Int.J. Behav. Med. (2014) 21:927–935 DOI 10.1007/s12529-013-9369-z
Age- and Cohort-Related Variance of Type-A Behavior Over 24 Years: the Young Finns Study Taina Hintsa & Markus Jokela & Laura Pulkki-Råback & Liisa Keltikangas-Järvinen
Published online: 14 November 2013 # International Society of Behavioral Medicine 2013
Abstract Background Over the recent decades, the incidence of cardiovascular and heart diseases has decreased while levels of typeA behavior, i.e., a potential risk factor, appear to have increased. However, the long-term developmental patterns of type-A behavior is poorly understood. Both age- and cohortrelated changes may be involved in these developments. Purpose The purpose of this study was to examine an ageand cohort-related changes of Hunter-Wolf type-A behavior from adolescence to adulthood. Methods Type-A behavior and its components (aggressiveness, leadership, hard driving, and eagerness energy) were assessed using the Hunter-Wolf A-B rating scale at five time points (1983, 1986, 1989, 2001, and 2007) in a populationbased sample consisting of six birth cohorts born between 1962 and 1977 (n =3,341, a total of 10,506 person observations). Development of type-A behavior and its components was examined with cohort-sequential multilevel modeling. Results Aggressiveness decreased with age, eagerness energy, hard driving, and global type-A behavior increased, and leadership exhibited no mean level changes. Younger cohorts had
Electronic supplementary material The online version of this article (doi:10.1007/s12529-013-9369-z) contains supplementary material, which is available to authorized users. T. Hintsa (*) : M. Jokela : L. Pulkki-Råback : L. Keltikangas-Järvinen IBS, Department of Personality, Work and Health Psychology, University of Helsinki, P.O. Box 9, 00014 Helsinki, Finland e-mail: [email protected] M. Jokela e-mail: [email protected] L. Pulkki-Råback e-mail: [email protected] L. Keltikangas-Järvinen e-mail: [email protected]
higher aggressiveness, lower hard driving, and global type-A behavior. Conclusions The findings suggest that in order to understand the health consequences of type-A behavior, both life span and societal changes should be considered. Keywords Type-A behavior . Aggression . Leadership . Eagerness energy . Hard driving . Cohort study
Introduction The original type-A behavior concept, firstly presented by Friedman and Rosenman in 1959, was characterized by intense ambition, competitive drive, constant preoccupation with job-related deadlines, and a sense of time urgency [1]. Later, type-A behavior has been defined as “an actionemotion complex that includes (1) behavioral dispositions, (2) specific behaviors, and (3) emotional responses [2]. In the late 1970s, Western Collaborative Group study (WCGS) [3, 4] and Framingham study [5] suggested that type-A behavior contributes to a development of coronary heart disease (CHD). Later, both positive [6, 7] and negative findings [8–10] have been reported, and it has been concluded that evidence of type-A behavior as a prognostic factor of CHD has not been scholarly proved although it might have a minor etiological role in CHD among healthy adults [11]. There are, however, aspects that may compromise the negative evidence. First, the assessment methods of type-A behavior vary a lot, and the same assessments have resulted in conflicting findings in comparable samples. For instance, in a sample of WCGS participants type-A behavior has been reported to be predictive of CHD incidence [12, 13] whereas in a sample from the Multiple Risk Factor Intervention Trial (MRFIT) Study it has been unrelated to CHD [2, 14]. The
928
evidence from studies using questionnaires to assess type-A behavior has also been contradictory [15]. Second, type-A behavior is a multidimensional construct, and different measures focus on different contents [2]. Because only some components such as hostility and time urgence have been suggested to be “toxic” [16–20], while some of the type-A behavior components to be protective or represent healthy competitive style, i.e., leadership [16, 21], the role of the instrument becomes emphasized. Thus, it has been suggested that the dimensions of type-A behavior should be examined separately instead of the global rating [17, 18]. Third, the samples of previous prospective studies have consisted mainly of male subjects [14, 22, 23], and gender differences have been shown in associations between type-A behavior and CHD [2]. We have previously shown that eagerness energy component of Hunter-Wolf type-A behavior predicted adulthood atherosclerosis over different developmental transitions from early adolescence onwards in men but not in women [24]. Hunter-Wolf component of hard driving has been shown to predict decreased risk of atherosclerosis in women but not in men [24]. Fourth, and perhaps most important, the prevailing social and environmental approvals may change the meaning and effect of specific personality trait, and thus, the cohort effect should be systematically taken into account. In addition to individual variance, the development of personality traits may be influenced by social trends and societal circumstances that prevail during person’s formative years. For instance, studies in American youth have documented increasing levels of anxiety, and decreasing levels of internal locus of control and need for social approval during the late twentieth century [25, 26]. Consequently, it is the prevailing values and attitudes in the environment that are likely to influence at least reporting of type-A behavior during the time of measurement. Thus, time of measurement may also partly determine the health relevance of type-A behavior in relation to CHD risk, and in studies of the health consequences of type-A behavior, both age-related and societal changes should be considered. Results from the WHO MONICA project [27], on data from 6,695 Danish men and women aged 30, 40, 50, and 60 years, show that between the period of 1982 and 1992, myocardial infarction incidence declined by 3–5 % per year while type-A behavior increased in both men and women at the same time. This has been interpreted to confirm the negative evidence of type-A’s contribution to CHD, but it may refer to cohort effects, and suggest that scores of different decades are no more comparable and that the associations should be studied within a given cohort. The course of type-A behavior might be subject to birth cohort effects, but there are no previous research addressing this issue. We followed in a population-based sample (a) agerelated changes of type-A behavior and its subcomponents from early adolescence to middle adulthood, and (b) cohort-
Int.J. Behav. Med. (2014) 21:927–935
related changes of type-A behavior during 24 years (i.e., societal changes of type-A behavior during two decades). The cohort-sequential design allows one to test cohort differences in developmental trajectories.
