Title: Ethnicity specific association of BMI levels at diagnosis of type 2 diabetes with cardiovascular disease and all-cause mortality risk
Short title: Mortality risk in BMI categories modified by Ethnicity
Background: Studies evaluating the ethnicity-specific cardiovascular disease (CVD) and mortality risks at different adiposity levels are limited. The objective of this study was to evaluate the risk of CVD and all-cause mortality at different body mass index (BMI) levels among a multi-ethnic population of patients with type 2 diabetes mellitus (T2DM).
Methods: Longitudinal study of 53,559 patients from UK primary care who were diagnosed with T2DM (after 1999), aged 18 – 70 years, and without a history of CVD, renal diseases, and cancer. Risk of CVD and all-cause mortality at different BMI levels among three ethnic groups was estimated using treatment effects models, adjusting and balancing for confounders. CVD was defined as the first occurrence of heart failure, stroke or ischemic heart disease.
Results: Overall, the mean age at diagnosis was 51 years, 55% were male, and the median follow-up of 7 years was the same across the ethnic groups. White Europeans (n=56,443), African-Caribbeans (n=4,370), and South Asians (n=8,844) were 52 , 48, and 47 years old with a mean BMI of 33.0, 31.0, and 30.0 kg/m2 at diagnosis, respectively. Among White Europeans, normal weight patients developed CVD significantly earlier by 0.5 years (95% CI: 0.1, 0.9 years; p=0.006) compared to obese patients (mean time to CVD 4.6 years). Furthermore, those with normal body weight at diagnosis were significantly more likely to die earlier by 0.7 years (95% CI: 0.2, 1.3 years; p=0.006) in the White European group and by 2.5 years (95% CI: 0.5, 4.4 years; p=0.012) in the South Asian group compared to their respective obese patients.
Conclusion: This study suggests significantly different patterns of association of BMI with cardiovascular and mortality risks in different ethnic groups. Normal weight White Europeans and South Asians appear to have significantly higher mortality risk compared to those who were obese at the time of T2DM diagnosis.
Keywords: Body mass index; Type 2 Diabetes; Cardiovascular Disease; Obesity; Race and Ethnicity; Weight Change Pattern
Recent studies have reported an inverse association of body mass index (BMI) with mortality risk among adults with type 2 diabetes mellitus (T2DM), where patients who were normal weight [BMI 18.5 – 24.9 kg/m2] at diagnosis had significantly elevated mortality risk compared to their obese counterparts [BMI ≥ 30 kg/m2] 1-6. While the explanation for this phenomenon, referred to as the obesity paradox in T2DM remains unclear, weight loss before the diagnosis of T2DM as a result of underlying/undiagnosed medical condition was postulated as one of the possible reasons 7-9.
However, an analysis of body weight changes over 3 years before diagnosis in patients with T2DM under different BMI categories have shown otherwise 10. It is possible that ethnicity might play an essential role in understanding the underlying mechanism, as the distribution of adiposity levels in relation to cardiovascular disease (CVD) and mortality risk has been shown to be different for different ethnic groups 11-18.
Previous studies have evaluated the incidence of CVDs either in different ethnic groups 19-23, or in relation to BMI 24, 25. However, these studies did not evaluate the possible difference in the BMI related risk paradigm in different ethnic groups. Among Asians, a pooled analysis of 20 prospective cohort studies in Asia reported increased cardiovascular mortality risk at lower BMIs 26. However, this study was conducted in the general population, adjusting for diabetes status where appropriate. Wright and colleagues (2016) reported significantly lower mortality risk in South Asian and African-Caribbean individuals with T2DM compared to White Europeans 27. However, this UK primary care based study did not evaluate the interplay of BMI in this context.
Studies evaluating the ethnicity-specific long-term cardiovascular and mortality risks at different adiposity levels are scarce, and to the best of our knowledge, only one study has examined the modifiable effect of ethnicity on the observed phenomenon of the obesity paradox in T2DM. Kokkinos and colleagues 28 used data from two Veteran Affairs Medical Centres in the US to assess the association between BMI, fitness, and mortality in African-Americans and Caucasians. However, this study was based on only male patients, and the BMI measures were not evaluated at the time of diagnosis of diabetes.
A better understanding of the potential role of ethnicity in the obesity paradox in both male and female patients with T2DM may be important as this would enable clinicians to better manage diabetes amongst patients of different ethnicity and adiposity. Therefore, to address these knowledge gaps, the aim of this study was to use a cohort of incident T2DM patients (males & females) from United Kingdom primary care database, to evaluate for each ethnic group, (1) the CVD and mortality rate in each BMI category, by weight change pattern before diagnosis and (2) the association of BMI categories at diagnosis with CVD and mortality risk, controlling for weight change pattern before diagnosis and other risk factors.
The data for this study were obtained from The Health Improvement Network (THIN) database, which is a large, anonymised longitudinal dataset derived from a network of more than 600 primary care providers across the UK. With longitudinal data on approximately 11 million individuals registered with the primary care system, the THIN database has been extensively used for academic research 29. The accuracy and completeness of this database have been previously described 30, 31. Notably, the database has a similar distribution of major chronic diseases including diabetes, heart failure, and obesity when compared to UK national statistics30. Clinically diagnosed diseases are recorded using Read codes 32, and with each diagnosis, an event date is entered. The THIN database provides comprehensive patient-level longitudinal information on demographic, anthropometric, clinical and laboratory measures, clinical diagnosis of diseases and events, along with complete information on prescriptions for medications with dates and doses. Formal access to the database has been obtained from the Independent Scientific Review Committee for the THIN database (Protocol Number: 15THIN030).
