Journal of Human Hypertension (2001) 15, 107–112  2001 Nature Publishing Group All rights reserved 0950-9240/01 $15.00 www.nature.com/jhh ORIGINAL ARTICLE Hypertension and its determinants among adults in high mountain villages of the Northern Areas of Pakistan

SMA Shah1,2, S Luby1, M Rahbar1, AW Khan3 and JB McCormick1 1Department of Community Health Sciences, The Aga Khan University, Karachi, Pakistan; 2Department of Epidemiology, UT School of Public Health, Houston, TX, USA; 3The Aga Khan Health Services, Domiyal Link Road, Gilgit, Pakistan

We studied the prevalence and determinants of hyper- antihypertensive medication) was 15%, increasing from tension among adults in mountainous rural villages in 4% in the 18–29 year age group to 36% among persons the Ghizar district Northern Areas of Pakistan, an area 60 years of age or older. The age-standardised preva- that recently has undergone substantial economic lence of hypertension was 14% (12.5% among men and development. We selected a stratified random sample of 14% among women). There was no significant difference 4203 adults (age Ͼ18 years) from 16 villages in Punial in prevalence of hypertension in males, and in females. Valley of Ghizar district where the number of study sub- Multivariate analysis revealed that age, and higher body jects from each village was proportionate to the size of mass index (overweight and obesity) were indepen- the village. We obtained blood pressure (BP) records by dently associated with higher prevalence of hyperten- taking the mean of the second and third BP measure- sion. People with hypertension were more likely to have ment, using a standard mercury sphygmomanometer, a first-degree relative with physician-diagnosed hyper- CI 1.49, 2). Hypertension is a %95 ,1.90 ؍ and assessed risk factors for hypertension in the study tension (OR .subjects. The mean ؎ s.d. blood pressures (mm Hg) significant health problem in rural northern Pakistan were 125 ؎ 19 systolic and 80 ؎ 12 diastolic in men and The primary health care programme in the Northern ,systolic and 78 ؎ 14 diastolic in women. The Areas of Pakistan needs to address this problem 22 ؎ 125 .systolic and 78 ؎ 14 diastolic in women. The especially identifying people at risk 22 ؎ 125 mm Hg, or systolic BP у140 mm Hg or currently taking Journal of Human Hypertension (2001) 15, 107–112

Keywords: high blood pressure; Pakistan; salt intake; snuff use; high mountain villages

