Saturday 3 June 1978
BLOOD-PRESSURE AND MORTALITY IN A RURAL JAMAICAN COMMUNITY M. T. ASHCROFT
Medical Research Council Laboratories,
University of the West Indies, Kingston 7, Jamaica and women aged 35-64 years in rural Jamaica were first examined in 1962-1963, re-examined 5 and 10 years later, and followed-up until 1976. Overall mortality in 13 years, taking into account blood-pressures at all 3 surveys, showed that mortality was significantly increased only at pressures above 180 mm Hg systolic or 110 mm Hg diastolic. It was estimated that without this degree of hypertension the total number of deaths between the ages of 45 and 69 years would have been reduced by about 17%. Unlike other reports, mortality showed no significant association with lower levels of blood-pressure; this difference may be due to a lower incidence in this community of coronary and cerebral thrombosis. The relation between blood-pressure and mortality cannot be assumed to be identical in populations with different profiles of cardiovascular pathology. 1065
ACTUARIAL1 and epidemiologicaP-4 studies have shown that the higher the blood-pressure, even at relatively low levels, the greater the mortality. These studies, however, have been undertaken in relatively affluent societies. The same relationships cannot be assumed to apply in poorer communities in which the incidence of coronary heart-disease, for which blood-pressure is only one of the potent risk factors, is much lower. The influence of blood-pressure on mortality of Jamaican men and women in such a community has been investigated prospectively over 13 years. Other aspects of the study have been described previously.5-7
Population and Methods A rural area around the village of Lawrence Tavern was defined geographically and a private census showed a population of about 8000, mainly of African (Black) origin. The steep, hilly terrain, typical of much of Jamaica, requires unusually heavy physical exertion for everyday activities, particularly by men, whose main occupation is crop farming. The diet was earlier shown to be relatively high in carbohydrate and low in fat and protein.8 200 subjects of each sex and in each of the age-decades
between 35 and 64 years were selected by stratified random sampling from the census and, in 1962-1963, asked to attend the local health-centre for examination. Those who did not attend were visited at home. 1071 of the 1200 in the sample were seen at this first examination, a participation-rate of 89.3%. 5 and 10 years later those living in the area were requested to attend for re-examination or were seen at home. At each examination, medical histories were obtained and a questionnaire on smoking habits was administered. Heights and weights were measured and relative weight (% of actual to standard weight for height) calculated by the use of international standards adapted from insurance-company data.9 Blood-pressure was measured after the subject had been seated for at least 5 minutes. For the first two examinations mainly random-zero sphygmomanometers were used but these were replaced at the third examination by orthodox instruments. At each examination 3 readings were usually taken, to the nearest even number, and their mean was recorded. Diastolic-pressure was taken at the muffling of sounds (phase 4). At the first examination venous blood was obtained for serum-cholesterol, which was estimated by a direct technique. 10 Results by this method were later" shown to average 0.54 mmol;1 above those by the indirect method used in international comparisons12 and this amount was therefore deducted.
