Longitudinal study of heart disease in a Jamaican rural population

Bull. Org. mond. Sante 1972, 46, 695-708 Bull. Wld Hlth Org. Longitudinal study of heart disease in a Jamaican rural population * 3. Factors influencing changes in serial electrocardiograms W. E. MIALL,' E. DEL CAMPO,2 J. FODOR,2 J. R. NAVA RHODE,2 L. RUIZ,2 & K. L. STANDARD 3 A longitudinal survey of heart disease in adults in a representative rural population in Jamaica provided an opportunity to study factors influencing the progression and inci- dence of electrocardiographic abnormalities in serial tracings taken at a 5-year interval. An analysis ofchanges occurring in those with ECG abnormalities compatible with ischaemia at the first survey showed that progression from a less severe to a more severe category was greater in men than in women, in hypertensive than in normotensive subjects, and, among men, in those with the amplitude criteria of left ventricular hypertrophy. The inci- dence ofabnormal Q/QS patterns was greater in men than in women, and that of all abnor- malities suggesting ischaemia was greater in hypertensive than in normotensive persons; in men, it was greater in those with high amplitude R waves. The incidence cases showed S-T and T wave abnormalities, the great majority of which were classified as showing features compatible with ischaemia rather than with strain secondary to hypertrophy. These findings, which are discussed in terms of their possible causes, seem to confirm that much of the heart disease in this Jamaican community has features of myocardial ischaemia despite other evidence that extramural coronary vessels tend to be sparedfrom such severe occlusive atheromatous disease as is found in many other populations. A longitudinal study of heart disease in a repre- sentative inland rural population in Jamaica has shown an unexpectedly high prevalence of cardiac symptoms and electrocardiographic signs of myo- cardial disorders (Miall et al., 1972a). This popu- lation is experiencing high mortality from cardio- vascular diseases, particularly heart disease, and it was shown that the symptoms and signs found in the initial survey were associated with a poor prognosis and correlated reasonably well with the certified * This study was supported by the Medical Research Council and Ministry of Overseas Development and by a grant from the World Health Organization. 1 Former Director, MRC Epidemiology Unit (Jamaica), University of the West Indies, Kingston, Jamaica. Present address: Epidemiology and Medical Care Unit, Medical Research Council and Department of Health and Social Security, Northwick Park Hospital, Harrow, Middlesex, England. 'WHO Fellow. ' Epidemiologist, MRC Epidemiology Unit (Jamaica). Present address: Professor & Head, Department of Social & Preventive Medicine, University of the West Indies, Kingston, Jamaica. causes of death and with autopsy findings in the few cases examined post mortem (Miall et al., 1971b). In this paper we examine the roles of sex, arterial pressure, and type ofelectrocardiographic abnormali- ty in determining changes in serial electrocardiograms and examine their influence on the incidence of new abnormalities. METHODS Altogether 1 067 adults aged 35-64 years, 526 men and 541 women, had been investigated in the initial survey and 854 of these were reexamined by the same techniques 5 years later. A further 35 were examined electrocardiographically only at the fol- low-up survey. Including the latter, 94.70% of the men and 96.4% of the women still living in the area were reinvestigated. Electrocardiograms were read by one highly trained technician, experienced in reading according to the criteria of the Minnesota Code, and were also read by two other observers familiar with the Code. Where 2848 695- W. E. MIALL ET AL. Fig. 1. Examples of S-T and T wave changes classified as primary (a-d) and secondary (g & h). Example (e) shows a juvenile T wave and (f) shows the pattern resulting from the administration of digitalis. disagreement occurred an agreed classification was reached in consultatien. A careful comparison of paired ECGs was undertaken whenever changes in codes were recorded in the serial tracings. The 245 tracings classified at one or other sur- vey as showing S-T segment depression or T wave abnormalities were also read in duplicate by one of us (L.R.) without reference to knowledge of age, sex, or clinical status. They were classified as showing either primary (generally ischaemic) or secondary (systolic overload/strain) patterns according to com- monly used clinical criteria. Fig. 1 illustrates exam- ples of changes that were recorded as primary (a-d) and secondary (g and h), secondary patterns being defined as those occurring in opposition to an increase in QRS area. Such classification was made easier by the examination of serial electrocardio- grams. Patterns classified as "juvenile" (e) and those due to digitalis (f) are also shown. Abnormali- ties not classifiable in these ways, for example iso- electric T waves, were considered nonspecific. Abnormal Q waves and S-T and T wave changes were also read for location and classified as lateral if present in any of the leads I, avL, and V5 and V6 or as " other" (i.e., posterior or anteroseptal) if in II, III, avF, and V1l4, and not in lateral leads. A total of 46 men and 71 women who had initially shown ECG abnormalities compatible with myo- cardial ischaemia were followed up. Their ECGs had