The performance of the Rose angina questionnaire in South Asian and European origin populations: a comparative study in Newcastle, UK

Transcription

1 International Epidemiological Association 2001 Printed in Great Britain International Journal of Epidemiology 2001;30: The performance of the Rose angina questionnaire in South Asian and European origin populations: a comparative study in Newcastle, UK Colin M Fischbacher, a Raj Bhopal, a,b Nigel Unwin, a,c Martin White a and KGMM Alberti c Background The Rose angina questionnaire has been extensively used in different cultural settings, but may not perform consistently in different ethnic groups. We set out to assess the performance of the Rose angina questionnaire in UK South Asians compared with Europeans. Methods Data on major ECG abnormalities, possible or definite Rose questionnaire angina and diagnosed angina were collected from an age- and sex-stratified random sample of 1509 adults from European, Indian, Pakistani and Bangladeshi ethnic groups. Results The ECG abnormalities were commoner in South Asians than Europeans (6% versus 2% in men). The prevalence in both South Asian and European men of possible Rose angina and diagnosed angina was 18% and 8%, respectively, but definite Rose angina was less common in South Asians (3% versus 6%). Definite Rose angina showed lower sensitivity for other measures in South Asians than in Europeans: sensitivity for a doctor s diagnosis was 21% in South Asian and 37% in European men. For possible Rose angina, the corresponding figures were 81% and 84%. Similar patterns were seen in women. Conclusions The performance of the Rose angina questionnaire was sufficiently inconsistent to warrant further work to achieve greater cross-cultural validity. Possible Rose angina performed more consistently across ethnic groups than definite Rose angina and pending further validation studies may be the most appropriate form to use. Keywords Angina pectoris, ethnic minorities, test performance, questionnaire, ischaemic heart disease, chest pain Accepted 15 June 2001 The Rose questionnaire (RQ) was introduced in 1962 as a standardized method of measuring angina and myocardial infarction in population surveys (Appendix). It defines angina as chest pain that limits exertion, is situated over the sternum or in the left chest and left arm, and is relieved within 10 minutes by rest. 1 The diagnosis of angina on RQ predicts morbidity a Department of Epidemiology and Public Health, The Medical School, University of Newcastle upon Tyne, Newcastle NE2 4HH, UK. b Department of Public Health Sciences, University of Edinburgh, Edinburgh EH8 9AG, UK. c Department of Diabetes, The Medical School, University of Newcastle upon Tyne, Newcastle NE2 4HH, UK. Correspondence: Dr Colin M Fischbacher, Department of Epidemiology and Public Health, The Medical School, University of Newcastle, Newcastle upon Tyne NE2 4HH, UK. c.m.fischbacher@ncl.ac.uk and mortality in European populations independently of other clinical findings, 2 5 and measures the burden of disease. The RQ has been widely used in its original and modified forms, 6 and in translation 7 10 to study the prevalence and natural history of ischaemic heart disease, 7 14 response to interventions, 15 and to make population comparisons. 16,17 There is no agreed gold standard which can be used to validate RQ measures of angina but studies have used expert clinical diagnosis, 1,2,18 21 ECG findings, exercise thallium scans 25,26 and coronary angiography. 9 Clearly without such a standard the terms sensitivity and specificity need to be used with caution. Validation studies have generally reported high specificity (80 95%) but variable sensitivity (19 83%). There is evidence that the RQ is less valid in women. 27,28 Most validation studies have used the traditional RQ definition of angina. However, there is evidence from population-based 1009

2 1010 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY studies 5,23,29 that exertional chest pain alone (RQP, P for possible) has a similar prognosis to definite RQ angina (RQD). Almost all validation studies of the RQ have been performed in European or North American white ethnic groups, the exceptions being a study including Americans of Japanese descent 20 and two that included both black and white US residents. 24,30 Only one study 31 has assessed the Rose questionnaire in a population of UK South Asians (of mainly Punjabi ethnic origin). This study, by Patel et al., suggested that the questionnaire was inaccurate in South Asians because of its weak agreement with the respondents history of cardiac disease and the assessment of a clinician. A comparison group was not included, so the question of whether the performance of the questionnaire was worse than in European populations could not be assessed. The cross-cultural performance of the RQ is important because coronary heart disease (CHD) is a major cause of morbidity 32 and the leading cause of mortality 33 in UK residents of South Asian ethnic origin, and several UK studies have used the RQ to report the prevalence of angina in South Asians (though there are no studies linking it to outcomes). We used data from the Newcastle Heart Project to study the comparative performance of the RQ in three South-Asian groups and a European ethnic group. In order to establish whether the RQ performed equally well in Europeans and South Asians, we tested the hypothesis that compared with diagnosed angina and ECG changes the RQ provided similar sensitivity and specificity, similar levels of agreement with these measures, and similar ethnic rankings of prevalence. Methods Interviews and ECG data The Newcastle Heart Project is a cross-sectional study of a population sample of 1509 Newcastle residents of European (n = 825), Indian (n = 259), Pakistani (n = 305) and Bangladeshi (n = 120) ethnic origin. Full details of the methods have been published. 