Introduction were hospital based.10,11 There is very little data on risk factors for hypertension within Pakistan. We Several epidemiological studies have shown that conducted this study of blood pressure in the adult hypertension is an independent risk factor for stroke 1–4 residents of the high mountain villages in the Ghizar and coronary heart disease. Hypertension con- Valley of Pakistan as this area has recently under- tinues to be a leading contributor to cardiovascular gone a rapid social and demographic transition. morbidity and mortality both in developed and 5,6 People in the area were not exposed to western developing countries. From a public health per- influence until the early 1970s when the construc- spective, not only the detection of hypertensive tion of the Karakurum highway linked Pakistan to patients but also determining the pathogenic risk China through the Northern Areas. In the past 30 factors is important for the primary prevention of years economic development has spread rapidly in hypertension and for the reduction of disease bur- these mountainous villages and currently the area’s 7 den in the community. There have been some stud- 90 000 people are exposed to the process of modern- ies about hypertension in Pakistan but most of the 8,9 isation and their lifestyle is changing rapidly. There studies were conducted among special groups or has been a rapid decline in the infant mortality rate (158 in 1986 to 40 per 1000 live births in 1995) fol- lowing the implementation of a primary health care Correspondence: Syed Mahboob Ali Shah, MBBS, MPH, PhD programme by the Aga Khan Health Services.12 With (Candidate) Department of Epidemiology University of Texas Health Sciences Center, 7900 Cambridge # 18–1H Houston TX better control of communicable disease mortality, 77054, USA. E-mail: shahȰsph.uth.tmc.edu some of the chronic diseases such as hypertension Received 22 January 2000; revised 7 July 2000; accepted 21 are emerging as a priority health problem in the pro- July 2000 gramme area. The objective of this study was to esti- Hypertension in high mountain villages of Pakistan SMA Shah et al 108 mate the prevalence and correlates of high blood ometer. Blood pressure recording took place after pressure among adults in Ghizar Valley. completion of the interview so that each participant had been in the sitting position for approximately 20 min before the first blood pressure recording was Methods performed under the standardised conditions given Study population by the US Joint National Committee on detection of hypertension.13 There were intervals of 3 min Ghizar is one of the districts in the Northern Areas between each blood pressure measurement. Each of Pakistan situated amid the Hindu Kush, the time the first, fourth and fifth phases of the Korotkoff Karakoram, and the Himalayas ranges of mountains sounds and the pulse rate were recorded. Three cuff at 8000 feet above sea level in a region which has sizes (12 × 23 cm, 12 × 28 cm, 14 × 35 cm) were used been called ‘Baam-e-Duniya’, the world’s roof. The according to the circumference of the right upper inaccessibility of the area has been a strong impedi- arm of the participant. All blood pressure data are ment to development. Salted tea is the favourite and based on the 1st and 5th phases of the Korotkoff the most commonly available beverage. Most of the sounds and on the calculation of the mean of the people use powdered iodized salt and on average a second and third blood pressure measurements on quarter tea spoon of salt is added to each cup of tea. each occasion. The room temperature was not The major administrative unit of the area, Ghizar recorded but it was warm as the study was conduc- district, is further subdivided into Yasin, Gupis, Ish- ted in the summer season of the year. We classified koman and Punial Valley. We randomly selected persons as hypertensive if he/she had a systolic Punial Valley of Ghizar district for the study, which blood pressure of 140 mm Hg or greater and a dia- included 16 villages. stolic blood pressure of 90 mm Hg or greater or cur- rently taking antihypertensive medication.13 Inter- Sampling viewers recorded body weight with the subject standing on a weighing scale without support and A complete list of the total number of households in with minimal clothing. They calibrated the weigh- each village was obtained from the Aga Khan Health ing scale with a known weight daily. Height was Services (AKHS). Due to continuous surveillance in measured to the nearest 0.5 cm by having the subject the region by the AKHS, baseline information was stand upright with bare feet against a portable available for our study. The main objective of this aluminum stand, calibrated in centimeters. Trained study was to estimate the prevalence of hyperten- national health workers obtained the other variables sion. Assuming a 95% level of confidence and an (except blood pressure and height and weight) by estimated prevalence of 10%, we needed approxi- administering a questionnaire. Interviewers used a mately 4000 people. Knowing that the average local events chart to obtain the correct age of household size was approximately 10 in this area study subjects. and approximately 50% of the population were above 18 years of age, we randomly enrolled 800 households, proportionate to the size of each village. Data analysis To select the households from each village, a map We precoded the questionnaire data where possible of the village with important landmarks was pre- for ease and accuracy of collection. We also double- pared. Shops, schools and mosques served as village checked the data for obvious errors of translation landmarks. A landmark in each of the village was before entry into the computer database. The data randomly selected serving as a starting point. In entry operators then double entered the data into the each of the starting points a soda bottle was rotated computer using Fox Pro.14 The body mass index and a household in the direction of open end of the (BMI) was calculated as weight in kg/(height)2 in m2. bottle was visited. The next house was that which We used WHO ‘old world standard population’15 to was nearest to the first household and this process estimate age-standardised prevalence. We used odds was repeated until the required number of houses ratios to evaluate the association between possible for that village was enrolled. risk factors and the hypertension, and the chi square test of independence at 5% level of significance. We ϭ Measurements entered variables significant at P 0.05 in the univ- ariate analysis into a logistic regression model, to To assure the quality of measurement the principal arrive at the most parsimonious model. In the multi- author cross-checked his measurement of blood variate model, the presence of hypertension was pressure with an experienced cardiologist at the Aga taken as the dependent variable and age, ever use of Khan University until consistency in measuring the entered variables significant at P ϭ 0.05 in the univ- blood pressure was obtained. Then three inter- asalted tea use, BMI and family history of hyperten- viewers, who had at least 2 years of experience in sion were taken as predictor variables. We divided measuring blood pressure, were recruited and age into five groups and created five dummy vari- trained to make consistent measurements of blood ables and kept the youngest age group (age Ͻ30 pressure with a standard mercury sphygmoman- years) as the reference and other age categories were

