good psychometric properties. For instance, internal consistency of 0·77 and above was reported for most subscales and three-week test–retest reliability ranged between 0·46–0·88 for some patient groups. The presence of logical relationships, indicating discriminant validity, was also found between subscales of the IPQ-R and those of the Positive and Negative Affect Schedule (Moss-Morris et al. 2002).

Statistical procedures and analyses

Data from questionnaires were entered into the Statistical Package for Social Sciences (spss version 15; SPSS Inc, Chicago, USA), for analyses. Preliminary statistical investigations were performed to check whether the data met assumptions for parametric statistical tests. Where variables did not meet the required assumptions for parametric tests, the appropriate non-parametric equivalent was used. Univariate relationships between variables were explored using bivariate correlations. The Pearson’s correlation coefficient statistic (r) was used for continuous variables which were normally distributed, whilst the Spearman’s rho statistic (ρ) was used for ordinal or non-normally distributed variables. Point-biserial correlation coefficients were used for dichotomised variables (microvascular complication status). As percentage HbA1c was largely skewed, it was converted into an ordinal variable (quintiles). Age and duration of diabetes were significantly related to most of the dependent and independent variables in our preliminary analyses. It was therefore deemed appropriate to control for their effect in the multivariate analyses. This was achieved using the hierarchical multiple regression technique. Hierarchical binary logistic regression was used for predicting diabetes complication status. Sample size determination prior to data collection was performed using the G-Power statistical software (Faul et al. 2007). Based on effect size of 0·3 for correlations between illness perceptions and self-management (Griva et al. 2000), 82 participants were required for each ethnic group to yield 80% power at a significance level of 0·05.

Ethical considerations

All study procedures were approved by the Brent and the East London Ethics Committees.


Demographic and clinical characteristics

Four hundred and twenty patients who attended clinics in the study sites were eligible and were asked to take part in the study. Three hundred and fifty-nine (86%) patients completed the questionnaires. Twenty-nine per cent (N = 104) of the participants were recruited from facilities in Brent (Central Middlesex hospital, Wembley and Willesden health centres) and the rest were recruited from Hackney (Homerton hospital). Ninety-eight per cent of all the participants were patients who attended the clinics for retinal screening, the rest were to see either the diabetes nurse or the diabetes consultant. The proportion of people who agreed to participate in the study (response rate) was higher among white-British (91%) compared to black-Caribbean (82%) and black-Africans (83%). Table 1 shows the demographic and clinical characteristics of participants according to self-reported ethnicity.

Table 1.   Demographic and clinical characteristics of participants by ethnicity and for the total sample
DescriptiveWhite-British (n = 137)Black-Caribbean (n = 123)Black-African (n = 99)Total sample (N = 359)
  1. SD, standard deviation; SR, self-reported; note percentages are within ethnic groups.

  2. *p ≤ 0·05; **p ≤ 0·01; ***p ≤ 0·001.

Mean age (SD)***61·54 (1·5)59·07 (11·9)52·16 (10·4)57·98 (12·4)
Percentage female44·5361·7944·4450·42
Percentage married/cohabiting41·0529·5158·7641·93
Percentage with university qualification***18·8012·5034·0021·10
Mean known duration of diabetes (SD)**8·36 (7·7)9·71 (8·0)6·10 (5·3)8·13 (7·3)
Mean HbA1c (SD) (%)*7·63 (12·6)8·15 (2·1)8·34 (2·3)8·01 (2·0)
Mean BMI (SD)33·44 (6·5)31·43 (7·1)31·35 (5·3)32·20 (6·4)
Percentage with any microvascular complications54·4447·5448·1550·43
Treatment type (SR)
 Percentage on diet & exercise**16·182·456·108·64
 Percentage on hypoglycaemic agents57·6666·1266·0063·00
 Percentage on insulin25·7431·4127·8428·33

Relationships between knowledge and illness perceptions/self-management

Table 2 shows the correlation coefficients between: (a) knowledge and self-management (Fig. 1, pathway A) and (b) knowledge and illness perceptions (Fig. 1, pathway B). Knowledge about diabetes did not correlate significantly with any aspect of self-management among white-British patients. However, high knowledge about diabetes was associated with poor dietary self-management among black-African and black-Caribbean patients. As expected, knowledge correlated significantly with various dimensions of patients’ perceptions about their diabetes. Among white-British patients, knowing more about diabetes was associated with experiencing more diabetes related symptoms and perceiving diabetes as a long-term condition, with serious consequences. Similarly, high knowledge scores correlated with long-term perceptions about diabetes in black-Caribbean and black-African patients. In addition, high knowledge scores were associated with unstable perceptions (timeline-cyclical) about diabetes in the black-Caribbean patients.