Identification of T2DM cohort
Patients diagnosed with T2DM between January 1990 and September 2014 (n=406,098) were identified using a robust machine learning algorithm, which uses a combination of Read codes 32, anti-diabetes medications, and lifestyle modification interventions 33. Those included in this study satisfied the following criteria: (1) complete data on age (18 – 70 years), sex (male & female), BMI (≥ 15 kg/m2) and date of diagnosis of T2DM (from January 2000), (2) ethnicity identified as White European, African-Caribbean or South Asian, and (3) no history of CVD, renal diseases or cancer at diagnosis. South Asians were defined as patients with Indian, Pakistani, Sinhalese, and Bangladeshi origin while African-Caribbeans were defined as patients with Black-African and Caribbean origin. White Europeans were patients with self-reported ethnicity as White, European, Caucasian, and New Zealand European. Those with Read codes for type 1 diabetes mellitus (T1DM) or gestational diabetes, and those who received insulin as the first antidiabetic drug (ADD) were excluded. A final cohort of 53,559 patients with T2DM was used for this study.
Demographic and longitudinal measurements
Data on deprivation score (based on residential address) was extracted where available, and the smoking status for individuals were classified as current, ex, or never smokers. Longitudinal clinical and laboratory measurements including BMI, body weight, glycated haemoglobin (HbA1c), blood pressure and lipids were extracted for all patients. All available measures at or within three months before diagnosis of T2DM were considered as the baseline measures. If more than one measurement existed within this interval, the closest to the T2DM diagnosis date was taken. After that, longitudinal measures before and after the T2DM diagnosis were arranged in six-monthly windows. Only the latest measurement within each window was preserved. BMI categories for White Europeans and African-Caribbeans were defined as normal weight (18.5-24.9 kg/m2), overweight (25-29.9 kg/m2), and obese (≥ 30 kg/m2). For South Asians, BMI in the ranges 18.5-22.9, 23-27.4, ≥ 27.5 kg/m2 were used to define normal weight, overweight and obese patients respectively 34.
As weight loss before clinical diagnosis of T2DM is a common clinical manifestation, it was hypothesized that a weight loss of at least 2 kg before was clinically significant 35. Therefore, using 6 possible longitudinal body weight measures over 36 months before diagnosis, we classified patients who lost body weight (LBW) by at least 2 kg before diagnosis (if average of 5 prior measurements minus the body weight measure in the 6 months prior to diabetes diagnosis was ≥ 2 kg) and those who did not lose body weight (NWL) – i.e., they remained on the same level or increased body. Complete records on the prescriptions for different classes of ADDs, antihypertensive drugs, weight lowering drugs, anti-depressant drugs, and lipid-modifying drugs were extracted along with the dates of prescriptions
Mortality and comorbidity data
Records of CVDs, renal diseases (including chronic kidney disease (CKD)), and cancer with dates of diagnoses before and after T2DM diagnosis date were obtained. Information on deaths with dates and possible reasons were extracted. A composite variable for CVD (any CVD) was defined as the occurrence of angina, myocardial infarction, coronary artery disease (including bypass surgery and angioplasty), heart failure, or stroke. Patients with a recorded diagnosis of cancer, any CVD, retinopathy, neuropathy, or CKD before the T2DM diagnosis date were considered to have a relevant disease history. Time to a specific disease event and time to death were calculated as the time from T2DM diagnosis date to the first occurrence of the disease event and date of death respectively. Patients who were still alive at the end of the study data collection (September 2014) or dropped out were censored on the respective end date or drop out date.
The basic summary statistics were presented by number (percentage), mean (SD) or median (first quartile, third quartile), by ethnicity as appropriate. Among patients without disease history who were identified to have lost body weight (LBW) or not (NWL), age-weighted CVD and all-cause mortality rates (per 1000 person-years) were estimated by BMI categories for each ethnic group. To account for the inherent differences in risk factors between our defined BMI categories, treatments effects modeling approach was used to provide robust inferences. This modelling approach uses the potential outcomes or counterfactual framework to allow comparison of survival time for CVD and all-cause mortality for patients with different BMI categories, separately for each ethnic group. Briefly, given an observed outcome (Y0), for a patient with normal weight, the potential outcome or the counterfactual (Y1) for this same patient is the outcome if the patient had belonged to another BMI category and vice versa. Therefore, the average of the difference between the observed outcomes given a specific BMI category and the potential outcome is the average treatment effect [i.e., average treatment effect (ATE) = average (Y1-Y0)] 36-39. Since the outcome of interest is survival time, a survival model with propensity score-based inverse-probability weight estimator was used to estimate ATE for each BMI category. Variables that were conditioned on include sex, weight change pattern before diagnosis, age at diagnosis, smoking status, the incidence of cancer and renal diseases post-diagnosis, and receipt of lifestyle advice before and after diagnosis.
Basic demographic and clinical characteristics
In this study of 56,443 White Europeans, 4,370 African-Caribbean, and 8,844 South Asians adults with T2DM, the median follow-up time was 7 years for all three ethnic groups. The demographic and clinical profiles of these patients at diagnosis of T2DM in the three ethnic groups are presented in Table 1. South Asians had the clinical diagnosis of T2DM at a younger age (47 years) and at lower BMI (30.0 kg/m2) compared to White Europeans (age: 52 years, BMI: 33 kg/m2) and African-Caribbean (age: 48 years, BMI: 31.0 kg/m2). White Europeans had the highest proportion (58%) of ever-smokers (defined as current or ex-smokers) and a higher proportion of patients with systolic blood pressure above 140 mmHg (Table 1).
While the proportions of obese patients were 70%, 60%, and 68% in the White European, African-Carribean and South Asian patients respectively, African-Caribbeans had higher proportions of patients in the normal weight (10%) and overweight (30%) groups, as well as highest LDL-cholesterol levels (129 mg/dl) at diagnosis compared to White European and South Asian patients. Furthermore, White Europeans were more likely to receive lifestyle advice before (32%) and after (68%) diagnosis of T2DM compared to African-Caribbeans (25% and 59%) and South Asians (27% and 59%) respectively (Table 1).
The distribution of selected clinical characteristics among T2DM patients with no established disease history at diagnosis, separately for each ethnic group within the three defined BMI categories are presented in Table 2. African-Caribbeans had similar levels of ever-smokers across BMI categories, while South Asians who were normal weight at diagnosis had significantly higher proportion of ever-smokers (35%) compared to their counterparts who were overweight (26%) and obese (25%) at diagnosis. Furthermore, the proportion of ever-smokers among White Europeans who were normal weight at diagnosis (59%), was significantly higher compared to their White Europeans obese (57%) counterparts (Table 2).