36 Europeans were screened between April 1993 and October 1994 and South Asians between May 1995 and March The response rate was 67% for South Asians and 64% for Europeans. Local ethical committee approval was obtained and all participants gave informed consent. Clinical data included ECG, biochemical, anthropometric and cardiovascular measures. A questionnaire (covering lifestyle and cardiovascular risk factors) was first translated into Hindi, Punjabi, Urdu and Bengali and then back-translated into English. Researchers conferred with translators and back-translators until agreement on translation was reached. South Asians were interviewed in the subject s home and in their preferred language by interviewers of their own ethnic group and gender. European respondents self-completed the questionnaire while attending for screening. Respondents completed the standard WHO Rose questionnaire (RQ) 37 and were asked whether a doctor had ever told them that they had angina. Those who fulfilled all the Rose criteria were classified as having definite RQ angina (RQD). Exertional chest pain (with or without other Rose criteria) was defined as possible RQ angina (RQP). Resting 12-lead ECG were Minnesota coded 38 by two trained and independent coders, and discrepant reports were resolved with a third independent coder. Probable CHD was defined as Minnesota coding (large Q and QS waves) and possible CHD as Minnesota coding 1.3 (small Q and QS), (ST-T depression), (flattened or inverted T waves) or (complete left bundle branch block). Statistical methods Prevalences were directly age standardized to the 1991 England and Wales population using Stata 7 (Stata Corporation, College Station, TX). Agreement between measures was estimated using Cohen s kappa statistic, which takes the value one when there is perfect agreement and zero when there is no agreement beyond that observed by chance. We used a chi squared test for heterogeneity (χ 2 HE) described by Donner 39 to test for differences by ethnic group and gender in the extent of agreement of Rose questionnaire findings with other measures. We used a test for the difference in two proportions to compare sensitivity and specificity. In recognition of the heterogeneity within the South-Asian groups we present data on Indians, Pakistanis and Bangladeshis separately while focusing on the comparison of Europeans and South Asians. Results Prevalence of measures of ischaemic heart disease The prevalence of diagnosed angina was the same in European and South-Asian men, although lower in Indian than in Pakistani and Bangladeshi men (Table 1). Abnormal ECG findings were commoner in South-Asian than in European men, particularly those showing major Q waves suggestive of probable CHD. The prevalence of RQP angina in South-Asian men was the same as in European men, but lower in Indians and Pakistanis compared to Bangladeshis. The prevalence of RQD angina, in contrast, was lower in South-Asian than in European men. The prevalence was similar in Indians, Pakistanis and Bangladeshis. The prevalence of diagnosed angina was similar in South- Asian and European women, but ECG abnormalities were commoner in South Asians (specifically Indian and Pakistani). The prevalence of RQD angina was also similar in European and South-Asian women (and less common among Pakistanis and Bangladeshis than Indians) but the prevalence of RQP angina was higher in South Asians. While RQP angina was about three times as common as RQD in Europeans (men and women) it was six times as common in South Asians. Agreement between measures Table 2 shows respondents who reported RQD (versus all those who did not fulfil RQD criteria) and those reporting RQP (versus those who did not fulfil RQP criteria) by recall of a doctor s diagnosis of angina. It shows that agreement (measured by kappa) between RQD and RQP and a doctor s diagnosis of angina was similar in South-Asian and European men. The capacity of RQD to identify people reporting a doctor s diagnosis of angina (sensitivity) was lower in South-Asian than in European men, but the capacity to identify those who reported no diagnosis (specificity) was slightly higher. In both ethnic groups sensitivity was much higher and specificity lower using RQP, and inter-ethnic differences were smaller than with RQD. Among women, sensitivity using RQD was also lower in South Asians than Europeans, but for RQP, sensitivity was similar.

