The Association between Type A Behavior and Change in Coronary Risk Factors among Young Adults THOMAS F. GARRITY, PHD, J. MORLEY KOTCHEN, MD, HARLLEY E. MCKEAN, PHD, DIANA GURLEY, PHD, AND MOLLY MCFADDEN, MS Abstract: The association of Type A/B behavior pattern and changes in blood pressure, total serum cholesterol, serum triglyceride, body mass, and smoking was estimated in a cohort of 375 young Black and White men and women from a rural county in Central Kentucky between 1978-79 and 1985-88. Type A participants experienced significant increases in systolic (2.90 ± 1.29 mmHg) and diastolic (3.80 ± 1.17 mmHg) blood pressure and in cigarette smoking (3.26 ± 0.89 cigarettes per day) over the eight-year follow-up period,
but Type B participants experienced no change. Type A and B individuals showed similar changes in total serum cholesterol, serum triglyceride, or body mass. Differences between behavioral types in blood pressure were present for women but not men, and for Blacks but not for Whites. These findings suggest a possible significance of the Type A pattern for the development of cardiovascular risk of young adults. (Am J Public Health 1990; 80:1354-1357.)
Introduction Cardiovascular epidemiology has established blood pressure, serum cholesterol, and smoking as the most powerful and consistent predictors of coronary risk in middle age and older adults. Studies of children suggest that coronary risk factors are already present at early ages,1,2 but relatively little epidemiologic data exist on the coronary risk status of young adults, especially those who have recently completed physical growth as they begin to enter typical adult roles in this society,3 a phase characterized by changes in lifestyle and responsibility. The student becomes a worker, the single person a spouse and parent, with changes in eating, drinking, sleeping, and exercise. The style of coping with these challenges may have a bearing on level of life stress and, through it, on markers of coronary risk. Type A, coronary-prone behavior pattern is believed to relate to risk for coronary heart disease.4 We followed a sample of men and women from small town and rural backgrounds from an average age of 21 years to an average age of 29 to determine whether Type A/B behavior pattern is associated with change in coronary risk factors. Previously, we have shown clustering of risk factors in individuals in this population.5
adults who had the highest and lowest systolic blood pressure readings when they were teenagers in 1973. Because some of the subjects had not yet achieved adult growth at baseline, all of the subjects who were less than 16 years old in 1973 were also contacted for re-examination in 1978. Of the remaining group, 15 percent were selected at random for follow-up. Further description of this sample has been published previously.5-7 In 1985-88, the attempt was made to contact the entire population studied at baseline in 1973. Of the original population, 253 were lost to the study; 219 could not be located; 19 refused; 15 had died. Follow-up data were successfully gathered on 891 subjects, or 78 percent of the original pool. The sample for the present study (N = 375) includes subjects for whom data are available both from the 1978 study and from the 1985-88 population follow-up. Subjects with very high or very low systolic blood pressure measurements when they were teenagers are overrepresented in this group, as are younger participants. As seen in Table 1, the typical 1985-88 participant was a White, married, full-time employed, high school graduate born in 1958. About half of the study group was male. At both 1978-79 and 1985-88 data collections, participants were located by searches of local public documents, contacts with relatives and friends of the participants, and by placing classified advertisements in local and regional newspapers. Measurements were usually made at the project office in Bourbon County. For participants not willing to travel to the project office, home visits were performed by project staff throughout the state of Kentucky. For two out-of-state participants, Physical Measurements, Inc was hired to provide home visits at which a blood sample, blood pressure, and other anthropometric measurements were obtained according to protocol. Questionnaires were also completed by these two participants. Blood pressure, anthropometric, and medical history data were collected at both waves. The Type A measure (the Jenkins Activity Survey, Form C)8 was administered only in 1985-88. Type A behavior was measured with the JAS Type A subscale comprising 21 items from the longer instrument. Each response was assigned a weight and summed according to the method of Jenkins.8 Blood lipid analyses and other cardiovascular assessments7 are available on a subset of the participants (N = 98). Quetelet index9: Q = weight in kilograms/(height in meters)2 was calculated. Blood pressure was measured in the right arm after the subject was seated for at least five minutes.