Table 2.   Correlation coefficients between knowledge and self-management/illness perceptions
  1. SMBG, self-monitoring of blood glucose.

  2. *p ≤ 0·05; **p ≤ 0·01; ***p ≤ 0·001.

(a) Self-management
 Foot care−0·130·00−0·02
(b) Illness perceptions
 Personal control0·070·150·22
 Treatment control0·150·03−0·10
 Illness coherence0·25*0·150·27*
 Emotional representation0·090·070·19
Causal perceptions
 Risk factor0·110·20−0·03

However, after controlling for age and duration of diabetes in the hierarchical multiple regressions knowledge of diabetes did not predict any aspect of self-management in any ethnic group. Knowledge only contributed 0–1% among the black-Caribbean and 0–4% among the black-African participants, respectively. In all cases, the contribution of knowledge in explaining self-management was not statistically significant (p > 0·05). For reasons of brevity the regression tables are not reported in this manuscript.

Relationships between illness perceptions and self-management

Correlation coefficients between illness perceptions and self-management (Fig. 1, pathway C) are shown in Table 3. In white-British patients perceiving diabetes with severe consequences was associated with less exercise self-management, but high frequency of SMBG. Also, high emotional worries about diabetes were associated with less exercise self-management. In black-Caribbean patients, perceiving diabetes with severe consequences was associated with less frequent feet management. However, high perceptions of personal control over diabetes were associated with frequent feet management. Similar to the other two groups, perceiving diabetes with severe consequences also correlated with less overall and dietary self-management among black-African patients. However, perceived personal confidence in controlling diabetes was associated with frequent overall self-management.

Table 3.   Correlation coefficients between illness perceptions and specific self-management regimens
VariableOverall self-managementDietExerciseSMBGFoot
  1. SMBG, self-monitoring of blood glucose.

  2. *p ≤ 0·05; **p ≤ 0·01.

(a) White-British
 Personal control−0·09−0·060·10−0·11−0·02
 Treatment control−0·11−0·05−0·040·14−0·02
 Illness coherence−0·040·090·030·080·12
 Emotional representation−0·04−0·18−0·21*0·12−0·17
Causal perceptions
 Risk factors−0·27**−0·25**−0·030·02−0·23*
(b) Black-Caribbean
 Personal control0·080·100·08−0·140·24*
 Treatment control0·04−0·01−0·04−0·050·13
 Illness coherence0·10−0·04−0·050·010·07
 Emotional representation−0·05−0·07−0·110·10−0·11
Causal perceptions
 Risk factors−0·20−0·04−0·10−0·11−0·15
(c) Black-African
 Personal control0·37**0·150·36**0·160·09
 Treatment control0·200·080·130·100·03
 Illness coherence0·150·040·170·03−0·12
 Emotional representation−0·04−0·20−0·080·040·08
Causal perceptions
 Risk factors−0·20−0·09−0·15−0·24*−0·04

As seen in Table 3, patients’ causal attributions also correlated significantly with different aspects of self-management for all ethnic groups. For example, risk factor causal beliefs was associated with less overall, dietary and foot management in white-British patients, and to infrequent SMBG in black-Africans. However, attributions of diabetes to psychological factors (such as stress, family problems and emotional state) or to external factors (such as chance/bad luck) were associated with high frequency of overall, SMBG or foot management in the black-Caribbean patients.

