Impact of childhood trauma and parental socialization on at-risk mental state in non-clinical adolescents

Abstract

BACKGROUND

Childhood trauma and parental socialization have been postulated as environmental factors of at-risk mental state (ARMS). Parental socialization is the process through which parents shape children’s self-regulation by providing guidance and protection. Although the impact of trauma on ARMS has been theorized, its clinical implications have not yet been fully clarified in adolescence, nor have explanatory models of parenting styles been established.

AIM

To investigate the role of traumatic experiences in the appearance of ARMS in the general adolescent population, considering the influence of parental socialization.

METHODS

A cross-sectional study of 697 adolescents aged 11-15 years was conducted, during which several questionnaires assessing childhood trauma, psychotic symptoms, and parenting styles were administered. The sample was divided into control, low-risk, medium-risk, and high-risk groups.

RESULTS

Some 2.8% (n = 19) of the adolescents presented ARMS and the presence of childhood trauma was associated with an increased risk of ARMS. Furthermore, the presence of abuse was greater in the high-risk and low-risk groups compared to controls. Regarding parental socialization, it was determined that a family socialization style based on greater affection–communication decreased the probability of ARMS. Finally, using PROCESS model 1 (regression-based path analysis that uses ordinary least squares regression), results suggested that low levels of affection and communication may mediate the relationship between childhood trauma and ARMS in adolescents.

CONCLUSION

These results highlight the importance of the early detection of trauma in preventing ARMS, without forgetting the importance of socialization styles.

Key Words: Early psychosis; Clinical high-risk; Trauma; Parenting styles; Adolescence

Core Tip: This study explores the associations between childhood trauma, parental socialization, and at-risk mental state (ARMS) in adolescents aged 11 to 15. Results suggest that trauma may be linked to a higher likelihood of ARMS, while parenting styles characterized by affection and communication appear to be associated with a lower risk. Low parental affection and communication may play a role in the relationship between trauma and ARMS, highlighting the potential importance of supportive family environments for early identification and prevention efforts.

INTRODUCTION

Adolescence represents a unique stage of human development characterized by rapid psychosocial growth. Mental disorders are one of the main causes of morbidity during this period[1]. Given that many mental disorders such as psychosis begin in adolescence, early detection could have a considerable impact[2]. Psychotic disorders are multifactorial[3]. Birchwood et al[4] identified a critical period of subthreshold affective, anxious, and attenuated psychotic symptoms that could constitute prodromes prior to the onset of frank psychosis[5]. This period of clinical high-risk was called at-risk mental state (ARMS). Its prevalence is a subject of debate[6] but the latest meta-analysis places it at 1.7%[6].

Childhood trauma has been studied as a risk factor for ARMS[7] and is defined as a stressful life event related to sexual or physical violence, emotional abuse, or neglect[8]. Such adverse childhood experiences include exposure to long-term environmental stressors such as child abuse, domestic violence, and interpersonal losses[9]. One in 4 children suffers abuse or neglect during their childhood, with 78% of cases being neglect in terms of care, 18% physical abuse, and 9% sexual abuse[9]. Psychotic symptoms in these patients are more long-lasting[10]. Of note, a meta-analysis of 6 studies on ARMS in adults found an 87% prevalence of childhood trauma, which was significantly higher than the rates recorded in the control cohorts (42%-60%)[11]. When approaching traumas, the following types are usually considered: Sexual abuse, physical and/or psychological abuse, and physical and/or emotional neglect[12]. In fact, Inyang et al[12] suggested that the general adolescent population should be screened to detect trauma in order to prevent ARMS. In this context, the Davidson Trauma Scale (DTS) showed satisfactory levels of internal consistency in the early detection of trauma[13].

However, most published peer-reviewed literature comes from cross-sectional or longitudinal studies with small sample sizes that did not consider specific trauma variables. Furthermore, previous research has focused on specific pathologies such as schizophrenia and the adult clinical population[14]. Finally, the application of the Clinical High-Risk for Psychosis paradigm in childhood and adolescence remains a topic of debate[15], because psychotic experiences are more common but have less predictive power for psychosis[16].

Parental socialization models are other relevant factors in psychosocial research on psychosis[17]. This is defined as the process through which parents shape the emotional growth, self-regulation, and discipline of children through protection, guidance, and teaching[18]. Although greater attention has been paid to family characteristics as indicators of relapse, evidence has also been published suggesting their role in the risk of psychosis[19]. For example, family criticism has been associated with an increased risk of psychotic symptoms[20]. Conversely, although less explored, positive aspects of family environments such as warmth, have also been identified and are considered protective factors[21]. Parenting styles are typically divided into two main dimensions: Affect and communication, or control and structure[22]. The first dimension highlights the emotional atmosphere in the parent-child relationship, such as warmth and communication quality, while the second focuses on the level of discipline.