Journal of Human Hypertension Hypertension in high mountain villages of Pakistan SMA Shah et al 109 Table 1 Age and sex distribution of study subjects, Ghizar, Pakis- the hypertensives had normal blood pressure when tan, 1995 examined but were on antihypertensive medication. The prevalence of hypertension among males was Age (years) Male n (%) Female n (%) 14% (193/1406) and 15% (433/2979) among females. The age-standardised prevalence of hyper- 18–29 360 (25.6) 1101 (39.4) 30–39 256 (18.2) 605 (21.6) tension, using the age structure of a WHO standard 40–49 288 (16.2) 447 (15.9) population,15 was 13% in men and 14% in women. 50–59 268 (19.1) 376 (13.4) The prevalence of hypertension increased from 4% у60 294 (20.9) 268 (9.6) in the 18–29 year age group to 36% among persons Total 2797 (100.0) 1406 (100.0) at least 60 years of age or older. Males and females had nearly equal prevalence until the 5th decade when the prevalence in females increased. Among the study subjects only 30% (1266) had entered into the final model. After obtaining the been checked before for high blood pressure. Among main effects model, we also checked for the effect the 626 hypertensives, 60 had normal blood press- modification between the covariates. We analysed ure with current hypertensive medication. Forty data using Epi-Info16 and the Statistical Package for 17 percent of the persons with hypertension (229/566) the Social Science software. had never before been told they had elevated blood pressure. Results Tobacco use was common, especially among males (60%). Among males 21% smoked cigarettes, We approached 800 households, from which 4203 Ͼ 17% chewed snuff and 22% used both cigarettes persons aged 18 years in 790 households agreed to and snuff. Among females only 19% used tobacco participate. There were 2797 females (mean age 36 in the form of snuff (12%), cigarettes (5%) and both years) and 1406 males (mean age 42 years). Many cigarettes and snuff (2%). Less than 15% of the males were away from home either serving in the population were overweight or obese. Consumption army or living down country to study. There were of salted tea was nearly universal (98%). however more males in the older categories of age People with hypertension were more likely to (Table 1). smoke cigarettes, use smokeless tobacco (snuff), and Mean systolic and diastolic blood pressure and consume wine, to be obese or overweight or to have standard deviations are given, for the people in the had a family history of hypertension. People with sample who had a blood pressure measurement, in hypertension were less likely to use salted tea Table 2, by age and sex. Both systolic and diastolic (Table 3). Mutivariate logistic regression analysis blood pressure increased with age in both the sexes showed that age, higher BMI (Ͼ25) and family his- but the increase was more gradual in diastolic blood tory of hypertension were independent risk factors pressure. Fifteen percent of study participants for hypertension (Table 4). (626/4203) were hypertensive. One percent (60) of Discussion Table 2 Mean systolic and diastolic blood pressures by sex and age among adults aged 18 years and older, Ghizar, Pakistan, 1995 We found a substantial number of people (15%) hav- ing hypertension in this remote rural region of Paki- Sex; age Subjects Mean blood pressure (and a stan, which was considerably higher than a previous (n) s.d. ), mm Hg report (8%) from rural Punjab,18 but was more con- sistent with the preliminary reports from the Systolic Diastolic National Health Survey of Pakistan, which found a 16% prevalence of high blood pressure among Males 19 18–29 360 11.8 (13.9) 76.2 (11.7) adults. The high prevalence of hypertension in the 30–39 256 120.3 (15.2) 79.9 (10.8 area has serious public health implications as the 40–49 228 122.5 (16.4) 80.3 (11.8) risk of cardiovascular disease mortality increases 50–59 268 128.7 (24.5) 82.3 (12.4) progressively with incremental increase in blood у60 294 133.6 (20.9) 82.0 (12.5) All 1406 124.5 (19.3) 79.8 (12.1) pressure from the optimal level of less than 120 mm Hg systolic and less than 80 mm Hg diastolic Females to the highest levels of systolic and diastolic blood 18–29 1101 117.6 (14.1) 73.3 (12.6) pressure with little evidence of a threshold in risk.20 30–39 605 118.7 (14.0) 75.5 (11.6) As compared to studies in the western countries 40–49 447 128.8 (22.5) 81.5 (14.9) 50–59 376 136.0 (26.1) 84.0 (14.6) where 40% of the population over age 65 have у60 268 146.3 (30.0) 86.7 (16.0) hypertension,21 the prevalence of hypertension All 2797 124.9 (21.7) 77.8 (14.3) found in this study population was somewhat lower Total 4203 124.8 (20.9) 78.5 (13.6) (34%). Factors which could explain this difference include the relatively larger proportion of elderly as.d., standard deviation. people (eg, age Ͼ75 years) who have higher preva-

