لینک اتیولوژیک ضعیف بین کنترل و رفتارهای برون سازی بزهکاری و سوء مصرف مواد در نوجوانان
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38630||2015||6 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Personality and Individual Differences, Volume 75, March 2015, Pages 179–184
Abstract Impulsive/disinhibitory personality traits have consistently been associated with externalizing symptomatology such as delinquency and substance use problems, often starting in adolescence. Yet the etiological nature of this co-occurrence is not well understood. Using a classic twin study design with self-report data from 717 male and female twin pairs, aged 15–18 years, a hierarchical psychometric model was examined. In this model the shared variance and etiological structure between control, delinquency and substance abuse symptoms, was modeled through a common externalizing factor. Model fitting indicated that the genetic and environmental influences differed in strength between male and female adolescents. The heritability of the externalizing factor was 45% in males and 10% in females, though neither was statistically different from zero. A statistically significant influence of shared environmental factors was seen for both sexes, 21% in males and 54% in females. In both sexes, the externalizing factor accounted for little variance in control, indicating a weak association and little shared etiology with externalizing liability. These results illuminate further that facets of impulsivity are differentially associated with vulnerability for externalizing symptomatology.
Introduction Research suggests that impulsive/disinhibitory traits, evident already in early childhood, are the basis or first manifestation of externalizing symptomatology and may be precursors to alcohol and drug experimentation and delinquent behavior once the opportunities for these behaviors arise, usually in adolescence (Caspi et al., 1996 and Masse and Tremblay, 1997). Although studies have consistently shown associations between impulsivity/disinhibition, antisocial behavior and substance use problems in adolescent populations (e.g. Chassin, Flora, & King, 2004), the etiological nature of this co-occurrence is not well understood. A critical limitation to previous research is the inconsistent operationalization of impulsivity/disinhibition across studies (see Sharma, Markon, & Clark, 2014 for a review). For further progress in the field, studies relating distinct facets of impulsivity to externalizing symptomatology are needed. The present study examines whether a clear operationalization of impulsivity, namely control, can be modeled as part of a latent externalizing liability factor together with delinquency and symptoms of substance abuse, as well as the etiological structure using a genetically informative sample of Norwegian adolescents aged 15–18. Recent efforts to distinguish different facets of impulsivity indicate that there are several weakly or moderately correlated traits that are frequently used to represent impulsivity/disinhibition, and that are differentially linked to various types of externalizing behavior (Dick et al., 2010 and Sharma et al., 2014). For example, low control has been directly associated with non-aggressive rule-breaking or antisocial behavior, but not with aggressive antisocial behavior (Burt & Donnelan, 2008). Findings of a broad latent externalizing factor is explained by a common underlying genetic risk to putatively separate disorders, so-called pleiotropic effects, i.e. the same genes may dispose to a range of similar traits or behaviors (Kendler, Prescott, Myers, & Neale, 2003). Few genetically informed studies have investigated whether the etiology of disinhibitory personality traits may be modeled as part of a latent externalizing factor. One genetically informed study of 12–18 year old twins found that sensation/novelty seeking could be modeled as part of an externalizing factor (Young, Stallings, Corley, Krauter, & Hewitt, 2000). Another study of 17-year-old twins combined multiple traits, including control, harm avoidance and traditionalism, and found support for an association with the externalizing factor (Krueger et al., 2002). Both Young et al.’s and Krueger et al.’s study support the supposition of an etiologic link between disinhibitory personality traits and psychopathology, and found a substantial genetic basis for the shared variance, a2 = .84 and .81, respectively. However, it remains unknown whether the distinct measure of control is an indicator of an underlying vulnerability to externalizing symptomatology. The current study examines adolescents aged 15–18, which is ideal for investigating the relationship between control, delinquency and substance abuse symptoms, as these traits/behaviors typically show an increase in prevalence and co-occurrence during the adolescent years (Collado et al., 2014 and Moffitt, 1993). We expected to find a shared etiological structure that could be modeled through a common externalizing factor. Due to the relative lack of studies investigating the etiology of externalizing symptoms in adolescence, and no previous studies that have included control as a distinct disinhibitory personality indicator, no specific hypotheses were made with regard to the relative strength of the etiological effects. However, based on previous findings of a strong genetic influence on externalizing liability, we did expect to find a highly heritable common factor also in the current study. Finally, epidemiological studies consistently show that adolescent males have greater prevalence of externalizing behaviors in general, they are involved in more delinquent acts and they abuse substances more often. However, it remains uncertain whether there are differences in the etiological structure of these behavioral patterns for adolescent males and females, as previous studies have provided inconsistent results (Hicks et al., 2007 and Krueger et al., 2002). We therefore investigated potential sex differences in the etiology.