Recent studies have examined the role of family support in the development of paranoia in individuals with a history of adverse experiences, highlighting the significance of ensuring protective family environments[23]. In addition, family attachment relationships shape beliefs about oneself and vulnerability to harm, playing a key role in shaping paranoid beliefs during adolescence[24]. Other articles deal with the importance of addressing parental safety behaviors in the evolution of paranoid experiences[25]. One of the validated questionnaires used to assess parental socialization with demonstrated strong reliability is the TXP Parental Socialization Questionnaire for Adolescents (TXP-A)[26].

Given all the above, the main objective of this current article was to establish the impact of traumatic life experiences in the emergence of ARMS in a non-clinical sample of adolescents, while also considering the type of trauma reported and the parental socialization model. Thus, we constructed the following hypotheses: (1) Individuals with ARMS have a greater traumatic burden than those in the control group; and (2) Parental socialization, specifically in terms of affect-communication, suggests its involvement in the debut of ARMS in adolescents with a history of trauma.

MATERIALS AND METHODS

Study type and sampling technique

This was a cross-sectional, analytical, case-controlled observational study. Non-random sampling was carried out in the general non-clinical adolescent population aged 11-15 years old, from 9 high schools in Castellón, Spain. Deliberate non-probabilistic sampling of the adolescent population was used. A sample of 697 adolescents from the first and second years of compulsory secondary education (ESO) was recruited. The research was conducted in high schools located in Castellón and its surrounding towns (Vila-real and Onda), and the sample included students and families from diverse cultural backgrounds, including South America, Morocco, Russia, China, and Ukraine.

Inclusion and exclusion criteria

The inclusion criteria were belonging to the first cycle of compulsory ESO, that is, the first or second year of ESO (corresponding to an age of 11-15 years) and having an adequate level of Spanish language. Students who had special educational needs were excluded. Most students who did not participate were excluded due to family refusal, given that only 3 were unable to participate because of an insufficient level of Spanish (n = 1) or an intellectual disability (n = 2).

Procedure

First, we contacted the directors at the educational centers included in this study. The sample was recruited between September and November 2022, during which time, self-administered questionnaires were individually completed by the adolescents during school hours. About 60-90 minutes were required to respond them. This phase was supported by 4 researchers trained in the tests. The psychometric tests employed were: (1) The Prodromal Questionnaire-Brief (PQ-B), comprising 21 items formulated in a dichotomous (true/false) format to assess prodromal symptoms of positive-type psychosis[27]. If the participant answered in the affirmative to an item, they were asked to indicate the degree of concern or discomfort this experience caused them, using a Likert-type scale with 5 response options. In terms of the psychometric properties of the PQ-B, the test presented an ordinal alpha of 0.95[28] and 0.92 in this sample; (2) The Youth Psychosis At-Risk Questionnaire Brief (YPARQ-B) comprises 28 items in a trichotomous (yes/no/unknown) format. The YPARQ-B assesses self-reported subclinical psychotic experiences in adolescents in the general population and was developed specifically to screen for ARMS symptoms during adolescence[29]. The internal consistency of the total YPARQ-B score was 0.94[28]; (3) The DTS is a self-administered questionnaire consisting of 17 items that evaluate the frequency and severity of post-traumatic stress disorder symptoms[13]. It allows 5 response alternatives each for frequency and severity. Adequate levels of internal consistency were found for this questionnaire, with a Cronbach alpha of 0.94[28,30]; (4) The TXP-A was designed for the Spanish population to evaluate parental socialization practices[26] considering affect and communication (including affective and communicative variables and low use of punishment) and control and structure (including roles and limits). The TXP-A showed high reliability (Cronbach Alpha of 0.87, 0.86 in this sample and a test-retest value of 0.94); and (5) In addition, the parents of the participants completed the Neurodevelopment in Adolescence-Parents 1 questionnaire. Our group created this test, which encompasses sociodemographic data (age and academic performance, among others) as well as information related to mental health history.

After requesting parental permission, the participants who exceeded the cut-offs from both the PQ-B and YPARQ-B (this criterion was used to reduce false positives) were interviewed in May and June of 2023, employing the comprehensive assessment of ARMS (CAARMS) individual interview. This semi-structured clinical interview assesses incipient positive symptoms, sensory-perceptive alterations, and thought and language disorders. Each item is rated on a scale of 1-6, where 3 is the cut-off point for ARMS and 6 corresponds to first psychotic episode. The CAARMS is a precise instrument with an area under the curve of 0.85 (95% confidence interval: 0.81-0.88) and a high sensitivity of 0.93 (95% confidence interval: 0.87-0.96)[31]. This interview was conducted by two researchers with experience in its use.

