Association of cardiac and non-cardiac chronic disease comorbidity on glycaemic control in a multi-ethnic population with type 1 and type 2 diabetes
- R L Mehta1,
- M J Davies2,
- S Ali3,
- N A Taub1,
- M A Stone1,
- R Baker1,
- P G McNally4,
- I G Lawrence4,
- K Khunti1
- 1Department of Health Sciences, University of Leicester, Leicester, UK
- 2Department of Cardiovascular Sciences, Leicester Royal Infirmary, Infirmary Close, University of Leicester, Leicester, UK
- 3Primary Care and Population Sciences, University of Southampton, Southampton, UK
- 4Department of Diabetes and Endocrinology, University Hospitals Leicester, Leicester Royal Infirmary, Infirmary Close, Leicester, UK
- Correspondence to R L Mehta, Department of Health Sciences, University of Leicester, 22–28 Princess Road West, Leicester LE1 6TP, UK;
- Received 1 July 2011
- Accepted 13 July 2011
- Published Online First 26 August 2011
Aims To determine the prevalence of chronic disease comorbidity in south Asians (SAs) and white Europeans (WEs) with diabetes and to quantify the relationship of cardiac disease comorbidity (CDCM) and non-cardiac disease comorbidity (NCCM) to glycaemic control in SAs and WEs with type 1 and type 2 diabetes mellitus.
Methods A cross-sectional study using a database of patients of SA (25.5%) and WE (74.5%) origin attending a specialist diabetes clinic in the UK between 2003 and 2005 (n=5664).
Results The prevalence of SAs and WEs with type 1 diabetes was 12% and 88%, respectively; for those with type 2 diabetes the prevalence was 30% and 70%, respectively. Overall, the prevalence of comorbidity in people with type 1 diabetes was 25.5% and with type 2 diabetes was 47%. NCCM was more prevalent in WEs than SAs (17.6% vs 12.8%, p<0.001). In type 2 diabetes, the prevalence of suboptimal glycaemic control was significantly greater in SAs compared to WEs with NCCM and CDCM (79% vs 62%, p<0.001; 78% vs 65%, p<0.001, respectively). SAs with type 2 diabetes and comorbidity had excess odds of suboptimal glycaemic control compared to WEs: OR 2.27 (95% CI 1.50 to 3.43) for those with NCCM and OR 1.91 (95% CI 1.49 to 2.44) for those with CDCM.
Conclusions The prevalence of CDCM is higher in SAs compared to WEs with type 2 diabetes, whereas the prevalence of NCCM is higher in WEs compared to SAs. Taking into account comorbidities, SAs (compared to WEs) with type 2 diabetes had an excess risk of having HbA1c ≥7% ranging from 1.86- to 2.27-fold. Further research is needed to identify the reasons for unfavourable metabolic conditions in SAs and also develop and evaluate interventions.
- glycaemic control
- diabetes mellitus
- chronic disease
- cardiac epidemiology
- coronary heart disease
- diabetes & endocrinology
- internal medicine
Diabetes mellitus affects at least 180 million people worldwide and it has been predicted that the prevalence will more than double by 2030.1 This rising prevalence has been associated with reduced life expectancy and linked to higher rates of obesity and a decrease in healthy lifestyle habits. Cardiovascular disease (CVD) death rates are either high or appear to be increasing in countries where diabetes rates are high.2–4 CVD is the major complication of type 2 diabetes mellitus (T2DM) and is responsible for between 50–80% of deaths as well as for substantial morbidity and loss of quality of life.5 6 The Diabetes Control and Complications Trial (DCCT)7 and the UK Prospective Diabetes Study (UKPDS)8 9 follow-up cohort studies have suggested that the control of glycosylated haemoglobin (HbA1c), to achieve targets below or near 7% after diagnosis of diabetes mellitus, is associated with a term reduction in the risk of macrovascular complications. Intensive glycaemic control in people with type 1 diabetes mellitus (T1DM) has been shown to exert a beneficial effect of 42% in relation to cardiovascular events and a 57% risk reduction in myocardial infarction, stroke or death from CVD.10 The UK Prospective Diabetes Study demonstrated that intensive treatment of diabetes can reduce morbidity and mortality, and lowering HbA1c reduces the incidence of microvascular complications in T2DM as it does in T1DM.9 11 However, advances in glycaemic management including the provision of insulin therapy have improved the life expectancy of people with both T1DM and T2DM, and each year of prolonged life increases the likelihood of eventual term cardiovascular and cardiovascular complications.12
