Inequalities in risk factors and cardiovascular mortality among Australia s immigrants

Transcription

1 MORTALITY IN IMMIGRANTS Laven GT, Brown KC. Nutritional status of men attending a soup kitchen: a pilot study. Am J Public Health 1985; 75: National Health and Medical Research Council. Summary repoll on the thiamine status ojthe Ausfralian people. Report of the eighty-third session. Canberra: Australian Government Publishing Service, 1977: Appendix XIV Nobile S, Woodhill JM. Testing for vitamin deficiencies in a developed country-australia. Food Techno1 Ausf 1976; 28: Price J. The Wernicke-Korsakoff syndrome in Queensland, Australia: antecedents and prevention. Alcohol Alcohol 1985; 20: Inequalities in risk factors and cardiovascular mortality among Australia s immigrants Stan A. Bennett National Centre for Epidemiology and Population Health, Australian National University Abstraa: Inequalities in biomedical and life-style risk factors for cardiovascular disease were examined for immigrants to Australia and people born in Australia, using data collected in the 1980,1983 and 1989 risk-factor prevalence surveys. After adjusting for age and study design, significant differences were identified between immigrant groups and the Australian-born reference group, particularly for systolic blood pressure, overall obesity and behavioural risk factors. There were few substantial differences in blood lipid concentrations and little evidence to suggest that total plasma cholesterol has played a major role in lower cardiovascular mortality among inlmigrants. Overall, the results suggested that profiles of risk factors commonly accepted as determinants of cardiovascular disease are an insufficient explanation of the lower standardised mortality ratios from cardiovascular disease which characterise immigrants in Australia. Systolic blood pressure best explained variation in cardiovascular mortality among male immigrants, and smoking prevalence among female immigrants. The acculturation process affected immigrant groups differently. Generally, systolic blood pressure increased with period in Australia. Body mass index increased among Asian immigrants, as did participation in physical activity during leisure time. (AuctJ Public Health 1993; 17:251-61) T e Australian population embraces a wide range of origins and cultures. Over 20 per cent of residents have been born overseas, representing over 100 different countries of birth. Inspection of recent settler arrival statistics shows that the European sources of postwar immigration such as, Yugoslavia and have been replaced by Asian countries, notably Hong Kong and Vietnam2 British immigrants now represent 20 per cent of recent arrivals compared with over 50 per cent in past years. Many epidemiological studies of the health of immigrants in Australia have focused on mortality data, often for specific causes of death or for particular immigrant groups. One exception is the analysis of mortality data by Young which covered relatively wide ranges of causes and immigrant gr~ups.~ The analysis showed that immigrants to Australia have generally experienced lower mortality rates (for a wide range of causes, including cardiovascular disease) than people born in Australia. A more recent analysis of cardiovascular mortality data for showed lower age-standartlised mortality ratios for immigrants from Vietnam,,, Malaysia, China, Philippines (women) and Central and South America (Table 1). High agestandardised mortality ratios were found for immigrants from Malta (women only), Poland and the Pacific I~lands.~ ~ ~~ Correspondence to Mr Stan A. Bennett. National Centre for Eoidemiofogy and Population Health, ANU, GPO Box 4, Canbe&, ACT Fax (06) The extent to which mortality and morbidity differentials have been influenced by differences in cardiovascular risk-factor levels is not clear, as there have been few comprehensive studies of risk-factor differentials among Australia s immigrant group^.^.^ Differentials are to be expected. Recent results from the World Health Organization s cardiovascular disease monitoring project (MONICA) demonstrated marked variation in cardiovascular risk-factor levels among population centres in 26 countries, many of which are significant sources of Australia s immigrants. The present analysis systematically compares the levels of biomedical and behavioural risk factors among a range of immigrant groups in Australia, using native-born Australians as a reference group. Variation in risk-factor levels with period of residence is examined. Also, risk-factor profiles for each immigrant group are compared with their cardiovascular mortality experience. The data have been provided by the National Heart Foundation s Risk Factor Prevalence Study which presently comprises three cross-sectional probability surveys conducted in 1980,1983 and Over Australians living in capital cities participated in these surveys, providing a rich source of relatively recent cardiovascular risk-factor data on over 6000 immigrants. Methods suruqr mthods Although there were minor differences, the three surveys used basically the same survey methods.8-10 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. I7 NO

2 BENNETT Table 1 : Cardiovascular disease age-standardised mortality ratios by birthplace group, 1987 to 1989 Birthplace group Men Women United Kingdom and Ireland Yugoslavia Malta Germany Netherlands Poland Lebanon Egypt Vietnam Malaysia Philippines India China Hong Kong and Macao South Africa Canada United States Central and Southern America New Zealand Other Oceania Australia 92' 62' 67' 87' ' 80' 90 30' ' 33' ' ' 103' 91 55' 55' 79' 24' 90 84' 17' ' 59' 56' ' ' 104' 'Rotio significontly different from 100 (P < 0.05). Source Young. In: lmmigronts in Ausho/io: o health profile.' Taking the 1989 survey as an example, a probability sample of adults aged 20 to 69 was selected for each state and territory capital city using federal electoral rolls (enrolment is compulsory for all Australian citizens older than 18 years). Data collection took place from June to December. Invitations were posted to prospective participants giving them a specific appointment time to attend the local survey centre for a free check of heart disease risk factors. A card explaining the purpose of the survey in Chinese, Croatian, Finnish, French, Greek, Italian, Macedonian, Portuguese, Serbian, Spanish and Vietnamese was included to assist those who could not read English. Participation was encouraged by the use of day-before reminder calls, reminder letters, home visits, opening outside normal working hours and establishing temporary clinics in areas of low response. Each survey