نتیجه گیری انگلیسی
3. Results 3.1. Descriptives Means and standard deviations of the study variables, and the results of linear regressions examining the effects of age and sex on each variable, are shown in Table 1. Males scored significantly higher on control (reverse scored) than females, while females scored significantly higher on substance abuse than males. The only significant age effect was found for substance abuse, showing a higher severity with increasing age. Table 1. Means and standard deviations, and regression coefficients for the effects of sex and age on each variable. Mean SD Range Sex effect Age effect B (95% CI) B (95% CI) Control 1.03 0.36 0–2 .06 (.02, .10)⁎⁎ −.01 (−.03, .01) Delinquency 0.16 0.23 0–3 .01 (−.01, .04) .01 (−.00, .02) Substance abuse 0.61 1.06 0–6 −.19 (−.30, −.08)⁎⁎ 25 (.20, .30)⁎⁎⁎ Note: SD = standard deviation; B = unstandardized coefficient; CI = confidence interval. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options Phenotypic correlations for males and females are shown in Table 2 on the left hand side. For both sexes, control demonstrated a small correlation with delinquency and substance abuse, while delinquency and substance abuse were moderately correlated. Within twin pair correlations (within and cross trait) are presented in Table 2 on the right hand side. For males, the larger within-trait correlations for MZ pairs, when compared with DZ pairs, suggested that genetic factors may be involved for all phenotypes. Similarly, for female control and delinquency, the within-trait correlations for MZ twins were larger than for DZ twins. In the case of females’ substance abuse, the within-trait MZ and DZ pair correlations were practically equal in size, suggesting a minimal role for genetic effects in the expression of female substance abuse, but an important role for shared environmental effects. For opposite sex DZ twins, the within-trait correlations were generally of equal scale as the same sex DZ twins. Thus, any significant qualitative sex specific effects were not likely. Table 2. Estimated phenotypic (within individual) and within twin pair correlations stratified by sex and zygosity. Sex Phenotypic correlations Within twin pair correlations: Within trait (diagonal) and cross trait (off diagonals) MZ DZ same sex DZ opposite sex CON DEL CON DEL SUB CON DEL SUB CON DEL SUB Males CON – .40 −.05 .08 DEL .19 – .08 .47 .10 .11 .14 .34 SUB .19 .43 .10 .28 .60 .13 .22 .39 .13 .23 .33 Females CON – .27 .05 – DEL .29 – .23 .60 .19 .42 – – SUB .17 .38 .10 .20 .46 .03 .25 .45 – – – Note: CON = control; DEL = delinquency; SUB = substance abuse symptoms. Table options 3.2. Multivariate model fitting The results of the multivariate genetic analyses are presented in Table 3. The fully saturated model (Model I), allowing covariation between all variables without any constraint to any parameter, constituted the baseline model. A Correlated Factors model with quantitative sex differences (Model II) did not show significantly worse fit than such a model also allowing for qualitative sex differences (ΔLL = 10.15, Δdf = 15, p = .81). Furthermore, parameters could not be set equal across the sexes without a significant drop in fit (ΔLL = 78.24, Δdf = 9, p = <.001). This implies that the same genes and environmental factors affect the expression of the traits in males and females, but that their effects differ in magnitude. Due to observed quantitative sex differences, further model fitting was performed with sex heterogeneity. Because the IP and CP models with heterogeneity are not nested in the Correlated Factors model, the AIC and BIC were used to compare model fit. While the CP model indicated a better fit than the IP model, the Correlated Factors model was somewhat more parsimonious. Overall, the differences in model fit were rather small to make definitive conclusions. Given that the Correlated Factors model is not a theoretical competitor of the CP model, the CP model was retained, and results from the Correlated Factors model presented in addition to the results of the CP model. Table 3. Multivariate behavior genetic model fitting. No. ACE models −2 LL df AIC