Journal of Human Hypertension Hypertension in high mountain villages of Pakistan SMA Shah et al 110 Table 3 Risk factors for hypertension in univariate analysis in Ghizar, Pakistan 1995

Characteristics Subjects (n) Hypertension n (%) OR (95% CI) P value

Gender male 1406 193 (13.7) 0.86 (0.72, 1.04) 0.132 female 2797 433 (15.4)

Age 18–29 1461 58 (03.9) 1.0 30–39 861 63 (07.3) 03.03 (2.19, 04.19) 0.000 40–49 675 128 (18.9) 07.79 (5.77, 10.51) 0.000 50–59 644 176 (27.3) 11.79 (8.71, 15.99) 0.000 у60 562 201 (35.8) 13.34 (9.53, 18.66) 0.000

BMI Ͻ20 585 70 (11.9) 1.0 20–25 2999 395 (13.2) 1.12 (0.85, 1.46) 0.428 25–30 478 117 (24.5) 2.38 (1.72, 3.30) 0.000 Ͼ30 141 44 (31.2) 3.34 (2.16, 5.16) 0.000

Use of snuff Yes 939 186 (19.8) 1.59 (1.31, 1.92) 0.000 No 3264 440 (13.5) 1.0

Salted tea use Yes 4125 604 (14.6) 0.44 (0.26, 0.72) 0.001 No 78 22 (28.2) 1.0

Smoking Yes 767 137 (17.9) 1.31 (1.06, 1.61) 0.011 No 3436 489 (14.2) 1.0

Use of wine Yes 1244 230 (18.5) 1.47 (1.23, 1.75) 0.000 No 2959 396 (13.4) 1.0

Family history of hypertensiona Yes 850 147 (17.3) 1.42 (1.14, 1.76) 0.001 No 2305 296 (12.8) 1.0

aA total of 1048 people did not know about a family history of hypertension and were treated as missing for this analysis. An odds ratio of 1 represents the reference category.

