The Isfahan Cohort Study: Rationale, Methods and Main Findings

Home , Isfahan

Journal of Human Hypertension (2011) 25, 545–553 & 2011 Macmillan Publishers Limited All rights reserved 0950-9240/11 www.nature.com/jhh ORIGINAL ARTICLE The Isfahan cohort study: Rationale, methods and main findings

N Sarrafzadegan, M Talaei, M Sadeghi, R Kelishadi, S Oveisgharan, N Mohammadifard, AR Sajjadieh, P Kabiri, T Marshall, GN Thomas and A Tavasoli Isfahan Cardiovascular Research Center (WHO-Collaborating Center for Research and Training in Cardiovascular Diseases Control, Prevention, and Rehabilitation for Cardiac Patients in the Eastern Mediterranean region), Isfahan University of Medical Sciences, Isfahan, Iran and Public Health, Epidemiology and Biostatistics, Health and Population Sciences, The University of Birmingham, Birmingham, UK

A 10-year longitudinal population-based study, entitled person-years for women. The respective risk of ischemic the Isfahan Cohort Study (ICS) is being conducted. The stroke was 2.3 (1.7–3.0), 2.3 (1.6–3.3) and 2.3 (1.5–3.2) ICS commenced in 2001, recruiting individuals aged per 1000 person-years. The risk of IHD was approxi- 35 þ living in urban and rural areas of three counties in mately 3.5-fold higher in the presence of hypertension, central Iran, to determine the individual and combined followed by diabetes mellitus and hypercholesterolemia impact of various risk factors on the incidence with near 2.5- and twofold higher risk, respectively. This of cardiovascular events. After 24379 person-years of cohort provides confirmatory evidence of the ethnic follow-up with a median follow-up of 4.8 years, we differences in the magnitude of the impact of various documented 219 incident cases of ischemic heart risk factors on cardiovascular events. The differences disease (IHD) (125 in men and 94 in women) and 57 may be due to varying absolute risk levels among incident cases of stroke (28 in men and 29 in women). populations and the existing ethnic disparities for using The absolute risk of IHD was 8.9 (7.8–10.2) per 1000 western risk equations to local requirements. person-years for all participants, 10.6 (8.8–12.5) per Journal of Human Hypertension (2011) 25, 545–553; 1000 person-years for men and 7.4 (6.0–9.0) per 1000 doi:10.1038/jhh.2010.99; published online 25 November 2010

Keywords: cardiology; cohort; developing country

Introduction cross-sectional associations. The INTERHEART study, the largest study to date, was conducted in Chronic non-communicable diseases, particularly 52 countries and revealed that the relative risks cardiovascular disease (CVD) are no longer limited associated with different CVD risk factors are to the industrialized world. According to the World consistent in both sexes and at all ages and in the Health Organization estimates; in 2003 low- and different regions. The main difference was because middle-income countries accounted for 86% of the of the underlying distribution of the risk factors.4 global CVD disease burden, and by 2010, CVD However, the impact of risk factors in various will be the leading cause of death in developing 1 ethnicities can be best verified by longitudinal countries. CVD has no geographic, gender or socio- studies as the distribution of risk factors and in demographic margins, and the major risk factors are turn their impact on CVD events may vary between well-documented. There is a large body of evidence populations. supporting strong ethnic differences in CVD risk The scarcity of such longitudinal evidence in low- factor levels and their synergetic effects on CVD, 2 and middle-income countries and the abovemen- even from an early age. tioned ethnic differences underscore the necessity It has been suggested that environmental-gene of conducting cohort studies to determine not only interactions along with socioeconomic and beha- the risk factor prevalence but also the impact of each vioral factors underlie ethnic disparities in CVD 3 or different combinations of risk factors on the risk-factor profiles. Most studies have described incidence of coronary and cerebrovascular events in diverse populations. Correspondence: Professor N Sarrafzadegan, Isfahan Cardiovas- Middle Eastern countries are of special concern in cular Research Center, Isfahan University of Medical Sciences, this context, as in the next two decades they face the Khorram Street, Isfahan 81465-1148, Iran. world’s greatest increment in the absolute burden of E-mail: [email protected]. Received 10 January 2010; revised 8 September 2010; accepted 12 diabetes and in turn other chronic non-communic- 5 September 2010; published online 25 November 2010 able diseases, especially CVDs. As one of these The isfahan cohort study N Sarrafzadegan et al 546 countries, Iran has experienced both a rapid epide- Baseline Survey of the IHHP miologic transition and change in the health and The baseline survey of the IHHP was conducted in a disease profile from predominantly infectious representative population of adults aged X19 years diseases to chronic non-communicable diseases.6 who were living in urban and rural areas of Isfahan, Several cross-sectional studies have documented Arak and Najafabad. These areas are shown on the alarming prevalence rates of CVDs and their risk map of Iran (Figure 1). Participants were selected by factors among the Iranian population,7–12 but there is multistage random sampling. The study population very limited data from longitudinal studies and those was first stratified by their living area (urban vs are limited to a metropolitan area of the capital city.13 rural) according to the regional population distribu- The Isfahan Cohort Study (ICS) is the first study of tion derived from a national population census its kind, not only in Iran, but also in the Eastern conducted in 1999. Census blocks were then Mediterranean region. It is a 10-year longitudinal randomly selected from each county with the study, conducted from 2001 in urban and rural areas probability of selection proportional to the expected of three counties in central Iran. It aims to determine number of households and divided into clusters of the individual and combined impact of various risk approximately 1000 households. Within each factors on the incidence of CVD events including fatal cluster approximately 5–10% of households were and non-fatal myocardial infarction (MI), fatal and randomly selected for enumeration. From each non-fatal stroke and sudden cardiac death. Here, we household one eligible individual aged X19 years report the methodology and some of the main results was randomly selected, provided they were of of this ongoing population-based cohort study. Iranian nationality, mentally competent and not pregnant. The response rate in home interviews was 98%; however, 95% attended the examination Subjects and methods clinic. Given that individual’s cardiovascular health was not considered in the exclusion criteria of the The ICS is a population-based, longitudinal ongoing IHHP sampling framework, the baseline participants study of 6504 adults aged equal or greater than included 108 (2.8%) cases with a history of MI, 35 years at baseline, living in urban and rural areas stroke or heart failure who were excluded from the from three counties in central Iran (Isfahan, Arak ICS baseline survey. and Najafabad, Figure 1) who had participated in The sample for the IHHP was recruited into the baseline survey of a community trial for CVD different age and sex groups to reflect the age, sex prevention and control, entitled Isfahan Healthy distribution of the community. It was estimated that Heart Program (IHHP).14,15 They were recruited from the prevalence of cardiovascular risk factors would January 2 to September 28, 2001 and will be be 0.2 in the control area and sample size for the followed up for at least 10 years. Figure 2 sum- IHHP was determined to have a 90% power to detect marizes the methodology of the ICS including the a relative risk of 0.75 at a significance of 0.05. Taking variables measured, sample sizes and follow-up. account of clustering, the overall sample size was