shown abnormal Q/QS patterns (Minnesota code 11_3), S-T depression (41-3), T wave inversion, or flattening (51-3). It is this group that was particularly selected for the study of factors influencing ECG changes; their age structure at the first survey is shown in Table 1. In the study of 5-year incidence we are particularly concerned with ECG changes compatible with myo- cardial ischaemia and have excluded the 46 men and 71 women mentioned above whose ECGs showed such abnormalities at the first survey; this left base- line populations of 387 men and 385 women for the study of the incidence of new 11-3, 41-3, and 51-3 code items. Of these groups 17 men and 8 women did not have blood pressure measurements at the second survey. The population samples for these incidence analyses are also shown in Table 1. The criteria for hypertension advocated by the World Health Organization-160/95 mm Hg-have been used for arbitrarily separating " normotensive " from " hypertensive " subjects, but analyses based on distributions of arterial pressure, body build, and serum cholesterol are included. RESULTS Factors influencing ECG changes in those with abnor- malities compatible with ischaemia at thefirst survey The ECG status at both surveys of the 46 men and 71 women whose electrocardiograms had shown ( a) ( b) ( c) ( d) ( e) I ( f ) I ( g) ( h) 696 HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 Table 1. Population sample for 5-year incidence studies, Lawrence Tavern, Jamaica No. of No. with Baseline No. without Baseline Sex and age at No. avaial subjects 1 1-3, 41-3, population BP measure- populationfirst survey forfalablewp followed up and 51-3 for ECG at ment at for BP & ECG graphically first survey second survey second survey study Males 35-44 145 136 9 127 8 119 45-54 162 150 16 134 5 129 55-64 150 147 21 126 4 122 all ages 457 433 46 387 17 370 Females 35-44 155 146 12 134 3 131 45-54 170 163 30 133 3 130 55-64 148 147 29 118 2 116 all ages 473 456 71 385 8 377 11-3, 41-3, 51-3 codes at the first survey are shown in Table 2. Eight of the 46 men (17 %) and 28 of the 71 women (39%) had no abnormalities within these categories at the second survey. Q/QS pattems (11_3) were relatively stable and were present at the second sur- vey in 11 of 14 men and 7 of 9 women showing these items at the first survey, only 1 man and 1 woman having ECGs reverting to normal. S-T depression and T wave changes (41-3+51-3) were more variable. Of the 15 men and 21 women with these codes at the first survey only 8 and 6, respectively, had ECGs in the same group; 2 men and 2 women developed Q/QS patterns; 3 men and 2 women lost their S-T depression but retained T wave abnormalities; and 2 men and 11 women had ECGs that by the second survey had reverted to normal. Those with T wave abnormalities alone also showed considerable varia- bility: 4 of 17 men and 8 of 41 women developed more serious changes whereas 5 men and 16 women had ECGs that reverted to normal. Regression of ECG abnormalities was more fre- quently observed in women than in men, occurred particularly with the less severe S-T segment and T wave abnormalities, and occurred somewhat more frequently in younger than in older subjects. Table 3 shows a similar analysis according to blood pressure status for the 43 men and 64 women who had these ECG abnormalities at the first survey and whose blood pressures were measured at both. Progression of ECG abnormalities was greater in hypertensive than in normotensive subjects and in men than in women; the regression to less severe codings or to normal was more frequently observed in those who were normotensive than in those who were hypertensive, and more frequently in women than in men. The same criteria for ECG progression have been used to examine the relationship between left ventri- cular hypertrophy and changes occurring in men who, at the first survey, showed 11_3, 413, 51-3 items (Table 4). Of the 17 ECGs with high amplitude R waves (31) coded at both surveys, 12 showed pro- gression and the pattern of change was independent of blood pressure status; 3 regressed to less severe categories. Only 4 of 17 without this evidence of left ventricular hypertrophy showed progression whereas in 7 the changes were to less severe grades. Changes in the type of ECG abnormalities were also analysed according to arterial pressure at each survey (Table 5). Twelve followed-up men were considered to have had primary (ischaemic) changes at the first survey, 6 of whom were normotensive and 6 hypertensive: 6 of them retained the same pattern, 2 developed secondary changes, and in 4 the abnormality developed nonspecific features. None 2 697 698 W. E. MIALL ET AL. O~~~~~~~~~ C4 oC OD )o OD 0 0 0 _r co o C1N Y OD CY) N r a) .