Subjects requiring medical attention, including hypertenwere referred to local clinics. Few hypertensives, however, persisted in attending regularly for therapy; this poor attendance has not been taken into account in the analysis. A record of all deaths and of changes of address was kept until 1976, i.e. 122 13iyears after subjects were first seen. By then, 30 (5.7%) men and 48 (8.9%) women had moved out of the area but were known by relatives and friends with whom they were in contact to be alive. We excluded 3 men and 3 women, of whom we lost trace, so that the study includes 1065 subjects instead of the 1071 subjects seen at first examination. Participation-rates of those alive at second and third examinations were 86-0% and 87.9% respectively. Causes of death could not always be ascertained accurately. Reliance could not be placed on medical certification of death sives,
and relatives of the deceased were often reluctant to allow necropsies. In the absence of complete coverage of causes of death, analysis has been restricted to overall mortality. Mortality-rates were calculated by the person-year method in which blood-pressure at any year of age was taken as the mean recorded at the last examination before that year. Systolic and diastolic pressures were arbitrarily divided into four categories-low, moderate, high, and severe (<135, 135-159, 160-179, and >179 mm Hg systolic and <80, 80-94, 95-109, mm Hg diastolic respectively). Person-years were and >109 divided into five 5-yearly age-groups from 45-69 (too few observations were available outside this range for adequate interpretation of results) and mortality-rates/1000 person-years in these groups were calculated for each of the four categories of systolic and diastolic pressure. Mortality-rates were also calculated taking account of both systolic and diastolic pressures in each subject, pressure being classified in the higher category if systolic and diastolic pressures fell in different categories. A man aged 46, for example, with a blood-pressure of 120/70 at 8075
1168 TABLE I-SAMPLE
Table i shows number of subjects, participation-rates first examination, percentage of survivors whose blood-pressure was measured at second and at third examination, and total number of deaths. Knowledge of number of survivors at the end of the 13-year survey or age at death during the survey was complete for all subjects who attended the initial examination except for the 6 of whom trace was lost. Mean systolic and diastolic pressure, relative weight, and serum-cholesterol at initial examination are shown in table n. The prevalence of raised blood-pressure was high, as had been found in other studies in Jamaica. Mean pressures of men were similar to those of Whites, although lower than those of Blacks, examined in the U.S. National Survey, while mean pressures of women were of the same order as those of American Blacks but were higher than those of Whites.’3 Comparisons of blood-pressure surveys must, however, be made cautiously because many different factors affect results. For example, the use of the 5th rather than the 4th phase would have resulted in a slightly lower recorded diastolic-pressure in the U.S. survey. Mean relative weights showed that men were leaner and that women were stouter than the standards. The range of the relative weight of women was greater than that of men as shown at
*Excluding of survey.
lost before end
the first and 140/70 at the second examination 5 years later, and who died after 7 years, would contribute 4 person-years to the low-pressure category in the 46-49 age-group, 1 personyear at low pressure in the 50-54 age-group, and 2 personyears and one death to the moderate-pressure category in the 50-54 age-group.
TABLE II-MEANS AND STANDARD DEVIATIONS
(S.D.) OF SYSTOLIC I AND DIASTOLIC BLOOD-PRESSURES, RELATIVE
SERUM CHOLESTEROL AT INITIAL EXAMINATION, BY AGE AND SEX
*% of international standards of weight for height9
TABLE III-MORTALITY RATES BY AGE AND SEX ACCORDING TO FOUR BLOOD-PRESSURE CATEGORIES
*If systolic and diastolic pressures fell in different categories the higher was used.
responding categories of diastolic pressure. Mortality for these four categories of pressure, when both systolic and diastolic pressures are considered in each person, show that severe blood-pressure was clearly associated with substantially higher mortality-rates at all ages and in both sexes (table ill and accompanying figure). In the 5-year age-groups 45-, 50-, 55-, 60-, and 65-69 years, the ratios of mortality-rates in the severe category to the mortality-rates in all the other categories of pressure combined were 3.8, 4.4, 2-1, 2.4, and 2.6 for men and 8.5, 3.3, 3.7, 1.8, and 2.7 for women respectively. In order to test the association between blood-pressure and mortality for statistical significance, the total number of expected deaths in each category of blood-pressure was calculated by adding up, expected deaths in each age-interval (table IV a). The X2 test showed a significant departure (p<0-01) of observed on was deaths princifrom expected which inspection pally due to high mortality in the severe blood-pressure category. Mortality in the high-pressure category was not consistently greater than that in the two lower pressure categories, with the exception of the rates in men aged 60-64 and 65-69 years. In order to test for differences in mortality between these three categories the same procedure was repeated but, on this occasion, all data in the severe category were excluded (table iv b). The difference between observed and expected deaths was now not significant. These tests suggested that mortality-rates in the various categories of blood-pressure differed significantly only if data in the severe category rates
Mortality per 1000 person-years in 5-year age-groups of and women in four categories of blood-pressure.