Methods Participants The participants were 3,341 individuals (1,586 men, 1,755 women) from the population-based random sample of 3,596 men and women the Cardiovascular Risk in Young Finns Study (YFS) comprising six age cohorts (born in 1977, 1974, 1971, 1968, 1965, and 1962; i.e., 3, 6, 9, 12, 15, and 18 years at the baseline in 1980) who have been followed for 27 years. In 2007, 4.8 % of the participants had mandatory education and 35.4 % had higher degree. Type-A behavior was assessed at five time points in 1983, 1986, 1989, 2001, and 2007. Description of the birth cohorts and study phases are given in Fig. 1. We included all participants who had data on type-A behavior at least in one follow-up phase (n =3,341, a total of 10,506 person-observations). The study was approved by the university ethics committee. Details for selecting the sample have been given by Åkerblom et al. [28] and Raitakari et al. [28, 29]. Type-A Behavior Type-A behavior was measured on the Hunter-Wolf A-B Rating Scale [30]. The scale consists of four components to be answered in a form of a seven point ladder: (1) impatience– aggression (seven items, e.g., “I like to argue” and “I find it difficult to wait”), (2) leadership (six items, e.g., “I always like to win” and “I am always a leader in activities”), (3) Hard driving (three items, e.g., “I am hard-driving and competitive” and “ I take things seriously”), and (4) Eagerness energy (seven items, e.g., “I always feel hurried” and “ I think many things simultaneously”). The Cronbach’s alpha reliabilities of global type-A behavior at different measurement points varied Study phase
1983
n 1645
Age 152) 121)
183)
214)
1986
2000
125)
15
18
21
24
1989
2705
126)
15
18
21
24
27
2001
2089
24
27
30
33
36
39
2007
2038
30
33
36
39
42
45
Cohorts, born in 1) 1971 n=1809
2) 1968 n=1720
3) 1965 n=1513
4) 1962 n=1285
5) 1974 n=1301
Fig. 1 Description of the cohorts and study phases
6)1977 n=793
Int.J. Behav. Med. (2014) 21:927–935
between .67 and .79, and Cronbach alpha reliabilities of the components from .58 to .72. Statistical Analyses To study the developmental trajectories of type-A behavior and its subcomponents, we used multilevel linear modeling with a cohort-sequential study design. In a cohort-sequential study design (also known as an accelerated longitudinal design), participants who belong to different birth cohorts are followed over the same period of time but over different age periods, and then their developmental trajectories over age are used to construct a single trajectory spanning from the earliest age of the youngest cohort to the latest age of the oldest cohort. In the present case, a 24-year follow-up resulted in age periods of 12–30, 12–33, 12–36, 15–39, 18–42, and 21– 45 for the six birth cohorts in order of increasing age at baseline (type-A behavior was assessed only in participants aged 12 or older). The cohort-sequential design also allows one to test cohort differences in developmental trajectories. We assessed age× cohort interaction effects with a method formulated by Miyazaki and Raudenbush [31], where two models are fit to the data: one including cohort×age interaction effects (the full model) and the other without cohort effects (the reduced model). If the chi-square fit of the full model is significantly better than the fit of the reduced model, then cohort effects are present. This implies that birth cohorts may differ in the average level or in the shape of developmental trajectory of the outcome, which are shown by the cohort×age interaction effects. If the full model does not improve the fit of the reduced model, there are no significant differences between cohorts and a single trajectory can be used to characterize the development of the outcome over age. Cohorts were modeled with dummy variables with the oldest cohort as the references group. As described by Miyazaki and Raudenbush [31], participant’s age was centered on her/his cohort’s median age in the full model and on the median age of the total sample in the reduced model. Preliminary analyses suggested nonlinear developmental trajectories, so the effect of age was modeled with a linear and quadratic term with an additional cubic term for aggressiveness.