Across the three ethnic groups, the proportion of patients with clinically diagnosed hypertension was smaller in normal weight patients compared to obese patients. African-Caribbeans who were normal weight at diagnosis underwent more lifestyle intervention than their overweight and obese colleagues, while South Asians with normal weight at diagnosis received similar levels of lifestyle intervention post diagnosis of diabetes compared to obese South Asians (Table 2). The use of statins was significantly higher in normal weight patients compared to obese patients across ethnic groups.
While similar proportion of patients lost at least 2 kg of body weight before diagnosis across the three ethnic groups, the proportion of normal weight patients who experienced this weight loss was almost double that of obese patients across BMI categories, irrespective of ethnicity.
Cardiovascular disease and mortality event rates
To avoid potential bias resulting from already existing severe disease that may independently induce weight loss in patients, cardiovascular and mortality risk assessments were carried out excluding patients with established diagnosis of cancer, any CVD, retinopathy, neuropathy, or chronic kidney disease (CKD) at diagnosis of T2DM. The age-weighted CVD and all-cause mortality rates per 1000 person-years (95 % CI), by BMI categories and weight change pattern prior to diagnosis in patients without disease history at diagnosis, separately for the three ethnic groups are presented in Figures 1 and 2. Among White Europeans, CVD event rates per 1000 person-years were significantly higher in normal weight patients (rate: 23.7; 95% CI: 21.3, 26.5) compared to obese patients (rate: 20.3; 95% CI: 19.6, 21.0), independent of weight change pattern before diagnosis. In no other ethnic group did CVD event rates vary across different BMI categories and weight change pattern before diagnosis. However, the CVD event rates in White Europeans with normal weight were significantly higher than the rates in African-Caribbeans with normal weight (rate: 11.6; 95 % CI: 7.6, 18.6), and similar to the rates in South Asians with normal weight (rate: 21.3; 95 % CI: 16.1, 28.7) (Figure 1).
Irrespective of weight change pattern before diagnosis (i.e. loss of at least 2 kg of body weight or not), mortality rates per 1000 person-years were significantly higher among White Europeans with normal weight (rate: 12.2; 95% CI: 10.6, 14.1) compared to obese White Europeans (rate: 7.6; 95% CI: 7.2, 8.0). Furthermore, these mortality rates among White Europeans with normal weight were about three-fold higher compared to African-Caribbean (rate: 3.2; 95% CI: 1.5, 8.4) and South Asians (rate: 4.6; 95% CI: 3.1, 7.0) with normal weight (Figure 2).
Association of BMI categories with survival time for CVD and mortality
The adjusted average time to first CVD event (95% CI) and adjusted average time to all-cause mortality (95% CI) in normal weight and overweight patients compared to obese patients with T2DM and without history of disease at diagnosis, separately for males and females within each ethnic group are presented in Table 3. Among White Europeans, compared to obese patients (mean time to CVD of 4.6 years), normal weight patients developed CVD significantly earlier by 0.5 years (95% CI: 0.2, 1.0 years). This pattern was significant in male White Europeans but not for female White Europeans. Furthermore, there was no significant difference between overweight White Europeans and obese White Europeans with regards to time to first CVD event (p > 0.05). The risk of developing CVD was not significantly higher in normal weight African-Caribbeans and South Asians, compared to their obese counterpart.
With a mean time to death of 7.0 and 7.3 years among obese White Europeans and South Asians respectively, those with normal body weight at diagnosis were significantly more likely to die earlier by 0.7 years (95% CI: 0.2, 1.3 years; p=0.006) in the White European group and by 2.5 years (95% CI: 0.5, 4.4 years; p=0.012) in the South Asian group. Among male White Europeans, those who were normal weight at diagnosis also died 0.8 years (95% CI: 0.1, 1.5 years; p=0.021) earlier than obese patients. No evidence of such difference was observed for female White Europeans. However, both male and female South Asians who were normal weight at diagnosis were more likely to die by 2.3 years (95% CI: 0.7, 3.9; p=0.043) and 2.0 years (95% CI: 0.9, 3.1; p<0.001) respectively compared to their obese counterparts.
The time to death was not different between overweight and obese male White Europeans. However, female White Europeans who were overweight at diagnosis were significantly more likely to die earlier by 0.6 years (95% CI: 0.06, 1.1 years; p=0.029) compared to their obese counterparts.
The novelty of this electronic medical record based study from a nationally representative primary care database in incident T2DM patients include assessment of risk profile for different ethnic groups at the time of clinical diagnosis of diabetes by different adiposity level, extensive exploration of weight change patterns prior to diagnosis of diabetes, a robust evaluation of the rates and risk of cardiovascular disease and all-cause mortality in different ethnic groups with different adiposity levels, and the possible gender specific variations in such risk assessments. In this longitudinal outcome study based on three well-defined ethnic groups we found that (1) the paradoxical association of lower BMI with high CVD rate appeared only in White Europeans, and this was not modified by weight change pattern before diagnosis, (2) normal weight White Europeans and South Asians appear to have significantly higher mortality risk compared to their obese counterparts, independent of weight change patterns prior to diagnosis of T2DM, (3) the observed cardiovascular and mortality risk in association with adiposity level appears to be different for male and female, and (4) the BMI at diagnosis was not associated with increased risk of CVD and death among African-Caribbeans.
Obesity is a strong risk factor for cardiovascular diseases in the general population and in some clinical populations. However, increasing evidence is pointing to a paradoxical phenomenon where overweight or obese patients may have better survival outcomes regarding developing heart failure or coronary heart disease, compared to normal weight patients 25, 40. Our analysis in patients with T2DM goes further to show that this paradoxical association between lower BMI and higher CVD risk was strong among White Europeans. In keeping with previous studies, our data shows higher proportions of current smokers among normal weight White Europeans 41, 42 and this could contribute to increased CVD risk among this group of patients. We previously reported that contrary to the notion that the observed obesity paradox could be due to weight loss from latent diseases, weight loss before diagnosis of T2DM was not associated with increased mortality in normal weight patients 10. The current study shows that the significantly higher event rates for CVD and mortality among normal weight patients was independent of weight change pattern before diagnosis of T2DM. This clearly supports the fact that weight change pattern before diagnosis does not impact on the observed obesity paradox in patients with T2DM.