(D LOCC~~~~ E ODU m rDq - v ( ao z°Lbo ca> ._ -qs -HII 6 CZ r. N dC _C ZM NU> co C,)~~~~~~~zC C) ~~~~0 ~~~~~~~~~LON co N N Cj H . - t eH -oLo) 0D 0 ) 6 0) co Ic LO OD CD 0 0~~~~ Co 0 0 .CU. CD CD~~~~~~~~~~~~II C)~~~~~~~6m I' In co .0~~~~~ CD~~~~~~~~~~c II~~~ ~ N (C) LO i~+ C)- C0 NIHN -'- 0 CD U,)'C 6 N1 (C) lg 0) 0 co co N CD co~~~~~~~DCD * 0~~~~~~~~~~~~~~~~~~1 Cu *~~~~~~0 0 0c HI u.w~~~~~R L HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 0~~~~~~ E 0 co ) N (V 00 a) 0~~~~~ 0 LE COLO coo~ > Nv 0 F-c -a.) co ~ 404 0 Cu U >f 0 + _j :30 ~~N IN N > to 7m, ~, +-H C E o~~ ~ ~ ~ ~~L 6 C oL z 0D a) >0 0 CD )LO ~ ~ ~ ~ ~ cn c) Y U,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~l E >- >0+ ~~ ~ ~ ~ ~ ~ oL z~~~~~~~~~~~~~~~~~~~~~~~~~~~~~c Cui- 0 -H ID .L ~ ( c r ~ ~ ~ L 699 700 W. E. MIALL ET AL. Table 4. Relationship between high amplitude R waves (31) and changes in ECGs with Minnesota Code items 11-3, 41-3, and 51-3 for males according to blood pressure status, Lawrence Tavern, Jamaica Second survey: ECG findings a High amplitude R Normotensive at either first or Hypertensive at Total wave codes (31) both surveys secondrfisurve both surveysToa + No - + No - + No - + No - change change change change 3i at both surveys 5 1 2 3 - - 4 1 1 12 2 3 3i atfirstonly - 1 2 - 1 - - 1 - - 3 2 3i at second only - - 1 1 - - - 1 - 1 1 1 3i at neithersurvey 2 3 6 2 - 1 - 3 - 4 6 7 a + = progression; - = regression. reverted to normal. Eight men were judged to have The ECGs of women showed more regression to had secondary changes, with or without ischaemic normal: 8 of 21 initially with primary changes, 1 of 6 features, 4 retained this pattern, 3 reverted to normal, with combined primary and secondary features, and and 1 developed nonspecific features. Of the 12 who 17 of 32 with nonspecific changes showed no abnor- had ECGs with nonspecific features at the outset, mality of these types at the follow-up survey. 4 developed primary changes, 2 developed combined Tracings with voltage criteria of left ventricular primary and secondary characteristics, and 4 reverted hypertrophy (3,) showed abnormalities more often in to normal; only 2 remained unchanged. the lateral leads than elsewhere and this was true Table 5. Changes in classifications of types of ST and T abnormalities between surveys Type of abnormality at second survey a Type of abnormality Normotensive at Hypertensive at either Hypertensive at Grand at first survey both surveys first or second survey both surveys total P S±P NS N Total P S±P NS N Total P S±P NS N TotalNo. No. No. Males primary 3 2 1 - 6 - - - - - 3 - 3 - 6 12 secondary i primary - - 1 2 3 - 3 - - 3 - 1 - 1 2 8 non-specific - - 2 3 5 3 - - 1 4 1 2 - - 3 12 total 3 2 4 5 14 3 3 - 1 7 4 3 3 1 11 32 Females primary 6 - 2 4 12 3 - 1 2 6 1 - - 2 3 21 secondary + primary - 2 - - 2 - 2 - 1 3 - 1 - - 1 6 nonspecific - - 1 8 9 1 - - 3 4 4 3 6 6 19 32 total 6 2 3 12 23 4 2 1 6 13 5 4 6 8 23 59 a P = primaryabnormality; NS = non-specific abnormality; S±P = secondarywith orwithout primary; N = no 11-3,41-3, or 51-3 item. HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 Table 6. Incidence cases of ECG items 11-3, 41-3, or 51-3, by age and sex at the 5-year follow-up, Lawrence Tavern, Jamaica Type of ECG abnormality at follow-up survey Sex & age at first survey No. at risk Q/QS with S-T or T Q/QS patterns alone: S-T & T changes: T wave(years) changes: changes alone: Total No. 11-3 + 41-3 or 51-3 11-2 13 41-3 + 51-3 51-2 53 Males 35-44 127 2 - 3 1 - 4 10 45-54 134 - 1 4 4 - 4 13 55-64 126 1 - 4 3 4 10 22 all ages 387 3 1 11 8 4 18 45 Females 35-44 134 1 - 2 2 2 8 15 45-54 133 - - 2 4 3 6 15 55-64 118 - 1 2 2 4 7 16 all ages 385 1 1 6 8 9 21 46 for both hypertensive and normotensive subjects. Among followed-up men who showed code 3, at either the first or the second survey 4 of 5 Q/QS patterns were in lateral leads indicating left ventri- cular damage, whereas among 10 men without 31 at either survey only 2 were in lateral leads, the remain- ing 8 indicating posterior or anteroseptal abnormali- ties. Similarly S-T and T wave changes almost al- ways involved lateral leads if 31 had been coded- 18 of 21 cases-but involved leads other than lateral leads in 5 of 11 ECGs without a 3, code. These findings also seemed independent of blood pressure status. Factors influencing the incidence of ECG abnormali- ties compatible with ischaemia Altogether 45 men and 46 women acquired ECG abnormalities with Minnesota Code categories 1_, 41-3, and 51-3 during the 5-year interval, having had no such abnormalities at the original survey. They are shown according to type of ECG abnormality, age, and sex in Table 6. Fifteen men and 8 women developed abnormal Q/QS patterns, which in 9 of the men and 4 of the women were restricted to lead avL. None of these had developed angina though one of the women had given a history of angina previously; likewise none gave a history of infarction. Those with Q/QS patterns involving other leads more frequently showed symptoms; 2 of the 6 men had acquired angina and another had suffered a severe infarction; 1 of the 4 women had acquired angina and another had previously complained of it. Eight men and 8 women developed S-T depression with T wave changes, and of these 1 man and 2 wo- men had angina. T wave inversion alone was more commonly acquired by women than by men, and none of the 4 men, but 4 of the 9 women, complained of effort pain. Flat T waves were acquired by 18 men and 21 women, of whom only 1 man had symptoms of angina. The incidence of new 113, 41-3, and 51-3 abnor- malities was three times as great in hypertensive as in normotensive men and twice as great in hyper-' tensive as in normotensive women (Table 7). Those whose blood pressures equalled or exceeded 160/95 mm Hg at one survey but not at both surveys had intermediate incidence rates in both sexes. Among the 60 men with hypertension at only one of the surveys were 19 whose pressures had fallen; 7 of these 19 men had acquired ECG abnormalities and it is possible that the myocardial damage contributed to the change in blood pressure. The 83 women in this group included 24 whose pressures had fallen but