Figure 1 Map of Iran and locations of three areas of study (Isfahan, Najafabad and Arak) and Tehran (Iran’s capital). Figure 2 Algorithm for the Isfahan Cohort Study.

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 547 4828 in each area, allowing for losses to follow-up it post-prandial (2hpp) glucose test were measured was intended to recruit 6300 in each area (12 600 immediately, and serum frozen at À20 1C, then by a subjects). A total of 12 514 individuals were in- 3-h ground transport with cold chain (À20 1C) they cluded in the baseline survey. were transported to the central laboratory and kept frozen until assayed within 72 h. Dyslipidemia was defined as if LDL-C X130 mg dlÀ1, Selection of subjects for the ICS TC X200 mg dlÀ1, triglycerides X150 mg dlÀ1 or HDL-C Ethical approval was obtained from the Ethics o40 mg dlÀ1 in men or o50 mg dlÀ1 in women.21 Committee of Isfahan Cardiovascular Research Diabetes mellitus was defined as if fasting blood Center, a World Health Organization-collaborating glucose X126 mg dlÀ1 or the patient was receiving center. Of the 12514 individuals from the baseline anti-diabetic agents; impaired glucose tolerance was survey, there were 6640 adults aged 35 years and identified if the 2-h postprandial X140 but less than over in the baseline survey who were enrolled into 200 mg dlÀ1.22 The body mass index value X25 kg mÀ2 the ICS. We intend to follow-up the volunteers for at but less than 30 kg mÀ2 was classified as overweight least 10 years. and those who had body mass index X30 kg mÀ2 were classified as obese.23 According to International Dia- betes Federation cut points, waist circumference (WC) Data collection X94 cm in men or X80 cm in women were defined as After obtaining informed written consent, a 30-min abdominal obesity.24 We also considered the cut points full structured interview was conducted by trained suggested for abdominal obesity in the Iranian popula- health professionals using a validated questionnaire tion of WC X91.5 cm and WC X85.5 cm in men and including questions on demographic characteristics, women, respectively.25 Waist-to-hip ratio X0.95 in socioeconomic status, behaviors, attitudes, skills men and X0.8 in women was considered as a high and knowledge about chronic non-communicable waist-to-hip ratio.23 Blood pressure X140/90 mm Hg or diseases, as well as related lifestyle behaviors if the patients were receiving antihypertensive drugs (including smoking, physical activity and nutri- was defined as hypertension.26 tional habits). Thereafter, participants were invited to the nearest health center, where a 15-min medical interview and a 20-min physical examination Follow-up surveys were conducted by trained physicians and nurses. After the baseline survey in 2001, follow-up of the Measurement of blood pressure and anthropometric volunteers has been carried out every 2 years. parameters were carried out following standard Telephone interviews were carried out in 2003 and protocols16,17 and using calibrated instruments. in 2005–2006. In 2007, full structured interviews; Body mass index was calculated as weight (kg) physical and biochemical measurements were divided by height squared (m2). A 12-lead electro- repeated in the same way as for the baseline survey. cardiogram was recorded at the primary health care Figure 2 shows the study algorithm. A third centers for all participants. telephone interview follow-up has been finished Fasting (12 h) blood samples (10 ml) were ob- recently and will be repeated in 2011. For the tained from all participants and were examined at telephone follow-up interviews, at least five at- the Isfahan Cardiovascular Research Center central tempts are made to contact all living participants laboratory with adherence to external national and or their first-degree relatives if they are deceased. international quality controls. Serum total choles- If telephone interviews are unsuccessful, the parti- terol (TC), triglycerides and fasting blood glucose cipants are visited at their home address for follow- were measured enzymatically,18 using an autoana- up. After confirmation of participants’ identity, lyzer (Eppendorf, Hamburg, Germany) and serum structured primary interviews were performed high-density lipoprotein-cholesterol (HDL-C) was on the basis of a questionnaire with three main determined after precipitation of low-density and questions; ‘is he/she alive?’, ‘has he/she been very low-density lipoproteins with dextran sulfate- hospitalized for any reason? (with specific focus magnesium.19 C-reactive protein level was deter- on cardiovascular and cerebrovascular events) and mined by the same autoanalyzer. Serum low-density ‘has the participant experienced any of the follow- lipoprotein-cholesterol (LDL-C) was calculated ing five neurological symptoms (hemiparesis, dysar- using the Friedwald equation in subjects with thria, facial asymmetry, imbalance and transient triglycerides less than 400 mg dlÀ1, otherwise were monoocular blindness)?’ measured using standard kits.20 A 75 gram oral If death, hospitalization or neurological symptoms glucose tolerance test was performed on non- have occurred, the date of the events, physician diabetic individuals to determine their 2-h post- diagnosis and the hospital’s name are obtained load plasma glucose level. Blood samples were during the interview. If any event has occurred, centrifuged immediately in each county, samples the related questionnaire was checked alongside the obtained in Isfahan and Najafabad were transported relevant health records. In the case of out of hospital to the central laboratory within about 1 h; in the deaths, death certificates were obtained from the reference area, fasting blood glucose and 2-hour provincial mortality database and verbal autopsies