Comorbidity is defined as the occurrence of one or more chronic conditions in the same individual and is common in people with diabetes mellitus.13 The range of comorbidities affecting patients with diabetes is not restricted to those commonly associated with diabetes—for example, additional comorbidities include periodontal disease and depression.14 15 Alongside the impact of population ageing on the number of people with diabetes mellitus, the prevalence of associated chronic disease comorbidity is also expected to increase. Previous studies involving subjects of white European (WE) origin have shown that chronic comorbidity is associated with poorer outcomes for those individuals with diabetes mellitus.16–18 The management of several conditions in the same individual can be difficult and requires a multifactorial, multidisciplinary approach; consequently, focusing on single disease management may not be appropriate in subjects with diabetes mellitus.19 The duration of diabetes mellitus, suboptimal glycaemic control (HbA1c ≥7%), hypertension, smoking status, albuminuria, and lipids abnormalities are associated risk factors for microvascular and macrovascular disease.20–22 It is not well established whether suboptimal glycaemic control (HbA1c ≥7%) is associated with non-cardiac disease comorbidity (NCCM), cardiac disease comorbidity (CDCM), ethnicity or type of diabetes mellitus.
Individuals of south Asian (SA) origin (predominantly originating from India, Pakistan, Bangladesh, and Sri Lanka) comprise a large minority group worldwide. People from some ethnic groups, including SAs and those from low socioeconomic backgrounds, are at a higher risk of developing T2DM.23 Furthermore, SAs develop T2DM at an earlier age compared to WEs.24 25
The extent of comorbidity may vary between different population groups, including people from different ethnic and socioeconomic groups. Examining these differences may assist with identifying cohorts at high risk, in order to target patterns of service delivery and health policies aimed at reducing the disease burden. The aim of this study was to determine the prevalence of chronic disease comorbidity in an ethnically diverse population of people with diabetes of SA and WE origin. A further aim was to assess the relationship between glycaemic control, chronic disease comorbidity and ethnicity in people with T1DM and T2DM.
Study design and methods
A cross-sectional study was conducted using the clinical database of patients attending a specialist outpatient diabetes clinic in Leicestershire, UK, between 2003 and 2005. Leicestershire has a population of approximately 1.04 million of whom 47 500 (4.6%) are registered of having diabetes. It has been estimated that approximately 37% of patients with diabetes are managed either exclusively or partly in secondary care,26 but there is a lack of data regarding possible differences between SAs and WEs in relation to primary and secondary care management. Referral from primary care to the specialist hospital clinic from which our data are derived is based on a combination of local guidelines and clinical judgement; however, we are unable to comment on whether there are variations in referral patterns according to patients' ethnic origin. Among all minority ethnic groups in the UK as a whole, those of SA origin constitute the largest minority group (50%) and people from this background are strongly represented in Leicestershire, particularly in the city of Leicester (25.7% of the city population).27 Data were extracted for all patients attending the clinic for which the principal diagnosis was recorded as T1DM or T2DM.
Patient characteristics and other data were extracted from the clinical workstation (CWS). The CWS is an electronic patient record system used for recording clinical data and routine correspondence, as well as for audit and research purposes. The database includes all routinely collected demographic and clinical data including diagnosis and treatment for all patients attending diabetes clinics either through referral or for their annual diabetes review. Data from all consultations, including assessment of comorbidity, using clinical notes, blood tests, radiology reports, and general practitioner referral, are collected routinely using a standardised patient clinic sheet completed by the diabetes specialist at each patient encounter. Once completed, forms are passed to the diabetes information analyst for data entry into the CWS. Data extracted for this study, described in detail elsewhere,28 included type of diabetes, ethnicity, duration of diabetes, age, gender, smoking status, body mass index (BMI), microvascular and macrovascular complications, non-cardiac related complications, metabolic conditions, and the patient's residential postcode. The extraction of patient residential postcodes was in order to derive Indices of Multiple Deprivation (IMD) scores as a proxy for socioeconomic status, with greater scores indicating higher levels of deprivation.29 Ethnicity was categorised as SA or WE based on that given in the patient's record or by analysis of their name, making use of the SA name recognition software ‘Nam Pechan’30 supplemented by a visual inspection of surnames and forenames. This method of identifying SA ethnicity has been shown to have a high level of reliability in UK populations.30–32 People classified in other ethnic groups were not included in the final analyses of ethnic differences as these groups were too small for any meaningful interpretation.