only 3 of these had acquired ECG " ischaemic" patterns. Though the incidence in both sexes was 701 702 W. E. MIALL ET AL. Table 7. Age-adjusted 5-year incidence rates of ECG items 11-3, 41-3, and 51-3 according to age, sex, and blood pressure status, Lawrence Tavern, Jamaica Normotensive a at Hvpertensive b at either Hypertensive at Total both surveys first or second survey both surveys survey (years) No. at Incidence cases No at Incidence cases No at Incidence cases No at Incidence cases risk No. % risk No. % risk No. % risk No. % Males 35-44 102 7 6.9 8 1 12.5 9 2 22.2 119 10 8.4 45-54 90 4 4.4 25 5 20.0 14 3 21.4 129 12 9.3 55-64 70 7 10.0 27 8 29.6 25 7 28.0 122 22 18.0 all ages 262 18 7.1 60 14 20.7 48 12 23.9 370 44 11.9 Females 35-44 101 8 7.9 21 5 23.8 9 2 22.2 131 15 11.5 45-54 72 6 8.3 27 4 14.8 31 5 16.1 130 15 11.5 55-64 45 6 13.3 35 2 5.7 36 7 19.4 116 15 12.9 all ages 218 20 9.8 83 11 14.8 76 14 19.2 377 45 12.0 a Normotensive-SP < 160 and 95 mm Hg. b Hypertensive-P > 160 and/or 95 mm Hg. higher in hypertensive than in normotensive subjects, 18 of the 44 new cases among men and 20 of the 45 among women occurred in people whose pressures were below 160/95 mm Hg at each survey. Table 8 shows the age-adjusted incidence of S-T and T wave abnormalities, classified by type of lesion. The ECGs of 16 men and 16 women showed new S-T and T wave changes that could not be classed as either primary or secondary patterns, and were considered nonspecific. Eighteen of the remain- ing 21 new cases in men and 18 of 22 in women were classified as primary changes probably due to myo- Table 8. Age-adjusted 5-year incidence rates of ECG items 11-3, 41-3, and 51-3 according to. blood pressure status and type of abnormality-primary (ischaemic) and secondary (systolic overloading/strain)-Lawrence Tavern, Jamaica Normotensive Hypertensive c Hypertensive c at at either at Total both surveys first or second survey both surveys Males Females Males Females Males Females Males Females (262) (218) (60) (83) (48) (76) (370) (377) Type of abnormality a No. % No. % No. % No. % No. % No. % No. % No. % primary 5 2.0 8 3.6 6 7.6 4 5.4 7 11.4 6 11.4 18 4.9 18 4.7, secondary - - 1 0.7 - - 1 1.6 1 3.7 - - 1 0.3 2 0.5 primary Esecondary - - - - 1 1.2 1 1.2 1 1.3 1 1.1 2 0.5 2 0.5 nonspecific 6 2.5 8 3.7 6 10.4 3 3.8 4 8.7 5 4.8 16 4.1 16 4.3 a For definitions see text. b Normotensive-BP < 160 and 95 mm Hg. c Hypertensive-3P > 160 and/or 95 mm Hg. HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 Table 9. Five-year incidence rates of ECG items 11-3, 41-3, and 51-3 according to initial values of systolic and diastolic pressure, serum cholesterol, Quetelet's index, and haemoglobin level, expressed in standard deviation units, Lawrence Tavern, Jamaica 1 S.D. or more 1 S.D. below mean Mean to 1 S.D. 2 S.D. above below mean to mean 1 S.D. above mean above mean mean Characteristic Sex Incidence Incidence Incidence Incidence Incidence No. cases No. cases No. cases No. cases No. cases at risk at risk at risk at risk at risk No. % No. % No. % No. % No. % systolic pressure M 60 4 6.7 178 16 9.0 108 15 13.9 33 7 21.2 8 3 37.5 F 52 2 3.9 188 22 11.7 92 11 12.0 41 6 14.6 12 5 41.7 diastolic pressure M 45 5 11.1 186 13 7.0 114 16 14.0 33 9 27.3 9 2 22.2 F 69 4 5.8 158 19 12.0 113 14 12.4 27 4 14.8 18 5 27.8 serum cholesterol M 46 7 15.2 160 18 11.3 98 9 9.2 39 6 15.4 13 - - F 46 3 6.5 157 18 11.5 99 11 11.1 35 6 17.1 15 3 20.0 Quetelet's index M 52 6 11.5 166 20 12.1 127 15 11.8 30 2 6.7 10 2 20.0 F 52 8 15.4 177 22 12.4 91 11 12.1 51 5 9.8 11 - - haemoglobin level M 41 1 3.1 128 17 13.3 165 18 10.9 41 4 9.8 3 - - F 43 6 17.6 144 11 7.6 134 19 14.2 41 6 14.6 6 - - cardial ischaemia, and 2 men and 2 women had acquired ECGs showing combined primary and se- condary features. Only 1 man and 2 women had acquired ECGs with features of systolic overload or strain alone. Recently acquired abnormalities, therefore, showed predominantly primary patterns. This contrasts with the observations in those who had already shown S-T and T wave abnormalities at the first survey, among whom 8 of 20 followed-up men and 6 of 27 women (disregarding those with nonspecific abnor- malities) had shown secondary changes at the original survey. These findings suggest that the primary or ischaemic pattern is indeed the earlier manifestation and that the changes indicating " strain" tend to develop subsequently. Newly acquired Q/QS patterns and S-T and T items were found predominantly in left ventricular leads in men, whereas in women they occurred in leads indicating damage in posterior and anterosep- tal areas as frequently as in left ventricular leads. In both sexes the sites of newly acquired abnormalities were similar in normotensive and hypertensive sub- jects. The incidence of ECG ischaemic patterns ( 41-3, 51-) was greater among men who by the follow- up survey had the voltage criteria for left ventricular hypertrophy (3,) than among men without these high-amplitude R waves. Hypertension increased the incidence, as would be anticipated, but in both normotensive and hypertensive men the attack rate was higher in the presence of left ventricular hyper- trophy. There were too few women with high ampli- tude R waves without hypertension to warrant a simi- lar analysis. Table 9 shows an analysis of the incidence of I,-s, 41-3, and 51-3 items according to initial measure- ments of systolic and diastolic pressure, serum chol- esterol, Quetelet's