Classification of the initial sample of 697 participants was performed based on the results of the YPARQ-B and PQ-B as well as the CAARMS (Figure 1). Seven participants (1%) were excluded from the study because, even though they exceeded the cut-off points for both questionnaires, their parents did not give permission for the CAARMS interview to be carried out, which prevented their classification into one of the groups. The remaining participants were divided into 4 groups: (1) Healthy controls: Who did not exceed the cut-off points on the PQ-B or YPARQ-B (n = 413; 59.9% of the initial sample); (2) Low-risk group: Who exceeded the cut-off point in either the PQ-B or YPARQ-B, but not both (n = 235; 34.1%); (3) Intermediate-risk group: Who exceeded the cut-off points on both the PQ-B and YPARQ-B but did not meet any diagnostic criteria for ARMS on the CAARMS interview (n = 23; 3.2%); and (4) High-risk group (ARMS): Who exceeded the cut-off points of both the PQ-B and YPARQ-B and also met at least one diagnostic criterion for ARMS in the CAARMS interview (n = 19; 2.8%).

Figure 1 The sample selection process and its division into the four study groups. Thus, this study was carried out on a final sample of 690 individuals. The participants were divided into 4 groups: (1) Healthy controls: Who did not exceed the cut-off points on the Prodromal Questionnaire-Brief (PQ-B) or Youth Psychosis At-Risk Questionnaire Brief (YPARQ-B) (n = 413; 59.9% of the initial sample); (2) Low-risk group: Who exceeded the cut-off point in either the PQ-B or YPARQ-B, but not both (n = 235; 34.1%); (3) Intermediate-risk group: Who exceeded the cut-off points on both the PQ-B and YPARQ-B but did not meet any diagnostic criteria for the at-risk mental state (ARMS) on the comprehensive assessment of ARMS interview (n = 23; 3.2%); and (4) High-risk group (ARMS): Who exceeded the cut-off points of both the PQ-B and YPARQ-B and also met at least one diagnostic criterion for the ARMS in the comprehensive assessment of ARMS interview (n = 19; 2.8%). PQ-B: Prodromal Questionnaire-Brief; YPARQ-B: Youth Psychosis At-Risk Questionnaire Brief; CAARMS: Comprehensive assessment of at-risk mental state; ARMS: At-risk mental state.

Statistical analysis

Epidat 3.1 software[32] was used to calculate that, to evaluate the prevalence of the ARMS, which is 1.7% in the general adolescent population, with a precision of 1% and considering participants aged 11-15 years living in Castellon as the reference population (n = 8249)[33], the required sample size would need to be 596 individuals. Likewise, G*Power software (version 3.1.9.4) was used to calculate that the sample size required for multivariate analysis of variance would need to be 296 participants when applying an effect size 0.02, power of 0.80, and alpha of 0.05, with 4 groups and 4 variables. The data were analyzed with SPSS software (version 27; IBM Corp., Armonk, NY, United States).

After the exploratory and descriptive analysis and verification that the statistical assumptions were met, the relationships between the variables evaluated were studied using χ² tests for categorical variables and correlations for quantitative variables. multivariate analysis of variance and subsequent analysis of variance tests were used to analyze the differences between the study groups (the control, low-risk, intermediate-risk, and high-risk of ARMS groups) for the quantitative variables.

Given the difference in sample size between the different study cohorts, Bonferroni correction was used and considering these results, ordinal logistic regression was then performed to try to predict inclusion in each of the study groups. The group variable to which the participant belonged was considered ordinal, given that each group had a higher ARMS score than the previous one. Likewise, the data were modelled using the PROCESS v3.4 plugin[34] for SPSS to assess the relationships between ARMS, trauma, and parental socialization. PROCESS uses regression-based path analysis as a means of estimating various effects of interest (direct and indirect, conditional, and unconditional), using ordinary least squares regression to estimate the model parameters[34]. Model 1 was used to test moderation and model 4 was used to test mediation.

RESULTS

Sociodemographic and clinical characteristics

Table 1 shows the sociodemographic and clinical characteristics of the adolescents included in this study. Some 51.7% were female (n = 356), and the mean sample age was 12.6 years (SD = 0.7). A total of 57.4% (n = 396) belonged to a public school, 40.1% (n = 277) to a charter (state-subsidized) school, while only 2.5% (n = 17) belonged to a private school. Regarding academic performance, the mean grades obtained by most of the sample (70.2%; n = 390) were a “pass” (25.9%; n = 144) or “exceptional” (44.3%; n = 246). A total of 19.2% (n = 107) had, at some point, received care from mental health services. As shown in Table 1, there were no differences between the groups studied for most of the sociodemographic variables, except that there were more boys in the control group than in the low-risk group.