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 548 were performed by a trained expert nurse in a software (SPSS Inc., Chicago, IL, USA; version 15.0). secondary interview with surviving family mem- Student’s t-test was used for the comparison of bers. The verbal autopsy used a pre-defined means of independent groups and the w2-test for the questionnaire including medical history, signs and comparison of proportions and categorical variables. symptoms before death. Expert nurses conducted The residency areas were included in the model as a additional secondary interviews for hospita- variable in six levels for the urban and rural areas of lized cases where information was incomplete or the three counties being studied. inconsistent. Cox proportional hazards modeling was used with time to outcome as the dependent variable and the presence of defined risk factors as independent Confirmation of end points dichotomous variables for the calculation of hazard The reported events were checked with the MI and ratios (HR) and 95% confidence intervals. Indivi- stroke registry database of the Surveillance Depart- duals were censored at the first cardiovascular ment, Isfahan Cardiovascular Research Center. The event. Kaplan–Meier analysis was used to evaluate registry is collected monthly in the three aforemen- time to outcome as a function of dichotomous- tioned counties. In the case of any inconsistency in independent variables with log-rank statistic for dates or diagnoses, or unobtainable records, original determining statistically significant differences. medical records were investigated. If hospitalization For all analyses, statistical significance was assessed data were not found in the registry database, trained at a level of 0.05 (two-tailed) and a P-value less nurses investigated relevant hospital medical than 0.1 was considered of borderline significance records. Two separate panels of specialists consist- (marginal significance). ing of four cardiologists and neurologists, reviewed all relevant documents of every patient (primary questionnaires, registry records, medical records, secondary interviews, verbal autopsies or death Results certificates) and made the final decision on all of The response rate for house interviews was 98%; the five main events (fatal and non-fatal MI, fatal however, 95% attended the examination clinic. and non-fatal stroke and sudden cardiac death) and Given that individual’s cardiovascular health was unstable angina (UA). not considered in the exclusion criteria of the IHHP The diagnosis of acute MI was based on the sampling framework, the baseline participants presence of at least two of the following criteria: (1) included 108 (2.8%) cases with a history of MI, typical chest pain lasting more than 30 min, (2) ST stroke or heart failure who were excluded from the elevation 40.1 mV in at least two adjacent electro- ICS baseline survey. cardiograph leads and (3) an increase in the serum From the 6640 individuals aged 35 years and level of cardiac biomarkers.27 The definition of UA older studied in the baseline cohort, 996 (15%) of required typical chest discomfort lasting more than telephone interviews were not successful, and 20 min within the 24 h preceding hospitalization address follow-ups were carried out. Of the whole and representing a change in the usual pattern of sample, 978 (15%) and 464 (7.1%) participants were angina or pain: occurring with a crescendo pattern, missed after first and second stage of follow-up, being severe and described as a frank pain.28 respectively. The baseline characteristics and the The diagnosis of UA might be new or be based on prevalence of CVD risk factors were not significantly dynamic ST-segment or T-wave changes in at least different among those participants lost to follow-up two adjacent electrocardiogram leads. Sudden car- compared those remained in the follow up surveys. diac death was defined as death within 1 h of onset, The questionnaires of the 1014 (20%) reports for a witnessed cardiac arrest, or abrupt collapse not events were investigated by a general practitioner preceded by 41 h of symptoms. IHD included to identify 620 (12.2%) relevant reports including definite or probable MI, UA and sudden cardiac 132 (2.6%) deaths, 396 (7.8%) hospitalizations and death. Moreover, the World Health Organization 92 (1.8%) reports of only neurological symptoms stroke definition was used, that is, stroke was and referred them for a further search of more defined as a rapid-onset focal neurological disorder authentic documents including medical records, persisting at least 24 h and had probable vascular death certificates and verbal autopsies. Of the total origin. The diagnosis of incident stroke was con- deaths, 77 (36%) occurred out of hospital. Expert ducted based on the clinical criteria. CVDs were nurses conducted additional secondary interviews defined as combination of IHD and stroke. Although for 196 (3.9%) of 5063 available hospitalized cases the in-hospital diagnoses of clinicians were taken with incompatible or incomplete information. into account, the final decisions of the panel were After 24379 person-years of follow-up (with made independently. median follow-up of 4.8 years, 4.6 and 5 years for the 25th and the 75th quartiles, respectively), Statistical analysis we documented 219 incident cases of IHD (125 in Data entry was carried out using EPI info. All men and 94 in women) and 57 incident cases of data were analyzed using the SPSS for Windows stroke (28 in men and 29 in women). IHD comprised