Methods The sample for the present study was derived from a 15-year longitudinal cohort study of young adults in a rural county of the Central Kentucky Bluegrass. In 1973, all high school students of Bourbon County, Kentucky (N = 1,142) were assessed for their blood pressures, body weights, and heights, as well as a few other demographic characteristics. Students' ages ranged from 14 to 19 years. In 1978-79, a sample (N = 472) was drawn from the original study population to include the 20 percent of young Address reprint requests to Thomas F. Garrity, PhD, Professor and Chairman, Department of Behavioral Science, College of Medicine Office Building, University of Kentucky, Lexington, KY 40536-086. Dr. Gurley is also with that department at the University; Dr. Kotchen is with the Department of Medicine, West Virginia Medical Center; Dr. McKean is with the Department of Statistics, University of Kentucky; Ms. McFadden of Cleveland Heights, Ohio was with the West Virginia Medical Center at the time of the study. This paper, submitted to the Journal September 5, 1989, was revised and accepted for publication April 2, 1990. © 1990 American Journal of Public Health 0090-0036/90$1.50
1 354
AJPH November 1990, Vol. 80, No. 11
TYPE A BEHAVIOR AND CORONARY RISK IN YOUNG ADULTS TABLE 1-Distributions of Selected Demographic Characteristics of Participants, Bourbon County, Kentucky, 1978-79 and 1985-88 (N = 375)*
TABLE 2Coronary Risk Factors in Participants, Means and Standard Deviations, Bourbon County, Kentucky, 1978-79 and 1985-88
N Age (years) Sex Male Female
1978-79
1985-88
20.7 ± 1.8 N (%) 200 (53.3) 175 (46.7)
28.6 ± 1.4
Martal Status Single Married Separated Divorced Widowed Employment Fulltime Parttime Not employed Education Some High School High School Diploma Some College College Diploma Race White Black
3 0
(67.1) (31.0) (1.1) (0.8) (0.0)
N 101 227 8 39 0
(26.9) (60.5) (2.1) (10.4) (0.0)
90 227 58
(24.0) (60.5) (15.5)
283 25 67
(75.5) (6.7) (17.9)
42 243 83 7
(11.2) (64.8) (22.1) (1.9)
42 232 53 48
(11.2) (61.9) (14.1) (12.8)
301 74
(80.3) (19.7)
251 116 4
(%)
*375 is the total number of participants. Smaller numbers are due to missing information on some variables.
Three measurements by a trained observer were taken over a 10-15 minute period using a standard mercury manometer and cuff calibrated for arm circumference; these readings were averaged. Blood samples were analyzed for total serum cholesterol and serum triglyceride values by the autoanalyzer method described in the Lipid Research Clinics Manual of Laboratory Operations.'0 We studied systolic and diastolic blood pressure, serum cholesterol, and serum triglyceride levels, body mass, and smoking behavior as indicators of risk. Change in risk status was measured as the difference in scores from 1978-79 to 1985-88 on each of these variables. Using change scores for each of the risk factors as outcomes, we calculated a three-way factorial model for all risk factors except smoking change, with behavior pattern, race and sex as the explanatory variables. Change in number of cigarettes smoked was treated as a categorical variable for log-linear modeling and contingency table analysis. Race, sex and body mass were included as potential control variables.'2 Relations between behavior pattern and risk variables were affected by race and sex, but not by change in body mass (data available on request) so change in body mass was removed from further
analyses. Results Coronary risk factor means and standard deviations at the two time periods are shown in Table 2. The three-way factorial analysis for change in systolic blood pressure revealed that only the main effect of behavior pattern accounted for significant variance in change in systolic blood pressure between 1978-79 and 1985-88 assessments. As shown in Table 3, Type A subjects experienced a mean 2.90 ± 1.29 (standard error) mmHg change, whereas Type B subjects showed a mean -1.47 ± 1.22 mmHg change in systolic blood pressure. The Type A subjects experienced greater increases in systolic blood pressure than Type B AJPH November 1990, Vol. 80, No. 11
Systolic Blood Pressure Diastolic Blood Pressure Serum Cholesterol Serum Triglyceride Body Mass Index JAS Type A Score Cigarettes/Day
375+ 375 98 98 371 375 369
1978-79 Means 120.27 ± 73.08 ± 167.21 ± 107.53 ± 24.17 ±
SD 14.90 11.35 37.32 77.81 6.48
7.24 ± 10.99
1985-88
Means 120.69 ± 75.28 ± 204.19 ± 168.36 ± 26.62 ± 211.28 ± 9.18 ±
SD 16.07 11.88 43.18 99.65 5.81
72.82 13.39
+Fewer than 375 cases on a varable are due to missing information (except for the serum cholesterol and serum triglyceride variables in which sample design called for fewer cases).