index (wt/ht2), and haemoglobu- lin level, using standard deviation units to allow the pooling of different age groups. It is clear that the incidence of 11-3, 41-3, and 51-3 items was strongly related to arterial pressure and in women possibly, but not significantly, related to serum cholesterol level but was independent of the other indices considered. DISCUSSION Prospective epidemiological surveys of the type described allow a study of the development of dis- orders that may clarify the interpretation of pre- valence data. These analyses of the changes in serial electrocardiograms have shown that abnormalities detected at the initial survey and coded as showing abnormal Q/QS patterns, S-T depression, or T wave changes tend to persist and to retain the same charac- teristics. Progression from a less severe to a more severe category was greater in men than in women and greater in hypertensive than in normotensive sub- jects, and among men, whether hypertensive or 703 W. E. MLALL ET AL. normotensive, the progression was greater in those with the amplitude criteria of left ventricular hyper- trophy. The incidence of Q/QS patterns was greater in men than in women, and of T wave changes in women than in men; the incidence of 11-3, 4_3, and 51-3 items together was greater in hypertensive than in normotensive persons, and among men was in- creased in those with high-amplitude R waves. The incidence was not significantly related to cholesterol level or body build. In both sexes the majority of cases of S-T de- pression and T wave abnormality were classified as showing features compatible with ischaemia rather than with strain secondary to ventricular hypertrophy, and they tended to retain these charac- teristics. Almost all the incidence cases showed pri- mary or ischaemic features, suggesting that progres- sion was from an ischaemic to a strain pattern. Abnormal Q/QS patterns, S-T depression, and T wave changes were usually detected in lateral leads if voltage criteria of left ventricular hypertrophy were met but were more frequently found in leads indicating interventricular septum or right ventri- cular damage when evidence of left ventricular hypertrophy was lacking. The site of abnormalities detected at the first survey, and of newly acquired abnormalities, seemed independent of blood pres- sure levels. These findings confirm that much of the heart dis- ease seen in this Jamaican community has features of ischaemia despite evidence that major coronary ves- sels tend to be spared from such severe occlusive disease as is found in many other populations, and support the conclusions drawn from the analyses of prevalence and mortality data. The possible role of impaired intramural blood supply, or small-artery disease, in contributing to the overall prevalence of myocardial disorders in Law- rence Tavern was mentioned previously (Miall et al., 1972a 1972b). James (1961, 1964, 1967, 1969, 1970) has emphasized the possible importance of the intra- mural vessels in the etiology of obscure cardiomyo- pathies and their role in myocardial infarction. The extent to which the small arteries are involved in coronary artherosclerosis is not known; some reports have suggested significant pathology in small vessels in association with conventional coronary atheroma (Saphir et al., 1956; More & Sommers, 1962; Dono- mae et al., 1962). The small coronary vessels are often narrowed in diabetes (Blumenthal et al., 1960; Siperstein et al., 1964). That small-artery disease is much commoner than generally recognized is shown by the careful autopsy study of Mitchell & Schwartz (1965), who found small-vessel pathology in 69 of 137 unselected hospital necropsies. In 18 of the 69 cases gross thickening was found in several sites. Though small-vessel involvement in Mitchell & Schwartz's series was found to be age-dependent it bore no close relationship with atheroma of the main coronary arteries and was no more prevalent in those with gross infarcts or severe coronary stenosis. Little is known about the prevalence of small-vessel pathology in different ethnic groups and the subject has received scant attention in epidemiological stu- dies of vascular disease. In the International Athero- sclerosis Project (McGill et al., 1968), for example, the possible role of the small intramural vessels was disregarded, partly, no doubt, because of the consi- derable difficulty in quantifying small-artery disease. There is some evidence to support the concept that inadequacies of intramural vasculature may be a factor of importance in rural Jamaica, but at present it is largely indirect and only compatible with this hypothesis. Several subjects who had given histories of effort pain or prolonged chest pain compatible with in- farction, or whose ECGs had shown changes sug- gestive of myocardial ischaemia, died and at autopsy showed no atheroma or mild atheroma in their major coronary arteries. The heart weights of 5 men in this category who were examined at post mortem were respectively 610 g, 450 g, 340 g, 580 g, and 625 g. Post-mortem evidence of small-artery disease was obtained in only 2 patients (one of whom had advanced atheromatous narrowing of his major coronary vessels also) but special techniques are usually needed to demonstrate it and these were not routinely applied. The details of these 2 cases are summarized below: V. I., a cultivator, aged 40 years when seen at the first survey, gave a history of angina of