Table 1 Sociodemographic and clinical characteristics of the adolescents included in this study, n (%).

Characteristics		Overall sample	Controls	Low-risk	Intermediate-risk	High-risk (ARMS)	χ2/F, P value	ES, CTR
Sex	Male	333 (48.3)	221 (53.5)	95 (40.6)	9 (39.1)	8 (42.1)	11.11, 0.011a	0.127, 3.3/-2.9, -3.3/2.9
	Female	356 (51.7)	192 (46.5)	139 (59.4)	14 (60.9)	11 (57.9)
Age (years)	-	12.6 (0.7)	12.6 (0.7)	12.6 (0.7)	12.6 (0.4)	12.6 (0.7)	0.25, 0.860	0.001
School type	Public	396 (57.4)	237 (57.4)	136 (57.9)	11 (47.8)	12 (63.2)	3.13, 0.792	0.048
	Charter	277 (40.1)	168 (40.7)	91 (38.7)	11 (47.8)	7 (36.8)
	Private	17 (2.5)	8 (1.9)	8 (3.4)	1 (4.3)	0 (0)
School year	1st CSE	350 (50.7)	205 (49.6)	121 (51.5)	14 (60.9)	10 (52.6)	1.22, 0.747	0.042
	2nd CSE	340 (49.3)	208 (50.4)	114 (48.5)	9 (39.1)	9 (47.4)
Academic performance	Fail	33 (5.9)	14 (4.3)	17 (8.7)	1 (5)	1 (6.7)	11.09, 0.269	0.082
	Pass	144 (25.9)	77 (23.8)	55 (28.1)	5 (25)	7 (46.7)
	Exceptional	246 (44.3)	148 (45.7)	85 (43.4)	8 (40)	5 (33.3)
	Outstanding	132 (23.8)	85 (26.2)	39 (19.9)	6 (30)	2 (13.3)
Prior mental health follow-up	No	451 (80.8)	269 (82)	153 (78.5)	17 (85)	12 (80)	1.23, 0.753	0.047
	Yes	107 (19.2)	59 (18)	42 (21.5)	3 (15)	3 (20)
Psychiatric medication	No	537 (95.9)	313 (95.4)	190 (96.4)	20 (100)	14 (93.3)	1.44, 0.727	0.051
	Yes	23 (4.1)	15 (4.6)	7 (3.6)	0 (0)	1 (6.7)
Mental disorder	No	614 (89)	376 (91)	202 (86)	20 (87)	16 (84.2)	4.51, 0.204	0.081
	Yes	76 (11)	37 (9)	33 (14)	3 (13)	3 (15.8)
Substance consumption	No	673 (97.5)	409 (99)	224 (95.3)	23 (100)	17 (89.5)	14.37, 0.015a	0.144, 3.1/-2.7/-2.3, -3.1/2.7/2.3
	Yes	17 (2.5)	4 (1)	11 (4.7)	0 (0)	2 (10.5)
Video game addiction	No	557 (99.5)	327 (99.7)	195 (99)	20 (100)	15 (100)	1.36, 0.639	0.049
	Yes	3 (0.5)	1 (0.3)	2 (1)	0 (0)	0 (0)
Bullying	Never	414 (75.1)	261 (80.6)	133 (68.6)	11 (57.9)	9 (64.3)	16.09, 0.015a	0.121, 3.5/-2.6, -2.1, -2.6/2.5
	Maybe	79 (14.3)	38 (11.7)	32 (16.5)	5 (26.3)	4 (28.6)
	Yes	58 (10.5)	25 (7.7)	29 (14.9)	3 (15.8)	1 (7.1)
Needs mental health care	No	459 (83)	283 (87.1)	154 (79)	13 (68.4)	9 (64.3)	12.90, 0.048a	0.108, 3, -2.7
	Maybe	57 (10.3)	24 (7.4)	26 (13.3)	4 (21.1)	3 (21.4)
	Yes	37 (6.7)	18 (5.5)	15 (7.7)	2 (10.5)	2 (14.3)

^aP < 0.05.

CSE: Compulsory secondary education; ARMS: At-risk mental state; ES: Effect size evaluated by Partial Eta Squared; CTR: Corrected typified residuals.

