key: cord-0903948-mc9vvr29
authors: Anderson, Jana J; Welsh, Paul; Ho, Frederick K; Ferguson, Lyn D; Welsh, Claire E; Pellicori, Pierpaolo; Cleland, John G F; Forbes, John; Iliodromiti, Stamatina; Boyle, James; Lindsay, Robert; Celis-Morales, Carlos; Gray, Stuart Robert; Katikireddi, Srinivasa Vittal; Gill, Jason Martin Regnald; Pell, Jill P; Sattar, Naveed
title: Ethnic differences in prevalence of actionable HbA1c levels in UK Biobank: implications for screening
date: 2021-08-05
journal: BMJ Open Diabetes Res Care
DOI: 10.1136/bmjdrc-2021-002176
sha: 1de08fb838ee6c41559481efebfe4a0fed8012a5
doc_id: 903948
cord_uid: mc9vvr29

INTRODUCTION: Early detection and treatment of diabetes as well as its prevention help lessen longer-term complications. We determined the prevalence of pre-diabetes and undiagnosed diabetes in the UK Biobank and standardized the results to the UK general population. RESEARCH DESIGN AND METHODS: This cross-sectional study analyzed baseline UK Biobank data on plasma glycated hemoglobin (HbA1c) to compare the prevalence of pre-diabetes and undiagnosed diabetes mellitus in white, South Asian, black, and Chinese participants. The overall and ethnic-specific results were standardized to the UK general population aged 40–70 years of age. RESULTS: Within the UK Biobank, the overall crude prevalence was 3.6% for pre-diabetes, 0.8% for undiagnosed diabetes, and 4.4% for either. Following standardization to the UK general population, the results were similar at 3.8%, 0.8%, and 4.7%, respectively. Crude prevalence was much higher in South Asian (11.0% pre-diabetes; 3.6% undiagnosed diabetes; 14.6% either) or black (13.8% pre-diabetes; 3.0% undiagnosed diabetes; 16.8% either) participants. Only six middle-aged or old-aged South Asian individuals or seven black would need to be tested to identify an HbA1c result that merits action. CONCLUSIONS: Single-stage population screening for pre-diabetes or undiagnosed diabetes in middle-old or old-aged South Asian and black individuals using HbA1c could be efficient and should be considered.

The prevalence of diabetes is rising worldwide, particularly in low-income and middleincome countries. 1 Diabetes has long been recognized as a major cause of morbidity and mortality. Most recently, diabetes has been identified as a risk factor for adverse COVID-19 outcomes, including hospitalization and death. 2 3 In 2017, it was estimated that the global prevalence of pre-diabetes was 7.7% and 4.2% for undiagnosed diabetes. 4 It is predicted that these numbers will rise significantly by 2045. 4 There is now clear evidence that people with pre-diabetes are not just at elevated risk of developing diabetes but also at risk of adverse cardiovascular outcomes. 5 Lifestyle changes and metformin can prevent progression to diabetes, 6 and diabetes prevention programs now operate in many countries, including the UK. 7 However, the scale of prediabetes and undiagnosed diabetes and the best way to identify people with these conditions are currently unclear.

The aim of the current study was to determine the prevalence of pre-diabetes and undiagnosed diabetes by ethnic group and the average number of people who would need to undergo testing to identify each case.

What is already known about this subject? ► Diabetes prevention programs now operate in many countries, including the UK. ► How to efficiently screen for people at elevated risk and in different ethnicities remains uncertain. All participants gave written informed consent before enrollment in the study.

Deprivation status was based on the Townsend Deprivation Index derived from the postcode of residence at recruitment. Body mass index (BMI) was calculated from weight/height 2 and categorized into underweight (<18.5 kg/m 2 ), normal weight (18.5-24.9 kg/m 2 ), overweight (25.0-29.9 kg/m 2 ), and obese (≥30.0 kg/m 2 ), according to the WHO classification. A lifestyle score, previously shown to be associated with all-cause mortality, was used to characterize overall lifestyle as described previously. 9 10 Lifestyle factors, self-reported at baseline, included smoking status (current, former, or never), physical activity (time spent doing moderate and vigorous physical activity per week, converted to metabolic equivalents (METs)-min/week and dichotomized into inactive (<600 METs-min/week) and active (≥600 METs-min/ week) according to the physical activity guidelines 11 ), sedentary time (watching television, using a computer, and non-work-related driving), adequate sleep, and optimal dietary intake of fruit and vegetables (≥5/day), red meat (<70 g/day) and processed meat (never or less than once a week), oily fish (≥1/week), and alcohol (<14 units/week). The overall lifestyle score ranged from 0 (most unhealthy; highest risk of all-cause mortality) to 9 (most healthy; lowest risk of all-cause mortality). Sampling procedures for UK Biobank biomarkers have been described and validated previously. 12 13 Briefly, biochemistry analyses were performed at a dedicated central laboratory on 480 000 samples between 2014 and 2017 and included plasma glycated hemoglobin (HbA1c) (VARIANT II TURBO Hemoglobin Testing System; Bio-Rad). Data were adjusted for preanalytical variables by the UK Biobank centrally before release.