subjects in the entire sample (p = 0.01), in women (p = 0.02) but not in men (p = 0.23), and in Blacks (p = 0.05) but not in Whites (p = 0.12). In the three-way factorial analysis for change in diastolic blood pressure, none of the explanatory variables or their interactions were statistically significant, but Type A individuals rose an average 3.80 ± 1.17 mmHg, whereas Type B individuals rose 0.50 ± 1.11 mmHg in diastolic blood pressure (Table 4). The Type A subjects experienced somewhat larger increase in diastolic blood pressure than Type B subjects in each contrast especially among Blacks and women. Factorial analysis of changes in total serum cholesterol, serum triglyceride, and body mass values showed no statistically significant main or interaction effects on change in any of these three risk variables. Log-linear modeling analysis was performed with smoking, the outcome, divided into participants who decreased, remained unchanged and increased cigarette consumption from 1978-79 to 1985-88, and with behavior pattern, sex and race the predictors. Only behavior pattern was significantly associated with smoking change. Table 5 shows that Type As increased smoking more than type Bs, and that Type Bs held their smoking at the same level to a greater extent than Type As. Log-linear analysis found that the behavior pattern/ smoking change association was statistically significant in Black females but not in the other three race/sex groups. Discussion In 1981, an authoritative panel examined the available evidence and judged the Type A behavior pattern to be a risk factor for coronary disease, comparable in impact to smoking, elevated cholesterol, and elevated blood pressure.4 Subsequent results call that conclusion into question.'3-'5 Recent research suggests that underlying hostility is the aspect of Type A behavior that is most strongly predictive of coronary heart disease endpoints.'6 A reanalysis of the Multiple Risk Factor Intervention Trial (MRFIT) data, one of the major recent studies failing to find a Type A/coronary disease association, has found indicators of hostility to significantly predict coronary disease incidence.17 Appropriate measurement of Type A behavior is uncertain, 18 but at best is only modestly associated with other coronary risk factors,'9.20 and has been seen as an independent risk factor. This study suggests that, in young adults, Type A behavior pattern is associated with more rapid increases in two well-accepted coronary risk factors. Type A behavior may operate differently at different ages to produce 1 355
GARRITY, ET AL. TABLE 3-Mean Change (+ standard error) in Systolic Blood Pressure for Sex/Race/Behavior-Pattern Groups* Males Type A
Type B
1.17
-0.66 (1.65) 0.73 (2.88) 0.04 (1.66)
White
Black Both Races
Females
(1.38) 4.93 (3.49) 3.05 (1.88)
A vs B (p-value)
Both Sexes
Type A
Type B
Type A
Type B
Avs. B (p)
1.72
-1.40 (1.58)
1.44
(1.09)
0.12
2.90
-1.03 (1.14) -1.91 (2.15) -1.47
(1.29)
(1.22)
(1.68)
-4.56 (3.19) -2.98 (1.78)
3.79 (3.10) 2.76
(1.76)
0.23
4.36 (2.34)
0.02
0.05
0.01
*Three-way analysis of variance found a significant (p
but Type B participants experienced no change. Type A and B individuals showed similar changes in total serum cholesterol, serum triglyceride, or body mass. Differences between behavioral types in blood pressure were present for women but not men, and for Blacks but not for Whites. These findings suggest a possible significance of the Type A pattern for the development of cardiovascular risk of young adults. (Am J Public Health 1990; 80:1354-1357.)
Introduction Cardiovascular epidemiology has established blood pressure, serum cholesterol, and smoking as the most powerful and consistent predictors of coronary risk in middle age and older adults. Studies of children suggest that coronary risk factors are already present at early ages,1,2 but relatively little epidemiologic data exist on the coronary risk status of young adults, especially those who have recently completed physical growth as they begin to enter typical adult roles in this society,3 a phase characterized by changes in lifestyle and responsibility. The student becomes a worker, the single person a spouse and parent, with changes in eating, drinking, sleeping, and exercise. The style of coping with these challenges may have a bearing on level of life stress and, through it, on markers of coronary risk. Type A, coronary-prone behavior pattern is believed to relate to risk for coronary heart disease.4 We followed a sample of men and women from small town and rural backgrounds from an average age of 21 years to an average age of 29 to determine whether Type A/B behavior pattern is associated with change in coronary risk factors. Previously, we have shown clustering of risk factors in individuals in this population.5
