Myocardial Infarction and Cardiovascular Risk Factors in Mexico City and San Antonio, Texas
Abstract Because the epidemiology of cardiovascular disease in Mexico is largely unknown, we examined the prevalence of cardiovascular risk factors and myocardial infarction (MI) in adults from low-income colonias in Mexico City. Data were collected as part of the Mexico City Diabetes Study, a population-based survey of diabetes and cardiovascular disease conducted between 1987 and 1992. Results were compared with those obtained from a comparable survey conducted previously among low-income Mexican Americans in San Antonio, Tex. A total of 2271 individuals between the ages of 35 and 64 years from Mexico City and 1143 adults of the same age range from San Antonio were studied. Despite being leaner and having lower levels of total cholesterol, Mexicans in Mexico City had markedly higher levels of triglycerides (P<.001) and lower levels of HDL cholesterol (P<.001) than Mexican Americans in San Antonio. MI was assessed by Minnesota-coded electrocardiograms and by a self-reported history of physician-diagnosed heart attack. In men, the prevalence of self-reported heart attack was significantly higher in San Antonio than in Mexico City (odds ratio, 5.85; P<.001), and in women, the prevalence of electrocardiogram-documented MI was significantly higher in San Antonio than Mexico City (odds ratio, 2.51; P<.001). The apparent excess of MI in San Antonio relative to Mexico City could be due to higher case-fatality in Mexico City or to a higher incidence in San Antonio.
- Received December 22, 1994.
- Accepted March 21, 1995.
Few population-based surveys of cardiovascular disease (CVD) have been conducted in Mexico. We have reported the prevalence of selected cardiovascular risk factors from a survey conducted among low-income adult residents of Mexico City.1 Among the findings from this survey were that Mexicans experience a high prevalence of hypertriglyceridemia and that the serum insulin levels were relatively high in this population, suggesting a high degree of insulin resistance. The triglyceride and insulin levels observed in Mexico were even higher than those in Mexican Americans in the United States, a population characterized as having hypertriglyceridemia, low HDL cholesterol (HDL-C), and hyperinsulinemia relative to US non-Hispanic whites.2 The higher triglyceride and lower HDL-C levels in Mexico City are compatible with their higher intake of carbohydrate (60% of total calories) and lower fat intake (18% of total calories) compared with Mexican Americans in San Antonio.1
The baseline phase of the Mexico City Diabetes Study has now been completed. The goal of this article is to describe the prevalence of CVD as documented by electrocardiogram (ECG) and by self-reported heart attack in this population as well as to update our findings regarding the prevalence of cardiovascular risk factors. These results are compared with those obtained from a comparable survey conducted among low-income Mexican Americans in San Antonio, Tex.
The Mexico City Diabetes Study was designed to provide information about the prevalence of diabetes, CVD, and CVD risk factors among low-income residents of Mexico City. The baseline component of this survey was conducted between 1987 and 1992. Six low-income colonias were identified from city census tracts; all households within these colonias were enumerated, and the age of all household residents was determined. All adults between the ages of 35 and 64 years, excluding pregnant women, were eligible to participate. Of the 3492 age-eligible individuals, 2271 (65.0%) attended a neighborhood medical clinic where they received a medical examination.
All participants were asked to fast for at least 12 hours prior to the examination. Subjects reporting that they had not fasted were rescheduled to be examined on another day. Blood samples were collected in both the fasting state and 2 hours after a standardized 75-g oral glucose load.3 Serum samples were centrifuged, divided into aliquots, and stored at −70°C. At approximately 4- to 6-week intervals, serum was shipped in dry ice to the Division of Clinical Epidemiology laboratory at the University of Texas Health Science Center, where concentrations of glucose, insulin, cholesterol, and triglycerides were determined.4 5
Diabetes was diagnosed on the basis of a 2-hour oral glucose tolerance test by using the plasma glucose criteria of the World Health Organization.3 Individuals who did not meet the plasma glucose criteria were also considered to have diabetes if they reported a history of diabetes and were currently taking either insulin or oral antidiabetic agents.