Known diabetes was defined as at least one of the following: type 1 or type 2 diabetes self-reported by the participant at baseline and/or taking insulin or other diabetes-related medications (metformin, sulfonylurea, meglitinides/glinides/prandial glucose regulators, alphaglucosidase inhibitors, thiazolidinedione/glitazones, dipeptidyl peptidase-4 (DPP-4) inhibitors/gliptins, incretin mimetics/glucagon-like peptide-1 (GLP-1) analogs, and amylin analogs). Participants with known diabetes were excluded from the study. Among the remaining participants, pre-diabetes and undiagnosed diabetes were ascertained from HbA1c concentrations and defined as HbA1c 42-47.9 mmol/mol and ≥48 mmol/mol, respectively. 14 Sensitivity analyses were conducted excluding participants with self-reported cardiovascular disease at baseline (heart attack, angina, stroke, or transient ischemic attack), since such individuals are likely to be already in receipt of preventive pharmaceutical or lifestyle interventions.

Participants were stratified by ethnicity and sex and their baseline characteristics summarized using percentages for categorical variables and mean and SD, or median and IQR, for continuous variables. Crude prevalence was calculated by dividing the total number of cases by the total number of the population in each category. Crude prevalence was derived overall and by age, gender, and ethnic subgroup. The 2011 Census data 15 were used to standardize the prevalence rates to the age, gender, and ethnic breakdown of the UK general population, within the age group recruited to the UK Biobank. Briefly, sex-specifc and ethnic-specific prevalence rates obtained from the UK Biobank were used to estimate the total cases in the UK general population, according to the total of men and women in each of the ethnic group in the 40-70 age group only, and the corresponding UK population prevalence rates for all 40-70 years old were recalculated. The yield (average number needed to test to detect a case) was derived from inversion of the prevalence.

UK Biobank participants who had missing data on HbA1c (n=36 135), had known diabetes at baseline (n=24 593), or classified themselves as mixed or other ethnic group (n=8081) were excluded from the study. The resultant study population comprised 433 856 participants: 419 512 (96.7%) white, 7400 (1.7%) South Asian, 5578 (1.3%) black, and 1366 (0.3%) Chinese. This compared with 94.0%, 3.2%, 2.4%, and 0.5%, respectively, in the UK general population within the same age range. White participants were older, black participants were more likely to live in the most deprived areas, and black women had the highest BMI (table 1). Lifestyle score was higher (healthier) among women than men but did not differ significantly by ethnic group.

Overall, the crude prevalence was 3.6% for prediabetes, 0.8% for undiagnosed diabetes, and 4.4% for either (table 2) . However, there were wide variations by ethnic group. The prevalence of pre-diabetes was only 3.3% among white participants compared with 11.0% and 13.8% in South Asian and black, respectively. Similarly, only 0.7% of white participants had undiagnosed diabetes compared with 3.6% of South Asian and 3.0% Table 1 Characteristics of UK Biobank participants by ethnic group and sex, excluding people with known diabetes at baseline .7) 108 ( (table 3) . When the results were recalculated for age and sex subgroups within each ethnicity group, there was a consistent trend in white and South Asian participants, whereby the prevalence of pre-diabetes or undiagnosed diabetes was higher in men than in women and increased with increasing age (tables 4 and 5). Among South Asian or black men over 55 years of age, only three to five individuals needed to be tested to identify an actionable HbA1c value.

The sensitivity analyses, which excluded 22 873 participants with self-reported cardiovascular disease at baseline, made little meaningful difference to the prevalence rates or the average number needed to be tested to identify cases both overall and by ethnic group (online supplemental table S1).

To investigate the impact of the characteristics that differed among the included ethnic groups on their increased risk of pre-diabetes and undiagnosed diabetes, we ran logistic regression models unadjusted and adjusted for sociodemographic variables (sex, age, deprivation) and lifestyle (lifestyle score). All ethnic groups had a much stronger risk of pre-diabetes or undiagnosed diabetes than white participants. After adjustment, the risk of both prediabetes and undiagnosed diabetes became even stronger for all ethnic groups but black participants, where it attenuated slightly (online supplemental table S2).