BICadj ΔLL Δdf P Comparison model I Saturated −3541.92 3873 −11287.92 −3082.94 – – – II Correlated het −3435.80 3975 −11385.80 −3323.61 105.92 102 .38 I III IP het −3442.73 3966 −11374.73 −3299.93 99.19 93 .31 I IV CP het −3429.55 3974 −11377.55 –3307.16 112.37 101 .21 I 13.18 8 .11 III Note: −2LL = negative 2 log-likelihood; df = degrees of freedom; AIC = Akaike’s Information Criterion; BICadj = sample size adjusted Bayesian Information Criterion; ΔLL = change in negative 2 log-likelihood; Δdf = change in degrees of freedom; SL = sex limitation; IP = independent pathway; CP = common pathway; Corr = correlated factors model; het = heterogeneity. Table options Fig. 1a and b shows the standardized and squared path estimates with confidence intervals for all factors in the CP model for males and females separately. “Externalizing behavior” was here modeled by means of a common latent factor with a moderate additive genetic component (45% of the common factor variance) for males and a small additive genetic component (10% of the common factor variance) for females. The remaining common factor variance was attributable to both shared (21% for males and 54% for females) and non-shared environment (34% for males and 36% for females). For both males and females, the additive genetic effect was non-significant, as indicated by the confidence intervals. The figures also show that the effect of the common latent factor was largest on the delinquency ratings, somewhat smaller on the substance abuse ratings, and smallest on the control ratings. Trait-specific effects were evident for all measured traits, with the strongest additive genetic effect on substance abuse for males and delinquency for females. (a and b) Final model with standardized and squared path estimates and ... Fig. 1. (a and b) Final model with standardized and squared path estimates and confidence intervals for males (a) and females (b). Ac = common additive genetic factor; As = phenotype-specific additive genetic factor; Cc = common shared environmental factor; Cs = phenotype-specific shared environmental factor; Ec = common non-shared environmental factor; Es = phenotype-specific non-shared environmental factor. Figure options Genetic and environmental correlations between each phenotype, as estimated in the Correlated Factors model, are presented in Table 4. The genetic correlations (rA) ranged from medium (.25) to large (.52), while the shared environmental correlations (rC) were all large (from .66 to 1.00), and non-shared environmental correlations ranged from small (.12) to medium (.36). As shown in Table 4, confidence intervals were mostly wide. Table 4. Genetic and environmental correlations (95% confidence intervals). rA rC rE CON DEL CON DEL CON DEL DEL .52 1 .12 (−.01, .97) (−1, 1) (−.02, .22) SUB .32 .25 .66 1 .14 .36 (−1, 1) (−1, .62) (.25, 1) (.53, 1) (.04, .23) (.25, .46) Note: CON = control; DEL = delinquency; SUB = substance abuse symptoms; rA = genetic correlation; rC = shared environmental correlation; rE = non-shared environmental correlation. Table options The importance of a particular rA, rC or rE on the covariance between phenotypes depends both on its magnitude and the specific genetic and environmental effects on the traits. In males, estimates for control were: h2 = .01, c2 = .15, e2 = .84, for delinquency: h2 = .28, c2 = .13, e2 = .59, and for substance abuse: h2 = .34, c2 = .22, e2 = .44. In females, estimates for control were: h2 = .21, c2 = .03, e2 = .77, for delinquency: h2 = .48, c2 = .13, e2 = .39, and for substance abuse: h2 = .02, c2 = .46, e2 = .52. Although genetic effects were overlapping between the traits, the explained proportion of phenotypic covariation due to common genetic effects was relatively low, explaining 10% between control and delinquency, 8% between control and substance abuse, and 18% between delinquency and substance abuse in males, and 57% between control and delinquency, 12% between control and substance abuse, and 6% between delinquency and substance abuse in females. Overall, environmental factors explained most of the covariance between the traits, as also found in the common pathway model.