3 PERFORMANCE OF ROSE QUESTIONNAIRE IN ETHNIC GROUPS 1011 Table 1 Prevalence of abnormal electrocardiogram (ECG) findings, Rose questionnaire (RQ) angina and reported doctor s diagnosis of angina by gender and ethnic group; percentages directly age standardized to the 1991 England and Wales population; figures are percentage (number). P-values are derived from χ 2 for differences in proportions. Totals for Indian, Pakistani and Bangladeshi respondents are included in the total for South Asian respondents European South Asian P-value a Indian Pakistani Bangladeshi P-value b Males n = 425 n = 325 n = 105 n = 156 n = 64 ECG findings Possible CHD (ST/T changes) 11 (63) 14 (54) (16) 14 (29) 14 (9) 0.99 Probable CHD (Q/QS changes) 2 (12) 6 (26) (4) 7 (16) 7 (6) 0.51 Definite RQ angina (RQD) 6 (29) 3 (13) (4) 3 (6) 3 (3) 0.88 Possible RQ angina (RQP) 18 (84) 18 (72) (14) 17 (35) 33 (23) Doctor s diagnosis of angina 8 (51) 8 (32) (7) 10 (20) 9 (5) 0.41 Females n = 400 n = 359 n = 154 n = 149 n = 56 ECG findings Possible IHD (ST/T changes) 13 (67) 17 (62) (26) 20 (27) 14 (9) 0.48 Probable IHD (Q/QS changes) 1 (6) 5 (18) (14) 3 (4) 0 (0) 0.02 Definite RQ angina (RQD) 5 (24) 4 (14) (10) 3 (4) 0 (0) 0.02 Possible RQ angina (RQP) 14 (66) 23 (85) (32) 26 (39) 21 (14) 0.38 Doctor s diagnosis of angina 5 (27) 6 (27) (10) 6 (10) 13 (7) 0.11 a For difference between Europeans and South Asians. b For difference between Indians, Pakistanis and Bangladeshis. Table 2 Numbers/totals (%) of respondents reporting definite or possible Rose questionnaire (RQ) angina by gender, (self-reported) doctor s diagnosis and ethnic group; kappa; sensitivity and specificity of Rose questionnaire for a doctor s diagnosis of angina and 95% CI for the difference. Totals vary because of missing responses Males Doctor s diagnosis of angina Definite RQ angina (RQD) Possible RQ angina (RQP) Difference Difference European South Asian (95% CI) European South Asian (95% CI) Yes 17/46 (37) 5/24 (21) 41/49 (84) 26/32 (81) No 12/332 (4) 7/276 (3) 39/340 (11) 37/283 (13) Agreement Kappa (95% CI) 0.42 (0.27, 0.55) 0.29 (0.11, 0.50) 0.57 (0.45, 0.66) 0.48 (0.33, 0.59) Test performance Females RQ sensitivity 37% 21% 16 ( 38, 5) 84% 81% 2 ( 19, 15) RQ specificity 96% 97% 1 ( 2, 4) 89% 87% 2 ( 7, 4) Doctor s diagnosis of angina Yes 6/25 (24) 2/24 (8) 19/26 (73) 20/27 (74) No 16/324 (5) 11/303 (4) 45/331 (14) 60/310 (19) Agreement Kappa (95% CI) 0.20 (0.06, 0.38) 0.06 ( 0.03, 0.24) 0.36 (0.21, 0.47) 0.29 (0.13, 0.38) Test performance * P = RQ sensitivity 24% 8% 16 ( 36, 4) 73% 74% 1 ( 23, 25) RQ specificity 95% 96% 1 ( 2, 4) 86% 81% 6 ( 11, 0)* Specificity was lower in South Asians than Europeans. The ethnic differences in sensitivity and specificity were generally smaller with RQP than with the RQD, the most significant being the ethnic difference in specificity in women (P = 0.05). In relation to doctor s diagnosis, differences in kappa between Europeans and South Asians in both sexes combined were of marginal significance (P = for RQD and P = for RQP) but combining both ethnic groups kappa was lower in women (P = for RQD, P = for RQP). Table 3 shows that agreement between RQD and ECG abnormalities was similar in South Asian and European men. Specificity was slightly higher, but sensitivity was lower. For RQP, agreement, sensitivity and specificity were higher in South-Asian men.