Results Figure 1 describes the cohorts and study phases. The number of participants in each cohort and mean values of type-A behavior at each measurement point are described in Table 1. To test whether type-A behavior was differentially associated with attrition between study waves, we fitted a series of logistic regression models by birth cohort and assessment time in which type-A behavior was used to predict the probability
929
of not participating in the next study wave. The odds ratios for the attrition odds are reported in supplementary Table 5. There were no systematic cohort differences in attrition related to type-A behavior. Table 2 shows results of the full model including cohort effects for each of the components and for the global scale. Table 3 shows the corresponding reduced models without cohort effects. Table 4 summarizes the deviance statistics of the full and reduced models and compares their fits for each outcome. With the exception of eagerness energy, the full model provided a significantly better fit to the data in the type-A traits, suggesting that there were cohort effects in the developmental trajectories of aggressiveness, leadership, hard driving, and global type-A scale. Aggressiveness followed a nonlinear developmental trajectory which decreased from age 12 to 24, increased temporarily up to age 33, and then followed a decreasing trend up to age 45. The average decrease from age 12 to age 45 was 0.34 points (0.37 SD). Cohort×age interaction effects (Table 2) indicated that the cohort differences in aggressiveness were related mainly to mean level differences, with younger cohorts having higher aggressiveness but largely similar developmental patterns than older cohorts over age. There was no clear developmental pattern in leadership, and the difference in leadership at age 12 and 45 was only 0.01 points (0.02 SD), indicating no age-related change over this period. And despite the better overall fit of the full model for leadership, there were no consistent cohort differences in average levels or age-dependent associations. Hard driving followed an increasing developmental trajectory that reached a plateau around age 27 and changed little after that. The change from age 12 to 27 was 0.64 points (0.60 SD) and the change from age 27 to 45 only 0.04 points (0.03 SD). Significant cohort main effects and cohort×age interaction effects indicated that younger cohorts had lower mean levels and their increasing developmental trajectories leveled off earlier than in older cohorts. Eagerness energy also increased from age 12 onwards, reached the highest value at age 33 (0.60 points or 0.79 SD compared to age 12) and decreased slightly from age 33 to 45 (0.12 points, 0.16 SD). There were statistically significant cohort interaction effects in eagerness, but given that their inclusion did not increase the overall fit of the full model over the reduced model, they are not interpreted as significant cohort effects. Thus, all cohorts followed essentially the same eagerness energy trajectory over age. The development of global type-A score reflected a combination of the development of its components, eagerness energy and hard driving in particular, so that it increased between age 12 and 33 (0.23 points, 0.42 SD) and began to decrease after that (0.10 points or 0.18 SD between ages 33 and 45). Cohort differences indicated that younger cohorts had lower mean scores than older cohorts.
930
Int.J. Behav. Med. (2014) 21:927–935
Table 1 Type A behavior in six cohorts by the measurement points from 1983 to 2007 Measurement time
1983
N, whole cohort (% males) n 1977 n =577 (51.1 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1974 n =583 (48.0 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1971 n =646 (50.0 %) 465–487 Aggression Leadership Hard driving Eagerness energy Global TAB 1968 n =651 (49.5 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1965 n =602 (47.7 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1962 n =537 (47.9 %) Aggression Leadership Hard driving Eagerness energy Global TAB
1986 Mean (SD)
n
1989
n Mean (SD) 457–459 12 years old na na 3.86 (1.02) 4.23 (0.84) 4.58 (1.08) 4.49 (0.65) 4.24 (0.51) na 472–475 12 years old 479–480 15 years old 3.76 (0.92) 3.71 (0.94) 4.18 (0.76) 4.22 (0.88) 4.65 (0.98) 4.65 (1.12) 4.48 (0.64) 4.68 (0.65) 4.20 (0.45) 4.26 (0.53) 12 years old 482–484 15 years old 497–498 18 years old 4.02 (0.89) 3.80 (0.87) 3.68 (0.95) 4.33 (0.87) 4.18 (0.86) 4.16 (0.88) 4.69 (1.08) 4.67 (1.03) 4.85 (1.09)
4.55 (0.68) 4.35 (0.51) 476–486 15 years old 412 3.87 (0.90) 4.21 (0.90) 4.68 (1.08) 4.68 (0.71) 4.31 (0.52) 380–381 18 years old 332 3.78 (0.91) 4.19 (0.97) 4.88 (1.06) 4.76 (0.69) 4.33 (0.53) 288–294 21 years old 300–301 3.75 (0.96) 4.29 (0.91) 5.13 (1.00) 4.85 (0.74) 4.41 (0.57)
Mean (SD)
2001
2007
n Mean (SD) 338–341 24 years old 3.74 (0.95) 4.35 (0.87) 5.15 (1.03) 4.93 (0.74) 4.45 (0.55) 352 27 years old 3.77 (0.94) 4.31 (0.88) 5.21 (0.99) 5.09 (0.76) 4.50 (0.59) 348–349 30 years old 3.77 (0.94) 4.40 (0.95) 5.26 (1.09)
n Mean (SD) 318–319 30 years old 3.67 (0.79) 4.33 (0.87) 5.15 (0.98) 4.95 (0.71) 4.42 (0.51) 325–326 33 years old 3.66 (0.85) 4.35 (0.89) 5.30 (0.96) 5.12 (0.75) 4.50 (0.52) 356–360 36 years old 3.58 (0.81) 4.31 (0.90) 5.21 (0.99)
4.63 (0.61) 4.26 (0.48) 18 years old 461 3.65 (0.81) 4.13 (0.85) 4.85 (1.01) 4.73 (0.66) 4.26 (0.50) 21 years old 430–431 3.53 (0.83) 4.19 (0.89) 5.10 (1.02) 4.86 (0.70) 4.31 (0.51) 24 years old 381–382 3.55 (0.87) 4.17 (0.89) 5.25 (0.96) 4.93 (0.73)
4.80 (0.70) 5.10 (0.78) 5.08 (0.78) 4.30 (0.54) 4.54 (0.58) 4.44 (0.52) 21 years old 363–365 33 years old 359–365 39 years old 3.61 (0.90) 3.80 (0.89) 3.56 (0.78) 4.10 (0.92) 4.24 (0.91) 4.21 (0.95) 5.13 (1.03) 5.24 (1.02) 5.25 (0.98) 4.89 (0.72) 5.08 (0.78) 5.01 (0.74) 4.33 (0.54) 4.50 (0.57) 4.39 (0.50 24 years old 376–379 36 years old 369–370 42 years old 3.49 (0.95) 3.80 (0.94) 3.57 (0.80) 4.17 (0.89) 4.21 (0.94) 4.30 (0.90) 5.26 (1.01) 5.38 (0.98) 5.27 (0.96) 4.92 (0.73) 5.08 (0.79) 5.00 (0.76) 4.33 (0.54) 4.50 (0.57) 4.42 (0.51) 27 years old 312–314 39 years old 311–313 45 years old 3.50 (0.90) 3.77 (0.97) 3.54 (0.80) 4.08 (0.88) 4.28 (0.93) 4.26 (0.84) 5.34 (0.99) 5.37 (1.03) 5.32 (0.91) 4.97 (0.79) 5.10 (0.81) 4.98 (0.76)
4.35 (0.55)
4.34 (0.58)
4.52 (0.60)
4.40 (0.49)
na not applicable, TAB type A behavior
Courses of aggression, leadership, hard driving, eagerness energy, and global type-A behavior by age and measurement time (age- and cohort-related variance) are shown in Fig. 2. Hard driving and eagerness energy increased until the age of 33 years. Aggression and leadership did not have a clear pattern of increase or decrease.
