The study included UK individuals of middle age and with no previous diagnosis of hypertension, diabetes and cardiovascular disease. The findings indicate that during the 1990s and in the study setting, individuals of South-Asian origin did not have an adverse risk factor profile compared with the Caucasian ethnic group. In fact for some risk factors the inverse is true (i.e. more favourable risk factor levels among South-Asians). The findings also appear to show a similar pattern of time trends in risk factor levels for the examined five risk factors between Caucasian and South Asian groups. Individuals of South Asian origin are at higher risk of impaired glycose tolerance and diabetes, in itself a strong independent risk factor of cardiovascular disease, but no data on diabetic status were available for analysis. Nevertheless, the findings would contrast with the hypothesis that increased cardiovascular disease burden in UK South-Asians is due to higher level of exposure to conventional cardiovascular risk factors (other than impaired glucose tolerance and diabetes). Other hypotheses, such as greater genetic susceptibility to cardiovascular disease, may be relevant. Further research should also continue to explore the likely importance of novel and as yet unknown risk factors, and the likely relevance of lower risk of competing morbidity.
The lower levels of total cholesterol and smoking prevalence levels observed in the South-Asian group are in general agreement with previous findings from UK cross-sectional surveys [20, 21], including the 1999 Health Survey for England (HSE) which focused on ethnic minority groups [4]. In particular, both the 1999 HSE and the present study indicate significantly lower prevalence of current smoking among South-Asian individuals particularly for women (although it should be noted that the 1999 HSE found higher smoking prevalence in one of the South-Asian ethnic subgroups – Bangladeshi men); similar or lower levels of cholesterol (particularly in relation to lower levels in South-Asian women); and higher levels of mean BMI in women. Conflictingly, a number of UK studies, previously expertly reviewed [22], have shown both higher (particularly in London) and lower levels of hypertension in South Asian compared to Caucasian populations -the present study being in broad agreement with the latter. In relation to BP, comparisons of the findings of the present study with the 1999 Health Survey for England area difficult, due to the variable findings in the latter survey, depending on South-Asian particular ethnic subgroup and sex.
The risk factor trends for total cholesterol, BMI and current smoking prevalence observed in this study are in agreement with those observed in UK population-based studies of the similar period (e.g. Health Survey for England (HSE) [3], Health Survey for Scotland [23]) i.e. reflecting downward trends for cholesterol and smoking status, and increasing trends for BMI. However, the lack of any considerable decrease in systolic and diastolic blood pressure for both men and women observed in the present study is in contrast to information from population-based epidemiological surveys [3, 23]. This difference raises the question whether the difference can be due to bias, a "real" effect or chance. Selection bias is theoretically possible, as patients with known diagnosis of hypertension and other cardiovascular conditions were excluded from the study, unlike population-based surveys such as the HSE that randomly include both individuals with and without known diagnosis of hypertension (or cardiovascular disease). This hypothesis would have meant that the downward population trends in mean BP observed in HSE surveys of the same period are largely due to strong treatment effects and secular improvement in BP control among known hypertensive participants, and this is judged unlikely. Artefactual explanations are also likely, for example systematic differences in measurement progressively taking place during the study period (e.g. timing allowed to achieve a resting state, body posture, observer technique and training, number of BP readings, calculation of mean values from more than one readings).
A crucial question in relation to the generalisability of the study findings is whether screening participants were representative of the Stockport population of 35–60 year olds who were free of hypertension, diabetes and other cardiovascular disorders. However, if the rigour and accuracy with which individuals were identified and excluded because of known hypertension, diabetes and other cardiovascular conditions was differential between the two ethnic groups, then this potentially could have biased the findings, and this possibility cannot be dismissed with the data that are available.
The study used a computerised package to ascribe ethnicity. Reliance on the same software for identification of ethnic names has been proven not fully reliable before, particularly for South-Asian populations residing outside Northern England [15]. Nevertheless the computerised process of ethnic group assignment was supplemented by visual inspection of both the positive and the negative cases by two independent researchers of diverse ethnic background, which has been shown to improve specificity and specificity [15]. Overall, 1.23% of all cases were assigned South-Asian origin, which corresponds with estimates of the Stockport population of South Asian origin (all ages) of 1.12% in the 1991 and 2.1% in the 2001 Census [13]. In any case, any misclassification errors in the ascertainment of ethnicity would tend to make the two ethnic groups similar, and hence would have made true differences between ethnic groups more difficult to detect is such differences truly existed. Therefore although there is a possibility of undetected differences, those ethnic differences that have been identified in this study are unlikely to be biased by the method used to assign ethnicity.
The fact that Stockport has a relatively small South-Asian community, in combination with the fact that a relatively greater proportion of South-Asian population are younger than 30, is reflected in the small proportion of South Asians included in the study. It is theoretically possible that because Stockport is neither affluent nor deprived, Stockport South Asians are not representative of the wider South Asian communities in the UK.
An important limitation of the study is that the South-Asian ethnic group category contains an important degree of heterogeneity in terms of ethnic group, religion, culture and country of origin. Previous research has showed considerable heterogeneity in relation to exposure to cardiovascular risk factors between UK South-Asian ethnic groups [24]. Nevertheless, aggregation of all South-Asian groups has helped increase the statistical precision of estimates. In the 1991 census, within Stockport South-Asian individuals aged over 30 years of age, the relative proportion of those categorised as of Indian, Pakistani and Bangladeshi origin was 43.4 %, 49.5% and 7.1% respectively (data available from MIMAS, http://www.mimas.ac.uk). In the 2001 census a similar overall breakdown of South-Asian groups was observed (31.3%, 49.3% and 5.8%). It is therefore likely that the relative proportion of South-Asian subgroups in the dataset is a reflection of the proportions of subgroups of South-Asian origin in the community, but no further detailed study of such subgroups was possible. More importantly, it is unlikely that the observed time trends in risk reflect changes in the ethnic sub-group composition of the South-Asian group. Individuals belonging to ethnic subgroups are known to have an overall lower access rate to preventive interventions. We therefore acknowledge that it is possible that screening participants of South-Asian origin (and of some South-Asian subgroups in particular) may not be representative of the overall community. Unfortunately quantifying this potential problem is not possible with the available data.