Table 4 Multivariate logistical regression analysis for association observed in western societies23,24 was similar in the of various risk factors with prevalence of hypertension present study. Prevalence of hypertension for females was lower than for males at younger ages Risk factor Odds ratio (95%CI) P but exceeded that of males after the age 35–40 years. Ͻ This sort of increase has been observed in the US Age (reference age 30 yrs) 25 30–39 2.76 (1.94, 3.92) 0.000 population, but happens at a later age. The expla- 40–49 6.40 (4.56, 8.99) 0.000 nation for this finding is at present obscure, 50–59 12.10 (8.55, 17.12) 0.000 although hormonal changes after the menopause26 у 60 13.21 (8.78, 19.86) 0.000 and selective survival27 have been advanced as Overweighta 2.01 (1.51, 2.67) 0.000 possibilities. (BMI 25–30) In this study, hypertension was observed more Obesity 2.11 (1.30, 3.40) 0.002 often in obese and overweight adults as compared (BMI Ͼ30)a to normal and underweight people. Clinical trials in A positive family history of 1.90 (1.49, 2.42) 0.000 hypertension hypertensive and normotensive persons have docu- mented that loss of excess weight reduces both sys- 28 aReference category is BMI Ͻ25. tolic and diastolic blood pressure. Thus, weight loss has the potential to be a powerful means of pre- venting hypertension. Studies have consistently lence of hypertension in the total adult population identified an association between obesity and hyper- of the western countries, and the difference in obes- tension, independent of age in clinical trials28,29 ity level between Punial Valley (2%) and western longitudinal studies30,31 and also in cross-sectional countries (12%).22 studies.32,33 If the economy continues to improve the The usual rise in blood pressure with age percentage of people in the higher BMI will be