In evaluating the association of BMI levels with mortality risk, we adjusted for weight loss pattern before diagnosis of T2DM, in addition to other known confounders in the risk estimation models and found the paradoxical association of lower BMI with higher mortality risk was more prominent in South Asians than White Europeans. Some studies in patients with T2DM have compared all-cause mortality in different ethnic groups and different BMI categories, but to the best of our knowledge, none have provided ethnicity-specific mortality risk estimates by BMI categories at diagnosis. While the study by Kokkinos and colleagues 28 reported significantly higher mortality risk among African-Caribbeans and Caucasians with normal weight compared to patients with BMI ≥ 35 kg/m2 (reference group), the BMI measure used in this study was not obtained at diagnosis of diabetes. It is important to note that for the purpose of evaluating the obesity paradox, adiposity measures should be obtained at diagnosis of diabetes in keeping with the operational definition of the obesity paradox in T2DM (i.e. phenomenon where patients who were normal weight at diagnosis had significantly elevated mortality risk compared to their obese counterparts). Our risk assessments were based on BMI measured at diagnosis of T2DM and a more pragmatic approach of estimating the time to the events under consideration compared at different adiposity levels with an average 7 years of follow-up time, rather than estimating the hazard ratios which might provide misleading inference under highly heterogeneous characteristics in different ethnic groups. We also ensured the exclusion of patients with known history of diseases that are associated with increased mortality risk.
One of the novel findings of this study was that South Asians male and female with normal body weight at diagnosis were significantly more likely to die earlier by about 2.5 years compared to their counterparts who were obese at diagnosis. One may argue that the proportion of ever-smokers in normal weight South Asians (35%) was significantly higher than that in the obese group (25%), while the distribution of ever-smokers was similar between normal weight and obese White Europeans. However, our analyses were propensity-score balanced for such possible baseline differences, while we do not know the change in smoking behaviour during the 7 years of median follow-up. Our result is contrary to that by Wright and colleagues 27 who reported longer life expectancy and reduced all-cause mortality for older South Asians with diabetes compared to White Europeans. Our current observation of lower prevalence of current or ex-smokers among South Asians was consistent with our previous study 11 and the study by Wright and colleagues 27. Nonetheless, our observation agrees in principle with the study by Bellary and colleagues 22 who reported mean age at death for South Asians to be significantly lower by seven years compared to that in White Europeans. Findings from our study also suggest disparities in the receipt of lifestyle intervention advice, prescription of antidiabetic and cardio-protective drugs among patients with T2DM. These disparities are mostly skewed towards the majority White European population and could be the reason the obesity paradox was more prominent in South Asians than White Europeans.
The current analysis shows significant difference in sexes for the association of BMI with CVD and mortality risk among White Europeans and South Asians. This might partially reflect the connection between sex, obesity and cardiovascular/ mortality outcomes. T2DM increases the risk of several cardiovascular disorders particularly for women 43. However, our data suggest that men were more likely to progress to CVD events among normal weight White Europeans and overweight South Asians. We also observed that White European men and women who were normal weight and overweight at diagnosis respectively, died earlier compared to their obese counterparts. Earlier findings have shown that men develop T2DM at lower BMI than women and hence may be at higher risk of CVD events at lower BMI (normal weight) as seen in our study 44.
Our findings should be interpreted considering the limitations of this study, which include: (1) availability of ethnicity data on a limited number of patients; (2) non-availability of longitudinal data on smoking cessation, and (3) potential for residual confounding as with all observational studies. Despite the issue of limited ethnicity data on some patients, previous work with this cohort by our research group showed that the distribution of sex, smoking status and BMI among persons with missing information on ethnicity was similar to the respective distributions among those with available information on ethnicity 11. Furthermore, we also attempted to minimize bias introduced by confounders by using the “treatment effect” modelling approach. With this approach, robust inferences are provided through appropriate adjustments and balancing of a detailed list of confounders. However, patient-level data from electronic health records still present challenges regarding accuracy and completeness.
In conclusion, our study confirms a paradoxical association BMI and mortality among patients with T2DM and provides new insight into the possible role of ethnicity in explaining the obesity paradox both regarding CVD and total mortality.
1. Carnethon MR, De Chavez PJD, Biggs ML, Lewis CE, Pankow JS, Bertoni AG, Golden SH, Liu K, Mukamal KJ, Campbell-Jenkins B and Dyer AR. Association of weight Status with mortality in adults with incident diabetes. JAMA. 2012;308:581-590.
2. Logue J, Walker JJ, Leese G, Lindsay R, Mcknight J, Morris A, Philip S, Wild S, Sattar N and on behalf of the Scottish Diabetes Research Network Epidemiology Group Association between BMI measured within a year after diagnosis of type 2 diabetes and mortality. Diabetes Care. 2013;36:887–893.
3. Mulnier HE, Seaman HE, Raleigh VS, Soedamah-Muthu SS, Colhoun HM and Lawrenson RA. Mortality in people with type 2 diabetes in the UK. Diabetic Medicine: a journal of the British Diabetic Association. 2006;23:516-521.
4. Thomas G, Khunti K, Curcin V, Molokhia M, Millett C, Majeed A and Paul S. Obesity paradox in people newly diagnosed with type 2 diabetes with and without prior cardiovascular disease. Diabetes, Obesity & Metabolism. 2014;16:317-25.
5. Zoppini G, Verlato G, Leuzinger C, Zamboni C, Brun E, Bonora E and Muggeo M. Body mass index and the risk of mortality in type II diabetic patients from Verona. International Journal of Obesity. 2003;27:281-285.
6. Zhao W, Katzmarzyk PT, Horswell R, Wang Y, Li W, Johnson J, Heymsfield SB, Cefalu WT, Ryan DH and Hu G. Body Mass Index and the Risk of All-Cause Mortality Among Patients with Type 2 Diabetes. Circulation. 2014;130:2143-51.