standard deviations. Mean
cholesterol, particularly of men, was relatively low. For example, mean serum-cholesterol levels of Jamaican, American
and American White
4-72, 5-62, and 5.91 mmol/1 respectively; values for women were 5 -13, 3-49, and 3 34 comparable 35-44
mmol/1 respectively. 14
Few subjects smoked heavily: of 476 men whose replies were available, 24 (3-0%) smoked 20 or more and 42 (8.8%) smoked 10-19 cigarettes a day, the remainder being irregular or occasional cigarette or pipe smokers. Of 486 women whose replies were available, 41 (8-4%) said they smoked 1-5 cigarettes, but only 1 smoked at least 20 cigarettes a day; 443 (91 .2%) women were non-smokers. No consistent relations were found between mortality over 13 years and initial relative weight or serum-cholesterol, thus confirming the results of the first 5 years of follow-up.6 Too few subjects smoked for analysis to be worthwhile. Age and sex-specific mortality-rates at low, moderate, high, and severe systolic pressures as calculated by the person-year method were about the same as those in cor,
Discussion Overall mortality is often neglected in prospective studies which are designed to determine factors associated with an increased incidence of morbidity or mortality from a specific condition. Once a risk factor for a
TABLE IV-NUMBERS OF OBSERVED AND EXPECTED DEATHS BY
*If systolic and diastolic pressures fell in different categories the higher was used. ’Expected values based on mortality in all blood-pressure categories. Total men, 20-06, p<0.01; total women, =14-53, p
1170 condition has been identified, however, there may be an unwarranted tendency to assume that its possession is, in general, disadvantageous. For example, the relation between high serum-cholesterol and high incidence of coronary heart-disease is well documented in the Framingham study. It is therefore surprising, in view of the high incidence of coronary heart-disease in Framingham, that overall mortality appears to be lower at higher cholesterol levels.3 Little attempt appears to have been made to explain these findings. The recording of overall mortality should not, if possible, be omitted in reporting results, especially as it can usually be recorded more accurately than other "events". Hypertension was clearly an important cause of mortality in this community. The number of deaths related to severe hypertension was estimated by calculating the expected number of deaths in the severe-pressure category according to mortality-rates in the three lowerpressure categories and taking age-intervals into account (tableiv). By this calculation there were 30 excess. deaths observed; this excess represents 17-1% of the 1755 actual deaths. The finding that increased mortality showed a statistically significant relation with severe hypertension (> 179 mm Hg systolic or > 109 diastolic) but not with lower of contrasts with actuarial data ranges blood-pressure and with results from other epidemiological studies. At least two explanations are possible. Firstly, the graduated relation between blood-pressure levels and mortality found in actuarial studies may be partly due to the calculation of mortality-rates over long observation periods but based on initial blood-pressure only. Blood-pressure often rises with age so that the higher the pressure at initial examination, even in lower ranges, the more likely it is to rise, in a long follow-up, to a level at which death may occur. When serial bloodpressure measurements are taken into account, as in this investigation, mortality-rates, instead of appearing to increase gradually, may rise steeply at certain ranges of
blood-pressure. Secondly, differences
in the levels at which bloodinfluence on mortality may be related pressure to the relative prevalence of other risk factors for coronary and cerebral thrombosis. Moderately elevated bloodpressure may be a risk factor for these thrombotic conditions only if high serum-cholesterol, heavy smoking, and sedentary occupation are also present. In communities such as Lawrence Tavern, with a low prevalence of these risk factors, cardiovascular deaths relative to hypertension may be mainly due to cerebral hoemorrhage and cardiac or renal failure, conditions which are usually caused by severe hypertension only. Coronary and cerebral thrombosis may be less common and therefore moderately elevated blood-pressure would have little effect on mortality. Information on causes of death was too incomplete to test this hypothesis. However, among confirmed causes of death, cerebrovascular accidents were frequent and coronary heart-disease was never recorded. These observations are consistent with other evidence from Jamaica. For example, in one hospital, coronary thrombosis was rarely seen in patients from rural areas but was more common in urban and affluent patients. In a study based on necropsy findings, haemorrhage was three times more frequent than thrombosis as a cause of cerebrovascular death.16 In an