In terms of the clinical variables (Table 1), there were significant differences in the perceptions of the parents of the participants regarding substance use, bullying, and the need for mental health help by their child. Specifically, the corrected standardized residuals indicated that the intermediate and high-risk (ARMS) groups presented a higher proportion of substance consumption than the control group. Compared to the low-risk group, a greater proportion of the control group replied “never” or “perhaps” when asked if they had suffered bullying and the low-risk group had experienced bullying in a higher proportion than the control group. Finally, in the control group, a greater proportion of the parents of participants considered that their child did not need mental health care, compared to those who thought their child “perhaps” needed this kind of support.

Trauma

Table 2 shows comparisons of the variables related to trauma. The control group scored lower on the DTS scale than all the other groups and the low-risk group scored lower than the intermediate and high-risk groups. In fact, the three risk groups (low, intermediate, and high) more frequently exceed the DTS cut-off point than the control group. In other words, the presence of trauma was greater in these three groups than in the control group. Regarding the specific traumas experienced (Table 2), the low-risk group had experienced physical or emotional neglect more frequently than the control group, while the low and high-risk groups had experienced a greater frequency of abuse than the control group. Finally, the control group comprised more individuals who had not experienced any traumas.

Table 2 Variables related to trauma, n (%).

Characteristics		Overall sample	Controls	Low-risk	Intermediate-risk	High-risk (ARMS)	χ2/F P value	ES CTR/significant differences
DTS (total score), mean ± SD		28.45 ± 27.32	16.06 ± 18.01	42.76 ± 25.44	69.91 ± 33.80	71.11 ± 31.15	128.03, < 0.001c	0.363, C < B/C < I/C < A/B < I/B < A
DTS (cut-off point)	No	480 (70.8)	363 (89.2)	111 (48.5)	4 (17.4)	2 (10.5)	186.90, < 0.001c	0.525, 12.9/-9.1/-5.7/-5.9 -12.0/9.1/5.7/5.9
	Yes	198 (29.2)	44 (10.8)	118 (51.5)	19 (82.6)	17 (89.5)
Physical/emotional neglect	No	646 (94)	395 (95.9)	212 (91)	20 (87)	19 (100)	9.59, 0.032a	0.118, 2.5/-2.4, -2.5/2.4
	Yes	41 (6)	17 (4.1)	21 (9)	3 (13)	0 (0)
Sexual abuse	No	681 (99.1)	411 (99.8)	228 (97.9)	23 (100)	19 (100)	6.62, 0.097	0.098
	Yes	6 (0.9)	1 (0.2)	5 (2.1)	0 (0)	0 (0)
Abuse	No	632 (92)	392 (95.1)	207 (88.8)	19 (82.6)	14 (73.7)	20.10, 0.001b	0.193, 3.7/-2.2/-3, -3.7/2.2/1.7
	Yes	55 (8)	20 (4.9)	26 (11.2)	4 (17.4)	5 (26.3)
Natural disaster	No	673 (98)	400 (97.1)	231 (99.1)	23 (100)	19 (100)	4.07, 0.203	0.077
	Yes	14 (2)	12 (2.9)	2 (0.9)	0 (0)	0 (0)
Interpersonal event	No	673 (98)	407 (98.8)	226 (97)	22 (95.7)	18 (94.7)	4.09, 0.180	0.077
	Yes	14 (2)	5 (1.2)	7 (3)	1 (4.3)	1 (5.3)
Grief of a death	No	560 (81.5)	324 (78.6)	197 (84.5)	21 (91.3)	18 (94.7)	7.34, 0.056	0.103
	Yes	127 (18.5)	88 (21.4)	36 (15.5)	2 (8.7)	1 (5.3)
Grief without a death	No	667 (97.1)	403 (97.8)	223 (95.7)	22 (95.7)	19 (100)	3.07, 0.363	0.067
	Yes	20 (2.9)	9 (2.2)	10 (4.3)	1 (4.3)	0 (0)
Interpersonal problems	No	635 (92.4)	382 (92.7)	215 (92.3)	20 (87)	18 (94.7)	1.18, 0.778	0.042
	Yes	52 (7.6)	30 (7.3)	18 (7.7)	3 (13)	1 (5.3)
Others	No	552 (80.3)	343 (83.3)	180 (77.3)	17 (73.9)	12 (63.2)	7.77, 0.049a	0.106, 3, -2.3
	Yes	135 (19.7)	69 (16.7)	53 (22.7)	6 (26.1)	7 (36.8)

^aP < 0.05.

^bP < 0.01.

^cP < 0.001.

DTS: Davidson Trauma Scale; ARMS: At-risk mental state; ES: Effect size evaluated by Partial Eta Squared; CTR: Corrected typified residuals.

Parental socialization

Table 3 presents the scores for the parental socialization questionnaire. The control group scored higher for affect and communication than the other three groups and in turn, the low-risk group had a higher score than the intermediate and high-risk groups. Additionally, the control group also scored higher for control and structure than the low-risk group.