7. Banack HR and Kaufman JS. The “Obesity Paradox” explained. Epidemiology (Cambridge, Mass). 2013;24:461-462.
8. Banack HR and Kaufman JS. Does selection bias explain the obesity paradox among individuals with cardiovascular disease? Annals of Epidemiology. 2015;25:342-349.
9. Hainer V and Aldhoon-Hainerová I. Obesity paradox does exist. Diabetes Care. 2013;36:S276-S281.
10. Owusu Adjah ES and Paul SK. Evaluating the Obesity Paradox in Type 2 Diabetes with Real-world Data from Primary Care System. Australian Diabetes Society and the Australian Diabetes Educators Association,Annual Scientific Meeting 2016. 2016.
11. Paul SK, Owusu Adjah ES, Samanta M, Patel K, Bellary S, Hanif W and Khunti K. Comparison of body mass index at diagnosis of diabetes in a multi-ethnic population: A case-control study with matched non-diabetic controls. Diabetes, Obesity and Metabolism. 2017;19:1014-1023.
12. Ganz ML, Wintfeld N, Li Q, Alas V, Langer J and Hammer M. The association of body mass index with the risk of type 2 diabetes: a case-control study nested in an electronic health records system in the United States. Diabetol Metab Syndr. 2014;6:50.
13. Hsu WC, Araneta MRG, Kanaya AM, Chiang JL and Fujimoto W. BMI Cut Points to Identify At-Risk Asian Americans for Type 2 Diabetes Screening. Diabetes Care. 2015;38:150-158.
14. Misra A. Ethnic-Specific Criteria for Classification of Body Mass Index: A Perspective for Asian Indians and American Diabetes Association Position Statement. Diabetes Technol Ther. 2015;17:667-71.
15. Misra A, Vikram NK, Gupta R, Pandey RM, Wasir JS and Gupta VP. Waist circumference cutoff points and action levels for Asian Indians for identification of abdominal obesity. International journal of obesity (2005). 2006;30:106-11.
16. Ntuk UE, Gill JMR, Mackay DF, Sattar N and Pell JP. Ethnic-Specific Obesity Cutoffs for Diabetes Risk: Cross-sectional Study of 490,288 UK Biobank Participants. Diabetes Care. 2014;37:2500-2507.
17. Shai I, Jiang R, Manson JE, Stampfer MJ, Willett WC, Colditz GA and Hu FB. Ethnicity, Obesity, and Risk of Type 2 Diabetes in Women: A 20-year follow-up study. Diabetes Care. 2006;29:1585-1590.
18. Tillin T, Sattar N, Godsland IF, Hughes AD, Chaturvedi N and Forouhi NG. Ethnicity-specific obesity cut-points in the development of Type 2 diabetes – a prospective study including three ethnic groups in the United Kingdom. Diabetic medicine : a journal of the British Diabetic Association. 2015;32:226-34.
19. Shah AD, Langenberg C, Rapsomaniki E, Denaxas S, Pujades-Rodriguez M, Gale CP, Deanfield J, Smeeth L, Timmis A and Hemingway H. Type 2 diabetes and incidence of cardiovascular diseases: a cohort study in 1·9 million people. The Lancet Diabetes & Endocrinology. 2015;3:105-113.
20. U.K. Prospective Diabetes Study Group. Ethnicity and Cardiovascular Disease: The incidence of myocardial infarction in White, South Asian, and Afro-Caribbean patients with type 2 diabetes (U.K. Prospective Diabetes Study 32). Diabetes Care. 1998;21:1271-1277.
21. Lanting LC, Joung IMA, Mackenbach JP, Lamberts SWJ and Bootsma AH. Ethnic Differences in Mortality, End-Stage Complications, and Quality of Care Among Diabetic Patients: A review. Diabetes Care. 2005;28:2280-2288.
22. Bellary S, O’Hare JP, Raymond NT, Mughal S, Hanif WM, Jones A, Kumar S and Barnett AH. Premature cardiovascular events and mortality in South Asians with type 2 diabetes in the United Kingdom Asian Diabetes Study – effect of ethnicity on risk. Current medical research and opinion. 2010;26:1873-9.
23. George J, Mathur R, Shah AD, Pujades-Rodriguez M, Denaxas S, Smeeth L, Timmis A and Hemingway H. Ethnicity and the first diagnosis of a wide range of cardiovascular diseases: Associations in a linked electronic health record cohort of 1 million patients. PLOS ONE. 2017;12:e0178945.
24. Lavie CJ, De Schutter A, Patel D, Artham SM and Milani RV. Body Composition and Coronary Heart Disease Mortality—An Obesity or a Lean Paradox? Mayo Clinic Proceedings. 2011;86:857-864.
25. Lavie CJ, Milani RV and Ventura HO. Obesity and the “Obesity Paradox” in cardiovascular diseases. Clinical Pharmacology & Therapeutics. 2011;90:23-25.
26. Chen Y, Copeland WK, Vedanthan R, Grant E, Lee JE, Gu D, Gupta PC, Ramadas K, Inoue M, Tsugane S, Tamakoshi A, Gao Y-T, Yuan J-M, Shu X-O, Ozasa K, Tsuji I, Kakizaki M, Tanaka H, Nishino Y, Chen C-J, Wang R, Yoo K-Y, Ahn Y-O, Ahsan H, Pan W-H, Chen C-S, Pednekar MS, Sauvaget C, Sasazuki S, Yang G, Koh W-P, Xiang Y-B, Ohishi W, Watanabe T, Sugawara Y, Matsuo K, You S-L, Park SK, Kim D-H, Parvez F, Chuang S-Y, Ge W, Rolland B, McLerran D, Sinha R, Thornquist M, Kang D, Feng Z, Boffetta P, Zheng W, He J and Potter JD. Association between body mass index and cardiovascular disease mortality in east Asians and south Asians: pooled analysis of prospective data from the Asia Cohort Consortium. BMJ : British Medical Journal. 2013;347.