achieved a response rate close to 75 per cent. The survey protocol was approved by the Ethics Committee of the Australian Institute of Health and Welfare. Measurement and definition of risk factors The following biomedical and behavioural cardiovascular risk factors were used in this analysis. Details of measurement protocols are given in the individual survey reports.a1o systolic blood pressure (SBP) diastolic blood pressure (DBP) total plasma cholesterol (TC) high density lipoprotein cholesterol (HDL) triglyceride (TC) low density lipoprotein cholesterol (LDL) HDL to TC ratio (HDLPC) body mass index (BMI) smoking status alcohol consumption exercise during leisure time Biomedical risk factors: Physical examinations were conducted by nursing sisters. Two consecutive SBP and DBP readings were taken, five minutes apart, using normal mercury sphygmomanometers. The average of the two readings was used as the final observation. AN subjects were included in the analysis, whether or not on treatment for hypertension. Morning blood specimens were analysed for TC, HDL and TG, at Flinders Medical Centre, Adelaide (1980, 1983) and the Institute of Medical and Veterinary Science, Adelaide (1 989). The laboratories satisfied the criteria for precision and accuracy as specified for standardisation by the Centers for Disease Control, Atlanta, Georgia. The ratio of HDL to TC was expressed as a percentage. LDL was calculated as TC-HDL-TG/5 if TG was lower than 4.5 mmol/l.' Fasting status was determined by inquiry. Body mass index (weight/height2 in kg/m2) was used as the indicator of overall obesity. One kilogram was deducted from the measured weight as an allowance for weight of clothing. Behavioural risk factors: Data on behavioural risk factors were obtained from the questionnaire. For smoking behaviour, the proportion of current smokers (cigarettes, cigars or pipe) was used as the risk-factor indicator for each immigrant group. Respondents were asked how often they usually drank alcohol and, on a day when they drank, how many drinks they usually had. These responses were used to calculate usual drinks per week. Alcohol intake was classified as never or occasional (less than one drink per week), light (1 to 27 drinks per week for men, 1 to 13 drinks per week for women) or moderate to heavy (28 drinks or more per week for men, 14 drinks or more per week for women). These definitions are consistent with National Health and Medical Research Council recommendations which regard fewer than 28 drinks per week (men) and fewer than 14 drinks per week (women) as responsible drinking behaviour. l2 Recent evidence suggests that modest alcohol consumption is linked with lower cardiovascular risk Accordingly, the prevalence of light alcohol intake, as defined above, has been used as a beneficial risk-factor indicator in this analysis. In the 1983 and 1989 surveys, respondents were asked about exercise taken for recreation, sport or health or fitness purposes in the previous two weeks. This included vigorous exercise (defined as causing breathlessness, puffing and panting), less vigorous exercise and walking. The present analysis used the prevalence of individuals who reported no leisuretime exercise of any kind as an indicator of cardiovascular risk. No account was taken of exercise at work. Definition of immigrant groups and period of resideme Each survey included the question 'Where were you born?' and respondents were asked to write state or 252 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO. 3

3 MORTALITY IN IMMIGRANTS territory if born in Australia, or country if born overseas. Responses to the 1989 survey were coded to the Australian Standard Classification of Countries for Social Statistics, released in 1990.*O The majority of responses to the 1980 and 1983 surveys could be successfully recoded. Aggregation of birthpla.ces into immigrant groups was necessary to provide reasonable sample sizes for the analysis (Table 2). Countries which had historically been a major source of migrants to Australia were separately identified. Immigrants from Scotland and Ireland, who have been identified as having higher overall mortality than Australian-born, were analysed separately from the rest of the United Kingd~m.~ It was possible to separate Asia, which covers such a diverse group of nations, into two broad groups. In addition to reporting country of birth, respondents born overseas were also asked how many years they had lived in Australia. To allow for the different waves of immigrants to Australia, the pattern of riskfactor levels by period of residence was estimated separately for four broad regions. The classification into regions and period of residence was the best possible given the relatively few recent immigrants from Europe in the sample, and few longer term Asian residents. Statistical analysis The purpose of the analysis was to detect differences in risk-factor levels between each immigrant group and Australian-born men and women. fie null hypotheses were that no such differences existed. The analysis was conducted separately for men and for women. Age, at 30 June of the survey year, was calculated from date of birth provided by the respondent. The analysis was conducted on the age range 25 to 64 years and used the SAS software package, version 6. For continuous variables (for example, blood pressure), inequalities between immigrant groups Table 2: Sample counts for eoch immigrant group Immigrant group Men Women England and Wales Scotlond and Ireland ltoly o b Eastern Europec Middle East and North Africad Southeast Asiae New Zeoland Q Insufficient informationh Australian-born Total Notes: la1 Includes Yugoslavia, Cyprus and Malta lb) Includes Germany, Netherlands and Austria Icl Northern Europe, Eastern Europe, the USSR and the Boltic States; includes Poland, Hungary and Finland Id) Includes Lebanon, Egypt ond Turkey lel Includes Vietnam, Indonesia, Malaysia and the Philippines If1 Includes Indio, Chino, Taiwan ond Hang Kong lgl Includes south Africa, Canada, United States and,papua-new Guinea lhl United Kingdom, Great Britain, etc. were examined by analysis of covariance with age as a continuous covariate. Survey year, location and their interaction were also included in the model to allow for the study design. The analysis shows, for each immigrant group, the crude (unadjusted) mean and the adjusted differential (Dl), which is the immigrant mean minus the Australian-born mean adjusted for age and study design factors. Statistically significant differences (2-sided tests) have been indicated. Body mass index, smoking status, alcohol consumption and oral contraceptive use (women) were treated