Table 3 Parental socialisation, mean ± SD.

Characteristics	Overall sample	Controls	Low-risk	Intermediate-risk	High-risk (ARMS)	F, P value	ES, significant differences
Affection and communication	75.18 ± 9.78	77.65 ± 7.82	72.02 ± 10.76	70.70 ± 13.99	66.05 ± 11.53	26.69, < 0.001b	0.106, C > B/C > I/C > A/L > A
Control and structure	36 ± 5.08	36.49 ± 5.01	35.18 ± 5.27	36.87 ± 3.87	34.42 ± 4.24	3.58, 0.014a	0.016, C > B

^aP < 0.05.

^bP < 0.01.

ARMS: At-risk mental state; ES: Effect size evaluated by Partial Eta Squared; CTR: Corrected typified residuals.

Ordinal logistic regression

As shown in Table 4, all the variables with differences between the groups could be individually used to predict the group to which the participant had belonged. However, when we evaluated the model adjusted for the remaining variables, only the affect and communication score, DTS score, and presence of other traumas had still maintained this predictive power.

Table 4 Ordinal logistic regression: Dependent variable group (control, low-risk, intermediate-risk, or high-risk of ARMS).

Variable	Unadjusted model			Adjusted model
	OR	95%CI	P value	OR	95%CI	P value
Sex (female)	1.659	1.225-2.247	0.001b	1.004	0.676-1.492	0.984
Substance use	3.568	1.489-8.550	0.004b	1.735	0.581-5.178	0.323
Bullying	2.110	1.246-3.571	0.005b	1.402	0.740-2.653	0.300
Potential mental health need	2.297	1.345-3.922	0.002b	1.543	0.840-2.836	0.162
Affect and communication	0.937	0.922-0.952	< 0.001c	0.971	0.946-0.997	0.029a
Control and structure	0.959	0.932-0.987	0.005b	1.017	0.969-1.067	0.499
DTS score	1.055	1.047-1.063	< 0.001c	1.046	1.037-1.056	< 0.001c
Physical-emotional neglect	2.037	1.121-3.702	0.020a	1.039	0.463-2.328	0.927
Abuse	2.996	1.756-5.110	< 0.001c	1.931	0.999-3.734	0.050
Other traumas	1.607	1.111-2.325	0.012a	1.785	1.091-2.919	0.021a

^aP < 0.05.

^bP < 0.01.

^cP < 0.001.

DTS: Davidson Trauma Scale; CI: Confidence interval; OR: Odds ratio.

Modelling the relationship between variables

To evaluate the role of parental socialization on the relationship between trauma (DTS score) and ARMS (according to the participant classification group), we tested whether the affect and communication variable acted as a moderator (model 1) or mediator (model 4) in said relationship. The data adjusted better to the mediation model shown in Figure 2. Of note, the quantitative variable of trauma was taken into account by employing the DTS score rather than the score from each specific trauma, given that the latter is a categorical variable that results in the same data.

Figure 2 Model of how parental socialization (affection and communication) mediated the relationship between trauma (Davidson Trauma Scale score) and an at-risk mental state (group into which the participants had been classified). DTS: Davidson Trauma Scale; ARMS: At-risk mental state.

DISCUSSION

In this study, we examined the potential risk of childhood trauma on having ARMS in a large sample of the general adolescent population while also considering parental socialization style. Likewise, the prevalence of ARMS in non-clinical adolescents was assessed. We found that 34.1% (n = 235) of adolescents in our sample presented attenuated symptoms, albeit with varying degrees of risk, which were classified as low, intermediate, or high. The high-risk group (ARMS) comprised 2.8% of adolescents (n = 19). The participants were classified into groups according to previously published academic literature: Fonseca-Pedrero et al[35] classified individuals into 6 groups of increasing latent risk, while others[36] classified them into 3 groups, to try to compare the impact of trauma and parental socialization regarding the severity of psychotic symptomatology.

The prevalence of ARMS in the general non-clinical adolescent population is a topic of epidemiological debate, with disparity because of the different evaluation methods employed[6]. Furthermore, some previous articles have only focused on the prevalence of transition to psychosis[37]. Classically, the prevalence of ARMS ranges from 1 to 8%[38], with some studies suggesting that it is 7.3%[16]. Nonetheless, the most recent meta-analysis[6] places it at 1.7%, one percentage point lower than the prevalence we report in this current work. A possible explanation could be the younger mean age in our sample. Another reason could be that the parents of healthy children declined to participate in order to avoid social stigma.