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 549 13 (8 in men and 5 in women) fatal and 45 (30 in value was higher in the participants with CVD men and 15 in women) non-fatal MI, 113 (54 in men events, whereas the corresponding figure was not and 59 in women) UA and 48 (33 in men and 15 in significantly different for the prevalence of obesity. women) sudden cardiac deaths. The risk of CVD was approximately 3.5-fold Ischemic stroke comprised 13 (7 in men and 6 in higher in the presence of hypertension, followed women) fatal and 44 cases (21 in men and 23 in by diabetes mellitus and hypercholesterolemia with women) of non-fatal stroke. Absolute risk of near 2.5- and twofold higher risk, respectively IHD was 8.9 (95% confidence intervals: 7.8–10.2) (Table 2). Multivariable adjustment attenuated the per 1000 person-years for total participants, 10.6 extent to which these risk factors increased the risk (8.8–12.5) per 1000 person-years for men and 7.4 of CVD but the ranking remained constant. Abdom- (6.0–9.0) per 1000 person-years for women. inal obesity according to International Diabetes The respective risk of ischemic stroke was 2.3 Federation and Iranian definitions yielded margin- (1.7–3.0), 2.3 (1.6–3.3) and 2.3 (1.5–3.2) per 1000 ally significant associations with similar HR. Never- person-years. theless, in the multivariable analysis, none of the Table 1 shows the baseline characteristics of definitions resulted in significant association with participants and CVD risk factors. In general, the an increased risk of CVD events. Moreover, a unit participants who suffered from defined CVD events increase in the proportion of TC to HDL-C ratio, were urban men from older age groups with higher considered as a quantitative variable, was associated rates of smoking, over 3 cm greater WC, higher with a significant increase in risk of CVD even in waist-to-hip ratio and a twofold higher prevalence of multivariable adjusted model. diabetes or hypertension than those without CVD Figure 3 shows the age- and sex-adjusted effect of event. Except HDL-C, all other lipid levels were multiple risk factors on the level of risk for CVD significantly higher in those with CVD events than events. Although adding high TC to the analysis those without an event. The mean body mass index did not change the HR of hypertension and

Table 1 Characteristics of the study participants according to the development of cardiovascular event: ICS, 2001–2006

Characteristics CVD event (N ¼ 276) No CVD events (N ¼ 5099) P-value

Age (years) 58.8±11.2 50.2±11.5 o0.001 Male (%) 153 (55.4) 2635 (51.7) 0.021 Fasting blood glucose (mg dlÀ1) 100±47 88±32 o0.001 2-h Post prandial glucose (mg dlÀ1) 129±73 110±56 o0.001 Systolic blood pressure (mm Hg) 138±25 121±20 o0.001 Diastolic blood pressure (mm Hg) 85±14 78±11 o0.001 Total cholesterol (mg dlÀ1) 227±52 211±49 o0.001 Triglycerides (mg dlÀ1) 218±118 189±102 o0.001 LDL-C (mg dlÀ1) 140±46 127±42 o0.001 HDL-C (mg dlÀ1) 46.5±10.3 46.9±10.4 0.534 Total /HDL-C 5.04±1.44 4.66±1.31 o0.001 C-reactive protein (mg dlÀ1) 3.32±1.30 3.31±1.49 0.965 Body Mass Index (kg mÀ2) 27.3±4.9 26.7±4.4 0.024 Waist circumference (cm) 97.6±12.6 94.5±12.2 o0.001 Waist-to-hip ratio 0.95±0.07 0.92±0.07 o0.001 Urban residence area 217 (78.6) 3680 (72.2) 0.019 Diabetes mellitusa 59 (21.8) 459 (9.1) o0.001 Impaired glucose testb 31 (12.6) 436 (8.9) 0.055 Hypertensionc 165 (59.8) 1395 (27.4) o0.001 Current smoking 71 (25.7) 1050 (20.6) 0.041 Dyslipidemiad 252 (92.3) 4355 (86.9) 0.009 Overweighte 112 (41.5) 2031 (40.8) Obesityf 65 (24.1) 1078 (21.6) Abdominal obesity (IDF cut points)g 206 (75.5) 3542 (69.7) 0.044 Abdominal obesity (Iranian cut-points)h 203 (75.2) 3438 (68.9) 0.030 High waist-to-hip ratioi 194 (71.3) 3322 (65.4) 0.046

Abbreviations: CVD, Cardiovascular disease; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol. aDiabetes mellitus: fasting blood glucose X126 mg dlÀ1 or receiving anti-diabetic agents. bImpaired Glucose Test: 2-h post prandial glucose (2hPP) X140 but less than 200 mg dlÀ1. cHypertension: systolic blood pressure X140 mm Hg, Diastolic blood pressure X90 mm Hg, or current treatment for hypertension. dDyslipidemia: LDL-C X130 mg dlÀ1, total cholesterol X200 mg dlÀ1, triglyceride X150 mg dlÀ1, HDL-C o40 mg dlÀ1 in men and o50 mg dlÀ1 in women. eOverweight: 25 kg mÀ2 pbody mass index o30 kg mÀ2. fObesity: body mass index X30 kg mÀ2. gAbdominal obesity (IDF cut-points): waist circumference X94 cm in men and X80 cm. hAbdominal obesity (Iranian cut-points): waist circumference X91.5 cm in men and X85.5 cm. iHigh waist-to-hip ratio: waist-to-hip ratio X0.95 in men and X0.8 in women. Mean±s.d. or proportion (%).

