In our population of over 3500 White European and South Asian individuals who were previously undiagnosed and would be eligible for the NHS Health Check Programme, the overall prevalence of screen-detected diabetes was approximately 5%, of IGR was approximately 10%, of high CVD risk was 18%, and of CKD was 8%. Furthermore, 34% of individuals had at least one of these risk factors for vascular disease, and only 0.4% had all of these risk factors. Generally, risk factor prevalence was higher among South Asians than White Europeans, except for CKD.
Strengths and limitations of the study
A strength of our study was that it was conducted on a cohort with a high proportion of individuals of South Asian origin. Whilst we acknowledge that the utility of ethnicity as a concept is limited, it is well-established that there are ethnic differences in the risk of certain diseases such as diabetes due to genetic and cultural risk factors. Furthermore, ethnicity-related inequalities in health care continue to exist. Understanding the prevalence of vascular risk factors in non-White groups is therefore highly important as these differences might reflect underlying biological mechanisms, but might also reflect differences in contact with the health care system.
A major limitation of our study is that it suffered from a low response rate with just over 20% of eligible patients attending screening appointments 
, and so is likely subject to some degree of selection bias. There may be a number of reasons for this but it is likely that the time consuming process of attending an oral glucose tolerance test deterred many people from attending the screening appointment. Also, many eligible patients were from deprived areas and/or ethnic minority populations which may have affected participation rates. The extent of the selection bias is probably small because the response rate is similar to other studies conducted in deprived areas or in people of minority ethnic groups 
, the patient characteristics of those participating in the study are similar to those of the Leicester population 
(for example, 24% of the study population was South Asian compared with 27% of adults in Leicester 
), and 18% of individuals were found to have a high CVD risk, a finding similar to that suggested in the NHS Health Check Programme 
, suggesting that similar selection biases are present in our study to those in the Programme. People who take part in studies tend be healthier than the general population. Thus, the likely consequence of a selection bias is that our prevalence estimates for the individual risk factors will be underestimated, which could also result in an underestimation of the overlap between the risk factors. It is unclear what the consequence of this would be on our estimates at the national level as it would depend on how much the overlap increased in relation to the individual components increasing.
We held detailed medical history for the study participants and so were able to closely replicate the inclusion criteria for the NHS Health Check Programme. However, we were unable to exclude people with previously diagnosed CKD and so our estimates of undiagnosed CKD might be artificially inflated. This is unlikely to affect the comparison between ethnic groups in terms of kidney disease prevalence.
A potential drawback of our study is that we used only one risk score (Ethrisk) to identify people at a high risk of CVD when there are several such measures available. NICE do not currently recommend one CVD risk score over another but the most commonly used ones are Framingham, Ethrisk (which is based on Framingham but with an ethnicity correction), and QRisk. It appears that Framingham overestimates CVD risk in comparison with QRisk, which would not affect the relative comparisons within our study, but could potentially have inflated the absolute estimates.
Multiple imputation, rather than complete case analysis, was used as both lead to negligible biases when data are missing at random, but multiple imputation is more efficient. Furthermore, the most data that are missing for any one variable is 3% so the multiple imputation is likely to have had a negligible effect on our findings.
Finally, the vast majority of the participants in ADDITION-Leicester were of White European or South Asian ethnicity; therefore, we could not reasonably estimate the prevalence of risk factors in other ethnic groups and so these individuals were excluded from the present analyses. A consequence of this is that when we standardised our prevalence rates to the UK population we had to make an assumption that people of other ethnicities had risk factor prevalence that was similar to White Europeans, South Asians or somewhere in between. If in fact the prevalence is much higher or much lower among other ethnic groups then our extrapolations may not be reasonable. Nevertheless, any effect is likely to be small because altogether these groups only comprise a small percentage (6.2%) of the UK population.