as potential explanatory variables, to explore possible reasons for differences between immigrant groups. Multiple logistic regression was used to analyse categorical variables (for example, current smoking). The model included all survey design factors and treated age as a continuous covariate. The parameter estimates from the model were used to derive prevalence odds ratios for each immigrant group, using Australian-born as the reference, together with 95 per cent confidence limits. Certain data were excluded from each analysis as appropriate. Data for pregnant women were excluded from all analyses which included body mass index. Anthropometric data collected in Adelaide in 1980 were excluded from the analysis because measurement procedures in that survey centre deviated from the study protocol. Data from respondents who had not been fasting were excluded from all blood lipid analyses. Three extreme values (over 40 mmol/l) were excluded from the TG analysis. Results Results for DBP and TG are not given in tabular form but statistically significant differentials are mentioned in the text. Blood pessure After adjusting for age and study design factors, mean SBP was lower in male immigrants from the Middle East (-5.4 mmhg), other Asia (-5.3 mmhg), Southeast Asia (-4.4 mmhg), (-3.6 mmhg), (-1.9 mmhg) and England and Wales (-1.9 mmhg) than their Australian-born counterparts (Table 3). Only men from Eastern Europe (+2.4 mmhg) had statistically higher SBP than Australianborn men. Differentials for DBP in men were largely non-existent. Women from the Middle East (-4.3 mmhg), Southeast Asia (-3.4 mmhg) and England and Wales (-2.3 mmhg) had lower SBP than Australian-born women. As for men, the differentials for DBP were noticeably less pronounced than those for SBP, and only those for women from the Middle East (-2.1 mmhg) and England and Wales (-1.5 mmhg) reached statistical significance. Although there was some variation between immigrant groups in the proportion on treatment for blood pressure, the differentials which have been identified were also apparent when the analysis was confined to those not on treatment. Overall obesity The results for BMI suggest marked differences in overall obesity among immigrant groups. Relative to native-born Australians, men from the Asian regions AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO

4 BENNET7 Table 3: Mean and adjusted mean systolic blood pressures (mmh 1 and body mass index (kg/m2) for immigrant groups compared with Austrafan-born Systolic blood pressure Body mass index" Immigrant group Unadiusted mean DI SEC D2d Unadiusted mean Dlb SEC D2d Men England ond Wales Scotland and Ireland Eastern Europe The Middle East Southeost Asia New Zeoland Australian-born Women England and Wales Scotland and Ireland Eastern Europe The Middle East Southeast Asia New Zeoland Australian-born t t -1.9' ' -5.4t -4.4t -5.3t t t -3.4t t -2.9t t -4.5t t t - 1. q 1.2t 0.8t 0.8t 0.9t -1.8t -1.6t - - -t - 2.7t 2.3t 2.3t 2.5t -1.3t -0.9' - -t ' 1.6t 1.3t 0.9t 0.8t t -1.7t -1.5t - - -t - 2.2t 2.lt 2.1 t 2.2t -1.7t -I.lt - Notes:(a) Excludes pregnant women. (bl Immigrant group mean minus the Australian-born mean, adjusted for age and study design factors. (c) Standard error of D1. (d) DI further adiusted for covariates. 'P<O.O5, tp<0.01. and from the United Kingdom had statistically significantly lower average levels of BMI (Table 3). Results were similar but less pronounced for women. Men and (particularly) women from Southern European regions and the Middle East each had higher adjusted BMI on average than their native-born counterparts, as did men from Eastern Europe. Blood lipidr Only men and women from and men from Western Europe had adjusted mean levels of TC which were significantly different from their Australianborn counterparts (-9, -5 and + 7 mmol/l respectively, Table 4). Statistically significantly lower levels of HDL were observed among men and women from and the Middle East, and among women from, other Southern Europe, and other Asia. Mean TG level among women from the Middle East was highly statistically significantly different ( -I- 6 mmol/l, P < 0.01) from their Australian-born counterparts. Other significant differentials (P < 0.05) were men and women from (-4 and mmol/l), and women from other Southern Europe ( + 0 mmol/l) and Southeast Asia ( 4-1 mmol/l). Men and women from had lower levels of LDL than their Australian-born counterparts (Table 5). Only men from had a higher adjusted mean level of LDL (4-0 mmol/l). Only men and women from New Zealand had more beneficial HDL/TC ratios than their Australian-born counterparts (+ 1.3 per cent and per cent respectively). Men from and the Middle East, and women from, other Southern Europe, the Middle East and other Asia all had statistically significantly lower ratios than Australian-born men and women. Smoking status The prevalence of current smoking of any form of tobacco was significantly higher among men and women from Scotland and Ireland (adjusted prevalence odds ratios 1.65 and 1.63 respectively, Table 6). Smoking prevalence odds ratios were also significantly higher among men from Southern Europe ( 1.65, 1.60, and other Southern Europe 1.80) and the Middle East (2.03). Among women, immigrants from (1), (3), Southeast Asia (3) and other Asia (6) had lower smoking-prevalence odds ratios than Australianborn women. Among women from the smoking-prevalence odds ratio was significantly higher (1.39) than for their Australian-born counterparts. Alcohol cansumption Light alcohol consumption (1 to 27 drinks per week), potentially a beneficial risk factor, was reported more commonly among male immigrants from England and Wales,, and Eastern Europe (adjusted prevalence odds ratios of 1.58, 1.60, 1.96 and 1.63 respectively). Prevalence odds ratios were significantly lower among male immigrants from the Middle East (1) and Southeast Asia (4). 254 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO. 3