effort for the preced- ing 5 years. He was found to have unexplained cardio- megaly (cardiothoracic ratio, 61), his blood pressure was 140/95mm Hg, his serum cholesterol was 224 mgm/100 ml, and his ECG showed impaired A-V conduction with PR intervals up to 0.35 seconds and with blocked beats and sinus arrest, and S-T depression and T inversion in leads Vc and Vs. Nine months later he had a severe chest pain and was in bed for 2 days; 4 months later he experienced a simi- lar episode lasting one night. These were thought to be due to small myocardial infarctions. From the second such illness he had a complete SA block with a heart rate of 36. He died suddenly 3 years later, and at autopsy was found to have a heart weight of 610 g with widely 704 HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 patent main coronary arteries, biventricular hypertrophy without gross evidence of fibrosis, and no evidence of valvular disease. Histological examination revealed medial and intimal thickening of intramural vessels some of which showed cystic medial necrosis. The SA node showed fibrotic changes; the nodal artery showed severe cystic medial necrosis and was almost occluded. The AV node and bundle of His appeared normal but the left bundle branch was almost completely fibrotic. Intramural vessels in the interventricular septum showed medial hypertrophy. This case illustrates the difficulty of distinguishing symptoms due to small-vessel pathology from those due to ordinary coronary atheroma. Three other surviving men, two with angina and one with a history of possible infarction also, have clinical features very similar to those described above. G. C. was another example of a man whose death may have been largely determined by small coronary artery disease. When seen at the first survey, at age 63 years, he had no symptoms, slight cardiac enlargement, a blood pres- sure of 150/95 mm Hg, and a serum cholesterol level of 164 mg/100 ml. His ECG showed no abnormality other than unusual linearity of the ST segment in V. At age 68 years, while working in his field, he collapsed with severe chest pain confirmed electrocardiographically as being the result of a large lateral and inferior myocar- dial infarction. He died 4 months later in congestive failure and at autopsy the heart was found to weigh 480 g and to be almost completely replaced by fibrous tissue with severe atherosclerosis of the small vessels, particul- arly in the septum, and atheromatous narrowing of the major coronary vessels. Small-artery disease has been implicated in a few other patients with cardiomyopathy who died in the University hospital in Jamaica (Campbell et al., 1971). The clinical features of small-artery disease exem- plified by hereditary medial necrosis-an inherited disorder of the small coronary vessels, which devel- op medial necrosis, intimal hyperplasia, narrowing, and occlusion-have been described by James (1969). Such patients " rarely have characteristic myo- cardial infarction, but they often have angina pec- toris, arrhythmias, conduction disturbances, syn- copal attacks, and sudden death and it is probable that all these represent focal myocardial ischaemia and necrosis due to occluded small coronary arte- ries". The angina occurs despite normal, or even larger than normal, main coronary vessels and is usually but not always related to exertion and is sometimes uninfluenced by nitroglycerine. The same features have been described in other cardiomyo- pathies (Goodwin et al., 1961) and are characteristic of asymmetrical hypertrophy of the heart-the so- called hypertrophic obstructive cardiomyopathy- where angina associated with unusually large calibre major coronary vessels is believed to result from relatively inadequate small-artery circulation (Teare, 1964) and where the response to nitrites is also dis- appointing (Burchell, 1964). Not only do the symptoms in Lawrence Tavern have some of the features described but several of the unusual characteristics of the electrocardiographic findings are compatible with the hypothesis that the fault may lie in the intramural vasculature. A high proportion of the abnormal Q waves were restricted to lead avL, a finding that is compatible with localized interference to the supply from the small left coronary artery branches. The high pro- portion of QS patterns in right precordial leads suggesting anteroseptal fibrosis is compatible with interference to the supply from terminal perforating branches of the left anterior descending artery. Two critical structures of the heart, the sinus node and the atrioventricular node, derive their blood supplies entirely from small arteries. The frequency of A-V conduction defects in men in this study (5.1 % at the first survey) is higher than that described in most other cardiovascular surveys; Ostrander et al. (1965), for example, reported a prevalence of 1.40% in men aged 40-59 years in the survey of Tecumseh, Michigan. Six of the 27 men with A-V conduction defects in Lawrence Tavern gave histories of effort pain or prolonged chest pain. It is of interest that first degree A-V block was found to be a significant predictor of subsequent coronary episodes in the ana- lysis of the 5-year follow-up populations studied in seven countries by Blackburn et al. (1970). The high prevalence of delayed left atrial acti- vation, as judged by Morris's index, has been repor- ted (Miall et al., 1972a), and is discussed more fully elsewhere (Ruiz