27. Wright AK, Kontopantelis E, Emsley R, Buchan I, Sattar N, Rutter MK and Ashcroft DM. Life Expectancy and Cause-Specific Mortality in Type 2 Diabetes: A Population-Based Cohort Study Quantifying Relationships in Ethnic Subgroups. Diabetes Care. 2016.
28. Kokkinos P, Myers J, Faselis C, Doumas M, Kheirbek R and Nylen E. BMI–mortality paradox and fitness in African American and Caucasian men with type 2 diabetes. Diabetes Care. 2012;35:1021-1027.
29. IMS Health Incorporated. The Health Improvement Network (THIN) database. 2017;2017.
30. Blak BT, Thompson M, Dattani H and Bourke A. Generalisability of The Health Improvement Network (THIN) database: demographics, chronic disease prevalence and mortality rates. Informatics in Primary Care. 2011;19:251-255.
31. Denburg MR, Haynes K, Shults J, Lewis JD and Leonard MB. Validation of The Health Improvement Network (THIN) database for epidemiologic studies of chronic kidney disease. Pharmacoepidemiology and drug safety. 2011;20:1138-1149.
32. Read J. The Read clinical classification (Read codes). Br Homoeopath J. 1991;80:14-20.
33. Owusu Adjah ES, Montvida O, Agbeve J and Paul SK. Data Mining Approach to Identify Disease Cohorts from Primary Care Electronic Medical Records: A Case of Diabetes Mellitus. The Open Bioinformatics Journal. 2017;10:16-27.
34. WHO Expert Consultation. Appropriate body-mass index for Asian populations and its implications for policy and intervention strategies. Lancet. 2004;363:157.
35. Ho AK, Bartels CM, Thorpe CT, Pandhi N, Smith MA and Johnson HM. Achieving Weight Loss and Hypertension Control Among Obese Adults: A US Multidisciplinary Group Practice Observational Study. American Journal of Hypertension. 2016;29:984-991.
36. Rubin DB. Estimating causal effects of treatments in randomized and nonrandomized studies. Journal of educational Psychology. 1974;66:688.
37. Austin PC and Stuart EA. The performance of inverse probability of treatment weighting and full matching on the propensity score in the presence of model misspecification when estimating the effect of treatment on survival outcomes. Statistical Methods in Medical Research. 2015;26:1654-1670.
38. Cattaneo MD. Efficient semiparametric estimation of multi-valued treatment effects under ignorability. Journal of Econometrics. 2010;155:138-154.
39. Lee M-j. Treatment effects in sample selection models and their nonparametric estimation. Journal of Econometrics. 2012;167:317-329.
40. Lavie CJ, McAuley PA, Church TS, Milani RV and Blair SN. Obesity and Cardiovascular Diseases. Journal of the American College of Cardiology. 2014;63:1345-1354.
41. Zheng Y, Song M, Manson JE, Giovannucci EL and Hu FB. Group-Based Trajectory of Body Shape From Ages 5 to 55 Years and Cardiometabolic Disease Risk in 2 US Cohorts. American Journal of Epidemiology. 2017;186:1246-1255.
42. Canoy D, Wareham N, Luben R, Welch A, Bingham S, Day N and Khaw KT. Cigarette smoking and fat distribution in 21,828 British men and women: a population-based study. Obes Res. 2005;13:1466-75.
43. Diabetes Mellitus: A Major Risk Factor for Cardiovascular Disease: A Joint Editorial Statement by the American Diabetes Association; the National Heart, Lung, and Blood Institute; the Juvenile Diabetes Foundation International; the National Institute of Diabetes and Digestive and Kidney Diseases; and the American Heart Association. Circulation. 1999;100:1132-1133.
44. Paul S, Thomas G, Majeed A, Khunti K and Klein K. Women develop type 2 diabetes at a higher body mass index than men. Diabetologia. 2012;55:1556-1557.
Table 1: Basic clinical and demographic characteristics of patients with T2DM by ethnicity.
|White European||African-Caribbean||South Asian|
|Age at diagnosis (years) †||52 (12)||48 (11)||47 (12)|
|age group *|
|≤40||7,347 (17)||995 (26)||2,339 (31)|
|41-50||9,916 (24)||1,187 (31)||2,152 (28)|
|51-60||13,401 (32)||952 (25)||1,919 (25)|
|61-70||11,512 (27)||673 (18)||1,166 (15)|
|Male *||23,441 (56)||1,948 (51)||4,062 (54)|
|Smoking status, *|
|Never smoker||17,767 (42)||2,623 (69)||5,590 (78)|
|Current smoker||10,179 (24)||499 (13)||993 (13)|
|Ex-smoker||14,143 (34)||676 (18)||954 (13)|
|Weight (kg) †||95.0 (19)||89.0 (16)||82.0 (15)|
|BMI (kg/m2) †||33.0 (6)||31.0 (5)||30.0 (5)|
|BMI categories *|
|Underweight||71 (0.2)||12 (0.3)||21 (0.3)|
|Normal weight||2,541 (6)||376 (10)||451 (6)|
|Overweight||10,272 (24)||1,124 (30)||1,988 (26)|
|Obese||29,292 (70)||2,295 (60)||5,116 (68)|
|SBP (mmHg)†||139 (16)||137 (16)||134 (16)|
|SBP ≥ 140 mmHg *||20,184 (48)||1,589 (42)||2,538 (34)|
|HBA1c (%)†||9 (2)||9 (2)||9 (2)|
|HBA1c ≥ 7.5%||29,714 (71)||2,802 (74)||5,465 (72)|
|LDL (mg/dl) †||125 (29)||129 (30)||125 (28)|
|HDL (mg/dl) †||45 (11)||47 (11)||44 (10)|
|Triglycerides (mg/dl) ‡||170 (135-212)||127 (97-159)||160 (125-201)|
|LBW prior to diagnosis *||7,926 (19)||694 (18)||1,276 (17)|
|Lifestyle advice *|
|Before diagnosis||13,308 (32)||944 (25)||2,075 (27)|
|After diagnosis||28,552 (68)||2,238 (59)||4,475 (59)|
|Follow-up (years)||7 (4-11)||7 (4-10)||7 (3-10)|
*: n (%); †: mean (SD); ‡: median (Q1, Q3);
BMI: Body mass index;
SPB: Systolic blood pressure;
LDL: Low-density lipoprotein cholesterol;
HDL: High-density lipoprotein cholesterol;
LBW: Lost at least 2kg body weight before diagnosis;
Any CVD: Cardiovascular disease defined as the occurrence angina, myocardial infarction, coronary heart disease (including bypass surgery and angioplasty), heart failure, and stroke on or before diagnosis of T2DM;
CKD: chronic kidney disease
Table 2: Distribution of clinical characteristics among patients with T2DM by ethnicity in each BMI category.