Height and weight were measured with the participant having removed his or her shoes and upper garments. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. The ratios of the waist to hip circumferences and the subscapular to triceps skin folds were measured as indicators of body fat distribution. Systolic and diastolic (5th phase) blood pressure was measured in the sitting position by using a random-zero sphygmomanometer.
Subjects were asked if they had ever been told by a physician that they had sustained a heart attack. Standard 12-lead ECGs were obtained from all subjects and analyzed by the ECG Coding Laboratory of the University of Minnesota according to the Minnesota Code criteria.6 Tracings were then classified on the basis of their Q and QS patterns as definite, possible, or unlikely myocardial infarction (MI). Definite MI included pathological Q and QS patterns (all 1.1 codes) and prominent Q and QS patterns accompanied by negative T waves (all 1.2 codes accompanied by codes 5.1 or 5.2, with the exception of 1.2.6 and 1.2.8). Possible MI included minor Q and QS patterns (all 1.3 codes and codes 1.2.6 and 1.2.8), and unlikely MI included all other codes. Internal quality control of ECG coding is conducted at the ECG Coding Laboratory, where samples of ECGs are periodically recirculated for recoding. The agreement across lead groups between independently coded ECGs is excellent as judged by a κ statistic ranging from .86 to .92.7
The prevalence of cardiovascular risk factors and MI in Mexico City was compared with that in Mexican Americans from San Antonio, Tex. The San Antonio population consisted of Mexican Americans who had participated in the baseline survey of the San Antonio Heart Study (SAHS) between 1979 and 1988. The sampling methodology for the SAHS was identical to that described for the Mexico City Diabetes Study in that a random sample of households within selected neighborhoods was deemed eligible for study. The response rate for participants of the SAHS was 65.3%. The prevalence of MI has been described in this population,8 although the present report includes only Mexican Americans who resided in the low-income barrios and were between the ages of 35 and 64 years (n=1143). The median level of education among the Mexican Americans in San Antonio was 7 years, compared with 6 years for the Mexicans in Mexico City. The medical examinations used identical protocols in San Antonio and Mexico City. The ECGs collected from San Antonio were also coded by the ECG Coding Laboratory of the University of Minnesota. ECGs were obtained from 2269 subjects from Mexico City (representing 99.9% of the study population) and 1086 subjects from San Antonio (representing 95.0% of the study population).
The prevalence of MI in Mexico City and San Antonio was estimated by using two different end points: ECG-documented criteria and self-report of a physician-diagnosed heart attack. Because of the relatively small number of confirmed MIs, the “definite” and “possible” categories were combined and considered to be an event.
Cardiovascular risk factors were compared between Mexico City and San Antonio by using the t test for continuous variables and the χ2 test for categorical variables. Because of the skewed distributions of triglycerides and insulin values, the median levels of these variables were compared between the two cities by using the Kruskal-Wallis test. A cardiovascular risk score was calculated for each subject based on the subject’s age, levels of total and HDL serum cholesterol, triglycerides, systolic blood pressure, diabetes status (present or absent), level of cigarette smoking (number of cigarettes smoked per day), and relative weight based on the Metropolitan Life Insurance Co tables. The equations used to calculate the cardiovascular risk scores were generated from the Framingham Study and were kindly provided by Dr Ralph D’Agostino (Boston University, written communication, 1989). The equations describe the 12- to 14-year cardiovascular mortality experience of men and women enrolled in the Framingham Study based on a set of cardiovascular risk factors measured at baseline. Although these equations do not necessarily predict cardiovascular mortality in San Antonio and Mexico City, we have used them to generate a summary measure of cardiovascular risk to facilitate comparisons between the two populations. We have used the Framingham risk equations to compare cardiovascular risk profiles between Mexican Americans and non-Hispanic whites in the SAHS.2
MI prevalence was compared between the two populations by using the Mantel-Haenszel procedure to adjust for age.9 Separate analyses were conducted for men and women. To determine whether city differences were influenced by diabetes status, separate analyses were also conducted for diabetics and nondiabetics. Multiple logistic regression was used to estimate the independent associations of city and diabetes status with MI.10
The study population consisted of 2271 Mexicans from Mexico City and 1143 Mexican Americans from San Antonio. A comparison of selected cardiovascular risk factors between Mexico City and San Antonio is presented in Table 1⇓. The Mexicans were slightly younger than the Mexican Americans (47 versus 50 years, respectively; P<.001). Mean levels of BMI, total cholesterol, and systolic blood pressure were higher in San Antonio than in Mexico City (all P<.001). The prevalence of non–insulin-dependent diabetes mellitus was also 1.5 to 2 times higher in San Antonio than in Mexico City. In contrast, median triglyceride levels were significantly higher and mean level HDL-C levels significantly lower in Mexico City than in San Antonio. Among nondiabetic women, the median level of fasting insulin was slightly but significantly higher in Mexico City than in San Antonio (12.6 versus 10.4 μU/mL; P<.001), although there were no significant differences in the median 2-hour insulin levels between women in the two cities nor in the fasting nor 2-hour insulin levels between men in the two cities.