One of our main findings was that the combination of childhood trauma and a parenting style with little affect and communication is associated with greater severity of psychotic symptoms. In line with previous findings[39], the results of this present study showed that 78.9% of adolescents with ARMS (n = 15) had reported some form of childhood trauma. Furthermore, in support of our first hypothesis, the distribution of reports of having experienced trauma across the groups was statistically significant, with each successively higher risk group showing a higher percentage of traumatic experiences, culminating in the ARMS group. It has been postulated that trauma experienced during childhood can lead to negative perceptions about oneself, others, and the environment[11]. In this sense, early bullying or physical abuse can foster mistrust towards others, which in turn, could contribute to the debut of paranoid ideas or self-referential thoughts[11]. Moreover, there is also evidence for the role of biological mechanisms in the appearance of symptoms. First, the stress vulnerability model[39] explains how biological changes related to stress exposure from trauma interact with underlying predisposing vulnerability, generating a cytokine imbalance[40]. Second, the neurodevelopmental trauma model focuses on the causal role of childhood trauma in the development of psychosis, suggesting that the increased stress sensitivity seen in psychotic patients is related to neurological brain changes after exposure to trauma[41].

From among the different types of childhood trauma, our study considered physical and emotional neglect, physical and psychological abuse, sexual abuse, trauma derived from adverse natural events, and trauma caused by grief and interpersonal trauma. Thus, we included trauma categories that are usually considered unusual in order to consider the main life events adolescents considered traumatic[12]. Here, we observed a higher prevalence of physical and ARMS emotional neglect and of abuse in the low-risk group when compared to the controls. Physical neglect is defined as the inability of the caregiver to provide basic physical care such as hygiene, food, shelter, and medical care[42]. In contrast, emotional neglect refers to inadequacy in terms of attending to emotional needs and providing individual attention[42]. To date, the prevalence of care neglect in children with ARMS has been little studied, despite its potential clinical impact[43].

Surprisingly, we did not obtain statistically significant differences between the groups regarding sexual abuse. In our sample, 0.9% (n = 6) of the adolescents reported having been sexually abused, all of whom were female. Sexual abuse is usually postulated as the type of trauma most associated with the severity of psychotic symptoms and so the result we obtained here may be the result of underdiagnosis, given that previous evidence suggests that the prevalence is higher (13.5% in women and 5.6% in men[44]). The low reported prevalence in our study may reflect underreporting, possibly due to cultural stigma surrounding disclosure of sexual abuse, which can inhibit honest reporting, especially in adolescent populations[45]. Additionally, methodological factors such as the sensitivity of the questionnaire used may have limited detection, because standardized measures might not fully capture the complexity or secrecy of sexual abuse experiences.

In this context, emotional neglect may be related to a certain style of parental socialization[46]. The influence of family from the first stages of life, through attachment, on the correct development and upbringing of children has already been described[47]. Attachment is defined as the emotional connection formed based on the unconditional love and care provided by parents to their children. In this sense, previous studies describe how insecure attachment promotes a greater risk of experiencing early psychotic experiences, as well as worse clinical variables in adulthood[48]. In fact, people with schizophrenia spectrum diagnoses have reported higher levels of insecure attachment[49]. Moreover, recent literature establishes that insecure attachment is related to psychotic experiences both in non-clinical populations[50] and clinical groups[51].

The results of the present study showed a negative relationship between parental socialization style based on affect and communication and ARMS. This supports the second of our proposed hypotheses, given that the use of this educational style by parents may act as a protective factor against ARMS. However, we did not detect any conclusive differences between the high-risk and control groups in terms of the parental socialization style based on control and structure. Moreover, in addition to the association between childhood trauma and the emergence of the early psychotic symptoms described above, extensive academic literature has linked the latter to having been parented with a precarious socialization style in childhood[52]. In this sense, and in agreement with our data, some studies have shown that solid family support characterized by high affect and communication would act as a protective element against psychotic symptoms[52].

Although this study was conducted in the specific sociocultural context of Spain -where public schools include a high proportion of culturally diverse students (mainly from Morocco, Romania, China, Ecuador, and Bolivia[53]) - certain cultural factors may have influenced the observed outcomes. Spanish parenting is generally characterized by familism, emotional closeness, and strong involvement in the educational process, combining affection with clear boundaries and fostering open, bidirectional communication[54]. This style, often described as democratic or indulgent, may have shaped adolescents’ responses in this study. In contrast, other cultural settings may emphasize different parenting norms - such as early autonomy in the United States and United Kingdom, egalitarian approaches in Nordic countries, or more authoritarian parenting in some Asian societies[55]. Latin American cultures share elements of familism, though with contextual differences[56]. Recent Spanish research also highlights adolescents’ perceptions of high maternal involvement and family control[57,58], which may further contextualize our findings. These cultural dynamics, together with the use of non-random sampling, underscore the importance of replicating this study in other populations to assess the generalizability of our results.