Discussion Our findings from a 24-year longitudinal analysis yield novel information on how Hunter-Wolf type-A behavior and its components, i.e., aggression, leadership, hard driving, and eagerness energy, change from adolescence to adulthood,
Int.J. Behav. Med. (2014) 21:927–935
931
Table 2 Estimation of fixed and random effect for the full model of cohort effects in trajectories of type A behavior and its four subscales Aggressiveness
Leadership
Hard driving
Eagerness
Global A type
3.579* (0.041) −0.189* (0.017) 0.000 (0.006) 0.537* (0.113) −0.294* (0.058)
4.056* (0.033) 0.210* (0.017) −0.009 (0.006) 0.101* (0.046) –
5.470* (0.039) −0.342* (0.020) 0.022* (0.007) −0.133* (0.052) –
5.040* (0.027) −0.115* (0.014) 0.020* (0.005) −0.110* (0.037) –
4.426* (0.020) −0.084* (0.010) 0.005 (0.004) −0.013 (0.028) –
(reference) −0.015 (0.055) 0.103 (0.054) 0.194* (0.053) 0.221* (0.059) 0.371* (0.060)
(reference) 0.020 (0.044) −0.012 (0.044) 0.056 (0.043) 0.061 (0.043) 0.073 (0.052)
(reference) −0.086 (0.051) −0.276* (0.051) −0.446* (0.050) −0.586* (0.050) −0.715* (0.061)
(reference) −0.053 (0.036) −0.111* (0.036) −0.196* (0.035) −0.317* (0.035) −0.497* (0.042)
(reference) −0.021 (0.027) −0.044 (0.027) −0.055* (0.026) −0.109* (0.027) −0.141* (0.032)
(reference) −0.001 (0.009) −0.004 (0.008)
(reference) 0.006 (0.008) 0.006 (0.008)
(reference) 0.020* (0.010) 0.023* (0.009)
(reference) 0.004 (0.007) 0.010 (0.007)
(reference) 0.005 (0.005) 0.006 (0.005)
1971 1974 1977 Birth year by Age2
−0.020* (0.008) −0.009 (0.014) −0.074* (0.021)
0.006 (0.008) 0.021 (0.011) 0.030 (0.016)
0.011 (0.009) 0.015 (0.012) 0.053* (0.018)
0.018* (0.007) 0.036* (0.009) 0.041* (0.013)
0.002 (0.005) 0.021* (0.007) 0.027* (0.010)
1962 1965 1968 1971 1974 1977 Birth year by age3 1962 1965 1968 1971 1974 1977 Random effects Intercept
(reference) 0.076 (0.153) −0.031 (0.149) −0.043 (0.148) −0.301 (0.256) 0.339* (0.119)
(reference) −0.048 (0.062) −0.050 (0.061) −0.036 (0.061) −0.118 (0.076) −0.186 (0.096)
(reference) −0.093 (0.070) −0.069 (0.069) 0.062 (0.069) 0.096 (0.085) −0.119 (0.107)
(reference) 0.002 (0.050) −0.014 (0.050) −0.004 (0.049) −0.068 (0.062) −0.088 (0.081)
(reference) −0.019 (0.037) −0.023 (0.037) 0.024 (0.037) −0.061 (0.046) −0.107 (0.059)
(reference) −0.029 (0.078) 0.017 (0.077) 0.061 (0.077) 0.174 (0.109) –
– – – – – –
– – – – – –
– – – – – –
– – – – – –
0.885* (0.024)
0.767* (0.021)
1.080* (0.029)
0.491* (0.013)
0.288* (0.008)
Age Age2 Age3 Deviance
0.001 (0.001) 0.431* (0.110) 0.000 (0.000) 27,285.040
0.002* (0.001) 0.265* (0.103) – 26,979.270
0.001 (0.001) 0.362* (0.143) – 29,960.220
0.001* (0.000) 0.131* (0.064) – 22,759.480
0.001* (0.000) 0.114* (0.037) – 16,443.460
Fixed effects Intercept Gender Age Age2 Age3 Birth year 1962 1965 1968 1971 1974 1977 Birth year by age 1962 1965 1968
*p
Age- and Cohort-Related Variance of Type-A Behavior Over 24 Years: the Young Finns Study Taina Hintsa & Markus Jokela & Laura Pulkki-Råback & Liisa Keltikangas-Järvinen
Published online: 14 November 2013 # International Society of Behavioral Medicine 2013
Abstract Background Over the recent decades, the incidence of cardiovascular and heart diseases has decreased while levels of typeA behavior, i.e., a potential risk factor, appear to have increased. However, the long-term developmental patterns of type-A behavior is poorly understood. Both age- and cohortrelated changes may be involved in these developments. Purpose The purpose of this study was to examine an ageand cohort-related changes of Hunter-Wolf type-A behavior from adolescence to adulthood. Methods Type-A behavior and its components (aggressiveness, leadership, hard driving, and eagerness energy) were assessed using the Hunter-Wolf A-B rating scale at five time points (1983, 1986, 1989, 2001, and 2007) in a populationbased sample consisting of six birth cohorts born between 1962 and 1977 (n =3,341, a total of 10,506 person observations). Development of type-A behavior and its components was examined with cohort-sequential multilevel modeling. Results Aggressiveness decreased with age, eagerness energy, hard driving, and global type-A behavior increased, and leadership exhibited no mean level changes. Younger cohorts had
Electronic supplementary material The online version of this article (doi:10.1007/s12529-013-9369-z) contains supplementary material, which is available to authorized users. T. Hintsa (*) : M. Jokela : L. Pulkki-Råback : L. Keltikangas-Järvinen IBS, Department of Personality, Work and Health Psychology, University of Helsinki, P.O. Box 9, 00014 Helsinki, Finland e-mail: [email protected] M. Jokela e-mail: [email protected] L. Pulkki-Råback e-mail: [email protected] L. Keltikangas-Järvinen e-mail: [email protected]