Journal of Human Hypertension Hypertension in high mountain villages of Pakistan SMA Shah et al 111 expected to rise, which would be expected to lead Chicago Heart Association Detection Project in Indus- to a higher prevalence of hypertension in this popu- try, 1967–1972. J Chron Dis 1975; 28: 527–548. lation. 3 Garcia-Palmieri MR, Costas R. Risk factors for coronary Moist snuff is a form of smokeless tobacco, which heart disease: a prospective epidemiological study in contains nicotine, sodium, and liquorice that are Perto Rico. Progress in Cardiology, vol. 14. Philadel- 34–36 phia 1986; pp 101–190. known to raise blood pressure. In our study both 4 Dyer AR. An analysis of the relationship of systolic males (39%) and females (14%) reported using blood pressure, serum cholesterol, and smoking to 14 moist snuff, and those who used it at younger age years mortality in the Chicago Peoples Gas Company were more likely to be hypertensive in our univari- Study. J Chron Dis 1975; 28: 571–78. ate analysis. It was no longer a significant risk factor 5 1988 Joint National Committee. The 1988 Report of the (as was true for wine use) when adjusted with age. Joint National Committee on Detection, Evaluation, Given the other health hazards associated with snuff and Treatment of High Blood Pressure. Special Article. such as oral cancer,37,38 efforts to decrease its use in Arch Intern Med 1988 ; 148: 1023–1038. the population should be encouraged. 6 Beaglehole R. International trends in coronary heart A high proportion of study participants (70%) disease mortality, morbidity and risk factors. Epide- miol Rev 1990; 12: 1–15. were never ever examined for blood pressure. About 7 National High Blood Pressure Education Program 40% of the hypertensives were not aware of their Working Group Report on Primary Prevention of status. This finding has serious public health conse- Hypertension. Special Article. Arch Intern Med 1993; quences because unless these people are diagnosed, 153: 186–207. treated, and their hypertension controlled, many 8 Ahmad I. Hypertension among shopkeepers and will go on to develop serious organ damage and clerks. JPMA 1976; 26: 180–185. death due to uncontrolled blood pressure. Early 9 Ilyas M et al. Peshawar hypertension study: epidemiol- detection and treatment of hypertension will lead to ogic profile of juvenile and in-service population. decrease in stroke and other complications.39,40 JPMA 1980; 30: 174–178. Unfortunately we could not evaluate the role of 10 Haider Z, Bano KA, Shahid M. Diagnostic evaluation of hypertension: pilot study in a hypertension clinic. dietary salt in the form of salted tea on hypertension JPMA 1977; 27: 375–378. as almost all of the study participants drank salted 11 Nishtar T, Qazilbash A. A study of hypertension in tea (98%). We found rather a protective effect of Hospital. JPMA 1984; 34: 4–8. salted tea use. This inverse relation may come from 12 The AKHSPFNORTH PHC Evaluation Report 1995. the consequence that doctors advised patients with AKHSP Main Office, Domiyal Link Road, Gilgit, hypertension not to use salted tea. Another study N.As, Pakistan. carried out among the salted-tea drinking popu- 13 The fifth report of the Joint National Committee on lation in Indian-held Kashmir showed an inde- Detection, Evaluation, and Treatment of High Blood 41 Pressure (JNC V). Arch Intern Med 1993; 153: 175. pendent effect of salt intake on blood pressure. + There is good evidence from other studies also that 14 Fox Pro 2.6 (C) 1989 1994 Microsoft Corporation (Engineering Prototype). an increasing dietary salt intake is partly responsible 42–48 15 World Health Statistics Annual. The World Health for the rising incidence of hypertension. Hyper- Organization: Geneva, 1993. tension is a major health problem in this com- 16 Dean AG et al. Epi Info, Version 6: a word processing, munity. This study suggests that control pro- database, and statistics program for epidemiology on grammes focusing on preventing obesity and microcomputers. Centers for Disease Control and Pre- identifying people with family history of hyperten- vention, Atlanta: Georgia, USA, 1994. sion might be particularly effective in reducing total 17 Statistical Package for the Social Science/PC 3.0 cardiovascular risk. Chicago: SPSS, Inc. (444 N. Michigan Ave). 18 Manzar S, Khatoon AB, Iqbal M, Rehan N. Pattern of blood pressure and prevalence of hypertension and Acknowledgements diabetes mellitus in various regions of Pakistan. Pakis- tan J Med Res 1990; 29: 140–150. Supported by the Seed Grant No 02410 from the 19 Pakistan Medical Research Council, National Health Dean’s office the Aga Khan University, Karachi and Survey of Pakistan, 1990–94. moral and material support by the Aga Khan Health 20 Neaton JD, Wentworth D. The Multiple Risk Factor Services and the government health staff for North- Intervention Trial Research Group. Serum cholestrol, ern Areas, Pakistan. We thank A Rashid Khan and blood pressure, cigarette smoking, and death from cor- Rizwan, medical students from the Aga Khan Uni- onary heart disease. Arch Intern Med 1992; 152: 56–64. 21 Cornoni-Huntley J, LaCroix Az, Havlik RJ. Race and versity, for their help to conduct the survey. sex differentials in the impact of hypertension in the United States. The National Health and Nutrition References Examination1. Epidemiologic follow up study. Arch Intern Med 1989; 149: 780–788. 1 Stamler J, Neaton JD, Wentworth DN. Blood pressure 22 Najjar MF, Rowland M. Anthropometeric reference systolic and diastolic, and risk of coronary heart dis- data and prevalence of over-weight, United States, ease. Hypertension 1989; 13 (Suppl 1): 2–12. 1976–80. Vital Health Stat [ii]. October 1987; no. 238. 2 Stamler J et al. Multivariate analysis of the relationship DHHS Publication No. PHS 87–1688. of seven variables to blood pressure. Findings of the 23 Havlik RJ et al. Evidence for additional blood pressure