5 MORTALITY IN IMMIGRANTS Table 4: Total plasma cholesterol and high-density lipoprotein cholesterol (mmolll) for immigrant groups compared with Australian-born Totol cholesterol" HDL cholesterol" Immigrant group Unadiusted mean Dlb SEC D2d Unadiusted mean Dlb SEC D2d Men Englond and Woles Scotland ond Ireland Eastern Europe The Middle Eost Southeost Asio New Zeolond Austrolion-born Women Englond and Woles Scotland and Ireland Eostern Europe The Middle Eost Southeost Asio New Zeolond Austrolion-born t t ' ' t t - -3t t - ot -0.09t t ' ' ' -0. lo' ' - Notes: la1 Fasting. Ibl Immigrant group meon minus the Australian-born mean, adiusted for age and study design factors. Icl Standard error of D1. Id1 DI further adiusted far cavariates. 'P < 0.05, tp < Table 5: Low-density lipoprotein (LDL) cholesterol, and high-densi lipoprotein (HDL) to total cholesterol (TC) ratio for immigrant groups compared wit x Australian-born LDL cholesterol (mmolll)" HDL/TCb(%)" Immigrant group Unodiusted mean D1 SEC D2d Unadiusted meon D1 SEC DZd Men Englond ond Woles Scotland ond Ireland Eostern Europe The Middle Eost Southeost Asio Other Asio New Zeolond Austrolion-born Women Englond and Woles Scotland and lrelond ltoly Eostern Europe The Middle Eost Southeost Asio Other Asio New Zeoland Australion-born ' ' t t ' I -t -2.6t ' t at t 0. I -1.8' 1.6' I 0.9 t -1.1' 1.O' ' t Nates: la1 Fasting. lb) Immigrant group mean minus the Australian-barn mean, adiusted far age and study design factors. Icl Standard error of DI. Id) DI further adjusted for covariates. *P< 0.05, tp<o.oi. AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO

6 BENNETT Among women, the prevalence odds ratios for light drinking (1 to 13 drinks per week) were higher among immigrants from England and Wales (1.34), and lower among immigrants from Southern European countries (5, 2, 0), the Middle East (9) and Asian countries (5 and 8). Exercise during leisure time Physical inactivity during leisure time was more prevalent among men and women from Southern European countries, Eastern European countries and the Middle East than their Australian-born counterparts (Table 6). Exercise was also relatively unpopular among men and women from Southeast Asia and women from other Asia; however, men from the latter Asian countries exercised as frequently as Australian-born men. Men from Scotland and Ireland also reported a higher prevalence odds ratio for physical inactivity. Adjustment fw covarides The extent to which BMI, smoking status, alcohol consumption, and use of oral contraception (women) modified the differentials for blood pressure and blood lipids can be seen by comparing D1 and D2 (Tables 3 to 5). Adjustment for the covariates sometimes amplified and sometimes explained the differentials. Higher BMI was associated with greater blood pressure, as was heavy alcohol consumption. The association between smoking status and blood pressure was stronger in women than men, with current smokers having lower SBP and DBP. There were significant positive associations of body mass index with TC, TG and LDL and significant negative associ- ations with HDL and HDL/TC. Other significant relationships were between alcohol consumption and HDL and HDL/TC (heavy intake associated with higher levels) and between smoking status and HDL and HDL/TC (current smokers had lower levels). For women, taking oral contraceptives was associated with statistically significantly higher TC, lower HDL, higher TG, higher LDL and lower HDL/TC. Period of residence The difference in SBP between shorter and longer stay male immigrants from the United Kingdom and Ireland ( mmhg) was statistically significant (P < 0.01) as was the increase in SBP for other Europeans (4-3.3 mmhg) (Table 7). Among women, the more recently arrived immigrants from Britain and from Asia had significantly lower SBP than Australian-born women, but not so the longer stay immigrants from these regions. None of the comparisons by period of residence showed statistically significant trends for women, except the trend for all immigrants, from -2.8 mmhg to -0.9 mmhg. None of the comparisons for association between period of residence and DBP were statistically significant. Several differentials for BMI were statistically significant; however, length of residence was associated with statistically significant increases only for men and women from Asia ( -I- 0.9 kg/m2, P < 0.01; kg/m2, P < 0.05, respectively). Differentials for TC levels were small and varied little by period of residence in Australia. Similarly, there was no evidence of change in HDL, TG or LDL level with period of residence. The trend in HDL/TC for European Table 6: Behavioural risk factors for immigrant groups compared with Australian-born Current smoking Light alcohol intakea Physical inactivityb Immigrant group Odds ratioc 95% Cld Odds ratio 95% CI Odds ratio 95% CI Men England and Wales Scotland and Ireland Eostern Europe The Middle East Southeast Asia New Zeoland Australian-born Women England and Wales Scotland and Ireland Eastern Europe The Middle East Southeast Asia New Zeoland Australian-born q 1.65t 1.60t 1.80t t o t 1 t 3t f o 0.88 to to to to to to to 6 3 to to to to to to to to 6 0 to 2 0 to to to 5 6 to to 5 8 to to to t t t 1.63t 1 t o 1.34t t 2t 0t t 5t 8t 3 8 o 1.32 to to to to to to to to 4 4 to to to to to to to 5 0 to 6 8 to to to to 4 8 to 4 7 to to to t 3.35t 3.92t 3.67t t 2.77t o t 4.78t t 3.71 t 3.00t 7 9 o 0.84 to to to to to to to to to to to to to to to to to to to to to to to to 2.30 Notes: la1 1 to 27 drinks a weak for men, 1 to 13 drinks a week for women. Ib) No leisure-time activity of any kind during the past 2 weeks. Icl Prevalence odds ratio adjusted for age and study design factors. Id1 95% confidence limits for the prevalence odds ratio. *P< 0.05, tp< AUSTRALIAN JOURNAL OF PU8LlC HEALTH 1993 VOL. 17 NO. 3

7 ~~ ~~~ ~ MORTALITY IN IMMIGRANTS males (from per cent to - per cent) was statistically significant (P < 0.01). None of the tests for association between smoking prevalence and period of residence were statistically significant. For light alcohol intake, longer period of residence was associated with a statistically significant fall in prevalence for men from the United Kingdom and Ireland and an increase in prevalence for men and women from Asia and women from Europe. Physical inactivity during leisure time was more common among immigrants from Europe and Asia than their Australian-born counterparts. The decreases in prevalence odds ratios between im&- grants with shorter and longer residential periods were statistically significant for men and women from Asia and women from Europe. Standardising across region of birth showed that exercise was more popular among longer stay immigrants than shorter stay immigrants. Discussion Although it is generally accepted that cardiovascular disease is related substantially to the risk factors considered in this analysis, the relationship is exceedingly complex and is made more so by the additional factors associated with immigration. The inequalities identified are likely to be a function of the risk-factor profile of the source populations, the selection effect in the migration process, environmental influences after immigration and the degree to which habits and customs have been maintained. Nevertheless, it is instructive to explore the extent to which differentials in biomedical and life-style risk factors among immigrant groups correlate with cardiovascular mortality. Th British British immigrants are not a homogeneous group. Immigrants from Scotland and Ireland reported higher prevalences of smoking and physical inactivity (men) and those from England and Wales had lower blood pressure levels and higher prevalences of light alcohol intake compared with the Australian-born. These results are consistent with previous immigrant mortality analyses which have shown higher total mortality and cardiovascular mortality among the Scots and Irish than the English and Wel~h.