Our findings suggest that 1 in 22 (4.7%) of individuals aged 40-70 years old in the UK have actionable HbA1c concentrations. More importantly, 1 in 6-7 individuals of black or South Asian ethnicity have actionable values and approximately 1 in 30 are living with undiagnosed diabetes. The risk of having pre-diabetes or undiagnosed diabetes is magnified by older age, male sex, living in deprived areas, and having unhealthy lifestyle. Given that diabetes prevention programs are now increasingly well developed, including the English program, 7 these data suggest that screening based on ethnicity and age would be extremely efficient at identifying people at high risk of developing type 2 diabetes. For white individuals, needing to test 25 individuals to detect an actionable HbA1c value suggests a two-stage process may be more cost-effective, in line with the current National Institute for Health and Care Excellence guidelines, 16 with the first stage being a simple risk score (which could easily be self-completed). Such a score would identify those at higher risk of prevalent diabetes and therefore at higher risk of being in the pre-diabetes stage since there is a continuum of risk. The benefit of detecting those at higher risk of diabetes to delay or prevent conversion to frank diabetes has been widely accepted. Follow-up of earlier prevention studies has now shown that diabetes prevention lessens cardiovascular outcomes and lowers mortality risk. 17 While most such studies used oral glucose tolerance test-based criteria to determine those at elevated risk of diabetes, higher HbA1c levels are well accepted in predicting incident cardiovascular disease 18 and microvascular damage, 19 at least as well as other glycemia measures. As HbA1c can be done any time of the day, irrespective of fasting status, it is a more feasible test for widespread community-based adoption.

Of course, HbA1c is more expensive than blood glucose and so its use for screening purposes should be weighed carefully against the cost of mass testing. That noted, given that one in six individuals of black ethnicity or one in seven of South Asian ethnicity in the 40-70 year-old age ranges have actionable HbA1c levels, it would potentially make widespread HbA1c testing in these populations cost-effective, and especially if in addition to preventing some progressing to diabetes, it is also possible to identify and treat those with undiagnosed diabetes. Many people are still not diagnosed until the development of serious complications. The importance of identifying people with undiagnosed diabetes is reinforced by the emergence of evidence-based approaches to diabetes remission and earlier intervention with pharmacological therapies that reduce the risk of progression of complications. That UK Biobank does not cover people older than 70 years of age may be seen as a limitation; however, it is notable that those with younger-onset type 2 diabetes lose more life years from diabetes than those who develop it when much older. 20 Therefore, there is greater merit in identifying younger people at risk of diabetes or with undiagnosed diabetes.

While there are debates about the use of HbA1c in different ethnicities, important data from the Outcome Reduction with Initial Glargine Intervention (ORIGIN) trial of 12 527 people reported that the strong relationship between A1C and fasting plasma glucose (FPG) in people with moderate dysglycemia (5.6-9.0 mmol/L) is not significantly affected by ethnic or geographical differences. 21 This range includes those with pre-diabetes, and as such the findings lend strong confidence that HbA1c levels in different ethnicities reflect similar dysregulation in glycemia levels.

In the UK, risk screening is supposed to link a risk score to identify those at highest risk, followed by formal glycemia testing either in fasted state (using fasting glucose) or any time of the day (HbA1c) in those with higher diabetes risk scores. Our work suggests that such risk scores may not be needed for middle-aged and old-aged South Asian or black individuals in whom the yield appears sufficiently high to consider singlestage mass screening using HbA1c. However, further studies are required to determine feasibility, uptake, and cost-effectiveness.

The work has some notable strengths, with the size and coverage of the study surpassing all prior studies in this area. Even so, we accept there are some limitations to our study. The sociodemographic representativeness of the UK Biobank is not identical to the general population. 22 We addressed this partially by standardizing our estimates against the UK general population in terms of age, sex, and ethnic group distribution and we obtained very similar results. However, estimates of prevalence in the UK general population should still be taken with caution, as UK Biobank participants are less likely to be from a deprived area, less likely to be obese and to have a better lifestyle, and there is evidence of a 'healthy volunteer' selection bias. 22 Indeed, it may be that the UK Biobank underestimates glycemia risks in some ethnicities. We had insufficient statistical power to include some ethnic groups, such as Arabs, and were obliged to amalgamate others. HbA1c values were also not available in a small proportion of people, but the level of missing data was comparable between ethnic groups and is unlikely to be systematic. We also recognize that HbA1c can sometimes be erroneous in people with certain hemoglobinopathies, but notably hemoglobin A1c was reportable in the presence of HbS, HbC, HbD, and HbE traits for the assay method used in the UK Biobank. We also recognize the a cut-off age of 40 years old means that younger people at risk of undiagnosed diabetes are not captured, but even so the results in other age groups should still be valid. Finally, the distribution of HbA1c was positively skewed and it is possible that those in the lower ranges of the HbA1c could be a dynamic group that may revert into 'normal' ranges.