adults who had the highest and lowest systolic blood pressure readings when they were teenagers in 1973. Because some of the subjects had not yet achieved adult growth at baseline, all of the subjects who were less than 16 years old in 1973 were also contacted for re-examination in 1978. Of the remaining group, 15 percent were selected at random for follow-up. Further description of this sample has been published previously.5-7 In 1985-88, the attempt was made to contact the entire population studied at baseline in 1973. Of the original population, 253 were lost to the study; 219 could not be located; 19 refused; 15 had died. Follow-up data were successfully gathered on 891 subjects, or 78 percent of the original pool. The sample for the present study (N = 375) includes subjects for whom data are available both from the 1978 study and from the 1985-88 population follow-up. Subjects with very high or very low systolic blood pressure measurements when they were teenagers are overrepresented in this group, as are younger participants. As seen in Table 1, the typical 1985-88 participant was a White, married, full-time employed, high school graduate born in 1958. About half of the study group was male. At both 1978-79 and 1985-88 data collections, participants were located by searches of local public documents, contacts with relatives and friends of the participants, and by placing classified advertisements in local and regional newspapers. Measurements were usually made at the project office in Bourbon County. For participants not willing to travel to the project office, home visits were performed by project staff throughout the state of Kentucky. For two out-of-state participants, Physical Measurements, Inc was hired to provide home visits at which a blood sample, blood pressure, and other anthropometric measurements were obtained according to protocol. Questionnaires were also completed by these two participants. Blood pressure, anthropometric, and medical history data were collected at both waves. The Type A measure (the Jenkins Activity Survey, Form C)8 was administered only in 1985-88. Type A behavior was measured with the JAS Type A subscale comprising 21 items from the longer instrument. Each response was assigned a weight and summed according to the method of Jenkins.8 Blood lipid analyses and other cardiovascular assessments7 are available on a subset of the participants (N = 98). Quetelet index9: Q = weight in kilograms/(height in meters)2 was calculated. Blood pressure was measured in the right arm after the subject was seated for at least five minutes.
Methods The sample for the present study was derived from a 15-year longitudinal cohort study of young adults in a rural county of the Central Kentucky Bluegrass. In 1973, all high school students of Bourbon County, Kentucky (N = 1,142) were assessed for their blood pressures, body weights, and heights, as well as a few other demographic characteristics. Students' ages ranged from 14 to 19 years. In 1978-79, a sample (N = 472) was drawn from the original study population to include the 20 percent of young Address reprint requests to Thomas F. Garrity, PhD, Professor and Chairman, Department of Behavioral Science, College of Medicine Office Building, University of Kentucky, Lexington, KY 40536-086. Dr. Gurley is also with that department at the University; Dr. Kotchen is with the Department of Medicine, West Virginia Medical Center; Dr. McKean is with the Department of Statistics, University of Kentucky; Ms. McFadden of Cleveland Heights, Ohio was with the West Virginia Medical Center at the time of the study. This paper, submitted to the Journal September 5, 1989, was revised and accepted for publication April 2, 1990. © 1990 American Journal of Public Health 0090-0036/90$1.50
1 354
AJPH November 1990, Vol. 80, No. 11
TYPE A BEHAVIOR AND CORONARY RISK IN YOUNG ADULTS TABLE 1-Distributions of Selected Demographic Characteristics of Participants, Bourbon County, Kentucky, 1978-79 and 1985-88 (N = 375)*
TABLE 2Coronary Risk Factors in Participants, Means and Standard Deviations, Bourbon County, Kentucky, 1978-79 and 1985-88
N Age (years) Sex Male Female
1978-79
1985-88
20.7 ± 1.8 N (%) 200 (53.3) 175 (46.7)
28.6 ± 1.4
Martal Status Single Married Separated Divorced Widowed Employment Fulltime Parttime Not employed Education Some High School High School Diploma Some College College Diploma Race White Black
3 0
(67.1) (31.0) (1.1) (0.8) (0.0)
N 101 227 8 39 0
(26.9) (60.5) (2.1) (10.4) (0.0)
90 227 58
(24.0) (60.5) (15.5)
283 25 67
(75.5) (6.7) (17.9)
42 243 83 7
(11.2) (64.8) (22.1) (1.9)
42 232 53 48
(11.2) (61.9) (14.1) (12.8)
301 74
(80.3) (19.7)
251 116 4
(%)
*375 is the total number of participants. Smaller numbers are due to missing information on some variables.