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 550 Table 2 Crude and adjusted hazard ratios of risk factors for cardiovascular events: ICS, 2001–2006

Characteristic Crude HR (95% CI) P-value Adjusted HR (95% CI)a P-value

Sex (Male) 1.40 (1.10–1.79) 0.005 1.41 (1.05–1.90) 0.021 Diabetes mellitusb 2.66 (1.97–3.57) o0.001 1.75 (1.27–2.39) o0.001 Impaired glucose tolerancec 1.57 (1.07–2.31) 0.020 1.16 (0.78–1.73) 0.452 Hypertensiond 3.42 (2.68–4.37) o0.001 2.10 (1.60–2.76) o0.001 Current smoking 1.37 (1.04–1.80) 0.022 1.41 (1.03–1.93) 0.032 Overweighte 1.05 (0.79–1.39) 0.718 0.97 (0.72–1.30) 0.867 Obesityf 1.13 (0.81–1.56) 0.462 1.04 (0.73–1.49) 0.789 Abdominal obesity (IDF cut-points)g 1.31 (0.99–1.73) 0.059 1.18 (0.85–1.63) 0.304 Abdominal obesity (Iranian cut-points)h 1.36 (1.02–1.80) 0.031 1.10 (0.80–1.50) 0.545 High waist to hip ratioi 1.27 (0.97–1.66) 0.78 1.18 (0.86–1.63) 0.297 Hypercholesterolemiaj 1.91 (1.46–2.49) o0.001 1.46 (1.09–1.96) 0.010 High LDL-cholesterolk 1.81 (1.39–2.35) o0.001 1.57 (1.20–2.07) 0.001 Low HDL-cholesteroll 1.03 (0.81–1.32) 0.769 1.26 (0.97–1.65) 0.076 Hypertriglyceridemiam 1.72 (1.32–2.25) o0.001 1.23 (0.92–1.65) 0.161 High total cholesterol/HDL-Cn 1.27 (1.17–1.39) o0.001 1.19 (1.09–1.31) o0.001 C-Reactive Protein 0.96 (0.85–1.08) 0.582 0.94 (0.84–1.06) 0.362

Abbreviations: CI, confidence interval; HDL, high-density lipoprotein; HDL-C, high-density lipoprotein-cholesterol; LDL, low-density lipoprotein. aAdjusted model including: residency, age, sex, smoking status, BMI categories, hypertension, diabetes, high cholesterol, high triglyceride, low HDL-cholesterol. bFasting blood glucose X126 mg dlÀ1 or being on anti-diabetic agents. c2-h postprandial (2hPP) X140 mg dlÀ1, but less than 200 mg dlÀ1. dSystolic blood pressure X140 mm Hg, Diastolic blood pressure X90 mm Hg, or current treatment for hypertension. e25 kgmÀ2 pbody mass index o30 kg mÀ2. fBody mass index X30 kg mÀ2. gIDF cut-points: waist circumference X94 cm in men and X80 cm. hIranian cut-points: waist circumference X91.5 cm in men and X85.5 cm. iWaist to hip ratio X0.95 in men and X0.80 in women. jTotal cholesterolX240 mg dlÀ1. kLDL-cholesterol X130 mg dlÀ1. lHDL-cholesterol o40 mg dlÀ1 in men and o50 mg dlÀ1 in women. mHypertriglyceridemia: triglyceride X150 mg dlÀ1. nHigh total cholesterol/HDL-cholesterol X5mgdlÀ1.

combination of hypertension, diabetes, high LDL-C and smoking were documented in nine participants, and resulted in what was considered an unstable HR of 7.41 with wide confidence intervals of 2.36– 23.27, and was thus not shown in the figure.

Discussion This large population-based longitudinal study provides confirmatory evidence about the ethnic differences in the magnitude of the impact of various risk factors on CVD events. Globally, in most population-based studies, the prediction of the absolute risk of CVD events has been based on risk prediction equations originating from the Framing- ham Heart Study, as the oldest longitudinal study for CVD risk.29 However, the limitations for the applicability of the Framingham risk function in diverse populations are well-documented. The Framingham Heart study has developed mathema- Figure 3 Age- and sex-adjusted hazard ratios for risk factors combinations*. tical functions for predicting the risk of CHD events. This cohort consisted of 2439 men and 2812 women free of CVD. The 5- and 10-year CHD event rates diabetes combination, high LDL-C with the previous were 3.7 and 8.0% for men and 1.4 and 2.8% for combination increased the sex- and age-adjusted HR women, respectively.28 This and most other cohorts to nearly three. A combination of the first two in the Western countries consist of white middle- variables with high triglycerides and low HDL-C class individuals; therefore, there are concerns level resulted in a HR of more than four. The about the generalizability of their findings to other