Our study uncovered the mediating role of low family affect and communication in the relationship between adverse childhood experiences and the genesis of ARMS. Consequently, poor affective family dynamics should be regarded as a warning sign. Here, it is essential to point out the difference between the mediating and moderating models[34], established by Baron and Kenny[59] in 1986. In a mediation model, variable X affects variable Y because X affects the mediator variable M, and this causal effect then transmits the effect of X to Y through the effect of M on Y[60]. Mediation analysis has been around in various forms for at least 70 years but Baron and Kenny[59] popularized an approach to using it employing easy to understand regression analysis principles[60]. Subsequently, there has been increasing emphasis on estimating the indirect effect using different methods, the most popular being the bootstrap confidence interval[61]. Whereas mediation analysis focuses on how a causal effect operates, moderation analysis is used to address, when, or under what circumstances, or for what types of people that effect exists and at what magnitude[60]. These models are also called interaction models and the most popular methods for testing them is simple slope analysis, the pick-a-point approach, or spotlight analysis[60]. An alternative approach is the use of the Johnson-Neyman technique[62].

In the context of the above, our moderation model examines when or under what conditions the relationship between trauma and ARMS changes based on parenting styles, while a mediation model focuses on explaining how trauma affects ARMS through parenting styles. In this sense, promoting an adequate family environment characterized by empathy and in which problems can be verbalized, can have positive effects on mental health. In other words, this aforementioned approach helps to prevent the opposite situation in which everyday problems are not openly expressed, allowing them to become chronic and a substrate for other, more difficult problems[63]. From this perspective, it becomes important to work with the family members of individuals with ARMS to promote adaptive family functioning.

In terms of prevention, the American Academy of Pediatrics recommends screening all adolescents at age 12 to identify those at risk of depression and/or suicidal ideation[64]. However, the ARMS model usually assesses the presence of psychosis in help-seeking subjects[65]. Given that not all individuals at risk of developing psychotic-spectrum pathologies present significant distress, this criterion leaves individuals with symptoms barely perceptible by community members, undetected. This is when family and school counsellors become more relevant. Nonetheless, it is important that this be done by optimizing screening strategies so that we do not inadvertently fall into overdiagnosis. In this sense, addressing the history of trauma as well as parental socialization would provide valuable information.

This present research is subject to some limitations. First, most of the metrics used relied on adolescent self-reports, which may have introduced bias, especially in the assessment of trauma and parenting styles. Perceptions can be affected by current symptoms or memory distortions, potentially leading to under or overreporting. Future studies should consider multi-informant approaches to improve the validity of these assessments. However, to minimize this bias, sociodemographic information provided by parents was available and the CAARMS clinical interview was conducted with the individuals from our sample at the highest risk. Furthermore, different studies have shown the reliability of self-reports in the adolescent population[66]. Another limitation was the disproportion between the study group sample sizes because, logically, as the risk of psychosis increases the prevalence in the general population decreases. We minimized this bias by applying the Bonferroni correction. However, this disproportion increases the likelihood of type II errors, so the results must still be interpreted with caution. Another potential source of bias may be the failure to assess, and therefore to include in the regression models, pre-study factors such as socioeconomic status, family structure, or comorbid mental disorders, which could influence the results. Also, the cross-sectional design of the study prevents us from establishing causal relationships or determining the temporal order of the variables (trauma, parental socialization, and ARMS). Although the associations are theoretically supported, it remains unclear whether parenting style influences ARMS, or whether emerging symptoms affect how parenting is perceived. Longitudinal studies are needed to clarify these relationships. The exclusion of 7 possible ARMS cases because the families refused to allow completion of the CAARMS interview was also a limitation. Furthermore, because of perceived social stigma, affected individuals could choose to hide their symptoms and discomfort and not seek help. Moreover, the fact that individuals with cognitive difficulties or low Spanish proficiency (e.g. immigrants) were excluded, was a limitation. Nonetheless, only 3 students were excluded for these reasons. Finally, future longitudinal studies will be needed to validate mediation pathways and track ARMS progression over time.

CONCLUSION

This study showed that low affect-communication parental socialization is likely involved in the debut of the ARMS in adolescents with a history of childhood trauma. Thus, the presence of traumatic events was associated with greater risk of ARMS while a parental approach characterized by high affect and communication was found to reduce the chances of developing ARMS. These findings highlight the relevance of identifying trauma and emotionally dysfunctional family dynamics - characterized by low emotional warmth, poor communication, and limited parental support or empathy - early in adolescence to avoid ARMS.