et al., unpublished data). Interatrial delay compatible with hypertrophy or fibrosis of the left atrium, was associated with an abnormal Morris's index in 53 men and 27 women in the study, of whom 17 men and 6 women also showed Minnesota Code _3, 4, 3, or 51-3 items. The prevalence of frequent premature ectopic beats, though not particularly high in the Lawrence Tavern electrocardiograms, was unusual in that the majority showed supraventricular foci, suggesting pathology in the atria or nodal areas. In over 90% of the cases coded by Blackburn et al., (1970) pre- 705 W. E. MIALL ET AL. mature beats were ventricular in origin; these also were shown to be significant predictors of subsequent coronary episodes. If small-vessel pathology makes an important contribution to the prevalence of heart disease in this community it seems unlikely that it is related to hereditary medial necrosis. Though it is possible that such communities are inbred and could manifest atypical prevalences of inherited disorders, familial factors have not been obvious among the cases studied, nor have those with unexplained heart dis- ease shown a convincing excess of associated mus- culoskeletal or neurological disorders. In men and women the prevalence, the incidence, and the progression rates of Minnesota Code 11-3, 41-8 and 51-3 items were higher in hypertensive than in normotensive subjects and in men all three rates were increased in those with the voltage criteria of left ventricular hypertrophy. Ventricular hypertrophy is common in Lawrence Tavern whether judged by electrocardiographic criteria, which are probably sub- ject to ethnic differences (Ashcroft et al., 1970), or by the heart weights of those from the study who have died. Heart size, as measured radiographically, was found to be similar to that of subjects 9 kg heavier in body weight in a comparable Welsh agricultural population. Cardiothoracic ratios were also consis- tently greater in Jamaican than Welsh adults in both sexes but the smaller thoracic diameter of people of African origin contributes to this difference and makes the cardiothoracic ratio an unsatisfactory index (Ash- croft & Miall, 1968). Occupational factors, the remarkably steep terrain, and hypertension are prob- ably the most important among many possible stimuli for cardiac hypertrophy in this population. Kannel et al. (1969, 1970) recently demonstrated the grave prognostic significance of left ventricular hypertrophy in the Framingham study. Left ventri- cular hypertrophy accompanied by S-T and T wave changes was found to be associated with a threefold excess incidence of subsequent clinically overt coro- nary disease even after adjusting for blood pressure levels. Indeed the risk of a new coronary episode was almost but not quite as great in those with this type of ECG abnormality as in those who had sur- vived an actual infarction. Subjects with the ECG voltage criteria of left ventricular hypertrophy but without S-T and T changes had a risk increased two- fold, but this was largely abolished by adjusting for differences in blood pressure. This evidence from Framingham supports the concept that the diagnosis based on voltage criteria indicates hypertrophy alone -in Framingham usually due to hypertension but in Jamaica frequently attributable to other causes- whereas the diagnosis based on voltage criteria and S-T and T wave changes indicates ischaemic myo- cardial involvement. The concept of relative ischaemia was suggested by Fishberg (1954) as an explanation for hypertensive congestive failure; he postulated that ischaemic changes were due to a failure of coronary blood flow to increase, pari passu, with the increased demands of a hypertrophied myocardium. Evidence support- ing this concept, reviewed by Mitchell & Schwartz (1965), is conflicting. It has been suggested that the compression of intramural vessels by a hypertrophied myocardium may interfere with the blood flow in these small arteries (James, 1969). The role of the intramural vascular supply in contributing to the myocardial disorders seen in this representative Jamaican rural community is at present unproved and will require a careful collaborative study of the clinical and pathological correlations. The data from Lawrence Tavern suggest that left ventricular hyper- trophy, with or without hypertension, may be one precursor of a myocardial disorder that, especially in its early stages, has many of the electrocardiographic characteristics of ischaemia, and some of the clinical and electrocardiographic features that are compa- tible with an impaired intramural circulation. The epidemiological findings with regard to pre- valence, incidence, progression, and mortality show considerable consistency. They have been found in a population where there are few of the risk factors associated with myocardial infarction in more industrialized societies and they are not necessarily incompatible with the firmly held clinical and patho- logical impressions that occlusive atheroma and classical infarction are relatively infrequent in Jamaica. ACKNOWLEDGEMENTS This prolonged study required the active cooperation of many people and we are particularly grateful to the people of Lawrence Tavern, Jamaica, for their cheerful assistance. We are also indebted to our colleagues in the MRC Epidemiology Unit, in particular Dr M. T. Ashcroft, Mrs P. Desai, and Mr H. G. Lovell for help, advice and 706 HEART DISEASE IN A JAMAICAN RURAL POPULATION. 