4 1012 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Table 3 Numbers/totals (%) of respondents reporting definite or possible Rose questionnaire (RQ) angina by gender, electrocardiogram (ECG) findings and ethnic group (South Asian or European); kappa; sensitivity and specificity of Rose questionnaire for probable coronary heart disease (CHD) (major Q waves). Totals vary because of missing responses Males ECG shows probable CHD Definite RQ angina (RQD) Possible RQ angina (RQP) Difference Difference European South Asian (95% CI) European South Asian (95% CI) Yes 4/11 (36) 4/22 (18) 5/11 (45) 15/26 (58) No 25/378 (7) 9/285 (3) 79/389 (20) 56/298 (19) Agreement Kappa 0.19 (0.04 to 0.37) 0.20 (0.05, 0.41) 0.06 ( 0.08, 0.10) 0.22 (0.07, 0.32) Test performance Females RQ sensitivity 36% 18% 18 ( 51, 14) 45% 58% 12 ( 23, 47) RQ specificity 93% 97% 3 (0.2, 7)* 80% 81% 2 ( 4, 7) ECG shows probable CHD Yes 2/6 (33) 1/16 (6) 3/6 (50) 8/18 (44) No 22/350 (6) 13/326 (4) 63/358 (18) 77/334 (23) Agreement Kappa 0.11 (0.0, 0.31) 0.02 ( 0.04, 0.23) 0.05 ( 0.08, 0.11) 0.08 ( 0.07, 0.13) Test performance RQ sensitivity 33% 6% 27 ( 67, 12) 50% 44% 6 ( 52, 41) RQ specificity 94% 96% 2 ( 1, 6) 83% 77% 5 ( 11, 0.5) * P = Among women, specificity was similar, but agreement and sensitivity using RQD was lower in South Asians compared with Europeans. For RQP, agreement was similar in women of both ethnic groups, while sensitivity and specificity was lower in South Asians. Combining both sexes, agreement using RQP was higher in South Asians (P = 0.051) and combining both ethnic groups it was lower in women (P = 0.16). Components of the Rose questionnaire Table 4 shows that among male respondents with exertional chest pain (RQP) the RQ questions least likely to receive a positive RQ response were those relating to site (among Europeans, Indians and Pakistanis) and related to relief of pain within 10 minutes of rest (particularly among Bangladeshis). Among women with exertional chest pain the RQ questions least likely to receive a positive response related to site (European and Pakistani women) and relief of pain within 10 minutes (Indian and Bangladeshi women). Figure 1 shows that in European, Indian and Pakistani men the site of chest pain made the largest contribution to the difference between definite and possible Rose angina. In Bangladeshi men relief of pain by rest and relief within 10 minutes of rest also contributed to the difference. Site was the most important contributor to the difference between RQD and RQP in women Table 4 Percentage (number) of respondents with RQP (exertional chest pain, possible Rose questionnaire angina) who reported additional components of full Rose questionnaire angina definition (RQD). Percentages directly age standardized to the 1991 England and Wales population European South Asian Indian Pakistani Bangladeshi Males n = 426 n = 325 n = 105 n = 156 n = 64 Reports exertional chest pain (RQP angina) 100% (84) 100% (72) 100% (14) 100% (35) 100% (23) Reports pain in appropriate Rose site 61% (38) 42% (23) 40% (4) 24% (11) 48% (8) Stops or slows down in response to pain 93% (79) 93% (66) 97% (13) 59% (31) 95% (22) Pain relieved by rest 82% (69) 66% (54) 93% (12) 65% (28) 45% (14) Pain relieved within 10 minutes by rest 79% (63) 54% (43) 85% (10) 46% (23) 35% (10) Proportion of RQP with all RQ criteria 36% (29) 17% (13) 37% (4) 10% (6) 4% (3) Females n =399 n = 359 n = 154 n = 149 n = 56 Reports exertional chest pain (RQP angina) 100% (66) 100% (85) 100% (32) 100% (39) 100% (14) Reports pain in appropriate Rose site 58% (33) 44% (31) 65% (17) 31% (10) 37% (4) Stops or slows down in response to pain 96% (61) 85% (70) 86% (27) 80% (30) 81% (13) Pain relieved by rest 73% (55) 59% (51) 61% (20) 67% (24) 31% (7) Pain relieved within 10 minutes by rest 69% (51) 48% (39) 55% (16) 61% (21) 7% (2) Proportion of RQP with all RQ criteria 29% (24) 16% (14) 40% (10) 10% (4) 0% (0)