Hierarchical regressions controlling for age and duration of diabetes found that illness perceptions significantly explained 19% of the variation in SMBG among white patients. Only personal control perceptions contributed significantly to the model (β = −0·44, t = −3·10, p < 0·01). Due to inadequate cases-to-independent variables ratios for individual African-origin groups (black-African and black-Caribbean), hierarchical regressions for predicting self-management from illness perceptions were conducted for the two groups combined. The decision to pull the two groups together was also based on the similarities in their ethnic origin and in illness perceptions. For example, compared to the white-British participants, both black-Caribbean and black-Africans perceived diabetes as a curable and short-term condition with lesser consequences (data not shown). In the combined African-origin group, illness perceptions significantly predicted all areas of self-management, except SMBG. Illness perceptions explained about 13% of the variations in overall, 10% of dietary, 10% of exercise and 12% of foot self-management. Patients’ perceptions of the consequences (β = −0·28, t = −2·43, p < 0·01) and emotional representations (β = 0·29, t = 2·01, p < 0·05) contributed to the variation in overall self-management. Timeline-acute/chronic (β = −0·25, t = −2·51, p < 0·01) and personal control (β = 0·19, t = 2·00, p < 0·05) accounted for the variations in diet self-management. Personal control (β = 0·29, t = 2·93, p < 0·01) and consequences (β = −0·22, t = −1·94, p < 0·05) contributed to the variations in exercise, whilst consequences (β = −0·34, t = −3·17, p < 0·01) accounted for the variations in feet management.

Self-management and metabolic-control outcomes

Correlation coefficients between self-management and metabolic-control outcomes for diabetes (Fig. 1, pathway D) are shown in Table 4. No aspect of self-management was associated with glycaemic control in the white-British or black-African populations. In the black-Caribbean participants, high adherence to dietary self-management recommendations was associated with better glycaemic control. In addition, no significant correlations between self-management and microvascular complication status were found among the white-British or black-Caribbean participants. However, high frequency of overall self-management and SMBG were associated with microvascular complications status among black-Africans.

Table 4.   Correlation coefficients between specific self-management and HbA1c/microvascular complication status by ethnicity
 Overall self-managementDietExerciseSMBGFoot
  1. SMBG, self-monitoring of blood glucose.

  2. *p ≤ 0·05; **p ≤ 0·01.

(a) White-British
 HbA1c (%)−0·070·10−0·050·02−0·13
 Microvascular complications status0·05−0·07−0·070·100·07
(b) Black-Caribbean
 HbA1c (%)−0·13−0·23*−0·14−0·01−0·02
 Microvascular complications status−0·08−0·200·010·130·07
(c) Black-African
 HbA1c (%)0·130·080·080·24−0·12
 Microvascular complications status0·32*0·160·180·42**0·12

In multiple regressions controlling for age and duration of diabetes, no relationship was found between self-management and glycaemic control in any of the three groups. Self-management, however, explained significant proportions of the variations in microvascular complications (Wald χ2 = 18·01, p < 0·01) in black-African patients. SMBG (exp (B) = 1·35, 95% CI; 1·07–1·70, p < 0·01) contributed to the variability in microvascular complications in the black-African patients. As patients treated with insulin may be required to perform some self-management activities (such as SMBG) more frequently than those on other treatment options, we investigated whether this may have any effect by adjusting for insulin use. This adjustment did not alter the relationship between self-management and glycaemic control in any of the ethnic groups. However, addition of insulin use into the model attenuated the relationship between SMBG and diabetes complications status among black-African patients, resulting in non-significant relationship (exp (B) = 1·25, 95% CI; 0·96–1·63, p = 0·11).


This study was designed to investigate the role of disease-specific perceptions or knowledge in determining self-management behaviour among African and European-origin patients. The study also examined whether self-management was associated with metabolic-control outcomes including glycaemic control and microvascular complication status.

We found no associations between knowledge and most areas of self-management (Fig. 1, relationship A). However, knowledge about diabetes was associated with perceiving it as a long-term condition in all ethnic groups. Knowledge was also associated with perceiving diabetes as a serious condition among white-British, and as an unstable condition among black-Caribbean patients.