|WE (n=10,272)||AC (n=1,124)||SA (n=1,988)||WE
|AC (n=2,295)||SA (n=5,116)|
|Age at diagnosis (yrs)†||55 (12)||49 (11)||49 (13)||55 (11)||50 (11)||49 (11)||51 (12)||47 (12)||46 (12)|
|age group *|
|≤40||376 (15)||95 (25)||136 (30)||1,025 (10)||228 (20)||520 (26)||5,924 (20)||671 (29)||1,674 (33)|
|41-50||425 (17)||119 (32)||113 (25)||1,941 (19)||344 (31)||583 (29)||7,538 (26)||718 (31)||1,453 (28)|
|51-60||721 (28)||85 (23)||105 (23)||3,551 (35)||317 (28)||526 (27)||9,117 (31)||547 (24)||1,282 (25)|
|61-70||1,019 (40)||77 (21)||97 (22)||3,755 (36.6%)||235 (21)||359 (18)||6,713 (23)||359 (16)||707 (14)|
|Male *||1,444 (57)||270 (72)||289 (64)||6,779 (66)||710 (63)||1,245 (63)||15,183 (52)||958 (42)||2,518 (49)|
|Current smokers||844 (33)||69 (18)||92 (20)||2,513 (25)||148 (13)||285 (14)||6,785 (23)||275 (12)||613 (12)|
|Ex-smokers||670 (26)||60 (16)||66 (15)||3,609 (35)||223 (20)||246 (12)||9,852 (34)||392 (17)||640 (13)|
|Never smokers||1,025 (40)||247 (66)||291 (65)||4,128 (40)||749 (67)||1,457 (73)||12,592 (43)||1,623 (71)||3,826 (75)|
|Hypertension *||845 (33)||130 (35)||111 (25)||4,167 (41)||462 (41)||614 (31)||12,024 (41)||898 (39)||1,626 (32)|
|Lifestyle advice*||1,732 (68)||236 (63)||265 (59)||7,058 (69)||662 (59)||1,259 (63)||19,716 (67)||1,333 (58)||2,943 (58)|
|LBW*||734 (29)||107 (29)||128 (28)||2,188 (21)||222 (20)||469 (24)||4,981 (17)||361 (16)||671 (13)|
|OAD||2,068 (81)||336 (89)||408 (91)||8,210 (80)||980 (87)||1,745 (88)||23,775 (81)||1,879 (82)||4,138 (81)|
|Metformin||1,897 (75)||306 (81)||376 (83)||7,896 (77)||930 (83)||1,679 (85)||23,196 (79)||1,828 (80)||4,001 (78)|
|Sulphonylureas||1,259 (50)||215 (57)||272 (60)||4,254 (41)||540 (48)||899 (45)||10,455 (36)||865 (38)||1,911 (37)|
|TZD||312 (12)||41 (11)||72 (16)||1,426 (14)||113 (10)||262 (13)||3,985 (14)||232 (10)||590 (12)|
|DPP4-i||296 (12)||49 (13)||52 (12)||1,360 (13)||125 (11)||245 (12)||4,271 (15)||268 (12)||648 (13)|
|GLP1-RA||9 (0.4)||1 (0.3)||1 (0.2)||185 (1)||12 (1)||11 (0.6)||2,052 (7)||75 (3)||130 (3)|
|SGLT2-i||11 (0.4)||1 (0.3)||–||56 (0.5)||5 (0.4)||13 (0.7)||357 (1)||10 (0.4)||47 (0.9)|
|Alpha-glucosidase||11 (0.4)||3 (0.8)||3 (0.7)||47 (0.5)||7 (0.6)||10 (0.5)||126 (0.4)||8 (0.3)||19 (0.4)|
|Meglitinide||29 (1)||5 (1)||5 (1)||68 (0.7)||8 (0.7)||14 (0.7)||191 (0.7)||20 (0.9)||41 (0.8)|
|Insulin||539 (21)||61 (16)||53 (12)||1,339 (13)||152 (14)||188 (10)||4,270 (15)||335 (15)||553 (11)|
|WE (n=10,272)||AC (n=1,124)||SA (n=1,988)||WE
|AC (n=2,295)||SA (n=5,116)|
|Other medications *|
|CPM||2,037 (80)||291 (77)||338 (75)||8,825 (86)||882 (79)||1,625 (82)||24,310 (83)||1,697 (74)||3,741 (73)|
|Diuretics||597 (24)||67 (18)||55 (12)||2,959 (29)||283 (25)||319 (16)||9621 (33)||603 (26)||997 (20)|
|Beta-blockers||505 (20)||49 (13)||56 (12)||2,426 (24)||172 (15)||329 (17)||6945 (24)||344 (15)||878 (17)|
|Calcium blockers||614 (24)||117 (31)||94 (21)||3,257 (32)||428 (38)||500 (25)||9065 (31)||948 (41)||1,309 (26)|
|Renin-angiotensin||1,182 (47)||177 (47)||196 (44)||5,925 (58)||572 (51)||967 (49)||17,264 (59)||1,096 (48)||2,430 (48)|
|Ace inhibitors||1,099 (43)||154 (41)||175 (39)||8,103 (79)||721 (64)||1,472 (74)||21,420 (73)||1,345 (59)||3,242 (63)|
|Statins||1,851 (73)||232 (62)||317 (70)||5,361 (52)||502 (45)||841 (42)||15,102 (52)||935 (41)||2,029 (40)|
|Lipid-modifiers||1,865 (73)||234 (62)||318 (71)||8,015 (78)||719 (64)||1,471 (74)||21,208 (72)||1,339 (58)||3,216 (63)|
|Anti-depressants||842 (33)||59 (16)||102 (23)||3,416 (33)||217 (19)||484 (24)||11,897 (41)||462 (20)||1,348 (26)|
*: n (%);
†: mean (SD);
Data are presented for patients without history of disease at diagnosis.