The mean cardiovascular risk scores for subjects in Mexico City and San Antonio are shown in Table 2⇓. In neither men nor women were there significant differences in the risk scores between Mexico City and San Antonio.
Table 3⇓ shows the number of subjects examined and the number of cardiovascular events observed for each end point. In Mexico City, 48 individuals had evidence for an ECG-documented MI and 31 reported a history of a heart attack. In San Antonio, there were 47 ECG-documented MIs and 44 histories of a self-reported heart attack.
The prevalence of MI in Mexico City and San Antonio is shown in Table 4⇓. Definite and possible ECG-documented MIs have been combined into a single end point for this and the remaining analyses. In men, the prevalence of ECG-documented MI was slightly but not significantly higher in San Antonio than in Mexico City (5.5% versus 3.2%; P>.10 adjusted for age). However, men in San Antonio were significantly more likely than men in Mexico City to report a prior heart attack (6.4% versus 1.1%; P<.001). Among women, the prevalence of ECG-documented MI was significantly higher in San Antonio than in Mexico City (3.6% versus 1.3%; P=.002 adjusted for age), although the prevalence of self-reported heart attack was only slightly higher in San Antonio than in Mexico City (2.4% versus 1.6%; P>.30).
As expected, the prevalence of MI was higher in diabetic than in nondiabetic individuals (Table 5⇓). The one exception was for women in Mexico City, for whom the prevalence of both ECG-documented MI and self-reported heart attack was very low in both groups. Among diabetic individuals, the prevalence of ECG-documented MI was at least twice as high in San Antonio as in Mexico City, although this difference did not quite achieve significance in men (11.8% versus 5.7%; P=.08). Diabetic men from San Antonio were, however, significantly more likely to report having had a heart attack than diabetic men from Mexico City.
Among nondiabetic men there was virtually no difference in the prevalence of ECG-documented MI between the two cities, although residents of San Antonio had a significantly higher prevalence of self-reported heart attack than did their counterparts in Mexico City. Among the nondiabetic women, there was a nonsignificant excess of MI in San Antonio. In neither sex was there evidence of a city×diabetes interaction effect.
The relation between MI and city of residence was estimated by adjusting simultaneously for the effects of age, BMI, total cholesterol, HDL-C, triglycerides, systolic blood pressure, and diabetes through multiple logistic regression. Table 6⇓ shows the odds ratios (ORs) for city and diabetes status obtained through these analyses. In men, residents of San Antonio were more likely to have an MI than residents of Mexico City, although these associations were statistically significant only for self-reported heart attack (adjusted OR, 6.93; P<.001). Women from San Antonio were significantly more likely than women from Mexico City to have an ECG-documented heart attack (adjusted OR, 3.22; P=.002). After controlling for other cardiovascular risk factors, diabetes status was associated with MI in men (adjusted OR, 2.2 to 2.3; P<.05) but not women.
The present study provides some of the first epidemiological data on cardiovascular end points in Mexico City. The prevalence of MI according to ECG criteria was higher in San Antonio, Tex, than in Mexico City, although the difference was significant for women only, in whom the odds of an MI were 2.51 times higher. Mexican American men in San Antonio, however, were far more likely than men in Mexico City to report a prior heart attack (OR, 5.85), although it is possible that this difference reflects greater access to health care among the men in San Antonio. Interestingly, there were no significant differences in the proportion of women in Mexico City and San Antonio who reported a prior heart attack.
The higher prevalence of MIs in San Antonio could be due to several factors. One possibility is that the case-fatality rates following a heart attack could be higher in Mexico City than San Antonio. This could be particularly true for individuals with diabetes, since the city differences appear to be more pronounced for diabetic individuals. Higher case-fatality in Mexico City could result in a lower prevalence in Mexico City, even if the incidence of MI were higher in that city. A second possibility is that the higher prevalence of MI in San Antonio may actually reflect a higher incidence. Prospective studies are currently underway in both populations to measure incidence of MI and to assess mortality among baseline cases.
The cardiovascular risk profile of Mexicans in Mexico City is distinct from that of Mexican Americans in San Antonio. Despite a lower prevalence of obesity and lower levels of total cholesterol, Mexicans in Mexico City have markedly higher levels of triglycerides (median triglyceride levels are 25 to 50 mg/dL higher) and lower levels of HDL-C (mean HDL-C levels are 12 to 13 mg/dL lower). It has been hypothesized that insulin resistance may underlie a syndrome that includes both elevated triglyceride and low HDL-C levels.11 But obesity, diabetes mellitus, and possibly hypertension are also key features of this syndrome, and none of these latter conditions is more common in Mexico City than in San Antonio. Moreover, there are no marked differences in median insulin levels between Mexico City and San Antonio, although fasting insulin levels were slightly higher among women in Mexico City compared with women in San Antonio.
It is surprising that the distribution of body fat appears less favorable in Mexico City than in San Antonio as judged by both a higher waist-to-hip ratio and, in men, a higher subscapular-to-triceps skinfold ratio. Although it is possible that the unfavorable body fat distribution is associated with the dyslipidemia also present in the Mexico City population, we cannot rule out the possibility that the differences might be due to systematic differences between examiners in the measurements of the skin folds and waist-to-hip circumferences. Prior to the start of the Mexico City Diabetes Study, however, training sessions for the Mexico City staff were conducted by the San Antonio staff to minimize the potential for interexaminer measurement bias. Unfortunately, it has been difficult to evaluate the extent to which systematic differences between examiners in the two cities could nevertheless have influenced these comparisons.
If one assumes that the higher prevalence of MI in San Antonio reflects a true difference in the incidence of MI rather than a difference in survival rates, then it is interesting to note that the high triglyceride and low HDL-C levels in Mexico City more or less completely offset the otherwise more favorable cardiovascular risk profile of Mexicans (ie, less diabetes, less obesity, lower total cholesterol, and lower blood pressure), with the result that the net effect of all cardiovascular risk factors, as judged by the Framingham risk equations, was virtually identical in the two cities. One explanation for the lower prevalence of MI in Mexico City despite a similar risk score as San Antonio is that the Framingham risk equations may overestimate the impact of high triglycerides and low HDL-C in Mexico City, ie, that high triglycerides and low HDL-C are somehow less atherogenic in Mexico City than in the United States (or, at any rate, in Framingham). Alternatively, the frequency of other unmeasured cardiovascular risk factors may be higher in San Antonio than in Mexico City.
We have reported striking differences between the diets of these two populations, with the Mexico City diet much lower in fat and higher in carbohydrate than the San Antonio diet.1 12 Thus, we have speculated that Mexicans have carbohydrate-induced hyperlipidemia leading to their high triglyceride-low HDL pattern. Conceivably, the atherogenic potential of this dyslipidemic pattern could be quite different depending on whether it develops in the context of a high-carbohydrate (Mexico) or a high-fat (United States) diet. In this regard, it is interesting to note that high-carbohydrate feeding induces increased secretion of VLDL triglycerides but has no effect on VLDL apoB secretion.13 There is also evidence that the resulting VLDL, rich in triglycerides and relatively poor in apoB, may be less atherogenic than less buoyant VLDL (ie, VLDL with higher apoB/triglyceride ratios).14 15
A strength of the present study is that both the Mexico City and San Antonio study populations were identified through population-based surveys. Moreover, identical protocols were used in the two studies to enhance comparability of results. There are several limitations, however, in using these CVD end points for between-population comparisons. First, ECG tracings may detect silent as well as symptomatic heart attacks. Thus, no conclusions can be drawn about the clinical severity of the MIs observed. Abnormal ECGs may also revert to normal over time so that ECG abnormalities detect only recent events. Self-reported heart attacks, on the other hand, may be more sensitive in identifying symptomatic heart attacks, but some of these reports will also not represent bona fide heart attacks. This end point is also a function of medical care access since it requires a prior physician diagnosis and access to care could vary in the two communities. Finally, it should be noted that it was not possible for this study to validate either of the MI end points through a review of hospital records.
In summary, these data suggest that Mexican Americans in San Antonio experience a higher prevalence of MI than low-income residents of Mexico City. There appear to be, in addition, some striking contrasts in the distribution of cardiovascular risk factors between the two populations, with the Mexico City population having a more pronounced pattern of dyslipidemia despite being leaner than the San Antonio population. These comparisons raise two major questions: is the course of MI similar in these two populations (ie, is case-fatality higher in one population than the other)? And is the atherogenic potential of these risk factors similar in both populations? Prospective data are currently being collected to address these issues.
This work was supported by grants from the National Heart, Lung, and Blood Institute (RO1-HL24799 and R37-HL36820). The Centro de Estudios en Diabetes has also received support from the Consejo Asesor en Epidemiología and the Consejo Nacional de Ciencia y Tecnología. Administrative support is provided by the Fundación Mexicana para la Salud. Dr Valdez was supported by the American Diabetes Association mentor-based Postdoctoral Fellowship Program.
Stern MP, González C, Mitchell BD, Villalpando E, Haffner SM, Hazuda HP. Genetic and environmental determinants of type II diabetes in Mexico City and San Antonio. Diabetes. 1992;41:484-492.
Mitchell BD, Stern MP, Haffner SM, Hazuda HP, Patterson JK. Risk factors for cardiovascular mortality in Mexican Americans and non-Hispanic whites: the San Antonio Heart Study. Am J Epidemiol. 1990;131:423-433.
World Health Organization Expert Committee. Second Report on Diabetes Mellitus. Geneva, Switzerland: World Health Organization, 1980. Technical report series No. 646.
Stern MP, Rosenthal M, Haffner SM, Hazuda HP, Franco LJ. Sex difference in the effects of sociocultural status on diabetes and cardiovascular risk factors in Mexican Americans: the San Antonio Heart Study. Am J Epidemiol. 1984;120:834-851.
Haffner SM, Stern MP, Hazuda HP, Rosenthal M, Knapp JA. The role of behavioral variables and fat patterning in explaining ethnic differences in serum lipids and lipoproteins. Am J Epidemiol. 1986;123:830-839.
Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Survey Methods. 2nd ed. Geneva, Switzerland: World Health Organization; 1982:125-143.
Mitchell BD, Hazuda HP, Haffner SM, Patterson JK, Stern MP. Myocardial infarction in Mexican Americans and non-Hispanic whites: the San Antonio Heart Study. Circulation. 1991;83:45-51.
Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959;22:719-748.
Dallal GE. LOGISTIC: a logistic regression program for the IBM PC. Amer Statistician. 1988;42:272.
Reaven GM. Role of insulin resistance in human disease: Banting Lecture 1988. Diabetes. 1988;37:1595-1607.
Melish J, Le N-A, Ginsberg H, Steinberg D, Brown WB. Dissociation of apoprotein B and triglyceride production in very-low-density lipoproteins. Am J Physiol. 1980;239:E354-E362.
Brunzell JD, Schott HG, Motulsky AG, Bierman EL. Myocardial infarction in the familial forms of hypertriglyceridemia. Metab Clin Exp. 1976;25:313-320.
Brunzell JD, Albers JJ, Chait A, Grundy SM, Groszek E, McDonald GB. Plasma lipoproteins in familial combined hyperlipidemia and monogenic familial hypertriglyceridemia. J Lipid Res. 1983;24:147-155.