higher aggressiveness, lower hard driving, and global type-A behavior. Conclusions The findings suggest that in order to understand the health consequences of type-A behavior, both life span and societal changes should be considered. Keywords Type-A behavior . Aggression . Leadership . Eagerness energy . Hard driving . Cohort study
Introduction The original type-A behavior concept, firstly presented by Friedman and Rosenman in 1959, was characterized by intense ambition, competitive drive, constant preoccupation with job-related deadlines, and a sense of time urgency [1]. Later, type-A behavior has been defined as “an actionemotion complex that includes (1) behavioral dispositions, (2) specific behaviors, and (3) emotional responses [2]. In the late 1970s, Western Collaborative Group study (WCGS) [3, 4] and Framingham study [5] suggested that type-A behavior contributes to a development of coronary heart disease (CHD). Later, both positive [6, 7] and negative findings [8–10] have been reported, and it has been concluded that evidence of type-A behavior as a prognostic factor of CHD has not been scholarly proved although it might have a minor etiological role in CHD among healthy adults [11]. There are, however, aspects that may compromise the negative evidence. First, the assessment methods of type-A behavior vary a lot, and the same assessments have resulted in conflicting findings in comparable samples. For instance, in a sample of WCGS participants type-A behavior has been reported to be predictive of CHD incidence [12, 13] whereas in a sample from the Multiple Risk Factor Intervention Trial (MRFIT) Study it has been unrelated to CHD [2, 14]. The
928
evidence from studies using questionnaires to assess type-A behavior has also been contradictory [15]. Second, type-A behavior is a multidimensional construct, and different measures focus on different contents [2]. Because only some components such as hostility and time urgence have been suggested to be “toxic” [16–20], while some of the type-A behavior components to be protective or represent healthy competitive style, i.e., leadership [16, 21], the role of the instrument becomes emphasized. Thus, it has been suggested that the dimensions of type-A behavior should be examined separately instead of the global rating [17, 18]. Third, the samples of previous prospective studies have consisted mainly of male subjects [14, 22, 23], and gender differences have been shown in associations between type-A behavior and CHD [2]. We have previously shown that eagerness energy component of Hunter-Wolf type-A behavior predicted adulthood atherosclerosis over different developmental transitions from early adolescence onwards in men but not in women [24]. Hunter-Wolf component of hard driving has been shown to predict decreased risk of atherosclerosis in women but not in men [24]. Fourth, and perhaps most important, the prevailing social and environmental approvals may change the meaning and effect of specific personality trait, and thus, the cohort effect should be systematically taken into account. In addition to individual variance, the development of personality traits may be influenced by social trends and societal circumstances that prevail during person’s formative years. For instance, studies in American youth have documented increasing levels of anxiety, and decreasing levels of internal locus of control and need for social approval during the late twentieth century [25, 26]. Consequently, it is the prevailing values and attitudes in the environment that are likely to influence at least reporting of type-A behavior during the time of measurement. Thus, time of measurement may also partly determine the health relevance of type-A behavior in relation to CHD risk, and in studies of the health consequences of type-A behavior, both age-related and societal changes should be considered. Results from the WHO MONICA project [27], on data from 6,695 Danish men and women aged 30, 40, 50, and 60 years, show that between the period of 1982 and 1992, myocardial infarction incidence declined by 3–5 % per year while type-A behavior increased in both men and women at the same time. This has been interpreted to confirm the negative evidence of type-A’s contribution to CHD, but it may refer to cohort effects, and suggest that scores of different decades are no more comparable and that the associations should be studied within a given cohort. The course of type-A behavior might be subject to birth cohort effects, but there are no previous research addressing this issue. We followed in a population-based sample (a) agerelated changes of type-A behavior and its subcomponents from early adolescence to middle adulthood, and (b) cohort-
Int.J. Behav. Med. (2014) 21:927–935
related changes of type-A behavior during 24 years (i.e., societal changes of type-A behavior during two decades). The cohort-sequential design allows one to test cohort differences in developmental trajectories.