~.~' Overall, the British also had lower body mass index than Australian-born men and women and their riskfactor profile was consistent with lower agestandardised mortality from cardiovascular disease (Table 1). Southern Europeans Greek and Italian male immigrants had lower systolic blood pressure than Australian-born men despite sig- Table 7: Risk-factor differentials according to period of residence in Australia" Systolic blood Body mass Total Light alcohol Physical n pressure index cholesterol HDL LDL HDLlTC Smoking intake inactivity mmha kalm' mmolll mmolll mmollt % ORb ORb ORb Men United Kingdom 8 Ireland < 15 years years 867 Europe' <I 5 years years Asiad I 5 years years 176 Othere I I5 years years 1 I7 All immigrants I1 5 years years Women United Kingdom 8 Ireland <I 5 years years 764 E;ropec I1 5 years years 949 Asiad II 5 years years 156 Othere <I 5 years years 111 All immigrants < 15 years years t ' -3.7t t -4.8t t -2.3t -2.9t t I.o -2.8t t t t -1.3t ' q ' ' t t ' ' t -0.06t t lOt -0.06t Ot ' t ' - 1.5' -' ' -1.3t - -' -t -1.3' -' - -' t t ' t 1 t t 1.31 t 2.55t ISlt 6t t I.30t 0' 1.33t 4t 5t 1 t 9t t 0.84' t 2.43t 3.07t t 1.24' ' 3.96t 2.14t 4.74t 1.74t t 1.27t Notes: lo1 The difference between eoch immigrant group estimate and the Austrolion-born group estimate, adiusted for age and study design factors. Differentiols for 'a11 immigronts' ore adiusted for immigrant group composition. ib) Prevolence odds ratios. lcl Excludes UK and Ireland. (dl Includes the Middle East and North Africa. lel Includes New Zealand. AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO

8 BENNETT nificantly higher body mass index. There was no evidence of lower systolic pressure among women from Southern European countries. Previous Australian studies have found lower levels among Italian immigrant~.~~9~~ Southern European immigrants in generalp4 and Italian immigrants in particularp3 are known to have higher mean body mass index than other immigrants and Australian-born men and women. This analysis demonstrated the characteristic independently for immigrants from, and other Southern Europe. The significantly lower levels of total cholesterol among Italian immigrants compared with Australian-born men and women is a new finding.p3 Women from each Mediterranean region had lower high-density lipoprotein levels than Australian-born women, consistent with their alcohol-drinking behaviour but not their smoking habits. Each Mediterranean region reported a significantly higher prevalence of leisure-time inactivity than Australian-born men and women. The marked sex differential in smoking prevalence among Italian immigrants has been noted previou~ly.~~ Women from each region reported significantly lower prevalences of light alcohol intake than Australianborn women. The men tended to have higher rates than Australian-born men although statistical significance was reached only for men from. Cardiovascular mortality rates have been persistently lower among immigrants from Southern European countries compared with Australian men and women and remain so (Table 1). For immigrants from, the risk-factor profile of men suggests that factors other than the traditional risk factors considered in this analysis are important. Greek men have advantageously lower systolic blood pressure, but lower high-density lipoprotein, higher body mass index, higher smoking prevalence and higher leisuretime inactivity. Greek women have a lower prevalence of smoking, all other significant risk-factor differentials being comparatively disadvantageous. Results for immigrants from correlate better with traditional wisdom. The men have advantageous differentials for systolic pressure, total cholesterol, low-density lipoprotein and light alcohol intake. The women have advantageous total cholesterol, lowdensity lipoprotein levels and lower smoking prevalence. These results are not entirely consistent with the hypothesis that the Mediterranean diet offers a protective mechanism against cardiovascular disease. The low mortality rates for those from the Mediterranean region coincides with a diet which is relatively high in fresh fruit and vegetables, and high in olive oil.the olive oil used in the diets of people in and is around 80 per cent oleic acidp5 and substitution of dietary saturated fatty acids with oleic acid results in a fall in low-density lipoprotein levels.26 Only amongst Italian immigrants was there evidence of lower low-density lipoprotein levels among Southern European immigrants in Australia. Overall, the blood lipid profile of immigrants from Southern European countries was not markedly different from that of men and women born in Australia. Wine also plays an important part in the main meals of Italians and Greeks and a recent New Zealand study found that light and moderate alcohol consumption reduced the risk of coronary heart disease.14 In this analysis, men from Southern Europe generally had a higher prevalence of light alcohol intake but the converse was true for women. Western, Northem and Eastern Europeans Age-standardised mortality ratios for Germany and the Netherlands, the most common sources of immigrants from, were marginally lower in than those for Australian-born men and women but only the difference for women from the Netherlands was statistically significant. The only significant beneficial differential in risk-factor levels for male immigrants from was a higher prevalence of light alcohol intake. Detrimental risk-factor differentials were higher total cholesterol and low-density lipoprotein, and lower ratio of high-density lipoprotein to total cholesterol. The only significant differential for women was a higher smoking prevalence. Eastern European male immigrants had a significantly higher systolic pressure and body mass index, and were less likely to exercise during leisure time than Australian-born men. Their prevalence of light alcohol intake was higher. The only significant differential for women from Eastern Europe was related to a higher prevalence of physical inactivity. Immigrants from Poland, the