Classification of chronic disease comorbidities
The CDCMs examined were any heart disease (International Classification of Disease (ICD10): I5–I9, I11, I13, I20–I27 and I30–I52), including myocardial infarction (ICD10: I21–I23), congestive heart failure (ICD10: I50), cerebrovascular disease (ICD10: I6), and hypertension (ICD10: I10–I15). The NCCMs examined were any chronic lower respiratory disease (ICD10: J40–J47), non-infective enteritis and colitis (K50–K52), other intestinal diseases (K55–K63), episodic and paroxysmal disorder (G40–G47), thyroid disorders (E00–E07), soft tissue disorders (M60–M79), and malignancies (C00–C97). The depression classification examined was mental disorders (F00–F99). We subsequently categorised comorbidity into three categories for analysis: any comorbidity, CDCM only, and NCCM only. A dichotomous variable of SA or WE was created for ethnic origin.
Ethics committee approval
This study was conducted with the approval of the Leicestershire, Northamptonshire and Rutland Research Ethics Committee, United Kingdom (reference 06/Q502/50).
To ascertain prevalence rates, the proportions of SAs and WEs with CDCM and NCCM within the sample population were determined. To compare prevalence rates in the two samples (SA and WE), the following statistical techniques were applied. Parametric tests for normally distributed variables were applied to assess differences between the characteristics of the two ethnic groups separately in those individuals with T1DM and T2DM. Assumptions of normal distribution were tested using the Shapiro and Wilk test,33 equality of variances was tested using Bartlett's test,34 and when these assumptions were violated, non-parametric techniques were applied. Differences in categorical variables were analysed using the χ2 test or Fisher's exact test as appropriate.
The relationship between glycaemic control (based on HbA1c), chronic disease comorbidity and ethnicity (SA vs WE) was investigated as follows: HbA1c values were categorised into two groups (<7%: optimal; ≥7%: suboptimal) based on the American Diabetes Association (ADA) guidelines.35 To investigate the effects of ethnicity and comorbidity on HbA1c, four different logistic regression models were built separately for patients with T1DM and T2DM, using explanatory variables as follows: (1) ethnicity; (2) ethnicity, either class of comorbidity and the corresponding interaction; (3) ethnicity, NCCM and their interaction; (4) ethnicity, CDCM and their interaction. To allow for different responses between ethnic groups, interactions between ethnicity and types of comorbidity were examined to see whether the relationship between comorbidities and glycaemic control varied between subjects of SA and WE origin. When a significant interaction was observed, implying that the association between suboptimal glycaemic control and comorbidity were not similar between ethnic groups, then logistic regression analyses were conducted separately for those with and without the comorbidity. Models were fitted with adjustment for potential confounders (age (years), gender, smoking status, BMI, duration of diabetes (years) and IMD) and also without adjustment, in order to test the robustness of the results. The significance of the variables in the model was assessed using the Wald χ2 test and determination of ORs with associated 95% CIs. Goodness of fit to the model was assessed using the Hosmer and Lemeshow test.36 Throughout the analysis, statistical significance was assessed at the 5% (two sided) level. All statistical analyses were conducted using SPSS16 (SPSS Inc).
Characteristics of the study sample
The overall study sample consisted of 5664 subjects: 1332 (23.5%) with T1DM and 4332 (76.5%) with T2DM. Furthermore, the prevalence of SAs and WEs with T1DM was 12% and 88%, respectively, and the prevalence of SAs and WEs with T2DM was 30% and 70%, respectively. Full details of the characteristics of the cohort according to type of diabetes and ethnicity are shown in table 1. Subjects of SA origin were significantly younger than those of WE origin, both for those with T1DM and those with T2DM (41.9 vs 45.3 years, p=0.023; 59.2 vs 66.2 years, p<0.001 respectively). The mean duration of diabetes was significantly longer in WEs with T1DM compared to SAs (20.1 vs 14.2 years, p<0.001) and significantly shorter in WEs with T2DM compared to SAs (9.1 vs 10.4 years, p<0.001). Comorbidities were present in 340 (25.5%) and 2036 (47.0%) of subjects with T1DM and T2DM, respectively. The prevalence of any comorbidity was not statistically different between SAs and WEs with T1DM (SA vs WE: 30.1% vs 24.9%; p=0.156) and T2DM (SA vs WE: 45.3% vs 47.7%; p=0.159). In those subjects with comorbidities, the prevalence of CDCM was not significantly different between SAs and WEs with T1DM and T2DM (15.3% vs 11.3%, p=0.133; 37.5% vs 37.3%, p=0.907 respectively). In people with T2DM, WEs had a significantly higher prevalence of NCCM (17.6% vs 12.8%, p<0.001) compared to SAs.
Comorbidities in SAs and WEs
Results comparing people of SA and WE origin in terms of diabetes related complications and glycaemic control are presented in table 2. In subjects with T2DM, the prevalence of having ≥2 comorbidities was significantly greater in WE individuals compared to SAs (16.1% vs 11.8%, p<0.001). In contrast, in subjects with T2DM, the prevalence of ischaemic heart disease was significantly greater in SA individuals compared to WEs (32.4% vs 29.1%, p=0.032). Conversely, the prevalence of both peripheral vascular disease and cerebrovascular disease were significantly greater in WEs compared to SAs (5.3% vs 1.8%, p<0.001; 7.2% vs 5.5%, p=0.037, respectively). For those individuals identified with microvascular complications and T1DM, the prevalence of both retinopathy and neuropathy was significantly greater in WEs compared to SAs (48.0% vs 38.7%, p=0.025; 27.8% vs 14.7%, p<0.001, respectively). In those subjects identified with T2DM and indication of microvascular complications, the prevalence of both retinopathy and nephropathy was significantly greater in SAs compared to WEs (34.7% vs 28.3%, p<0.001; 10.8% vs 7.9%, p=0.002, respectively). On the other hand, the prevalence of neuropathy was significantly greater in WEs compared to SAs (36.4% vs 28.9%, p<0.001); however, it should be noted that there is a subjective element to the assessment of neuropathy which may have influenced this finding and the comparison relating to neuropathy in T1DM. In addition, the proportion of SAs with suboptimal glycaemic control and T2DM was significantly greater than for WEs with T2DM (73% vs 65%, p<0.001). Furthermore, the prevalence of suboptimal glycaemic control in subjects with T2DM was significantly greater in SAs compared to WEs for those with NCCM (79% vs 62%, p<0.001) and also for those with CDCM (78% vs 65%, p<0.001).
Results from the logistic regression models are shown in table 3. In model 1, with respect to T1DM, the excess odds of having HbA1c ≥7% was 69% in SAs relative to WEs (OR 1.69, 95% CI 0.99 to 2.87; p=0.054); borderline significance, after adjustment for age, gender, BMI, smoking status, duration of diabetes, and IMD. In model 1, with respect to T2DM, the excess odds of having HbA1c ≥7% was 47% in SAs relative to WEs (OR 1.47, 95% CI 1.27 to 1.70; p<0.001). Adjustment for confounding factors slightly attenuated the effect (OR 1.32, 95% CI 1.11 to 1.56; p<0.001) in subjects with T2DM. In models 2, 3, and 4 with interaction terms, considering patients with T2DM, the excess odds of having HbA1c ≥7% ranged from 1.86–2.27 in SAs with comorbidity relative to WEs. In model 3 with adjustment for confounders, with respect to T2DM, the excess odds of having HbA1c ≥7% was 24% in SAs relative to WEs (OR 1.24, 95% CI 1.03 to 1.48; p<0.05). The Hosmer and Lemeshow goodness of fit test for models 2, 3, 4 and adjusted models 1, 2, 3, 4 for those identified with T1DM suggests no evidence of lack of fit (p>0.05). In the cohort identified with T2DM, the unadjusted models 2, 3, and 4 also suggest no evidence of lack of fit. In contrast, adjusted models 2, 3, and 4 depicted lack of fit (p<0.05), suggesting that the cause may be due to a non-linearity effect in a predictor variable. Furthermore, in each of the adjusted models, age and duration of diabetes (but not gender, BMI, smoking, and deprivation) emerged as significant confounders contributing to poor levels of glycaemic control.
Summary of key findings
In our sample of people with diabetes, prevalence rates for any comorbidity, and for CDCM, were not statistically significant when comparing people of SA and WE origin; this was true for both T1DM and T2DM. However, the prevalence of NCCM was greater among WEs compared to SAs with T2DM (17.6% vs 12.8%). Mean HbA1c values for SAs were greater than for WEs, and logistic regression modelling, including ethnicity, any comorbidity, NCCM, and CDCM with adjustment for potential confounders, indicated that, compared to WEs, SAs with type 2 diabetes had an excess risk ranging from 1.86- to 2.27-fold of having HbA1c≥7%.
Comparison with other studies
Our finding of a lack of increased prevalence of overall comorbidity and CDCM in people of SA origin differs from the observations reported in some previous studies. As early as 1985, Mather et al noted that SAs with diabetes have a greater risk of cardiovascular complications compared to WEs,37 and previous research has also shown that patients with T2DM have an increased risk of developing CVD.38 39 However, this may be partly explained by the fact that SAs in our sample were younger than WEs.
Strengths and limitations
To our knowledge, the present study offers the first insight into the prevalence of chronic disease comorbidity in a large multi-ethnic population with both T1DM and T2DM. Previous studies have assessed the association between CVD and HbA1c,40–42 but none have investigated this link specifically in SAs. The data collected were derived from a large specialist diabetes clinic and the cohort of SAs in this study was large, hence we were able to utilise and apply robust statistical techniques for our analyses.
However, it is acknowledged that the study also has a number of limitations. First, we used a cross-sectional design which can provide only a simplistic view of a complex disease process. We are not, therefore, able to provide data regarding the sequence of events, nor establish the direction of causation between glycaemic control and comorbidity. In addition, the data are from a single hospital and thus the results may not be generalisable to other healthcare settings or localities. Furthermore, our prevalence rates associated with CDCM and NCCM are dependent on the completeness of routinely collected information and may therefore be underestimated. In considering the association between glycaemic control and comorbidity, we were unable to take account of frequency of attendance including ‘did not attend’ rates, as this information was not available for inclusion in our dataset. Finally, our categorisation of ethnicity represents broad groups within which there will be heterogeneous subgroups.
This study identified some ethnic disparities in individuals with diabetes, in relation to the association between comorbidity and glycaemic control. Although our study was unable to determine the direction of causation for this association, our findings suggest that there may an enhanced need for strategies to improve glycaemic control in SA populations in order to help reduce these inequalities. Further research is needed in order to identify the underlying reasons for unfavourable metabolic conditions in SAs—for example, in terms of identifying cultural barriers to effective self management and the development and evaluation of interventions aimed at addressing these barriers. There is also a need for further research using a longitudinal approach in order to examine the long term effects of chronic disease comorbidity. These studies should, optimally, include consideration of the severity of these comorbidities.
In a sample of people with type 2 diabetes, there was a higher prevalence of non-cardiac disease comorbidities linked to white Europeans, but south Asians in the sample were significantly younger.
However, in type 2 diabetes, south Asians with comorbidities were more likely to have poor glycaemic control compared to white Europeans.
Current research questions
How can links between ethnicity, comorbidity and glycaemic control be explained?
How are these links affected by severity of comorbidities?
How can ethnic disparities such as those resulting from poor glycaemic control be minimised through appropriate patient management?
We would like to thank Ismail Gangat, Clinical Information Analyst, from the Department of Diabetes and Endocrinology for his assistance.
Competing interests None.
Ethics approval Leicestershire, Northamptonshire and Rutland Research Ethics Committee, UK.
Provenance and peer review Not commissioned; externally peer reviewed.