Dimensions of illness perceptions correlated significantly with various aspects of self-management in both European-origin (white-British) and African-origin (black-African and black-Caribbean) groups (Fig. 1, relationship C). In multivariate analysis, patients’ illness perceptions about their diabetes explained one-fifth of the variations in SMBG among white-British and between one-tenth and one-eighth of various aspects of self-management among the African-origin patients. In the white-British patients, personal control perceptions were the main determinant of the variations in self-management. However, several dimensions of illness representations, including personal control, timeline-acute/chronic, consequences and emotional representations contributed significantly to the variations in self-management in the African-origin group.

Adherence to self-management recommendations, particularly SMBG, significantly explained the variations in microvascular complications among black-Africans (Fig. 1, relationship D). However, the statistical significance disappeared when the analysis was adjusted for insulin use.

It is expected that enhancing patient knowledge about diabetes through education would enable them appreciate the need to engage in health-related behaviours. We therefore envisaged that people who know more about diabetes will report higher adherence to self-management recommendations, than those who know less. This was, however, not the case for our sampled population, as no significant relationships could be detected between the two measures. The lack of relationship between knowledge and self-management is not surprising, as similar findings have been reported elsewhere (Chan & Molassiotis 1999, West & Goldberg 2002). Understandably, knowledge about diabetes does not necessarily mean that patients have the requisite skills, the motivation or the self-efficacy to perform the recommended self-management activities (Nuovo 2007). Indeed, less than one-third of the participants in this study scored correctly on key practice-based questions, such as the food items not suitable for treating hypoglycaemia (29%); time period HbA1c measurements reflect (19%); and food items with less impact (free foods) on their blood glucose levels (24%). The ADA’s recent model for diabetes self-management education emphasises the need to integrate patients’ education with self-management support skills (ADA 2008). Self-management support refers to measures that are aimed at empowering patients to self-manage. Such measures include building patients’ self-confidence and equipping them with decision-making and problem-solving skills. The ADA’s approach may therefore represent an effective way to bridge the knowledge-practice gap. The inverse association between knowledge and adherence to dietary recommendations among the black-African and black-Caribbean patients was a rare finding, which may have arisen by chance. It is also possible that the knowledgeable people in these two populations of African-origin reported accurate and realistic estimates, whilst the less knowledgeable ones overestimated their dietary adherence. Given the complexness of diabetes self-management, minority patient groups are more likely to give inaccurate estimates due to poor knowledge, miscommunication or non-intentional adherence (Rothschild 1998, Lanting et al. 2008).

Although modest, the finding that illness perceptions in the African-origin populations explained significant proportions of the variation in most aspects of their self-management indicates the relevance of the CS-SRM for this population. This suggests that CS-SRM framework can be used as a basis to formulate interventions for promoting self-management in the African-origin population (Keogh et al. 2007). For instance, the finding that personal control perceptions contributed substantially to predicting most aspects of self-management suggests that interventions that build patients confidence (self-efficacy) may be used to promote self-management in patients of African ancestry. It was surprising, however, that treatment control beliefs were not significantly associated with any aspect of self-management in any ethnic group. The lack of a relationship may be problem associated with the questionnaire rather than the concept. Measurement of treatment control may be more sensitive if it is captured using behaviour-specific items such as “my daily exercise will be effective in controlling my diabetes”. Indeed, some studies have investigated similar concepts using items from the personal models of diabetes interviews. In such cases treatment control beliefs have significantly contributed to variations in self-management (Glasgow et al. 1997, Searle et al. 2007).

In univariate analysis, high adherence to dietary self-management recommendations was associated with better glycaemic control among black-Caribbean whilst overall self-management and SMBG were associated with microvascular complication status among black-Africans. After adjusting for age and duration of diabetes, however, self-management significantly predicted microvascular complications only among black-African patients. The lack of relationship between self-management and metabolic-control outcomes has been reported in some other studies too. For example, investigators in New Zealand found no associations (in their multivariate analyses) between self-management and metabolic outcomes among Asian, European and Pacific Island patients (Bean et al. 2007). Similar findings have also been reported among Chinese, Dutch and North American patients (Chan & Molassiotis 1999, Lanting et al. 2008). All these studies, including ours, are however, cross-sectional and used self-report measures for determining adherence to self-management recommendations. It is important to recognise that HbA1c measurements reflect diabetes control for the past two to three month period, whilst microvascular complications may take even longer to develop and manifest. Yet, this study only measured self-management for the past seven days. It is therefore likely that the methods are not precise enough to detect any such temporal relationships. Also, apart from adherence to self-management recommendations, several other important factors, including demographic and psychological factors, could potentially have explained the variations in metabolic control in our study population (Tsenkova et al. 2007).