most common source of immigrants from these regions, were one of the few groups to have had a significantly higher agestandardised mortality compared with Australianborn men and women (Table 1). Immigrants from t h Middle East and Nmth Africa The men had a detrimental risk-factor profile comprising a relatively low high-density lipoprotein, low high-density lipoprotein to total cholesterol ratio, high body mass index, high smoking prevalence, low prevalence of light alcohol intake and high prevalence of leisure-time physical inactivity compared with Australian-born men. Despite this, their systolic pressures were significantly lower. The cardiovascular age-adjusted mortality for Lebanon and Egypt was lower than for Australian-born men (Table 1). The women were characterised by disadvantageous levels of high-density lipoprotein, triglyceride, highdensity lipoprotein to cholesterol ratio, body mass index, light alcohol intake and physical inactivity. As for the men, their systolic and diastolic blood pressures were lower. The mortality for women from Lebanon and Egypt was higher than for Australian women although the differences were not statistically significant. Asians Although basically similar, the risk-factor profiles of immigrants from Southeast Asia and immigrants from other Asia differed in important aspects. For men, both immigrant groups had lower systolic blood pressure and lower body mass index, but only immigrants from Southeast Asia had a lower prevalence of light alcohol intake (due to a higher prevalence of non-drinkers). Men from Southeast Asia also had a significantly higher rate of leisure-time inactivity. Asian women tended to have lower body mass index, lower smoking prevalence, lower prevalence of light 258 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 17 NO. 3

9 MORTALITY IN IMMIGRANTS alcohol intake and lower rates of physical inactivity. Women from Southeast Asia had lower systolic pressure and higher triglyceride; women from other Asia lower high-density lipoprotein and high-density lipoprotein to total cholesterol ratio. Standardised mortality ratios for immigrants from Southeast Asia (Vietnam, Malaysia, Philippines) and from other Asia (India, China, Hong Kong and Macao) were all lower than those for Australian-born men and women (Table 1). With the exception of a relatively higher prevalence of leisure-time physical inactivity, this is consistent with their generally advantageous risk-factor profiles. The data suggest that blood lipids have played little part in their favourable cardiovascular mortality differential as the lipid profiles of Asian immigrants were generally similar to those of Australian-born men and women. New Zealanders Immigrants from New Zealand had a very similar risk-factor profile to that of Australian-born men and women, the only statistically significant differences being higher high-density lipoprotein to cholesterol ratios for men and for women. The data suggest that their alcohol consumption pattern may also be different from that of Australian-born men and women (Table 6). At the beginning of the 198Os, immigrants from New Zealand had relatively lower cardiovascular mortality but this increased to a point where their standardised mortality ratios were not statistically different from those for Australian-born men and women in (Table 1). However, this does not necessarily reflect a deleterious change in riskfactor levels. The increase could reflect the higher proportion of Maoris in the migration stream from New Zealand in recent years4 Period of resia!eme Generally, analyses which have examined the influence of period of residence in Australia on mortality or on risk-factor levels have not taken into account the confounding effect of the marked variation in birthplace composition of immigrants over time. This present analysis attempts to allow for this by examining changes in risk-factor levels with duration of residence for broad immigrant groups and also by standardising across immigrant groups. Caution is necessary in interpretation however. Evidence of acculturation within a particular immigrant group may simply reflect the fact that longer stay immigrant residents had a different risk-factor profile on arrival in Australia than those who arrived more recently. Lack of evidence may indicate that the period-ofresidence dichotomy used in this analysis is insensitive to the acculturation effect. Risk-factor differentials which existed at the time of immigration may have been moderated by general environmental factors operating immediately after settlement in Australia and a finer classification of period in Australia may be required to detect this effect. Nevertheless, the results suggest that acculturation has had minimal effect on many of the traditional risk factors for cardiovascular disease. There is evidence that immigrants have generally had lower blood pressure on arrival in Australia which, after adjusting for age, increased as their time in Australia increased. There is also evidence that light alcohol intake becomes more prevalent with length of stay, particularly among Asian immigrants and women from Europe. There is also evidence of positive increases in exercise during leisure time. There is no evidence of any marked changes in the blood lipid profile of immigrants with increasing stay in Australia nor any real change in smoking prevalence. Mortality analysis indicates increasing age-standardised mortality for cardiovascular disease with period of residence in Australia for most immigrant group^.^ Methodological issues Although difficult to quantify, it is important to identify the aspects of the survey methods which have influenced the analysis. Despite the inclusion of an explanation in eleven languages with the letter of invitation in 1989, immigrants represented 35 per cent of the target population but only 27 per cent of respondents.6 Using the electoral roll as the sampling frame would have led to immigrants being underrepresented in the sample, particularly recent immigrants. It is also possible that the response rate among immigrants was lower than among nativeborn Australians. Markedly different response rates between immigrant groups would have confounded the analysis if propensity to respond was associated with risk-factor characteristics. Since no oversampling strategies were adopted to increase the number of immigrants in each survey, it was necessary to combine the three surveys so that tests for differentials had adequate statistical power. The minimal detectable difference, given a two-sided significance level of 0.05 and a,power of 80 per cent, was around 5 per cent of the Australian