5 PERFORMANCE OF ROSE QUESTIONNAIRE IN ETHNIC GROUPS 1013 but relief of pain by rest and relief within 10 minutes also contributed substantially among Pakistani and Bangladeshi women. P-values less than 0.05 for comparisons with the European reference group are shown in the Figure. Site of chest pain Figure 2 shows that European men and women mostly reported chest pain in RQD qualifying sites (i.e. sternum or left chest and left arm). In contrast, there were substantial numbers of South- Asian respondents with left-sided chest pain without left-sided arm pain. This pattern was particularly clear in Indians. Rightsided chest pain was commoner in Pakistanis and the pattern of pain was diffuse in Bangladeshis of both sexes, in whom left chest pain was never combined with left arm pain. There was some evidence of ethnic variation in reporting of right-sided chest pain in men (χ 2 = 8.8, P = 0.03), but no evidence of statistically significant variations at other sites. Discussion Study strengths and limitations This is a study of a representative population sample. The response rate was similar across ethnic groups and acceptably high. The survey was developed in close consultation with the community by researchers familiar with the relevant cultures and languages; it used standardized methods, interviewers from the same ethnic group and gender as the respondents, and (for South Asians) face-to-face interviews conducted in the preferred language of interviewees. Europeans completed the questionnaire themselves because we did not have the resources to interview them in the absence of a clear need. Self-completion has been associated with more frequent positive responses, but without differences in the associations with abnormal ECG findings or subsequent coronary mortality. 37 This weakness in our study design, moreover, cannot explain differences between Indians, Pakistanis and Bangladeshis. Figure 1 Prevalence of angina in relation to components of the Rose questionnaire, from Rose questionnaire angina possible (RQP) (1) to Rose questionnaire angina definite (RQD) (5). Percentages are directly age standardized to the 1991 population of England and Wales. P-values in Figure are for comparisons with European reference group Comparative performance of the Rose questionnaire Comparing RQ with ECG findings has evident limitations, e.g. angina, may show no abnormalities on a resting ECG, while diagnosed angina, though based on questions similar to the RQ, is influenced by access to health care and diagnostic practices. Complete agreement between these measures and RQ would be neither expected nor desirable. Our results show that the RQ s known low sensitivity for, and agreement with, ECG and diagnosed angina in Europeans, also applies in South-Asian populations. The RQ performs less well in European women, and the same seems to be true for South-Asian women. In this respect the criticisms of the RQ of Patel et al. are substantiated by our data. The differences between ethnic groups in the performance of the RQ should be interpreted with caution, for reasons discussed above and since the low power of the study means that the performance measures in different ethnic groups had wide and overlapping confidence intervals. Overall RQD performed less consistently than RQP, the latter being considerably more sensitive than RQD, though less specific. The inter-ethnic differences for RQP were much smaller than those seen with the RQD. While RQP, diagnosed angina and ECG abnormalities all suggested that South-Asian men had a similar or higher