We conclude that HbA1c is extremely efficient at identifying pre-diabetes and undiagnosed diabetes in middleaged and old-aged South Asian and black individuals. Therefore, consideration should be given to single-stage mass screening of these high-risk populations using HbA1c.

Global prevalence of diabetes: estimates for the year 2000 and projections for 2030

Associations of type 1 and type 2 diabetes with COVID-19-related mortality in England: a wholepopulation study

Living risk prediction algorithm (QCOVID) for risk of hospital admission and mortality from coronavirus 19 in adults: national derivation and validation cohort study

IDF diabetes atlas: global estimates of diabetes prevalence for 2017 and projections for 2045

Glycated hemoglobin, prediabetes, and the links to cardiovascular disease: data from UK Biobank

Preventing type 2 diabetes: systematic review of studies of cost-effectiveness of lifestyle programmes and metformin, with and without screening, for prediabetes

Early outcomes from the English National health service diabetes prevention programme

Uk Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age

Traditional and emerging lifestyle risk behaviors and all-cause mortality in middle-aged and older adults: evidence from a large population-based Australian cohort

Men across a range of ethnicities have a higher prevalence of diabetes: findings from a cross-sectional study of 500 000 UK Biobank participants

World Health Organization. Global recommendations on physical activity for health. World Heal Organ

The UK Biobank sample handling and storage protocol for the collection, processing and archiving of human blood and urine

Uk Biobank showcase: biomarker assay quality procedures: approaches used to minimise systematic and random errors (and the wider epidemiological implications). version 1.2, 02/04/2019

Definition and diagnosis of diabetes mellitus and intermediate hyperglycemia

Risk identification and interventions to prevent type 2 diabetes in adults at high risk: summary of NICE guidance

Morbidity and mortality after lifestyle intervention for people with impaired glucose tolerance: 30-year results of the dA Qing diabetes prevention outcome study

Glycated hemoglobin measurement and prediction of cardiovascular disease

Glycemic thresholds for diabetes-specific retinopathy: implications for diagnostic criteria for diabetes

Age at diagnosis of type 2 diabetes mellitus and associations with cardiovascular and mortality risks

Relationship between A1c and fasting plasma glucose in dysglycemia or type 2 diabetes: an analysis of baseline data from the origin trial

Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population

Acknowledgements The authors would like to thank the participants of the UK Biobank and the UK Biobank for access to these data. The authors thank Liz Coyle, University of Glasgow, for her assistance in the preparation of this article.Contributors JJA, JPP, and NS conceived the idea, with useful methodological discussions among all authors. JJA analyzed the data. JJA, JPP, and NS cowrote the first draft. All other authors contributed to discussions and reviewed/edited the manuscript. NS is the guarantor of this work and as such had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Competing interests NS reports grant and personal fees from Boehringer Ingelheim, and personal fees from Amgen, AstraZeneca, Eli Lilly, Merck Sharp & Dohme, Novartis, Novo Nordisk, Pfizer, and Sanofi, outside the submitted work.Patient consent for publication Not required.Ethics approval The UK Biobank received ethical approval from the North West Multi-centre Research Ethics Committee (REC reference: 11/NW/0382). The study was conducted in accordance with the principles of the Declaration of Helsinki.Provenance and peer review Not commissioned; externally peer reviewed.Data availability statement Data may be obtained from a third party and are not publicly available. UK Biobank data can be requested by bona fide researchers for approved projects, including replication, through https://www. ukbiobank. ac. uk/.Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise. Paul Welsh http:// orcid. org/ 0000-0002-7970-3643 Frederick K Ho http:// orcid. org/ 0000-0001-7190-9025 Stamatina Iliodromiti http:// orcid. org/ 0000-0001-6453-6654 Carlos Celis-Morales http:// orcid. org/ 0000-0003-2612-3917 Stuart Robert Gray http:// orcid. org/ 0000-0001-8969-9636 Naveed Sattar http:// orcid. org/ 0000-0002-1604-2593