OAD: use of oral antidiabetic drug; CPM use of cardio-protective medications; SPB: Systolic blood pressure;
LDL: Low density lipoprotein cholesterol; HDL: High density lipoprotein cholesterol; LBW: lost at least 2kg body weight before diagnosis; TZD: Thiazolidinedione; DPP4-i: Dipeptidyl peptidase 4 inhibitors; GLP1-RA: Glucagon-like peptide-1 receptor agonists; SGLT2-i: sodium-glucose transport protein 2 inhibitors; WE: White European; AC: African-Caribbean; SA: South Asian.
Table 3: Adjusted average time to first CVD event (95% CI) and adjusted average time to all-cause mortality (95% CI) in normal weight and overweight patients compared to obese patients with T2DM, separately for males and females within each ethnic group. Patients were without history of disease at diagnosis.
|White European (n=42,176)||African-Caribbean (n=3,807)||South Asian (n=7,576)|
|Mean time (years) – Obese||4.6 (4.5, 4.8)||4.5 (3.9, 5.2)||5.0 (4.6, 5.3)|
|Normal weight vs Obese||-0.5 (-0.9, -0.1)||0.006||-0.6 (-2.1, 1.0)||0.495||-0.2 (-1.3, 0.9)||0.719|
|Male||-0.5 (-1.0, -0.04)||0.032||0.3 (-1.4, 2.0)||0.710||-0.3 (-1.7, 1.0)||0.636|
|Female||-0.5 (-1.2, 0.1)||0.093||-2.3 (-4.8, 0.3)||0.086||-0.2 (-1.7, 1.4)||0.838|
|Overweight vs Obese||0.04 (-0.2,0.3)||0.734||1.2 (0.1, 2.2)||0.028||-0.6 (-1.2, 0.0)||0.052|
|Male||0.1 (-0.1, 0.4)||0.304||2.0 (0.7, 3.4)||0.003||-0.9 (-1.6, -0.1)||0.019|
|Female||-0.1 (-0.6, 0.2)||0.393||0.5 (-1.4, 2.3)||0.613||-0.1 (-1.2, 1.0)||0.918|
|Mean time (years) – Obese||7.0 (6.8,7.2)||6.6 (5.6, 7.7)||7.3 (6.6, 7.9)|
|Normal weight vs Obese||-0.7 (-1.3, -0.2)||0.006||-1.0 (-2.5, 0.5)||0.211||-2.5 (-4.4, -0.5)||0.012|
|Male||-0.8 (-1.5, -0.1)||0.021||-1.3 (-2.9, 0.2)||0.085||-2.3 (-3.9, -0.7)||0.043|
|Female||-0.6 (-1.5, 0.2)||0.145||-1.0 (-3.1, 1.1)||0.348||-2.0 (-3.1, -0.9)||0.000|
|Overweight vs Obese||-0.3 (-0.7, 0.01)||0.063||-0.1 (-1.7, 1.5)||0.909||0.1 (-1.1, 1.3)||0.906|
|Male||-0.2 (-0.6, 0.3)||0.469||0.3 (-1.8, 2.4)||0.807||0.3 (-0.8, 1.5)||0.427|
|Female||-0.6 (-1.1, -0.06)||0.029||0.4 (-2.9, 2.1)||0.749||1.0 (-0.06, 2.6)||0.233|
Any CVD: Cardiovascular disease defined as the occurrence angina, myocardial infarction, coronary heart disease (including bypass surgery and angioplasty), heart failure, and stroke post diagnosis of T2D.
Figure 1: Age-weighted CVD event rates per 1000 person-years (95% CI) by BMI categories and weight change pattern before diagnosis in patient without disease history at diagnosis separately for three ethnic groups. CVD: Cardiovascular disease defined as the occurrence angina, myocardial infarction, coronary heart disease (including bypass surgery and angioplasty), heart failure, and stroke post diagnosis of T2D;
Figure 2: Age-weighted all-cause mortality rates per 1000 person-years (95% CI) by BMI categories and weight change pattern before diagnosis in patient without disease history at diagnosis separately for three ethnic groups. ACM: All-cause mortality; rates were not calculated for events less than or equal to 5
Figure 1: Age-weighted CVD event rates per 1000 person-years (95% CI) by BMI categories and weight change pattern before diagnosis in patient without disease history at diagnosis separately for three ethnic groups. Any CVD: Cardiovascular disease defined as the occurrence angina, myocardial infarction, coronary heart disease (including bypass surgery and angioplasty), heart failure, and stroke post diagnosis of T2DM;
Figure 2: Age-weighted all-cause mortality rates per 1000 person-years (95% CI) by BMI categories and weight change pattern before diagnosis in patient without disease history at diagnosis separately for three ethnic groups. ACM: All-cause mortality; rates were not calculated for events less than or equal to 5
Cite This Work
To export a reference to this article please select a referencing stye below:
Related ServicesView all
Related ContentAll Tags
Content relating to: "Nutrition"
Metabolic Profiling of Severely Malnourished Children Treated with Gut-microbiota Modifying Therapeutic Feeds
Metabolic profiling of severely malnourished children treated with gut-microbiota modifying therapeutic feeds. INTRODUCTION Millions of the world’s children suffer from undernutrition, which is...
Dissertation introduction effectiveness Intensive Nutrition Education
Outline of the PhD dissertation This PhD dissertation on the effectiveness of intensive nutrition education (INE) ......
DMCA / Removal Request
If you are the original writer of this dissertation and no longer wish to have your work published on the UKDiss.com website then please: