American Journal of Epidemiology Copyright © 1992 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved
Vol. 136, No. 1 Printed in U.S.A.
Leukocyte Count and Risk of Major Coronary Heart Disease Events
Andrew N. Phillips,12 James D. Neaton,3 Derek G. Cook,1 Richard H. Grimm,4 and A. Gerald Shaper1
The association between leukocyte count and subsequent risk of major coronary heart disease events was examined using data from three prospective cohort studies— two from the United States and one from Great Britain. A total of 28,181 middle-aged men were followed for 6-12 years. A total of 1,768 men had a nonfatal myocardial infarction or died of coronary heart disease. In all three cohorts, there was a positive, statistically significant relation between baseline leukocyte count and risk of subsequent major coronary heart disease events after adjustment for age, serum total cholesterol, diastolic blood pressure, and number of cigarettes smoked per day (relative odds = 1.32 (p < 0.0001), 1.15 (p = 0.0001), and 1.14 (p = 0.003), corresponding to a 2,000/ mm3 difference in leukocyte count). The associations persisted when all nonsmokers (former smokers plus never smokers) and never smokers alone were considered and when those with evidence of preexisting coronary heart disease at baseline were excluded. Leukocyte count appears to be an indicator of a person's future risk of major coronary heart disease events. Am J Epidemiol 1992; 136:59-70. blood cell count; coronary disease; leukocyte count; myocardial infarction; prospective studies
Reports from several prospective studies of population samples have suggested that leukocyte count, or white blood cell count, is positively related to subsequent risk of myocardial infarction (1-5). Two major issues arising from these findings remain unresolved. The first concerns the extent to which the higher risk of myocardial infarction experienced by those with high leukocyte counts is due to the fact that they are
more likely to smoke cigarettes. Cigarette smokers have leukocyte counts averaging 20-25 percent higher than those of former smokers or never smokers (6, 7). Results from all of the above studies suggest that there remains an association between leukocyte count and risk of myocardial infarction after standardization, or adjustment, for cigarette smoking. However, such adjustment, which effectively pools estimates of
Received for publication May 1, 1991, and in final form January 21, 1992. 1 Department of Public Health and Primary Care, Royal Free Hospital School of Medicine, London NW3 2PF, England. 2 Current address: Academic Department of Genitourinary Medicine, University College and Middlesex School of Medicine, London W1N 8AA, England. 3 Coordinating Centers for Biometric Research, Division of Biostatistics, University of Minnesota, Minneapolis, MN 4 Division of Epidemiology, University of Minnesota, Minneapolis, MN. Reprint requests to Dr. Andrew N. Phillips, Academic Department of Genito-Urinary Medicine, University College and Middlesex School of Medicine, James Pringle House, 73-75 Charlotte Street, London W1N 8AA, England
The British Regional Heart Study is conducted by a British Heart Foundation Research Group and is also supported by the British Department of Health and the Chest, Heart, and Stroke Association. Much of this work was conducted while Dr. Andrew N. Phillips was a visiting scientist at the Coordinating Centers for Biometric Research, University of Minnesota. The authors are grateful to the principal investigators and senior staff of the clinical, coordinating, and support centers; the National Heart, Lung, and Blood Institute, and members of the Multiple Risk Factor Intervention Trial Policy Advisory Board and Mortality Review Committee. The authors thank Dr. Goya Wannamethee (Royal Free Hospital School of Medicine) for helpful comments and suggestions.
59
60
Phillips et al.
the association found within each cigarette smoking category, may not always be adequate. The method is based on the assumption that the magnitude of the association is the same in each category—for example, the same in current smokers as in never smokers. Clearly, this may not be the case. Indeed, it has been reported that the association between leukocyte count and risk of myocardial infarction was present only in current smokers who said they inhaled (2). Those researchers concluded that the association between leukocyte count and subsequent myocardial infarction is explained by the fact that leukocyte count is a better index of exposure to cigarette smoke than is reported number of cigarettes smoked per day. As the authors of these previous reports have been aware, to study whether the association between leukocyte count and subsequent myocardial infarction is truly independent of cigarette smoking, the association should ideally be studied in people who have never smoked cigarettes. However, large numbers of subjects would be required for such an exercise. If it can be demonstrated that an association that is independent of cigarette smoking exists between a raised leukocyte count and risk of myocardial infarction, a second major question arises concerning the direction of causality. Does the association arise because leukocyte count reflects the existing degree of coronary damage? This question will be difficult to resolve definitively given the data available in most prospective studies. In this paper, we present data on the relation between leukocyte count and risk of myocardial infarction from three large prospectively followed population samples from the United States and Great Britain in order to confirm the existence of a positive association and, in particular, to address the above issues. MATERIALS AND METHODS The British Regional Heart Study cohort
The British Regional Heart Study includes 7,735 men aged 40-59 years randomly se-
lected from the age-sex registers of one representative group general practice in each of 24 towns in England, Wales, and Scotland. The men were examined in 1978-1980. The criteria for selecting the towns, general practices, and subjects, as well as the methods of data collection, have been reported previously (8). Measurements. Full details of biochemical and hematologic measurements and measurement of blood pressure have been published elsewhere (9, 10). Blood samples were taken throughout the day in nonfasting subjects, were allowed to stand for 30 minutes, and were then spun for 10 minutes. All samples were stored vertically at 4°C and dispatched overnight to the Wolfson Research Laboratories (Birmingham, England) and the Department of Haematology, Queen Elizabeth Medical Centre (Birmingham, England), where biochemical and hematologic estimations were completed by noon the following day. Leukocyte count was analyzed by means of an automated cell counter (Coulter Electronics, Inc., Luton, England). A leukocyte count was done for only 7,346 of the men. Smoking. A standard smoking questionnaire was completed by each man at screening. Information on previous smoking and on current cigarette, pipe, and cigar smoking was obtained. Coronary heart disease at initial examination. The presence of coronary heart disease at initial examination ("preexisting coronary heart disease") was assessed in three ways. 1) The World Health Organization (Rose) questionnaire on chest pain, which covered both angina and possible myocardial infarction, was administered by a nurse (11). 2) A three-orthogonal-lead electrocardiogram was performed (12). 3) The subject was asked whether a doctor had ever diagnosed angina or a heart attack (myocardial infarction, coronary thrombosis) (13). The men were grouped in three groups of preexisting coronary heart disease status (14). Group I contained men with no evidence of preexisting coronary heart disease; group II comprised men with evidence of preexisting coronary heart disease short of a definite myo-
Leukocyte Count and Major Coronary Events
cardial infarction, according to either the electrocardiogram (possible or definite ischemia or possible myocardial infarction), the chest pain questionnaire (angina or possible myocardial infarction), or recall of a doctor's diagnosis of angina; and group III was men with a definite previous myocardial infarction, according to an electrocardiogram or recall of a doctor's diagnosis of a heart attack. Follow-up. All 7,735 men initially examined in 1978-1980 were followed through their primary care physicians for 8 years so that information on the occurrence of major coronary heart disease events (nonfatal myocardial infarction and coronary heart disease death) could be obtained. The men were also followed up through the British National Health Service Central Register for information on mortality. Contact has been maintained with 99 percent of the surviving cohort (15). Endpoint definition. Coronary heart disease death was defined as International Classification ofDiseases, Ninth Revision, codes 410-414 entered as the cause of death on the death certificate, without contradiction by the medical history or postmortem examination. Sudden death for which no other cause was apparent and which was certified as being due to ischemic heart disease was also included in this category. Any clinical report of a myocardial infarction which included at least two of the following manifestations was classified as a nonfatal myocardial infarction: 1) the infarction's being preceded by severe, prolonged chest pain, 2) electrocardiographic evidence of myocardial infarction, and 3) cardiac enzyme changes associated with myocardial infarction (8). The Multiple Risk Factor Intervention Trial 1 cohort
A group of 361,662 men aged 35-57 years were screened between 1973 and 1976 for eligibility status for the Multiple Risk Factor Intervention Trial, a multicenter US study of the effect of lowering coronary risk factors (16, 17). A total of 25,545 men found to be in the top 10-15 percent of risk for future
61
coronary heart disease, based on a single measurement of serum cholesterol, diastolic blood pressure, and self-reported number of cigarettes smoked daily, were invited to a second screening examination so that exclusion criteria not addressed at the initial screening could be applied (18). Those with evidence of major preexisting diseases or events (e.g., coronary heart disease, diabetes, stroke) or those who were unlikely to comply with the nutritional intervention were excluded. Of those invited, 22,080 men attended the second screening examination. After a third eligibility visit, 12,866 men were randomized to either a special intervention group or a usual care group. These 12,866 men constitute the Multiple Risk Factor Intervention Trial 1 cohort. Men in the special intervention group underwent intense intervention designed to decrease their risk factors for cardiovascular disease. Men in the usual care group were informed of their risk status and were referred to their customary source of medical care. At annual examinations, blood testing and other special tests were carried out on each participant. Earlier results from this study population on leukocyte count and risk of coronary heart disease, cancer, and all-cause mortality in the usual care group have been published (4). This analysis considers all men in the trial, whether in the special intervention group or the usual care group, since the relation between leukocyte count and risk of myocardial infarction does not differ importantly between the two groups. Measurements. Leukocyte counts were measured locally at the study centers at the beginning of the trial and at annual visits. The method used most frequently was the automated blood cell (Coulter) counter. A few centers performed the leukocyte count on-site using hemocytometers. Total cholesterol was measured in serum according to previously described methods (17, 18). Smoking. Cigarette smoking was ascertained at the first two eligibility visits, and a detailed smoking history questionnaire, including former smoking status, inhalation pattern, and use of other tobacco products, was completed at the third eligibility visit.
62
Phillips et al.
Coronary heart disease at initial examination. Men with any history or presence of coronary heart disease were excluded from the trial (18). Follow-up. The investigators followed and saw the men regularly for an average of 7 years, up to February 28, 1982, to obtain information on fatal and nonfatal myocardial infarctions and to assess risk factor changes. Endpoint definition. Deaths were systematically ascertained by clinic staff, and cause of death was assigned by a three-member panel of cardiologists. This ascertainment was made after a thorough review of hospital records, the physician's reports, interviews of relatives/household members, death certificates, and autopsy reports (if available). Nonfatal myocardial infarctions were diagnosed on the basis of either a significant serial electrocardiogram change between baseline and the annual visit or the review of medical records by cardiologists. All hospital records that were associated with heart problems were reviewed, and diagnoses were classified as "definite myocardial infarction," "possible myocardial infarction," "other cardiac event," or "noncardiac event." Definite myocardial infarction or serial electrocardiogram change was used as the nonfatal coronary heart disease endpoint. The Multiple Risk Factor Intervention Trial 2 cohort
A total of 22,080 men attended the second screening examination for the Multiple Risk Factor Intervention Trial. Leukocyte counts were taken at this visit. The 8,016 men who had a leukocyte count done at this second stage but did not participate in the trial form the third cohort. Measurements. Techniques used for leukocyte counts and measurements of total cholesterol were similar to those used for the baseline screening, as described above. Smoking. Basic data on cigarette smoking were obtained, but no distinction was made between former smokers and never smokers. Coronary heart disease at initial examina-
tion. For 34 percent of this cohort, the reason for exclusion from the trial was evidence of preexisting cardiovascular disease. This information was derived from the resting electrocardiogram, the World Health Organization (Rose) chest pain questionnaire, medical examination, or the fact that the subject had a history of using certain antihypertensive or lipid-lowering drugs (18). Follow-up. The men have been followed for an average of 12 years, up to December 1986, for mortality data only. Endpoint definition. Mortality follow-up was accomplished using the US National Death Index. A coronary heart disease death was defined as any death classified under International Classification of Diseases, Ninth Revision, codes 410-414 or 429.2. Statistical methods
Relative odds estimates in figures 2-4 were obtained by fitting a multiple logistic model. Besides smoking status, which was considered in detail, adjustment was made for age, diastolic blood pressure, and serum cholesterol. None of these latter variables had a large confounding effect. In figures 3 and 4, only three leukocyte groups are used, instead of the six shown in figure 2, because of the smaller numbers of men involved. Men with leukocyte counts below 1,000/ mm3 or above 30,000/mm3 were excluded {n = 2 in the British Regional Heart Study, n = 21 in Multiple Risk Factor Intervention Trial 1, and n = 24 in Multiple Risk Factor Intervention Trial 2). Pooled relative odds estimates were obtained by pooling the log relative odds estimates, weighted by precision (1/variance). RESULTS
Table 1 contains a summary description of the three cohorts. "Number of men" refers to the number for whom a leukocyte count was available. In total, over the three cohorts, there were 28,181 men, of whom 1,768 experienced a specified coronary heart disease event. Although we wish to focus on within-cohort rather than between-cohort
Leukocyte Count and Major Coronary Events
63
TABLE 1. Characteristics of the three cohorts used in a study of the relation of leukocyte count to myocardial infarction* MRFIT 1
BRHS
Average length of follow-up (years) Type of event
No. of ment No of events No. of nonsmokersj No. of events No. of never smokers§ No. of events % with preexisting coronary heart disease || Mean leukocyte count (/mm3) Mean age (years) Mean serum total cholesterol (mg/dl) Mean diastolic blood pressure (mmHg)
7.0
8.0 Coronary heart disease death or nonfatal myocardial infarction 7,344 485 3,566 174 1,546 56 25 7,202(1,771)11 50(6) 242 (40) 82.2(13.2)
Coronary heart disease death or nonfatal myocardial infarction 12,845 814 3,615 185 1,395 75 0
MRFIT 2
12.0 Coronary heart disease death 7,992 469 2,106 121
34
7,181 (2,092)
7,464(2,198)
46(6) 253 (37)
47(6) 254 (37)
99.2 (7.7)
99.1 (8.0)
•BRHS, British Regional Heart Study, MRFIT 1, Multiple Risk Factor Intervention Trial 1; MRFIT 2, Multiple Risk Factor Intervention Trial 2. t Subjects excluded because their leukocyte count was less than 1,000/mm3 or greater than 30,000/mm3: BRHS, n = 2; MRFIT 1,n = 21;MRFIT2, n = 24. $ Definition of a nonsmoker: BRHS and MRFIT 1—not currently smoking cigarettes, a pipe, cigars, or ciganllos; MRFIT 2—not currently smoking cigarettes. § Definition of a never smoker: BRHS and MRFIT 1—nonsmokers (see footnote 3) who have never smoked cigarettes. || BRHS—any evidence of preexisting coronary heart disease according to any of the three methods used (questionnaire, electrocardiogram, doctor diagnosis). MRFIT 2—men excluded from the trial because of the presence of cardiovascular disease. 11 Numbers in parentheses, standard deviation.
differences, some comparisons are worthy of note. The leukocyte counts were similar in the British Regional Heart Study and Multiple Risk Factor Intervention Trial 1 cohorts even though there was a higher proportion of current smokers (cigarettes, a pipe, or cigars) in the Multiple Risk Intervention Trial 1 cohort (51 percent vs. 72 percent); this is due to the fact that the men were selected because of high coronary heart disease risk, of which cigarette smoking was a component. The average leukocyte count was higher in the Multiple Risk Factor Intervention Trial 2 men. The serum cholesterol and diastolic blood pressure levels were higher in the two Multiple Risk Factor Intervention Trial cohorts, again because the men were selected on the basis of high levels of these factors. In figure 1, mean leukocyte count is shown by smoking status for the Multiple Risk Factor Intervention Trial 1 and British
Regional Heart Study cohorts. (Former smoking status was not ascertained for men in the Multiple Risk Factor Intervention Trial 2 cohort.) Leukocyte counts are lower in the Multiple Risk Factor Intervention Trial 1 cohort than in the British Regional Heart Study cohort for every smoking category. For both cohorts, the lowest levels are seen in men who have never smoked cigarettes, and the highest are seen in current cigarette smokers. Levels in former cigarette smokers are considerably lower than those in current smokers, but they appear to remain slightly higher than those in persons who had not smoked cigarettes for at least 5 years. Leukocyte counts in pipe/cigar smokers are intermediate between those of never smokers and current cigarette smokers, with secondary pipe/cigar smokers (former cigarette smokers) having higher levels than pipe/cigar smokers who had never smoked cigarettes. Thus, differences in leukocyte
64
Phillips et al.
9,000 _
8,000 .
7,000 .
2
i i i
6,000 .
5,000 Never Pipe/ Secondary Cigar pipe / cigar
£10
5-9
1-4
Vol. 136, No. 1 Printed in U.S.A.
Leukocyte Count and Risk of Major Coronary Heart Disease Events
Andrew N. Phillips,12 James D. Neaton,3 Derek G. Cook,1 Richard H. Grimm,4 and A. Gerald Shaper1
The association between leukocyte count and subsequent risk of major coronary heart disease events was examined using data from three prospective cohort studies— two from the United States and one from Great Britain. A total of 28,181 middle-aged men were followed for 6-12 years. A total of 1,768 men had a nonfatal myocardial infarction or died of coronary heart disease. In all three cohorts, there was a positive, statistically significant relation between baseline leukocyte count and risk of subsequent major coronary heart disease events after adjustment for age, serum total cholesterol, diastolic blood pressure, and number of cigarettes smoked per day (relative odds = 1.32 (p < 0.0001), 1.15 (p = 0.0001), and 1.14 (p = 0.003), corresponding to a 2,000/ mm3 difference in leukocyte count). The associations persisted when all nonsmokers (former smokers plus never smokers) and never smokers alone were considered and when those with evidence of preexisting coronary heart disease at baseline were excluded. Leukocyte count appears to be an indicator of a person's future risk of major coronary heart disease events. Am J Epidemiol 1992; 136:59-70. blood cell count; coronary disease; leukocyte count; myocardial infarction; prospective studies
Reports from several prospective studies of population samples have suggested that leukocyte count, or white blood cell count, is positively related to subsequent risk of myocardial infarction (1-5). Two major issues arising from these findings remain unresolved. The first concerns the extent to which the higher risk of myocardial infarction experienced by those with high leukocyte counts is due to the fact that they are
more likely to smoke cigarettes. Cigarette smokers have leukocyte counts averaging 20-25 percent higher than those of former smokers or never smokers (6, 7). Results from all of the above studies suggest that there remains an association between leukocyte count and risk of myocardial infarction after standardization, or adjustment, for cigarette smoking. However, such adjustment, which effectively pools estimates of
Received for publication May 1, 1991, and in final form January 21, 1992. 1 Department of Public Health and Primary Care, Royal Free Hospital School of Medicine, London NW3 2PF, England. 2 Current address: Academic Department of Genitourinary Medicine, University College and Middlesex School of Medicine, London W1N 8AA, England. 3 Coordinating Centers for Biometric Research, Division of Biostatistics, University of Minnesota, Minneapolis, MN 4 Division of Epidemiology, University of Minnesota, Minneapolis, MN. Reprint requests to Dr. Andrew N. Phillips, Academic Department of Genito-Urinary Medicine, University College and Middlesex School of Medicine, James Pringle House, 73-75 Charlotte Street, London W1N 8AA, England
The British Regional Heart Study is conducted by a British Heart Foundation Research Group and is also supported by the British Department of Health and the Chest, Heart, and Stroke Association. Much of this work was conducted while Dr. Andrew N. Phillips was a visiting scientist at the Coordinating Centers for Biometric Research, University of Minnesota. The authors are grateful to the principal investigators and senior staff of the clinical, coordinating, and support centers; the National Heart, Lung, and Blood Institute, and members of the Multiple Risk Factor Intervention Trial Policy Advisory Board and Mortality Review Committee. The authors thank Dr. Goya Wannamethee (Royal Free Hospital School of Medicine) for helpful comments and suggestions.
59
60
Phillips et al.
the association found within each cigarette smoking category, may not always be adequate. The method is based on the assumption that the magnitude of the association is the same in each category—for example, the same in current smokers as in never smokers. Clearly, this may not be the case. Indeed, it has been reported that the association between leukocyte count and risk of myocardial infarction was present only in current smokers who said they inhaled (2). Those researchers concluded that the association between leukocyte count and subsequent myocardial infarction is explained by the fact that leukocyte count is a better index of exposure to cigarette smoke than is reported number of cigarettes smoked per day. As the authors of these previous reports have been aware, to study whether the association between leukocyte count and subsequent myocardial infarction is truly independent of cigarette smoking, the association should ideally be studied in people who have never smoked cigarettes. However, large numbers of subjects would be required for such an exercise. If it can be demonstrated that an association that is independent of cigarette smoking exists between a raised leukocyte count and risk of myocardial infarction, a second major question arises concerning the direction of causality. Does the association arise because leukocyte count reflects the existing degree of coronary damage? This question will be difficult to resolve definitively given the data available in most prospective studies. In this paper, we present data on the relation between leukocyte count and risk of myocardial infarction from three large prospectively followed population samples from the United States and Great Britain in order to confirm the existence of a positive association and, in particular, to address the above issues. MATERIALS AND METHODS The British Regional Heart Study cohort
The British Regional Heart Study includes 7,735 men aged 40-59 years randomly se-
lected from the age-sex registers of one representative group general practice in each of 24 towns in England, Wales, and Scotland. The men were examined in 1978-1980. The criteria for selecting the towns, general practices, and subjects, as well as the methods of data collection, have been reported previously (8). Measurements. Full details of biochemical and hematologic measurements and measurement of blood pressure have been published elsewhere (9, 10). Blood samples were taken throughout the day in nonfasting subjects, were allowed to stand for 30 minutes, and were then spun for 10 minutes. All samples were stored vertically at 4°C and dispatched overnight to the Wolfson Research Laboratories (Birmingham, England) and the Department of Haematology, Queen Elizabeth Medical Centre (Birmingham, England), where biochemical and hematologic estimations were completed by noon the following day. Leukocyte count was analyzed by means of an automated cell counter (Coulter Electronics, Inc., Luton, England). A leukocyte count was done for only 7,346 of the men. Smoking. A standard smoking questionnaire was completed by each man at screening. Information on previous smoking and on current cigarette, pipe, and cigar smoking was obtained. Coronary heart disease at initial examination. The presence of coronary heart disease at initial examination ("preexisting coronary heart disease") was assessed in three ways. 1) The World Health Organization (Rose) questionnaire on chest pain, which covered both angina and possible myocardial infarction, was administered by a nurse (11). 2) A three-orthogonal-lead electrocardiogram was performed (12). 3) The subject was asked whether a doctor had ever diagnosed angina or a heart attack (myocardial infarction, coronary thrombosis) (13). The men were grouped in three groups of preexisting coronary heart disease status (14). Group I contained men with no evidence of preexisting coronary heart disease; group II comprised men with evidence of preexisting coronary heart disease short of a definite myo-
Leukocyte Count and Major Coronary Events
cardial infarction, according to either the electrocardiogram (possible or definite ischemia or possible myocardial infarction), the chest pain questionnaire (angina or possible myocardial infarction), or recall of a doctor's diagnosis of angina; and group III was men with a definite previous myocardial infarction, according to an electrocardiogram or recall of a doctor's diagnosis of a heart attack. Follow-up. All 7,735 men initially examined in 1978-1980 were followed through their primary care physicians for 8 years so that information on the occurrence of major coronary heart disease events (nonfatal myocardial infarction and coronary heart disease death) could be obtained. The men were also followed up through the British National Health Service Central Register for information on mortality. Contact has been maintained with 99 percent of the surviving cohort (15). Endpoint definition. Coronary heart disease death was defined as International Classification ofDiseases, Ninth Revision, codes 410-414 entered as the cause of death on the death certificate, without contradiction by the medical history or postmortem examination. Sudden death for which no other cause was apparent and which was certified as being due to ischemic heart disease was also included in this category. Any clinical report of a myocardial infarction which included at least two of the following manifestations was classified as a nonfatal myocardial infarction: 1) the infarction's being preceded by severe, prolonged chest pain, 2) electrocardiographic evidence of myocardial infarction, and 3) cardiac enzyme changes associated with myocardial infarction (8). The Multiple Risk Factor Intervention Trial 1 cohort
A group of 361,662 men aged 35-57 years were screened between 1973 and 1976 for eligibility status for the Multiple Risk Factor Intervention Trial, a multicenter US study of the effect of lowering coronary risk factors (16, 17). A total of 25,545 men found to be in the top 10-15 percent of risk for future
61
coronary heart disease, based on a single measurement of serum cholesterol, diastolic blood pressure, and self-reported number of cigarettes smoked daily, were invited to a second screening examination so that exclusion criteria not addressed at the initial screening could be applied (18). Those with evidence of major preexisting diseases or events (e.g., coronary heart disease, diabetes, stroke) or those who were unlikely to comply with the nutritional intervention were excluded. Of those invited, 22,080 men attended the second screening examination. After a third eligibility visit, 12,866 men were randomized to either a special intervention group or a usual care group. These 12,866 men constitute the Multiple Risk Factor Intervention Trial 1 cohort. Men in the special intervention group underwent intense intervention designed to decrease their risk factors for cardiovascular disease. Men in the usual care group were informed of their risk status and were referred to their customary source of medical care. At annual examinations, blood testing and other special tests were carried out on each participant. Earlier results from this study population on leukocyte count and risk of coronary heart disease, cancer, and all-cause mortality in the usual care group have been published (4). This analysis considers all men in the trial, whether in the special intervention group or the usual care group, since the relation between leukocyte count and risk of myocardial infarction does not differ importantly between the two groups. Measurements. Leukocyte counts were measured locally at the study centers at the beginning of the trial and at annual visits. The method used most frequently was the automated blood cell (Coulter) counter. A few centers performed the leukocyte count on-site using hemocytometers. Total cholesterol was measured in serum according to previously described methods (17, 18). Smoking. Cigarette smoking was ascertained at the first two eligibility visits, and a detailed smoking history questionnaire, including former smoking status, inhalation pattern, and use of other tobacco products, was completed at the third eligibility visit.
62
Phillips et al.
Coronary heart disease at initial examination. Men with any history or presence of coronary heart disease were excluded from the trial (18). Follow-up. The investigators followed and saw the men regularly for an average of 7 years, up to February 28, 1982, to obtain information on fatal and nonfatal myocardial infarctions and to assess risk factor changes. Endpoint definition. Deaths were systematically ascertained by clinic staff, and cause of death was assigned by a three-member panel of cardiologists. This ascertainment was made after a thorough review of hospital records, the physician's reports, interviews of relatives/household members, death certificates, and autopsy reports (if available). Nonfatal myocardial infarctions were diagnosed on the basis of either a significant serial electrocardiogram change between baseline and the annual visit or the review of medical records by cardiologists. All hospital records that were associated with heart problems were reviewed, and diagnoses were classified as "definite myocardial infarction," "possible myocardial infarction," "other cardiac event," or "noncardiac event." Definite myocardial infarction or serial electrocardiogram change was used as the nonfatal coronary heart disease endpoint. The Multiple Risk Factor Intervention Trial 2 cohort
A total of 22,080 men attended the second screening examination for the Multiple Risk Factor Intervention Trial. Leukocyte counts were taken at this visit. The 8,016 men who had a leukocyte count done at this second stage but did not participate in the trial form the third cohort. Measurements. Techniques used for leukocyte counts and measurements of total cholesterol were similar to those used for the baseline screening, as described above. Smoking. Basic data on cigarette smoking were obtained, but no distinction was made between former smokers and never smokers. Coronary heart disease at initial examina-
tion. For 34 percent of this cohort, the reason for exclusion from the trial was evidence of preexisting cardiovascular disease. This information was derived from the resting electrocardiogram, the World Health Organization (Rose) chest pain questionnaire, medical examination, or the fact that the subject had a history of using certain antihypertensive or lipid-lowering drugs (18). Follow-up. The men have been followed for an average of 12 years, up to December 1986, for mortality data only. Endpoint definition. Mortality follow-up was accomplished using the US National Death Index. A coronary heart disease death was defined as any death classified under International Classification of Diseases, Ninth Revision, codes 410-414 or 429.2. Statistical methods
Relative odds estimates in figures 2-4 were obtained by fitting a multiple logistic model. Besides smoking status, which was considered in detail, adjustment was made for age, diastolic blood pressure, and serum cholesterol. None of these latter variables had a large confounding effect. In figures 3 and 4, only three leukocyte groups are used, instead of the six shown in figure 2, because of the smaller numbers of men involved. Men with leukocyte counts below 1,000/ mm3 or above 30,000/mm3 were excluded {n = 2 in the British Regional Heart Study, n = 21 in Multiple Risk Factor Intervention Trial 1, and n = 24 in Multiple Risk Factor Intervention Trial 2). Pooled relative odds estimates were obtained by pooling the log relative odds estimates, weighted by precision (1/variance). RESULTS
Table 1 contains a summary description of the three cohorts. "Number of men" refers to the number for whom a leukocyte count was available. In total, over the three cohorts, there were 28,181 men, of whom 1,768 experienced a specified coronary heart disease event. Although we wish to focus on within-cohort rather than between-cohort
Leukocyte Count and Major Coronary Events
63
TABLE 1. Characteristics of the three cohorts used in a study of the relation of leukocyte count to myocardial infarction* MRFIT 1
BRHS
Average length of follow-up (years) Type of event
No. of ment No of events No. of nonsmokersj No. of events No. of never smokers§ No. of events % with preexisting coronary heart disease || Mean leukocyte count (/mm3) Mean age (years) Mean serum total cholesterol (mg/dl) Mean diastolic blood pressure (mmHg)
7.0
8.0 Coronary heart disease death or nonfatal myocardial infarction 7,344 485 3,566 174 1,546 56 25 7,202(1,771)11 50(6) 242 (40) 82.2(13.2)
Coronary heart disease death or nonfatal myocardial infarction 12,845 814 3,615 185 1,395 75 0
MRFIT 2
12.0 Coronary heart disease death 7,992 469 2,106 121
34
7,181 (2,092)
7,464(2,198)
46(6) 253 (37)
47(6) 254 (37)
99.2 (7.7)
99.1 (8.0)
•BRHS, British Regional Heart Study, MRFIT 1, Multiple Risk Factor Intervention Trial 1; MRFIT 2, Multiple Risk Factor Intervention Trial 2. t Subjects excluded because their leukocyte count was less than 1,000/mm3 or greater than 30,000/mm3: BRHS, n = 2; MRFIT 1,n = 21;MRFIT2, n = 24. $ Definition of a nonsmoker: BRHS and MRFIT 1—not currently smoking cigarettes, a pipe, cigars, or ciganllos; MRFIT 2—not currently smoking cigarettes. § Definition of a never smoker: BRHS and MRFIT 1—nonsmokers (see footnote 3) who have never smoked cigarettes. || BRHS—any evidence of preexisting coronary heart disease according to any of the three methods used (questionnaire, electrocardiogram, doctor diagnosis). MRFIT 2—men excluded from the trial because of the presence of cardiovascular disease. 11 Numbers in parentheses, standard deviation.
differences, some comparisons are worthy of note. The leukocyte counts were similar in the British Regional Heart Study and Multiple Risk Factor Intervention Trial 1 cohorts even though there was a higher proportion of current smokers (cigarettes, a pipe, or cigars) in the Multiple Risk Intervention Trial 1 cohort (51 percent vs. 72 percent); this is due to the fact that the men were selected because of high coronary heart disease risk, of which cigarette smoking was a component. The average leukocyte count was higher in the Multiple Risk Factor Intervention Trial 2 men. The serum cholesterol and diastolic blood pressure levels were higher in the two Multiple Risk Factor Intervention Trial cohorts, again because the men were selected on the basis of high levels of these factors. In figure 1, mean leukocyte count is shown by smoking status for the Multiple Risk Factor Intervention Trial 1 and British
Regional Heart Study cohorts. (Former smoking status was not ascertained for men in the Multiple Risk Factor Intervention Trial 2 cohort.) Leukocyte counts are lower in the Multiple Risk Factor Intervention Trial 1 cohort than in the British Regional Heart Study cohort for every smoking category. For both cohorts, the lowest levels are seen in men who have never smoked cigarettes, and the highest are seen in current cigarette smokers. Levels in former cigarette smokers are considerably lower than those in current smokers, but they appear to remain slightly higher than those in persons who had not smoked cigarettes for at least 5 years. Leukocyte counts in pipe/cigar smokers are intermediate between those of never smokers and current cigarette smokers, with secondary pipe/cigar smokers (former cigarette smokers) having higher levels than pipe/cigar smokers who had never smoked cigarettes. Thus, differences in leukocyte
64
Phillips et al.
9,000 _
8,000 .
7,000 .
2
i i i
6,000 .
5,000 Never Pipe/ Secondary Cigar pipe / cigar
£10
5-9
1-4