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 551 populations and societies particularly in low- and and 2.4, respectively. The relative risk ranged from middle-income countries. To our knowledge, ICS is 1.1 to 2.5 in men and 0.5–1.8 in women studied in the first community-based study of CVD risk from Framingham and that higher levels of blood pressure any Eastern Mediterranean country that is being are typically associated with abnormalities in other conducted in a geographic area with a large, stable, risk factors and increased prevalence of CVD events.34 well-defined population living in urban and rural The Atherosclerosis Risk in Communities study areas with high participation rates and using valid revealed that among black women, the relative risk case ascertainment. The number lost to follow-up, (RR) of hypertension was significantly higher (8.9) especially in the third phase, was low, and than that documented in the populations studied in it necessitated much technical effort and cost to the Framingham Heart Study and other cohorts, achieve this. A major reason for loss to follow-up in however, in other ethnic groups, as Hispanic, Japa- the earlier phases was changes in telephone num- nese-Americans and native Americans, the RR of CVD bers introduced by the Government. This was a part events was not significantly different in hypertensive of network capacity development and did not have individuals.34 Contrary to most other cohort studies special distribution, so it was likely to be random that underscored the role of hypertension on stroke, and thus not bias the follow-up of the volunteers. in our study it significantly increased the risk of MI, The baseline characteristics and the prevalence of and this shows the importance of conducting long- CVD risk factors were not significantly different itudinal studies in different ethnic groups. among those participants lost to follow-up com- Among lipids disorders, high TC and LDL-C had pared those remained in the follow-up surveys, and significant relative risk to CVD events even after this strengthen the ICS findings. adjustment for other factors. Low HDL-C was Although after correction for length of follow-up, not associated with CVD events even in the crude the observed number of CVD events in ICS was not analyses. The prevalence of low HDL-C in the significantly different from the expected number of population is high, which may be explained by a events derived from Framingham study, but CVD number of measures including genetic predisposi- events documented in ICS, that is, 5.4% among men tion, sedentary lifestyle and adverse quality of fat and 4.7% among women, showed a considerably intake in Iranians. higher rate of events among women when compared In our study, when hypertension was added to with the Framingham Heart Study. However, this other risk factors, there was an exponential increase difference might be because of higher admission in the RR for CVD events, which reached levels as rates for UA that are not included among CVD high as 7.4 when considering a combination of high events in some other studies. blood pressure combined with diabetes mellitus, As described before, the ICS was designed to high LDL-C and smoking. Overall, 78% of all CVD evaluate the relative risk of CVD risk factors in events happened in those individuals who suffered Iranians with its possible extrapolation to the greater from hypertension, diabetes and high LDL-C. Our Eastern Mediterranean population. Available evi- findings are consistent with the results of the dence suggests that absolute risk varies among INTERHEART study in regards of the association different populations independent of their major of risk factors with MI, which showed the effect risk factors. For instance, the absolute risk among of multiple risk factors increase the risk of MI. South Asians (Indians and Pakistanis) living in It showed that major CVD risk factors as dyslipide- Western countries appears to be about two times mia, smoking, hypertension, diabetes mellitus and higher than that of whites, even when the two abdominal obesity, as well unhealthy lifestyle populations are matched for major CVD risk fac- behaviors account for most of the risk of MI world- tors.30 Moreover, it is possible that some populations wide in both sexes and at all ages in all regions. would have lower baseline levels of the risk factors Furthermore, it found that smoking, hypertension than those observed in the Framingham Heart Study and diabetes mellitus increase the odds ratio for MI population. Results from the Honolulu Heart Study up to near 12-fold compared with those without and the Seven Countries study showed that the these risk factors.4 Although this global study found population of Eastern Asian ancestry and Japanese similar results in various regions, the associations exhibit a lower risk for CHD for a given set of risk were cross-sectional. In contrast, variations in the factors in comparison with other populations.31 HR and distribution of risk factors between popula- These marked differences between various populations in the current cohort study, more importantly tions provide sufficient evidence for adjustments of verified the longitudinal impact of the risk factors. The absolute risk in different racial and ethnic groups; Prospective Collaborative Study Group, which pooled however, the relative risk estimates can probably be 61 observational studies in more than one million reliable across various groups.32 volunteers with a collective experience of more than 12 High blood pressure was seen in 29% of the million person-year showed a powerful predictive population studied in ICS in comparison with value of systolic and diastolic blood pressure for 31.3% in Americans in 1999–2000.33 Hypertension vascular death down to levels of 115 mm Hg and in both univariate and multivariate analysis had the 75 mm Hg, respectively.35 A recent analysis of the same most powerful effect on CVD events with HR of 3.4 study evaluated the joint importance of cholesterol and