6 1014 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Figure 2 Percentage of those with exertional chest pain (Rose questionnaire angina possible, RQP) reporting chest pain at each of ten upper body sites, by ethnic group and sex prevalence of angina than European men, RQD suggested that angina was less prevalent. Similarly, in women all the measures except RQD suggested a higher prevalence of angina in South Asians than Europeans. Although the small numbers in individual ethnic groups do not allow firm conclusions to be drawn, RQD also gave lower estimates relative to other measures in Pakistanis and Bangladeshis than in Indians. While Patel et al. s criticisms are correct, our analysis shows that the RQP performs as well (compared to ECG changes and diagnosed angina) in South Asians as in Europeans. Site of pain in respondents reporting exertional chest pain The anatomical location of pain was more variable in South Asians than in Europeans, and classical elements of the history such as left arm pain were less frequent than in Europeans, while atypical features such as right-sided chest pain were more common in Pakistanis and Bangladeshis. Bangladeshis with chest pain mostly failed to report the other features required for RQD. Few South Asians reported that particular combination of left chest and left arm pain required to fulfil the RQD definition. Our data raise the possibility of systematic differences between ethnic groups in the site of reported chest pain. However, given the small number of cases and the fact that ethnic differences in site of reporting were not statistically significant, firm conclusions must await confirmation from further studies. Explanations for inter-ethnic differences The lack of power of the study to estimate differences precisely means that chance is a possible explanation, though the general consistency of findings in both sexes and using different measures argues against this. Despite the fact that the questionnaire was back-translated and checked, and respondents interviewed in their own language, the translations are not likely to be culturally equivalent. Chest pain may be experienced, understood, interpreted and described differently in different cultures. 40 The variations may reflect problems with the way those questions were translated, with differing interpretations of symptoms, or with differing pathophysiological processes. Educational and socioeconomic factors may influence the way chest pain is reported; more South Asians than Europeans had had only primary education (36% versus 17%) but more had had further education (33% versus 23%). Studies to develop a cross-culturally compatible version of the RQ for South Asian and Chinese populations are underway in Edinburgh University. Conclusion In comparison with ECG and diagnosed angina the RQD performs differently in South Asians than in Europeans, but the RQP performs similarly. In Europeans RQP has been shown to be as discriminating as RQD in prognostic terms; 5,23,29 and though this remains to be demonstrated in South Asians, RQP is presently the preferred measure of angina in comparative studies involving this population, since it is simpler to translate and administer and gives more consistent results. Acknowledgements Funding: Barclay Trust, British Diabetic Association, Newcastle Health Authority, research and development directorate of the Northern Regional Health Authority, Department of Health, and British Heart Foundation. We thank Margaret Miller, Mavis Brown, Amanda McEwan, Heather Armstrong, Afroz Qureshi,

7 PERFORMANCE OF ROSE QUESTIONNAIRE IN ETHNIC GROUPS 1015 Ayesha Motala, Kaushik Ramaiya and Dilip Singh for help at the screening sessions; Peter Stevenson for both screening and analysis of electrocardiograms; Nan Keen for coding electrocardiograms; Louise Walker for help with data preparation and analysis; Julie Yallop for managing the recruitment and screening for the South Asian study; Jane Harland for contributing to the design of the European study and Lisa Hanna and Salma Ayis for helpful comments. KEY MESSAGES Although the Rose questionnaire has been used extensively and in many different cultural settings there is little evidence about its cross-cultural validity. In a direct