international study, raised atheromatous lesions of the coronary 17 and cerebral’* vessels were less prevalent in Jamaica than in the U.S.A. Deductions from this study about the advisability of treating patients with symptomless and uncomplicated essential hypertension must be made cautiously because overall, not cardiovascular, mortality was recorded. For example, it is in theory possible that those with high blood-pressure (160-179 systolic or 95-109 mm Hg diastolic) may have had greater mortality from cardiovascular causes than those with lower levels but this was offset by a lesser mortality from other unidentified causes, so that the overall mortality was similar in all these categories of pressure. If that was the true situation, then antihypertensive therapy might beadvantageous, but no reasonable explanation for a lower mortality from unidentified causes could be found. The decision whether or not to treat patients with essential hypertension rests on many clinical and social factors, all of which must be carefully assessed. It would be unfortunate, however, if guidelines for antihypertensive therapy, based on experiences in populations with different cardiovascular risk profiles, were to be accepted uncritically in Jamaica, where medical facilities are already over-extended. There could be no doubt that severe hypertension was an important cause of morbidity and mortality but, although the size of this investigation was limited, the findings suggest that, in general, the treatment of patients over 45 years with pressures lower than 180 systolic or 100 mm Hg diastolic would confer little benefit. Patients with high but not severe pressures, without symptoms or complications, could be examined regularly to detect any rise in pressure.
We thank Dr W. E. Miall, who initiated the survey, our medical, and field colleagues, particularly Mr A. Chung, the people of Lawrence Tavern for their friendly cooperation, and Mr A. K. Sahney for help with statistics.
Requests for reprints should be addressed to M.T.A., Medical Research Council Laboratories, University of the West Indies, Hona, Kingston 7, Jamaica W.I.
Society of Actuaries, Build and Blood-Pressure Study, vol. 1. Chicago, 1959. Paul, O. Br. heart J. 1971, 33, suppl. 116. Shurtleff, D. in Framingham Study, 18-year Follow-up, Section 30. U.S. Department of Health, Education and Welfare, 1974. 4. Miall, W. E., Chin, S. Br. med. J. 1974, ii, 595. 5. Miall, W. E., Del Campo, E., Fodor, J., Nava Rhode, J. R., Ruiz, L., Standard, K. L., Swan, A. V. Bull. Wld Hlth Org. 1972, 46, 429. 6. Miall, W. E., Del Campo, E., Fodor, J. Nava Rhode, J. R., Ruiz, L., Standard, K. L., Swan, A. V. ibid. p. 685. 7. Ashcroft, M. T. West Indian med. J. 1976, 25, 216. 8. Cruickshank, E. K., Fox, H. in Dietary Survey of a Rural Area in Jamaica. 1961, Scientific Research Council, Jamaica. Unpublished Report. 9. Jelliffe, D. B. in The Assessment of the Nutritional State of the Community, World Health Organisation, Geneva, 1966. 10. MacIntyre, I., Ralston, M. Biochem. J. Proc. 1954, 56, 43. 11. Florey, C. du V., McDonald, H., Miall, W. E., Milner, R. D. G. J. chron. Dis. 1973, 26, 85. 12. Abell, L. L., Levy, B. B., Brodie, B. B., Kendall, F. E. J. biol. Chem. 1952, 1. 2. 3.
195, 357. 13. National Health Survey—Blood-Pressure of Adults by Race and Area, United States, 1960-62. U.S. Department of Health, Education and Welfare. National Centre for Health Statistics, series 11, no. 5, 1964. 14. National Health Survey—Serum-cholesterol Levels of Adults, United States, 15. 16. 17. 18.
1960-62, ibid. series 11, no. 22, 1967. Ashcroft, M. T., Stuart, K. L. West Indian med. J. 1973, 22, 60. Cole, F. M., Cole, H. L. ibid. 1969, 18, 202. McGill, H. C. et al. Lab. Invest. 1968, 18, 498. Solberg, L. A., McGarry, P. A., Moosy, J., Tejada, C., Løken, A. C., Robertson, W. B., Donoso, S. ibid. p. 604.