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 552 hypertension on vascular mortality, and revealed that Conflict of interest TC was positively associated with IHD mortality in both middle and old age and at all blood-pressure levels.36 The authors declare no conflict of interest. Our findings describing the combined effects of risk factors are in-line with other large cohorts with long-term follow-up. For instance with the 38 year Acknowledgements follow-up of 19 000 men in the Whitehall study that documented a threefold higher rate of dying from This cohort study was conducted by ICRC affiliated vascular disease, a twofold higher rate of dying from with the Isfahan University of Medical Sciences. We non-vascular causes and a nearly 10-year shorter life are thankful to the team of the ICRC, Isfahan expectancy at age 50 for those individuals with three Provincial Health Center, Najaf-Abad Health Office CVD risk factors compared with those without any and Arak University of Medical Sciences. We would risk factor.37 Such findings confirm that an approach like also to extend our sincere thanks to the for risk assessment of CVD events by considering the ICS team, especially to Mrs Mansoureh Boshtam, combined effect of risk factors instead of focusing on Mr Hossein Balouchi, Dr Hossein Heidari and single risk factors would account for the multi- Dr Ahmad Bahonar for their technical assistance. factorial origin of CVD, and would also be practical for the comprehensive management of patients at high risk. In our cohort, a combination of hyperten- References sion and diabetes with dyslipidemia in terms of high LDL-C levels increased the HR of CVD events, 1 http://www.who.int/chp/ncd_global_status_report/en/, whereas the corresponding figure was not signifi- accessed Oct 15 2009. cant for high TC levels alone. It is also worth 2 Winkleby MA, Robinson TN, Sundquist J, Kraemer HC. mentioning that the combination of the abovemen- Ethnic variation in cardiovascular disease risk factors tioned risk factors, that is, hypertension, diabetes among children and young adults: findings from the and high LDL-C in the presence of smoking resulted Third National Health and Nutrition Examination Survey, 1988–1994. JAMA 1999; 281(11): 1006–1013. to the highest risk. 3 Schaefer BM, Caracciolo V, Frishman WH, Charney P. Gender, ethnicity and genetics in cardiovascular disease: part 1: Basic principles. Heart Dis 2003; 5(2): Conclusion 129–143. The ICS provides an optimal base for assessing 4 Yusuf S, Howken S, Ounpuu S, Avezum A, Dans T, determinants of cardiovascular risk in Iranians, Lanas F et al. Effect of potentially modifiable risk which would likely be pertinent to other countries factors associated with myocardial infarction in 52 in the region. This study used internationally countries (the INTERHEART study): case control study. recognized standard methodology and quality con- Lancet 2004; 364: 937–952. 5 Wild S, Roglic G, Green A, Sicree R, King H. Global trol protocols both in the baseline survey and in the prevalence of diabetes: estimates for the year 2000 periodic follow-up studies. The study was con- and projections for 2030. Diabetes Care 2004; 27: ducted in a large population of urban and rural 1047–1053. residents, and also used a valid case ascertainment. 6 Naghavi M, Abolhassani F, Pourmalek F, Lakeh M, Similar to other cohort studies in various popula- Jafari N, Vaseghi S et al. The burden of disease and tions, the ICS confirmed the impact of traditional injury in Iran 2003. Popul Health Met 2009; 7:9. risk factors, notably hypertension, and the different 7 Sarraf-Zadegan N, Sayed-Tabatabaei FA, Bashardoost combinations of risk factors on CVD events. Future N, Maleki A, Totonchi M, Habibi HR et al. The analyses by gender and living area (urban vs rural) prevalence of coronary artery disease in an urban might result in different HR and impact on CVD population in Isfahan, Iran. Acta Cardiol 1999; 54(5): 257–263. events; nonetheless as a general assumption, this 8 Sarraf-Zadegan N, Boshtam M, Rafiei M. Risk factors cohort confirms the need for risk estimation in for coronary artery disease in Isfahan, Iran. Eur J Public different populations, particularly in this region Health 1999; 9(1): 41–44. where data are currently minimal. 9 Esteghamati A, Abbasi M, Alikhani S, Gouya MM, Delavari A, Shishehbor MH et al. Prevalence, aware- ness, treatment, and risk factors associated with What is known about topic hypertension in the Iranian population: the national K The impact of cardiovascular disease (CVD) risk factors survey of risk factors for non-communicable diseases varies in different populations. of Iran. Am J Hypertens 2008; 21(6): 620–626. What this study adds 10 Sarrafzadagan N, Amininik S. Blood pressure pattern K ‘Isfahan Cohort Study’ is the first long term cohort study in urban and rural areas: Isfahan Hypertension Study. that has been conducted in urban and rural areas of not J Human Hyperten 1997; 11(7): 424–428. only Iran, but also in the Eastern Mediterranean region, 11 Kelishadi R, Alikhani S, Delavari A, Alaedini F, Safaie aimed at assessing the relative risk of each or combination A, Hojatzadeh E. and associated lifestyle behaviours in of CVD risk factors among men and women aged X35 Iran: findings from the First National Non-communic- years. able Disease Risk Factor Surveillance Survey. Public Health Nutr 2008; 11(3): 246–251.