Journal of Human Hypertension Hypertension in high mountain villages of Pakistan SMA Shah et al 112 correlates in adults 20–56 years old. Circulation 1980; 37 Goud-ML et al. Epidemiological correlates between 61: 710–715. consumption of Indian chewing tobacco and oral can- 24 Editorial. Why does blood pressure rise with age? cer. Eur J Epidemiol 1991; 7: 93–97. Lancet 1981; I: 289–290. 38 Sankaranarayanan R et al. Tobacco chewing, alcohol 25 US Department of Health and Human Services. Health and nasal snuff in cancer of the gingiva in Kerala, United States National Centre for Health India. Br J Cancer 1989; 60: 638–643. Statistics/CDC, 1996. 39 Cumming KM et al. Prevalence, awareness, treatment 26 Green MS, Jucha E, Luz Y. Inconsistencies in the corre- and control of hypertension in the inner city. Pre Med lates of blood pressure and heart rate. J Chron Dis 1982; II: 571–582. 1986; 39: 261–270. 40 Wilber JA. The problem of undetected and untreated 27 Schmieder RE, Messerli FH, Ruddel H. Risks for hypertension in the community. Bull NY Acad Med arterial hypertension. Cardiol Clinic 1986; 4: 57–66. 1973; 49: 10–20. 28 The Trials of Hypertension Prevention Collaborative 41 Mir MA, Mir F, Khosla T, Newcombe R. The relation- Research Group. The effects of non pharmacologic ship of salt intake and arterial blood pressure in salted- interventions on blood pressure of persons with high tea drinking Kashmiris. Int J Cardiology 1986; 13: normal levels. Results of the trials of hypertension pre- 279–288. vention phase-1. JAMA 1992; 267: 1213–1220. 42 Hypertension Detection and Follow-up Programme 29 Scholtee DE, Stunkand AJ. The effect of weight Cooperative Group. Blood pressure studies in 14 com- reduction on blood pressure of 302 obese patients. munities: a two stage screen for hypertension. JAMA Arch Intern Med 1990; 150: 1701–1704. 1977; 237: 2385–2391. 30 Neser WB et al. Obesity and hypertension in a longi- 43 Australian National Health and Medical Research tudinal study of black physician. J Chron Dis 1986; Council Dietary Salt Study Management Committee. 150: 1701–1704. Fall in blood pressure with moderate reduction in 31 Hsn PH, Mathewson FAL, Rokin SW. Blood Pressure dietary salt intake in mild hypertension. Lancet 1989; and Basal Metabolic Index (BMI) Pattern: A longitudi- I: 399–402. nal study. J Chron Dis 1977; 30: 93–113. 44 Elliot P. Observational studies of salt and blood press- 32 Dyer AR, Elliot P, Shipleny M. For the INTERSALT ure. Hypertension 1991; 17 (Suppl): 13–18. Cooperative Research Group. Body mass index versus 45 Cutler JA et al. An overview of randomized trials of height and weight in relation to the blood pressure: sodium reduction and hypertension. Hypertension findings of the INTERSALT study. AMJ Epidemiol 1991; 17 (Suppl 1): 127–133. 1990; 131: 589–596. 46 INTERSALT Cooperative Research Group. an inter- 33 Stamier R et al. Weight and blood pressure: findings in national study of electrolyte excretion and blood hypertension screening of 1 million Americans. JAMA pressure. Results of 24-hour urinary sodium and pot- 1978; 240: 1607–1610. assium excretion. BMJ 1988; 297: 319–328. 34 Hampson NB. Smokeless is not salt less. N Engl J Med 47 Oliver WJ, Cohen EH, Neil JV. Blood pressure, sodium 1985; 312: 919–920. intake, and sodium related hormones in the Yamom- 35 Stewart P, Wallace A, Valentino R. Mineralocorticoid amo Indians, a “non-salt” culture. Circulation 1975; activity of liquorice: 11-beta hydroxysteriod dehydrog- 52: 146–151. enese deficiency comes of age. Lancet 1987; 2: 821– 48 Stewart T. Diet and hypertension. BMJ 1985; 291: 824. 125–127. 36 Moris D, Davis E, Latif S. Licorice, tobacco chewing and hypertension. N Engl J Med 1990; 322: 849.

Journal of Human Hypertension