comparison in European and South Asian ethnic groups, the Rose questionnaire did not have a consistent relationship with other measures: there were differences in sensitivity and specificity, in the site of pain and in the pattern of response to individual Rose questions. While none of these differences were individually conclusive, our findings suggest caution in the use of the Rose questionnaire in ethnic groups other than the one in which it was originally developed. The broader category of possible Rose angina performed more consistently across ethnic groups and pending further work may the most appropriate form to use. Please note, because of the correction of minor coding errors, some of the figures in this paper differ slightly from those previously given in reference 36. References 1 Rose G. The diagnosis of ischaemic heart pain and intermittent claudication in field surveys. Bull World Health Organ 1962;27: Cederlof R, Jonsson E, Lundman T, Stockholm M. On the validity of mailed questionnaires in diagnosing angina pectoris and bronchitis. Arch Environ Health 1966;13: Krueger D, Rogot E, Blackwelder W, Reid D. The predictive value of a postal questionnaire on cardio-respiratory symptoms. J Chron Dis 1970;23: Rose G, Hamilton P, Keen H, Reid D, McCartney P, Jarrett R. Myocardial ischaemia, risk factors and death from coronary heart disease. Lancet 1977;i: Lampe F, Whincup P, Wannamethee S, Ebrahim S, Walker M, Shaper A. Chest pain on questionnaire and prediction of major ischaemic heart disease events in men. Eur Heart J 1998;19: Zeiner-Henricksen T. Cardiovascular disease symptoms in Norway a study of prevalence and a mortality follow-up. J Chron Dis 1971;24: Raman Kutty V, Balakrishnan K, Jayasree A, Thomas J. Prevalence of coronary heart disease in the rural population of Thiruvananthapuram district, Kerala, India. Int J Cardiol 1993;39: Singh R, Sharma J, Rastogi V et al. Prevalence of coronary artery disease and coronary risk factors in rural and urban populations of north India. Eur Heart J 1997;18: Erikssen J, Forfang K, Storstein O. Angina pectoris in presumably healthy middle aged men. Eur J Cardiol 1977;6: Krogh V, Trevisan M, Panico S, Farinaro E, Mancini M, Menotti A. Prevalence and correlates of angina pectoris in the Italian nine communities study. Epidemiology 1991;2: Bainton D, Baker I, Sweetnam P, Yarnell J, Elwood P. Prevalence of ischaemic heart disease: the Caerphilly and Speedwell surveys. Br Heart J 1988;59: Reeder B, Liu L, Horlick L. Sociodemographic variation in the prevalence of cardiovascular disease in Saskatchewan: results from the Saskatchewan Heart Health Survey. Can J Cardiol 1996;12: Netterstrom B, Laursen P. Incidence and prevalence of ischaemic heart disease among urban bus drivers in Copenhagen. Scand J Soc Med 1981;9: Rossouw J, Weich H, Steyn K, Kotze J, Kotze T. The prevalence of ischaemic heart disease in three rural South African communities. J Chronic Dis 1984;37: Reid D, Brett G, Hamilton P, Jarrett R, Keen H, Rose G. Cardiorespiratory disease and diabetes among middle aged civil servants. A study of screening and intervention. Lancet 1974;i: Marmot MG, Syme SL, Kagan A, Kato H, Cohen JB, Belsky J. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: prevalence of coronary and hypertensive heart disease and associated risk factors. Am J Epidemiol 1975;102: Rose GA, Ahmeteli M, Checcacci L et al. Ischaemic heart disease in middle-aged men. Prevalence comparisons in Europe. Bull World Health Organ 1968;38: Heyden S, Bartel A, Tabesh E et al. Angina pectoris and the Rose questionnaire. Arch Intern Med 1971;128: Friedman L, Byington R, the Beta Blocker Heart Attack Trial Research Group. Assessment of angina pectoris after myocardial infarction: comparison of Rose questionnaire with physician judgment in the beta-blocker heart attack trial. Am J Epidemiol 1985;121: Blackwelder W, Kagan A, Gordon T, Rhoads G. Comparison of methods