Methods Participants The participants were 3,341 individuals (1,586 men, 1,755 women) from the population-based random sample of 3,596 men and women the Cardiovascular Risk in Young Finns Study (YFS) comprising six age cohorts (born in 1977, 1974, 1971, 1968, 1965, and 1962; i.e., 3, 6, 9, 12, 15, and 18 years at the baseline in 1980) who have been followed for 27 years. In 2007, 4.8 % of the participants had mandatory education and 35.4 % had higher degree. Type-A behavior was assessed at five time points in 1983, 1986, 1989, 2001, and 2007. Description of the birth cohorts and study phases are given in Fig. 1. We included all participants who had data on type-A behavior at least in one follow-up phase (n =3,341, a total of 10,506 person-observations). The study was approved by the university ethics committee. Details for selecting the sample have been given by Åkerblom et al. [28] and Raitakari et al. [28, 29]. Type-A Behavior Type-A behavior was measured on the Hunter-Wolf A-B Rating Scale [30]. The scale consists of four components to be answered in a form of a seven point ladder: (1) impatience– aggression (seven items, e.g., “I like to argue” and “I find it difficult to wait”), (2) leadership (six items, e.g., “I always like to win” and “I am always a leader in activities”), (3) Hard driving (three items, e.g., “I am hard-driving and competitive” and “ I take things seriously”), and (4) Eagerness energy (seven items, e.g., “I always feel hurried” and “ I think many things simultaneously”). The Cronbach’s alpha reliabilities of global type-A behavior at different measurement points varied Study phase
1983
n 1645
Age 152) 121)
183)
214)
1986
2000
125)
15
18
21
24
1989
2705
126)
15
18
21
24
27
2001
2089
24
27
30
33
36
39
2007
2038
30
33
36
39
42
45
Cohorts, born in 1) 1971 n=1809
2) 1968 n=1720
3) 1965 n=1513
4) 1962 n=1285
5) 1974 n=1301
Fig. 1 Description of the cohorts and study phases
6)1977 n=793
Int.J. Behav. Med. (2014) 21:927–935
between .67 and .79, and Cronbach alpha reliabilities of the components from .58 to .72. Statistical Analyses To study the developmental trajectories of type-A behavior and its subcomponents, we used multilevel linear modeling with a cohort-sequential study design. In a cohort-sequential study design (also known as an accelerated longitudinal design), participants who belong to different birth cohorts are followed over the same period of time but over different age periods, and then their developmental trajectories over age are used to construct a single trajectory spanning from the earliest age of the youngest cohort to the latest age of the oldest cohort. In the present case, a 24-year follow-up resulted in age periods of 12–30, 12–33, 12–36, 15–39, 18–42, and 21– 45 for the six birth cohorts in order of increasing age at baseline (type-A behavior was assessed only in participants aged 12 or older). The cohort-sequential design also allows one to test cohort differences in developmental trajectories. We assessed age× cohort interaction effects with a method formulated by Miyazaki and Raudenbush [31], where two models are fit to the data: one including cohort×age interaction effects (the full model) and the other without cohort effects (the reduced model). If the chi-square fit of the full model is significantly better than the fit of the reduced model, then cohort effects are present. This implies that birth cohorts may differ in the average level or in the shape of developmental trajectory of the outcome, which are shown by the cohort×age interaction effects. If the full model does not improve the fit of the reduced model, there are no significant differences between cohorts and a single trajectory can be used to characterize the development of the outcome over age. Cohorts were modeled with dummy variables with the oldest cohort as the references group. As described by Miyazaki and Raudenbush [31], participant’s age was centered on her/his cohort’s median age in the full model and on the median age of the total sample in the reduced model. Preliminary analyses suggested nonlinear developmental trajectories, so the effect of age was modeled with a linear and quadratic term with an additional cubic term for aggressiveness.