Three measurements by a trained observer were taken over a 10-15 minute period using a standard mercury manometer and cuff calibrated for arm circumference; these readings were averaged. Blood samples were analyzed for total serum cholesterol and serum triglyceride values by the autoanalyzer method described in the Lipid Research Clinics Manual of Laboratory Operations.'0 We studied systolic and diastolic blood pressure, serum cholesterol, and serum triglyceride levels, body mass, and smoking behavior as indicators of risk. Change in risk status was measured as the difference in scores from 1978-79 to 1985-88 on each of these variables. Using change scores for each of the risk factors as outcomes, we calculated a three-way factorial model for all risk factors except smoking change, with behavior pattern, race and sex as the explanatory variables. Change in number of cigarettes smoked was treated as a categorical variable for log-linear modeling and contingency table analysis. Race, sex and body mass were included as potential control variables.'2 Relations between behavior pattern and risk variables were affected by race and sex, but not by change in body mass (data available on request) so change in body mass was removed from further
analyses. Results Coronary risk factor means and standard deviations at the two time periods are shown in Table 2. The three-way factorial analysis for change in systolic blood pressure revealed that only the main effect of behavior pattern accounted for significant variance in change in systolic blood pressure between 1978-79 and 1985-88 assessments. As shown in Table 3, Type A subjects experienced a mean 2.90 ± 1.29 (standard error) mmHg change, whereas Type B subjects showed a mean -1.47 ± 1.22 mmHg change in systolic blood pressure. The Type A subjects experienced greater increases in systolic blood pressure than Type B AJPH November 1990, Vol. 80, No. 11
Systolic Blood Pressure Diastolic Blood Pressure Serum Cholesterol Serum Triglyceride Body Mass Index JAS Type A Score Cigarettes/Day
375+ 375 98 98 371 375 369
1978-79 Means 120.27 ± 73.08 ± 167.21 ± 107.53 ± 24.17 ±
SD 14.90 11.35 37.32 77.81 6.48
7.24 ± 10.99
1985-88
Means 120.69 ± 75.28 ± 204.19 ± 168.36 ± 26.62 ± 211.28 ± 9.18 ±
SD 16.07 11.88 43.18 99.65 5.81
72.82 13.39
+Fewer than 375 cases on a varable are due to missing information (except for the serum cholesterol and serum triglyceride variables in which sample design called for fewer cases).
subjects in the entire sample (p = 0.01), in women (p = 0.02) but not in men (p = 0.23), and in Blacks (p = 0.05) but not in Whites (p = 0.12). In the three-way factorial analysis for change in diastolic blood pressure, none of the explanatory variables or their interactions were statistically significant, but Type A individuals rose an average 3.80 ± 1.17 mmHg, whereas Type B individuals rose 0.50 ± 1.11 mmHg in diastolic blood pressure (Table 4). The Type A subjects experienced somewhat larger increase in diastolic blood pressure than Type B subjects in each contrast especially among Blacks and women. Factorial analysis of changes in total serum cholesterol, serum triglyceride, and body mass values showed no statistically significant main or interaction effects on change in any of these three risk variables. Log-linear modeling analysis was performed with smoking, the outcome, divided into participants who decreased, remained unchanged and increased cigarette consumption from 1978-79 to 1985-88, and with behavior pattern, sex and race the predictors. Only behavior pattern was significantly associated with smoking change. Table 5 shows that Type As increased smoking more than type Bs, and that Type Bs held their smoking at the same level to a greater extent than Type As. Log-linear analysis found that the behavior pattern/ smoking change association was statistically significant in Black females but not in the other three race/sex groups. Discussion In 1981, an authoritative panel examined the available evidence and judged the Type A behavior pattern to be a risk factor for coronary disease, comparable in impact to smoking, elevated cholesterol, and elevated blood pressure.4 Subsequent results call that conclusion into question.'3-'5 Recent research suggests that underlying hostility is the aspect of Type A behavior that is most strongly predictive of coronary heart disease endpoints.'6 A reanalysis of the Multiple Risk Factor Intervention Trial (MRFIT) data, one of the major recent studies failing to find a Type A/coronary disease association, has found indicators of hostility to significantly predict coronary disease incidence.17 Appropriate measurement of Type A behavior is uncertain, 18 but at best is only modestly associated with other coronary risk factors,'9.20 and has been seen as an independent risk factor. This study suggests that, in young adults, Type A behavior pattern is associated with more rapid increases in two well-accepted coronary risk factors. Type A behavior may operate differently at different ages to produce 1 355
GARRITY, ET AL. TABLE 3-Mean Change (+ standard error) in Systolic Blood Pressure for Sex/Race/Behavior-Pattern Groups* Males Type A
Type B
1.17
-0.66 (1.65) 0.73 (2.88) 0.04 (1.66)
White
Black Both Races
Females
(1.38) 4.93 (3.49) 3.05 (1.88)
A vs B (p-value)
Both Sexes
Type A
Type B
Type A
Type B
Avs. B (p)
1.72
-1.40 (1.58)
1.44
(1.09)
0.12
2.90
-1.03 (1.14) -1.91 (2.15) -1.47
(1.29)
(1.22)
(1.68)
-4.56 (3.19) -2.98 (1.78)
3.79 (3.10) 2.76
(1.76)
0.23
4.36 (2.34)
0.02
0.05
0.01
*Three-way analysis of variance found a significant (p