for diagnosing angina pectoris: the Honolulu heart study. Int J Epidemiol 1981;10: Lundman T, Liljefors I, Cederlof R, Friberg L. The validity of the questionnaire diagnosis. Angina pectoris. Arch Environ Health 1971;22: Wilcosky T, Harris R, Weissfeld L. The prevalence and correlates of Rose questionnaire angina among women and men in the lipid research clinics program prevalence study population. Am J Epidemiol 1987; 125: Cook D, Shaper A, MacFarlane P. Using the WHO (Rose) angina questionnaire in cardiovascular epidemiology. Int J Epidemiol 1989;18: Smith K, McGraw S, Crawford S, McKinlay J. Do blacks and whites differ in reporting Rose questionnaire angina? Results of the Boston Health Care Project. Ethnicity and Disease 1993;3: Bass E, Follansbee W, Orchard T. Comparison of a supplemented Rose questionnaire to exercise thallium testing in men and women. J Clin Epidemiol 1989;42:

8 1016 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY 26 Garber C, Carleton R, Heller G. Comparison of Rose questionnaire angina to exercise thallium scintigraphy: different findings in males and females. J Clin Epidemiol 1992;45: Harris R, Weissfeld L. Gender differences in the reliability of reporting symptoms of angina pectoris. J Clin Epidemiol 1991;44: Nicholson A, White I, Macfarlane P, Brunner E, Marmot M. Rose questionnaire angina in younger men and women: gender differences in the relationship to cardiovascular risk factors and other reported symptoms. J Clin Epidemiol 1999;52: LaCroix A, Guralnik J, Curb D, Wallace R, Ostfeld A, Hennekens C. Chest pain and coronary heart disease mortality among older men and women in three communities. Circulation 1990;81: Sorlie P, Cooper L, Schreiner P, Rosamond W, Szklo M. Repeatability and validity of the Rose questionnaire for angina pectoris in the atherosclerosis risk in communities study. J Clin Epidemiol 1996;49: Patel D, Winterbotham M, Sutherland S, Britt R, Keil J, Sutton G. Comparison of methods to assess coronary heart disease prevalence in South Asians. Natl Med J India 1997;10(5): Nazroo J. The Health of Britain s Ethnic Minorities. London: Policy Studies Institute, Wild S, McKeigue P. Cross sectional analysis of mortality by country of birth in England and Wales, Br Med J 1997;314: McKeigue PM, Ferrie JE, Pierpoint T, Marmot MG. Association of early-onset conronary heart disease in South Asian men with glucose intolerance and hyperinsulinemia. Circulation 1993;87: Williams R, Bhopal R, Hunt K. The health of a Punjabi ethnic minority in Glasgow: a comparison with the general population. J Epidemiol Community Health 1993;47: Bhopal R, Unwin N, White M et al. Heterogeneity of coronary heart disease risk factors in Indian, Pakistani, Bangladeshi and European origin populations: cross sectional study. Br Med J 1999;319: Rose G, McCartney P, Reid D. Self administration of a questionnaire on chest pain and intermittent claudication. Br J Prev Soc Med 1977; 31: Prineas R, Crow R, Blackburn H. The Minnesota Code Manual of Electrocardiographic Findings. London: John Wright, Donner A, Klar N. The statistical analysis of kappa statistics in multiple samples. J Clin Epidemiol 1996;49: Chaturvedi N, Rai H, Ben-Shlomo Y. Lay diagnosis and health-careseeking behaviour for chest pain in south Asians and Europeans. Lancet 1997;350: Appendix Criteria for Rose angina questionnaire 1. Have you ever had any pain or discomfort in your chest? 2. Do you get it when you walk uphill or hurry? 3. Do you get it when you walk at an ordinary pace on the level? 4. What do you do if you get it while you are walking? 5. If you are standing still, what happens to it? 6. How soon? 7. Will you show me where it is? Angina is defined as being present in subjects who answer as follows: Q1: yes; Q2 or Q3: yes; Q4: stop or slow down; Q5: relieved; Q6: 10 minutes or less; Q7: either (a) sternum (upper or middle or lower) or (b) left anterior chest and left arm.