Comparison with other studies
The ethnic composition of our study population allowed us to investigate the prevalence of various vascular risk factors in a population that comprised a higher proportion of South Asians than previous studies. Our prevalence estimates for the individual risk factors that we considered are in agreement with existing estimates 
. There are less data available for the commonality of these risk factors. In our study, while many people had at least one risk factor, few people had all of them. Only approximately 10% of the people with screen-detected diabetes also had CKD. This estimate is much lower than that observed in other studies 
. This might be because there were a higher proportion of South Asians in our study than in previous studies, and South Asians had a lower prevalence of CKD than White Europeans, or because these diseases might develop during different time frames and so are not all present cross-sectionally, or because only screen-detected diabetes was included whereas the overlap might have been greater had prevalent diabetes also been considered.
There are several possible reasons why South Asians had a lower prevalence of CKD than White Europeans. For example, it might be because we used the MDRD equation to estimate GFR which might underestimate CKD among South Asians 
. We chose to use the MDRD equation, rather than alternatives such as CKD-EPI, because MDRD is most commonly used in clinical practice. However, it is a known limitation of all eGFR equations that currently none have been adequately validated in South Asian populations and tend to underestimate CKD. An alternative explanation for the differences in CKD prevalence is related to ethnic differences in diabetes prevalence. GFR tends to be raised in people with newly diagnosed Type 2 diabetes compared with people with normoglycaemia, and so it might be expected that people diagnosed with diabetes through the NHS Health Check Programme would be at a reduced risk of CKD.
We observed greater overlap between high CVD risk and both screen-detected diabetes and CKD. This is consistent with the increased risk of CVD observed in people with diabetes and/or CKD 
, and with the observed association between risk markers of CKD 
and cardiovascular events.
Our findings have important implications for the NHS Health Check Programme, which was introduced by the Department of Health in April 2009 
. From our population sample of 6749 individuals, 33% were excluded because they had had a previous cardiovascular event or some other condition that means they should already be monitored through an existing care pathway. This number is in fact slightly higher because people with diabetes were already excluded from our population sample. It is possible that the people who attended screening were more likely to have a pre-existing condition. Nevertheless, this high percentage of exclusions indicates a high burden of existing monitoring that is likely to be substantially increased by the Programme. The low degree of risk factor commonality in our study suggests that a greater number of individuals will require management than if the degree of commonality was high. Our projections for the number of new diagnoses of diabetes, high CVD risk and CKD are much higher than the previously quoted figure of at least 20,000 diagnoses 
, even when the projections were based on a screening uptake of only 20%. This might in part be a consequence of us being unable to exclude people with previously diagnosed CKD from our analyses since this information was not collected in the ADDITION-Leicester study. However, the effect of this is likely to be small because our projections for diabetes and CVD risk only varied by a few thousand when we used the extreme assumption that all of the people that were diagnosed with CKD in ADDITION-Leicester were already known to have the disease and thus would be ineligible for the NHS Health Check Programme (data not shown). Our higher projections suggest that the burden of risk factors diagnosed though the NHS Health Check Programme on primary care trusts might be greater than previously anticipated.
The NHS Health Check Programme offers a real opportunity to make significant contributions to changing health inequalities, including ethnic, socio-economic and sex inequalities. However, it requires individual primary care trusts to ensure that their approach is appropriate for their own community 
. In view of the differences in risk factor prevalence between ethnic groups that were highlighted in our study, this is particularly important in a population comprising different ethnic communities where individuals may not get the services they need because of differences in language, literacy and culture 
. Screening of individuals at risk of vascular disease will thus need to be adapted to provide for individual needs. It may be that part of the screening programme ensures that individuals are aware of their ethnic-specific risks so that they are more likely to access screening programmes. Individual ethnic groups may also need tailored intervention programmes. For example, systematic reviews have shown that levels of physical activity are lower in all South Asian groups 
and so it may be beneficial to advocate exercise within this community.
Future research and Summary
Among White Europeans and South Asians, the prevalence of screen-detected diabetes, IGR, high CVD risk and CKD is high, but there is little overlap between these vascular risk factors. Future research should focus on outcomes in the group of individuals who have all of these risk factors as they are potentially a very high risk group. Additionally, we have highlighted ethnic differences in prevalence of vascular risk factors. Our findings suggest that the burden of risk factors diagnosed through the NHS Health Check Programme on general practitioners may be greater than previously expected, and emphasise the importance of offering the Programme in a wide range of settings with a view to decreasing health inequalities between ethnic groups.