mean for most risk factors, which suggests that sample sizes were adequate for the analysis. The adjustment for age assumed a linear relationship between age and the dependent variable and that the same relationship held for each immigrant group and each survey. A more complex model would be difficult to justify given the relatively small sample size for some immigrant groups. Also, to the extent that immigrant groups congregate in different cities, adjusting for the city effect may also have partly adjusted for immigrant group risk-factor differentials. The aggregation of individual countries into broad regions may have masked important differences. Regarding the body mass analysis, perhaps the issue is whether body mass index is an appropriate indicator of cardiovascular disease risk across such a diverse range of immigrant groups. Several measures of body fat distribution are used in the literature and they are known to have different properties.2a90 The 1989 survey also collected waist and hip circumferences and the result of analysing waist to hip ratios was similar to those based on body mass index with the following exceptions. Women from Asia had lower body mass indices but higher waist-hip ratios and women from had higher waisthip ratios but not body mass indices, indicating different body shapes. There is debate about the role of alcohol consump tion in cardiovascular disease and whether light to moderate drinking is protective. A recent review of AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. I7 NO

10 BENNETT the evidence concluded that two drinks a day was associated with no cardiovascular harm and might be protective against coronary heart disease, and that above two drinks a day there was evidence of harm, biological as well as ~0cial.I~ A recent analysis of MRFIT data suggests that, in men, the protective effect may be related to high-density lipoprotein levels, which increase with alcohol consumption.ig The indicator in this analysis, light alcohol intake, was based on the definition of responsible levels of alcohol consumption recommended in Australia. * There is no generally accepted instrument for the measurement of physical activity in population surveys. Studies have tended to focus on leisure-time activity, but recent evidence suggests that both leisure-time exercise and exercise at work reduce risk of heart disease, although their relative contributions are unclear, and may well differ for different immigrant gr~ups.~ *~~ Regarding leisure-time physical activity, recent years have seen a shift in emphasis away from the more vigorous aerobic exercise towards the health benefits of less vigorous exerci~e.~ ~~ The 1991 Heart Week in Australia emphasised the benefits of walking for exerci~e.~ The indicator used in this analysis, no leisure-time exercise of any kind, implicitly considered vigorous exercise, less vigorous exercise and walking all to be valued forms of exercise to reduce heart disease risk. Concluding remarks This analysis has identified differences in risk-factor levels among Australia s immigrants when compared with their Australian-born counterparts. However, it would appear that the lower age-standardised mortality ratios from cardiovascular disease which characterise immigrants in Australia cannot be explained solely on the basis of their profiles of risk factors which are commonly accepted as determinants of disease. Of the traditional risk factors, systolic blood pressure is the best single explanatory factor for variation in cardiovascular mortality among men, and smoking prevalence the best among women. The relationship between risk-factor and mortality differentials for immigrant groups is complex however, and a full explanation is likely to involve interaction among social, economic, cultural, environmental, biological and genetic factors, as well as factors relating specifically to the migration process.38 It may well be that the lead time between exposure to major coronary risk factors and subsequent effects on mortality also varies among immigrant groups. One of the most interesting findings was the lack of clear differences in blood lipids. A person s blood lipid profile may be influenced by factors such as physical activity, alcohol consumption, use of oral contraceptives, aspects of diet such as dietary fibre and saturated fats, as well as genetic factors. These, and other factors, will combine to determine the blood lipid levels of a particular immigrant group. It seems that in Australia, such factors have largely balanced each other out and that blood lipids, including total cholesterol, have played little part in explaining cardiovascular mortality differences. The evidence suggests that, after adjusting for the aging effect, systolic blood pressure increases with length of stay in Australia. Body mass index increases among Asians despite increased participation in recreational exercise with longer length of stay. There is no evidence that blood lipid profiles are affected by the process of acculturation. Acknowledgments The author thanks the Risk Factor Prevalence Study Management Committee for permission to analyse data from the study, and Dr S. Wilson, Dr E. Kliewer, Dr P. Solomon and Dr A. Worsley for their guidance and comments. References 1. Australian Bureau of Statistics Census ofpopulation and housing-mierofihe CX0003. Canberra: Australian Government Publishing Service, Bureau of Immigration Research. Immigration updute, December quarter Canberra: Australian Government Publishing Service, Young CM. Selection and survival. Immigrant mortality in Australia. Canberra: AGPS and Department of Immigration and Ethnic AfFairs, Young CM. Mortality: the ultimate indicator of survival. The differential experience between birthplace groups. In: Donovan J. d Espaignet E, Merton C, van Ommeron M, editors. Immigrants in Aus!ralia: a health profik. Canberra: Australian Institute of Health and Australian Government Publishing Service, Australian Bureau of Statistics Nation01 Health Survey. Cardiovascular and related conditions. Cat. no CanberIa: ABS, Young CM, Coles A. Women s health, use of medical services, medication, lifestyle and chronic illness. Some findings from the National Health Survey. In: Donovan J, despaignet E, Merton C, van Ommeron M, editors. Immigrants in Australia: a health profile. Canberra: Australian Institute of Health and AGPS, WHO MONICA Project/Keil U, Kuulasmaa K, WHO Monica Project: risk factors. Int J Epidemiol 1989; 18(Suppl 1): s Risk Factor Prevalence Study Management Committee. Risk fiorpevalence study: suruey no. 3,1989. Canberra: National Heart Foundation of Australia and Australian Institute of