Results Figure 1 describes the cohorts and study phases. The number of participants in each cohort and mean values of type-A behavior at each measurement point are described in Table 1. To test whether type-A behavior was differentially associated with attrition between study waves, we fitted a series of logistic regression models by birth cohort and assessment time in which type-A behavior was used to predict the probability
929
of not participating in the next study wave. The odds ratios for the attrition odds are reported in supplementary Table 5. There were no systematic cohort differences in attrition related to type-A behavior. Table 2 shows results of the full model including cohort effects for each of the components and for the global scale. Table 3 shows the corresponding reduced models without cohort effects. Table 4 summarizes the deviance statistics of the full and reduced models and compares their fits for each outcome. With the exception of eagerness energy, the full model provided a significantly better fit to the data in the type-A traits, suggesting that there were cohort effects in the developmental trajectories of aggressiveness, leadership, hard driving, and global type-A scale. Aggressiveness followed a nonlinear developmental trajectory which decreased from age 12 to 24, increased temporarily up to age 33, and then followed a decreasing trend up to age 45. The average decrease from age 12 to age 45 was 0.34 points (0.37 SD). Cohort×age interaction effects (Table 2) indicated that the cohort differences in aggressiveness were related mainly to mean level differences, with younger cohorts having higher aggressiveness but largely similar developmental patterns than older cohorts over age. There was no clear developmental pattern in leadership, and the difference in leadership at age 12 and 45 was only 0.01 points (0.02 SD), indicating no age-related change over this period. And despite the better overall fit of the full model for leadership, there were no consistent cohort differences in average levels or age-dependent associations. Hard driving followed an increasing developmental trajectory that reached a plateau around age 27 and changed little after that. The change from age 12 to 27 was 0.64 points (0.60 SD) and the change from age 27 to 45 only 0.04 points (0.03 SD). Significant cohort main effects and cohort×age interaction effects indicated that younger cohorts had lower mean levels and their increasing developmental trajectories leveled off earlier than in older cohorts. Eagerness energy also increased from age 12 onwards, reached the highest value at age 33 (0.60 points or 0.79 SD compared to age 12) and decreased slightly from age 33 to 45 (0.12 points, 0.16 SD). There were statistically significant cohort interaction effects in eagerness, but given that their inclusion did not increase the overall fit of the full model over the reduced model, they are not interpreted as significant cohort effects. Thus, all cohorts followed essentially the same eagerness energy trajectory over age. The development of global type-A score reflected a combination of the development of its components, eagerness energy and hard driving in particular, so that it increased between age 12 and 33 (0.23 points, 0.42 SD) and began to decrease after that (0.10 points or 0.18 SD between ages 33 and 45). Cohort differences indicated that younger cohorts had lower mean scores than older cohorts.
930
Int.J. Behav. Med. (2014) 21:927–935
Table 1 Type A behavior in six cohorts by the measurement points from 1983 to 2007 Measurement time
1983
N, whole cohort (% males) n 1977 n =577 (51.1 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1974 n =583 (48.0 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1971 n =646 (50.0 %) 465–487 Aggression Leadership Hard driving Eagerness energy Global TAB 1968 n =651 (49.5 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1965 n =602 (47.7 %) Aggression Leadership Hard driving Eagerness energy Global TAB 1962 n =537 (47.9 %) Aggression Leadership Hard driving Eagerness energy Global TAB
1986 Mean (SD)
n
1989
n Mean (SD) 457–459 12 years old na na 3.86 (1.02) 4.23 (0.84) 4.58 (1.08) 4.49 (0.65) 4.24 (0.51) na 472–475 12 years old 479–480 15 years old 3.76 (0.92) 3.71 (0.94) 4.18 (0.76) 4.22 (0.88) 4.65 (0.98) 4.65 (1.12) 4.48 (0.64) 4.68 (0.65) 4.20 (0.45) 4.26 (0.53) 12 years old 482–484 15 years old 497–498 18 years old 4.02 (0.89) 3.80 (0.87) 3.68 (0.95) 4.33 (0.87) 4.18 (0.86) 4.16 (0.88) 4.69 (1.08) 4.67 (1.03) 4.85 (1.09)
4.55 (0.68) 4.35 (0.51) 476–486 15 years old 412 3.87 (0.90) 4.21 (0.90) 4.68 (1.08) 4.68 (0.71) 4.31 (0.52) 380–381 18 years old 332 3.78 (0.91) 4.19 (0.97) 4.88 (1.06) 4.76 (0.69) 4.33 (0.53) 288–294 21 years old 300–301 3.75 (0.96) 4.29 (0.91) 5.13 (1.00) 4.85 (0.74) 4.41 (0.57)
Mean (SD)
2001
2007
n Mean (SD) 338–341 24 years old 3.74 (0.95) 4.35 (0.87) 5.15 (1.03) 4.93 (0.74) 4.45 (0.55) 352 27 years old 3.77 (0.94) 4.31 (0.88) 5.21 (0.99) 5.09 (0.76) 4.50 (0.59) 348–349 30 years old 3.77 (0.94) 4.40 (0.95) 5.26 (1.09)
n Mean (SD) 318–319 30 years old 3.67 (0.79) 4.33 (0.87) 5.15 (0.98) 4.95 (0.71) 4.42 (0.51) 325–326 33 years old 3.66 (0.85) 4.35 (0.89) 5.30 (0.96) 5.12 (0.75) 4.50 (0.52) 356–360 36 years old 3.58 (0.81) 4.31 (0.90) 5.21 (0.99)
4.63 (0.61) 4.26 (0.48) 18 years old 461 3.65 (0.81) 4.13 (0.85) 4.85 (1.01) 4.73 (0.66) 4.26 (0.50) 21 years old 430–431 3.53 (0.83) 4.19 (0.89) 5.10 (1.02) 4.86 (0.70) 4.31 (0.51) 24 years old 381–382 3.55 (0.87) 4.17 (0.89) 5.25 (0.96) 4.93 (0.73)
4.80 (0.70) 5.10 (0.78) 5.08 (0.78) 4.30 (0.54) 4.54 (0.58) 4.44 (0.52) 21 years old 363–365 33 years old 359–365 39 years old 3.61 (0.90) 3.80 (0.89) 3.56 (0.78) 4.10 (0.92) 4.24 (0.91) 4.21 (0.95) 5.13 (1.03) 5.24 (1.02) 5.25 (0.98) 4.89 (0.72) 5.08 (0.78) 5.01 (0.74) 4.33 (0.54) 4.50 (0.57) 4.39 (0.50 24 years old 376–379 36 years old 369–370 42 years old 3.49 (0.95) 3.80 (0.94) 3.57 (0.80) 4.17 (0.89) 4.21 (0.94) 4.30 (0.90) 5.26 (1.01) 5.38 (0.98) 5.27 (0.96) 4.92 (0.73) 5.08 (0.79) 5.00 (0.76) 4.33 (0.54) 4.50 (0.57) 4.42 (0.51) 27 years old 312–314 39 years old 311–313 45 years old 3.50 (0.90) 3.77 (0.97) 3.54 (0.80) 4.08 (0.88) 4.28 (0.93) 4.26 (0.84) 5.34 (0.99) 5.37 (1.03) 5.32 (0.91) 4.97 (0.79) 5.10 (0.81) 4.98 (0.76)
4.35 (0.55)
4.34 (0.58)
4.52 (0.60)
4.40 (0.49)
na not applicable, TAB type A behavior
Courses of aggression, leadership, hard driving, eagerness energy, and global type-A behavior by age and measurement time (age- and cohort-related variance) are shown in Fig. 2. Hard driving and eagerness energy increased until the age of 33 years. Aggression and leadership did not have a clear pattern of increase or decrease.
Discussion Our findings from a 24-year longitudinal analysis yield novel information on how Hunter-Wolf type-A behavior and its components, i.e., aggression, leadership, hard driving, and eagerness energy, change from adolescence to adulthood,
Int.J. Behav. Med. (2014) 21:927–935
931
Table 2 Estimation of fixed and random effect for the full model of cohort effects in trajectories of type A behavior and its four subscales Aggressiveness
Leadership
Hard driving
Eagerness
Global A type
3.579* (0.041) −0.189* (0.017) 0.000 (0.006) 0.537* (0.113) −0.294* (0.058)
4.056* (0.033) 0.210* (0.017) −0.009 (0.006) 0.101* (0.046) –
5.470* (0.039) −0.342* (0.020) 0.022* (0.007) −0.133* (0.052) –
5.040* (0.027) −0.115* (0.014) 0.020* (0.005) −0.110* (0.037) –
4.426* (0.020) −0.084* (0.010) 0.005 (0.004) −0.013 (0.028) –
(reference) −0.015 (0.055) 0.103 (0.054) 0.194* (0.053) 0.221* (0.059) 0.371* (0.060)
(reference) 0.020 (0.044) −0.012 (0.044) 0.056 (0.043) 0.061 (0.043) 0.073 (0.052)
(reference) −0.086 (0.051) −0.276* (0.051) −0.446* (0.050) −0.586* (0.050) −0.715* (0.061)
(reference) −0.053 (0.036) −0.111* (0.036) −0.196* (0.035) −0.317* (0.035) −0.497* (0.042)
(reference) −0.021 (0.027) −0.044 (0.027) −0.055* (0.026) −0.109* (0.027) −0.141* (0.032)
(reference) −0.001 (0.009) −0.004 (0.008)
(reference) 0.006 (0.008) 0.006 (0.008)
(reference) 0.020* (0.010) 0.023* (0.009)
(reference) 0.004 (0.007) 0.010 (0.007)
(reference) 0.005 (0.005) 0.006 (0.005)
1971 1974 1977 Birth year by Age2
−0.020* (0.008) −0.009 (0.014) −0.074* (0.021)
0.006 (0.008) 0.021 (0.011) 0.030 (0.016)
0.011 (0.009) 0.015 (0.012) 0.053* (0.018)
0.018* (0.007) 0.036* (0.009) 0.041* (0.013)
0.002 (0.005) 0.021* (0.007) 0.027* (0.010)
1962 1965 1968 1971 1974 1977 Birth year by age3 1962 1965 1968 1971 1974 1977 Random effects Intercept
(reference) 0.076 (0.153) −0.031 (0.149) −0.043 (0.148) −0.301 (0.256) 0.339* (0.119)
(reference) −0.048 (0.062) −0.050 (0.061) −0.036 (0.061) −0.118 (0.076) −0.186 (0.096)
(reference) −0.093 (0.070) −0.069 (0.069) 0.062 (0.069) 0.096 (0.085) −0.119 (0.107)
(reference) 0.002 (0.050) −0.014 (0.050) −0.004 (0.049) −0.068 (0.062) −0.088 (0.081)
(reference) −0.019 (0.037) −0.023 (0.037) 0.024 (0.037) −0.061 (0.046) −0.107 (0.059)
(reference) −0.029 (0.078) 0.017 (0.077) 0.061 (0.077) 0.174 (0.109) –
– – – – – –
– – – – – –
– – – – – –
– – – – – –
0.885* (0.024)
0.767* (0.021)
1.080* (0.029)
0.491* (0.013)
0.288* (0.008)
Age Age2 Age3 Deviance
0.001 (0.001) 0.431* (0.110) 0.000 (0.000) 27,285.040
0.002* (0.001) 0.265* (0.103) – 26,979.270
0.001 (0.001) 0.362* (0.143) – 29,960.220
0.001* (0.000) 0.131* (0.064) – 22,759.480
0.001* (0.000) 0.114* (0.037) – 16,443.460
Fixed effects Intercept Gender Age Age2 Age3 Birth year 1962 1965 1968 1971 1974 1977 Birth year by age 1962 1965 1968
*p