The data for this study were collected concurrently at one time-point. It is therefore not feasible to determine the direction of the relationships between determinants and outcomes such as illness perceptions and self-management. For example, it is not possible to determine in this study, whether patients’ perceptions about diabetes preceded their self-management behaviours or vice versa. Similarly, the cross-sectional nature of the study makes it impossible to investigate any causal relationships between any two variables.

Data for self-management were based on patients’ own reports and may have been influenced by social desirability bias as well as memory and recall bias. In line with their preference, study questionnaires were administered (face-to-face) to majority (about 95%) of the black-African and black-Caribbean participants. There was therefore the potential for information (interviewer) bias. However, to minimise this bias the researcher, who administered the questionnaires, strictly read questions to patients without any influence. Indeed, no further explanation was given by the researcher, unless it was absolutely necessary. Although patients were not randomly sampled for participation, the likelihood of any selection bias due to the convenience sampling may have been minimised by the high participation rate achieved. However, other forms of selection biases cannot be ruled out. For example, apart from recruiting patients from only two boroughs in London, only patients referred to the study sites could be reached and contacted for participation. It is possible that recruiting from primary care, where majority of the UK’s diabetes cases are managed, would have resulted in a more representative sample than what is achieved in this study. Other forms of selection bias, including non-consent and bias due to the exclusion criteria could limit the generalisability of the findings in this study. In addition, our statistical analysis did not adjust for multiple testing hence some of the findings could have occurred by chance. Whilst we recognise this non-adjustment as a limitation, it is important to note that our study just met the minimum required sample size for the analysis (albeit not for all regressions) hence procedures such as Bonferroni corrections could have reduced the power to detect minimal effects and led to type II error (Voss & George 1995, Perneger 1998).

Relevance to clinical practice

In spite of its shortcomings, this study provides valuable information for practice. For example, the finding that illness perceptions predict patients’ self-management behaviours suggest the need for healthcare professionals (HCPs) to take cognizance of their patients’ own beliefs as an opportunity to deliver appropriate services. Also, routinely screening and identifying patients’ perceptions about illnesses can help HCPs work with individual patients to develop personal action plans aimed at promoting adherence to self-management recommendations. For patients whose perceptions may be inconsistent with existing medical facts, and are likely to impede self-management, HCPs can work with them to synchronise their perceptions, or correct any maladaptive beliefs.


This study demonstrates that illness perceptions are important determinants of the way African-origin patients adhere to diabetes self-management recommendations. Thus, interventions guided by illness perceptions, especially those informed by the Leventhal’s CS-SRM can be used to promote adherence to self-management recommendations among African-origin patients.

The findings in this study also indicate that patients of different ethnic and cultural-origin may differ in their perceptions about a common long-term condition. Consequently, the influence of such perceptions may impact differently on their health-seeking behaviours (in this case self-management).

This study could not find significant relationship between self-management behaviour and metabolic-control outcomes (glycaemic control or microvascular complication status) in either European-origin or African-origin patients. However, given the limitations of our study, we recommend the need for longitudinal studies using real-time data collection procedures to confirm or annul the findings.


We are grateful to Professor Stewart Mercer and Dr Josie Evans for their helpful comments on aspects of this work. We also thank Nancy Hallet and Professor Jane Anderson of Homerton University Hospital, Rickie Banarsee, Don McLeod, Sylvia Sadeghian and Dr Gillian Vafidis at Brent Primary Care Trust, as well as Drs Mary Wells and Thilo Kroll of the University of Dundee, for the significant roles they played in making this study successful.


Data collection for this study was funded by a small grant from the Alliance for Self-Care Research Consortium.


Study design: ARA, MJ, WL, AK; data collection and analysis: ARA, MJ, WL, AK, DD, JA and manuscript preparation: ARA, MJ, WL, AK.

Conflict of interest

None declared.