Health, National Heart Foundation of Australia. Risk fiorprevalence study, no. 2, Canberra: NHFA, National Heart Foundation of Australia. Riskfactorprevalence study, no. 1, Canberra: NHFA, National Cholesterol Education Program (US). Repor! of!he expert panel on detection, evaluation and treatmen! of high blood choksterol in adults. No Bethesda: National Institutes of Health, National Health and medical Research Council. Is!here a safe he1 of daily consump!ion of alcohol for men and women? Canberra: Australian Government Publishing Service, Miller GJ, BeckJes GLA, Maude GH, Carson DC. Alcohol consumption: protection against coronary heart disease and risks to health. Int J Epidemiol 1990; 19: Jackson R, Scragg R, Beaglehole R. Alcohol consumption and risk of coronary heart disease. BMJ 1991; 303: Marmot M, Brunner E. Alcohol and cardiovascular disease: the status of the U shaped curve. BMJ 1991; 303: Rimm EB, Giovannucci EL, Willett WC, Colditz GA, et al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet 1991; 338: Razay G, Heaton KW, Bolton CH et al. Alcohol consumption and its relation to cardiovascular risk factors in British women. BMJ 1992; 304: Wilkins PS. Medicinal virtues of alcohol in moderation. Med J Aud 1992; Suh I. Shaten BJ, Cutler JA, Kuller LH, for the Multiple Risk Factor Intervention Trial Research Group. Alcohol use and mortality from coronary heart disease: The role of highdensity lipoprotein cholesterol. Ann I- Med 1992; 116: Australian Bureau of Statistics. Australian standard clnrrificdia of countries for social statistics. Cat. no Canberra: ABS, AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 VOL. 1 7 NO. 3

11 INFORMAL CAREGIVERS 21. Stenhouse NS, McCall MG. Differential mortality from cardiovascular disease in migrants from England and Wales, Scotland and, and native-born Australians. J Chronic Dis 1970; 23: Ulman R, Abernethy JD. Blood pressure and length of stay in Australia of Italian immigrants in the Australian National Blood Pressure Study. Znt J Epidemiol 1975; 4: Armstrong BK, Margetts BM, Masarei JRL, Hopkins SM. Coronary risk factors in Italian migrants to Australia. Am J Epidnniol 1983; 118: English RM, Bennett SA. Overweight and obesity in the Australian community. J Food Nutr 1985; 42: Keys A, Menotti A, Karvonen MJ et al. The diet and 15-year death rate in the seven countries study. Am J Ep demiol1986; 124: Carleton RA, Dwyer J, Finberg L, et al. Report of the Expert Panel on Population Strategies for Blood Cholesterol Reduction. A statement from the National Cholesterol Education program, National Heart, Lung and Blood Institute, National Institutes of Health. Circulation 1991; 83: World Health Organization. Monica manual, version 1.1. Geneva: WHO, Bjorntorp P. The associations between obesity, adipose tissue distribution and disease. Acto Med Scad Suppl 1988; 723: Himes JH, Bouchard C, Pheley AM. Lack of correspondence amongmeasuresidentifjmgtheobese.amjrmmed : Mueller WH, Wear ML, Hank CL, Emerson JB, et al. Which measurement of body fat is best for epidemiological research? AmJ Epidemiol 1991; 133: Salonen JT, Slater JS, Tuomilehto J, Rauramaa R. Leisure time and occupational physical activity: risk of death from ischemic heart disease. Am J Epidemiol 1988; 127: Zimmet PZ, Collins VR, Dowse GK, Alberti KGMM, et al. for the Mauritius Noncommunicable Disease Study Group. The relation of physical activity to cardiovascular disease risk factors in Mauritians. Am J Epidemiol 1991; 134: Blair SN, Kohl HW, Paffenbarger RS, Clark DG. Physical fitness and allcause mortality. A prospective study of healthy men and women.jama 1989; 262: debusk RF, Stenestrand U, Sheehan M, Haskell WL. Training effect of long versus short bouts of exercise in healthy subjects. Amj Cardiol 1990; 65: Oldenburg B, Bauman A, Booth M, Owen N. Increasing levels of physical activity in the Australian community. Healfh RmtJAM 1991; 1: National Heart Foundation of Australia. Exercise and head disease: pol? stoument M. 5. Canberra: NHFA, National Heart Foundation of Australia. Make it pa? ofyour day fkajlct]. Advicefrom the heart fkajlct]. So you want to get w e exercise [jmmfihkt]. Canberra: NHFA, Polednak AP. Racial and ethnic differmrr in disease. Oxford Oxford University Press, The health and wellbeing of informal caregivers: a review and study program Helen Herman, Bruce Singh and Hilary Schofleld Department of Psychiatv, University of Melbourne Robin Eastwood Departments of Psychiatry and Preventive Medicine and Biostatistics, University of Toronto Philip Burgess and Virginia Lewis Psychiatric Epidemiology and Services Evaluation Unit, Health Department Victoria Richard Scotton Professnial Fellow, Public Sector Management Institute, Monash University, Melbounze Abstract: Informal caregivers are the families and other unpaid caregivers in the home who support people of all ages with severe and chronic mental or physical disabilities. Home care of this sort has been increasing over the past 30 years because of the reduced number of beds in hospitals and nursing homes and increased outpatient and community care. Moreover, with an aging population and increasing rates of disability, the demand for family caregiving will continue to rise. This has important implications for the development of health, community service and social policy. At the same time, however, very little is known about the impact such changes are having on the caregivers of various ages and in various circumstances. The Victorian Health Promotion Foundation is funding a research and intervention program in Melbourne to promote wellbeing and prevent ill-health in caregivers. (ArrstJ Public Health 1993; 17: 261-6) F amilies, and other unpaid caregivers in the home, are important sources of support for people of all ages with severe disabilities and longterm illness. * In our community and elsewhere in the western world the work of these informal caregivers has increased over the past 30 years. Their work will expand further not only as a result of our Correspondence to Professor Helen Herrman, Department of Psychiatly, St Vincent s Hospital, 41 Victoria Parade, Fitzroy. Vic Fax (03) aging population and increasing rates of disability, 12 but also as changes in the organisation of services continue, %16 more specifically with the growth in community care. The caregiver role is clearly of growing importance and is becoming a major focus of research interest ove~seas~j~-~~ and, more recently, in Au~tralia.~.~ -~~ In recent years informal carers have been gaining some recognition in government policy at the state and federal levels AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1993 vot. 17 NO