Journal of Human Hypertension The isfahan cohort study N Sarrafzadegan et al 553 12 Sarrafzadegan N, Kelishadi R, Baghaei A, Hussein Working Group on Risk and High Blood Pressure. Sadri G, Malekafzali H, Mohammadifard N et al. Hypertens 1985; 7(4): 641–651. Metabolic syndrome: an emerging public health 27 Luepker RV, Apple FS, Christenson RH, Crow RS, problem in Iranian Women: Isfahan Healthy Heart Fortmann SP, Goff D et al. Case definitions for acute Program. Int J Cardiol 2008; 131(1): 90–96. coronary heart disease in epidemiology and clinical 13 Harati H, Hadaegh F, Saadat N, Azizi F. Population- research studies: A statement from the AHA Council based incidence of Type 2 diabetes and its associated on Epidemiology and Prevention; AHA Statistics risk factors: results from a six-year cohort study in Iran. Committee; World Heart Federation Council on Epi- BMC Public Health 2009; 9: 186. demiology and Prevention; the European Society 14 Sarrafzadegan N, Baghaei A, Sadri G, Kelishadi R et al. of Cardiology Working Group on Epidemiology and Isfahan Healthy Heart Program: Evaluation of Prevention; Centers for Disease Control and Preven- comprehensive. Community-based interventions for tion; and the National Heart, Lung, and Blood none-communicable disease prevention. Prevention Institute. Circulation 2003; 108(20): 2543–2549. and control 2006; 2: 73–84. 28 Braunwald E, Antman EM, Beasley JW, Califf RM, 15 Sarraf Zadegan N, Sadri G, Malek-Afzali H, Baghaei M, Cheitlin MD, Hochman JS et al. ACC/AHA guideline Mohammadi Fard N, Shahrokhi S et al. Isfahan update for the management of patients with unstable Healthy Heart Program: a comprehensive integrated angina and non-ST-segment elevation myocardial community-based programme for cardiovascular infarction—2002: summary article: a report of the disease prevention and control. Design, methods American College of Cardiology/American Heart As- and initial experience. Acta Cardiol 2003; 58: sociation Task Force on Practice Guidelines (Commit- 309–320. tee on the Management of Patients With Unstable 16 National Institutes of Health. The practical guide Angina). Circulation 2002; 106(14): 1893–1900. identification, evaluation and treatment of overweigh 29 D’Agostino RB, Grundy S, Sullivan LM, Wilson P. and obesity in adults. NIH Publication, 9: 2000. Validation of the Framingham Coronary Heart Disease 17 Chobanian AV, Bakris GL, Black HR, Cushman WC, prediction scores: Results of a multiple ethnic groups Green LA, Izzo JL et al. National Heart, Lung, investigation. JAMA 2001; 286(2): 180–187. and Blood Institute Joint National Committee on 30 Bhopal R, Fischbacher C, Vartiainen E, Unwin N, Prevention, Detection, Evaluation, and Treatment White M, Alberti G. Predicted and observed cardio- of High Blood Pressure; National High Blood Pressure vascular disease in South Asians: application of Education Program Coordinating Committee. The FINRISK, v Framingham and SCORE models to New- Seventh Report of the Joint National Committee on castle Heart Project data. J Public Health 2005; 27(1): Prevention, Detection, Evaluation and Treatment of 93–100. High Blood Pressure: the JNC 7 report. JAMA 2003; 31 Gordon T, Garcia-Palmieri MR, Kagan A, Kannel WB, 289(231): 235–240. Schiffman J. Differences in coronary heart disease in 18 Mc Namara JR, Schaefer EJ. Automated enzymatic Framingham, Honolulu and Puerto Rico. J Chronic Dis standardized lipid analyses for plasma and lipid 1974; 27(7–8): 329–344. lipoprotein fractions. Clin Chem Acta 1987; 166: 1–8. 32 Grundy SM, Pasternak R, Greenland P, Smith Jr S, 19 Warnick GR, Benderson J, Albers JJ. Dextran sulfate- Fuster V. Assessment of cardiovascular risk by use of magnesium precipitation procedure for quantitation of multiple-risk-factor assessment equations: a statement high-density lipoprotein cholesterol. Clin Chem 1982; for health care professional from the American Heart 28: 1379–1382. Association and the American College of Cardiology. 20 Friedewald WT, Levy RI, Fredrickson DS. Estimation Circulation 1999; 100: 1481–1492. of the concentration of low-density lipoprotein cho- 33 Fiels LE, Burt VL, Cutler JA, Hughes J, Roccella EJ, lesterol in plasma without use of the preparative Sorlie O. The burden of adult hypertension in the ultracentrifuge. Clin Chem 1972; 18: 499–502. United States 1999 to 2000: a rising tide. Hypertension 21 National Institutes of Health. Third Report of The 2004; 44(4): 398–404. National Cholesterol Education Program Expert Panel 34 Brand RJ, Rosenman RH, Sholtz RI, Friedman M. on Detection, Evaluation, and Treatment of High Blood Multivariate prediction of cardiovascular disease in Cholesterol in Adults (Adult Treatment Panel III). the Western Collaborative Group Study compared to National Institutes of Health: Bethesda, MD, 2001, the findings of the Framingham study. Circulation NIH Publication 01-3670. 1976; 53: 348–355. 22 WHO draft protocol and manual of operations popula- 35 Lewington S, Clarke R, Qizilbash N, Peto R, Collins R. tion survey for cariovascular disease risk factors in the Prospective study collaboration Age specific relevance Eastern Mediterranean Region. World Health Organi- of usual blood pressure to vascular mortality: a meta- zation: Alexandria (Egypt), 1995, pp 1–35. analysis of individual data for one million adults 23 WHO. Obesity: preventing and manageing the global in 61 prospective studies. Lancet 2002; 360(9394): epidemic, WHO/NUT/98 World Health Organization: 1903–1913. Geneva, Switzerland, 1998. 36 Lewington S, Clarke R, Qizilbash N, Peto R, Collins R. 24 International Diabetes Federation. The IDF consensus Prospective study collaboration Blood cholesterol and worldwide definition of the metabolic syndrome vascular mortality by age, sex and blood pressure: a www.bibalex.org/Supercourse/metabolic/IDF. meta-analysis of individual data from 61 prospective 25 Esteghamati A, Ashraf H, Rashidi A, Meysamie A. studies with 55 000 vascular deaths. Lancet 2007; Waist circumference cut-off points for the diagnosis 370(9602): 1829–1839. of metabolic syndrome in Iranian adults. Diabetes 37 Clarke R, Emberson J, Fletcher A, Breeze E, Marmot M, Research and Clinical Practice 2008; 82: 104–107. Shipley MJ. Life expectancy in relation to cardiovas- 26 An epidemiological approach to describing risk asso- cular risk factors: 38 year follow-up of 19 000 men in ciated with blood pressure levels Final Report of the the Whitehall study. BMJ 2009; 339: b3513.

Journal of Human Hypertension