3 707 computing assistance, and to Mrs V. Craig for much sec- retarial work. Mrs V. Jutsum kindly read and coded all the electrocardiograms. Credit for the successful response rates should go to the Unit's field research team and to Nurse Carmen Atkinson and the staff at Lawrence Tavern clinic. We are also grateful to Professor K. L. Stuart and Professor G. Bras and their colleagues in the Departments of Medicine and Pathology at the University of the West Indies for many useful discussions and for active colla- boration, and to Dr Henry Blackburn whose comments on some of the electrocardiographs first led to the hypo- thesis that has been described in this paper. R1ESUME ETUDE LONGITUDINALE DES CARDIOPATHIES DANS UNE POPULATION RURALE DE LA JAMAYQUE: 3. FACTEURS INFLUANT SUR LES MODIFICATIONS DES ELECTROCARDIOGRAMMES SUCCESSIFS Au cours de l'enquete sur les cardiopathies dans une population rurale de la Jamaique, on a etudie le r6le du sexe, de la tension art6rielle et du type d'anomalies dans les modifications affectant les traces successifs et leur influence sur l'apparition d'anomalies nouvelles. Sur les 1067 adultes examines initialement, 854 ont e revus apres 5 ans; 35 personnes ont subi le 20 examen seulement. L'analyse des modifications de l'electrocardio- gramme a porte sur 46 hommes et 71 femmes qui pre- sentaient a l'examen initial des traces compatibles avec une ischemie du myocarde (anomalies des ondes Q/QS, abaissement de S-T, anomalies de l'onde T, bloc de branche gauche complet). On a constate que les anomalies decelees lors du ler examen avaient tendance a persister avec les memes caracteristiques. L'aggravation des signes electrocardio- graphiques a e plus frequente chez les hommes que chez les femmes, chez les sujets hypertendus que chez les sujets ai tension normale et, parmi les hommes (hypertendus ou non), chez ceux dont le trace denotait l'existence d'une hypertrophie ventriculaire gauche. Des anomalies du trace sont apparues chez 45 hommes et 46 femmes qui presentaient un electrocardiogramme normal lors du ler examen. L'incidence des alterations des ondes Q/QS a e olus elevee chez les hommes que chez les femmes, celle des anomalies de l'onde T plus elevee chez les femmes que chez les hommes. On a not6 une apparition plus frequente de traces d'ischemie chez les sujets hypertendus que chez les sujets a tension normale, et chez les hommes, chez ceux qui etaient atteints d'hypertrophie ventriculaire gauche. I1 n'existait aucune correlation nette entre la survenue des anomalies electrocardiographiques et la teneur du serum en choles- terol, la valeur de l'indice de Quetelet ou le taux d'hemo- globine. Dans les deux sexes, la majorit6 des anomalies de S-T et de l'onde T ont e classees comme anomalies primaires denotant une ischemie du myocarde plut6t que comme anomalies secondaires a une hypertrophie ventriculaire. Ces observations confirment qu'une forte proportion des cardiopathies d6celees dans cette population jamal- quaine offrent des caracteristiques compatibles avec une ischemie myocardique, bien qu'il semble prouve que chez ces sujets le risque d'obliteration des grosses arteres coronaires extramurales est tres faible. II apparait que dans certains cas l'hypertrophie ventriculaire gauche, associee ou non a l'hypertension, est un trouble pre- curseur de la lesion myocardique laquelle, des points de vue clinique et electrocardiographique, presente des aspects compatibles avec des desordres de la circulation coronarienne intramurale. REFERENCES Ashcroft, M. T. & Miall, W. E. (1968) Amer. J. Epidem., 89, 161-167 Ashcroft, M. T. et al. (1970) Bull. Wld Hlth Org., 42, 205-223 Blackburn, H. et al. (1970) In: Keys A., ed., Coronary heart disease in seven countries, New York, American Heart Association Inc., pp. 154-161 (American Heart Association Monograph No. 29) Blumenthal, H. T. et al. (1960) Arch. Path., 70, 13 Burchell, H. B. (1964) In: Wolstenholme, G. E. W. & O'Connor, M., ed., Cardiomyopathies, London, Chur- chill, pp. 29-42 (Ciba Foundation Symposium) Campbell, M. et al. (1971) Brit. Heart J., 33, 193-202 Donomae, I. et al. (1962) Jap. Heart J., 3, 423-441 Fishberg, A. M. (1954) Hypertension and nephritis, Philadelphia, Lea & Febiger 708 W. E. MILL ET AL. Goodwin, J. F. et al. (1961) Brit. med. J., 1, 69-79 James, T. N. (1961) Anatomy of the coronary arteries, New York, Paul B. Hoeber James, T. N. (1964) Progr. cardiovasc. Dis., 7, 43-64 James, T. N. (1967) Amer. J. Cardiol., 20, 679-691 James, T. N. (1969) Circulation, vol. 40, No. 5, Suppl. No. 4, pp. IV-13-IV-17 James, T. N. (1970) Circulation, 42, 189-191 Kannel, W. B. et al (1969) Ann. int. Med., 71, 89-105 Kannel, W. B. et al. (1970) Ann. int. Med., 72, 813-822 McGill, H. C. et al (1968) Lab. invest., 18, 498-502 Miall, W. E. et al. (1972a) Bull. Wid Hith Org., 46, 429-441 Miall, W. E. et al. (1972b) Bull. WidHith Org., 46,685-694 Mitchell, J. R. A. & Schwartz, C. J. (1965) Arterial dis- ease, Oxford, Blackwell Scientific Publications More, B. M. & Sommers, S. C. (1962) Amer. Heart J., 64, 323-333 Ostrander, L. D. et al. (1965) Circulation, 31, 888-898 Saphir, 0. et al. (1956) Arch. Path., 62, 159 Siperstein, M. D. et al. (1964) Small blood vessel involve- ment in diabetes mellitus, Washington, D.C., American Institute of Biological Sciences Teare, R. D. (1964) In: Wolstenholme, G. E. W. & O'Connor, M., ed., Cardiomyopathies, London, Chur- chill, pp. 11-28 (Ciba Foundation Symposium)