Momentary Self-esteem as a Process Underlying the Association Between Childhood Trauma and Psychosis: Experience Sampling Study

Background: Exposure to childhood trauma is associated with an increased risk of developing and maintaining psychotic symptoms later in life. Self-esteem might be an important psychological process underlying the association between childhood trauma and psychosis, but there is only limited evidence to support this claim, especially in daily life. Objective: In this study, we aimed to investigate whether exposure to childhood trauma (physical, emotional, and sexual abuse and physical and emotional neglect) moderates the cross-sectional and temporal associations between self-esteem and psychotic experiences in patients with psychotic disorders, their first-degree relatives, and controls. Methods: We assessed momentary self-esteem and psychotic experiences in daily life using the experience sampling method in 139 patients with psychotic disorders, 118 first-degree relatives of patients with psychotic disorders, and 111 controls. Childhood trauma was measured using the Childhood Trauma Questionnaire. We fitted linear mixed models and added two-way and three-way interaction terms to test the hypotheses. Results: The association between momentary self-esteem and psychotic experiences in daily life was modified by prior exposure to high versus low levels of several types of childhood trauma, that is, physical ( χ 22 =24.9, family-wise error-corrected P <.001) and sexual abuse ( χ 22 =15.9, P <.001) and physical neglect ( χ 22 =116.7, P <.001). Specifically, momentary self-esteem was associated with more intense psychotic experiences in patients exposed to high versus low levels of physical neglect, in relatives exposed to high versus low levels of physical abuse, and in relatives and controls exposed to high versus low levels of sexual abuse. When investigating temporal order, the results showed no evidence that childhood trauma modified the temporal associations between self-esteem at t n-1 and psychotic experiences at t n or those between psychotic experiences at t n-1 and self-esteem at t n . Conclusions: The association between self-esteem and psychotic experiences in daily life was found to be stronger in those exposed to high versus low levels of several types of childhood trauma (ie, physical abuse, sexual abuse, and physical neglect).


Background
Exposure to childhood trauma can have persistent adverse effects on an individual's well-being, social development, and physical and mental health [1].Childhood trauma refers to potentially harmful experiences of physical, emotional, and sexual abuse and physical and emotional neglect during childhood [2].Accumulating evidence suggests that childhood trauma is associated with psychotic disorders [3][4][5][6] and increases the risk of developing and maintaining psychotic symptoms later in life [3,7,8].The study by Norman et al [9] showed that the prevalence of childhood trauma in patients with psychotic disorders is very common; 82 of their 100 participants (patients receiving treatment from an early psychosis clinic) reported exposure to childhood trauma.Similarly, other studies have shown that patients with psychotic disorders reported having experienced disproportionate levels of childhood trauma [3,7,8,[10][11][12].Furthermore, the results of the meta-analysis conducted by Varese et al [3] suggested that 1 out of 3 patients might not have developed psychosis if adversity were to be eliminated as a risk factor (assuming causality), which indicates that childhood trauma plays a prominent role in the development of psychosis.It has also been demonstrated that psychotic symptoms are more severe in patients exposed to childhood trauma [9,13].A prospective cohort study indicated that the incidence of psychotic experiences decreased significantly when exposure to trauma ceased over the course of the study [14].Finally, a systematic review revealed that exposure to childhood adversity was associated with the persistence of psychotic symptoms in both the general population and clinical studies [15], suggesting that childhood adversity may play an important role in the maintenance of psychotic symptoms in patients with enduring psychotic disorders.
Evidence for familial liability to psychosis [16][17][18] shows that first-degree relatives of patients with psychosis have an increased risk of developing psychotic disorders [17,19].In addition, although exposure to childhood trauma is higher in patients with psychotic disorders, patients and their siblings share a degree of exposure to childhood trauma [20], as they also share many sociodemographic, parental, and developmental characteristics [21].
Several putative psychological mechanisms and processes have been posited to be involved in the association between childhood trauma and psychosis [22,23].Self-esteem is one such process.A systematic review indicated that exposure to childhood trauma may contribute to low self-esteem [24].Low self-esteem has also been found to be common in patients with psychotic disorders [25][26][27][28].Moreover, self-esteem has been shown to be involved in the development and maintenance of psychotic symptoms [28][29][30][31][32]. Lower levels of momentary self-esteem have been found to be associated with an increased intensity of psychotic experiences in daily life [33].Prior exposure to childhood trauma may affect cognition [34], which contributes to the development of a complex system consisting of negative views of one's self, the world, and the future [35].Some researchers have suggested that self-esteem is involved in the pathway from negative life events (eg, childhood trauma) to psychosis, but evidence to support this claim remains limited [29,36,37].A few studies have investigated self-esteem and closely related processes such as negative self-schemas, or negative beliefs about self and others, as an underlying process in the association between childhood trauma and psychotic experiences [38][39][40][41][42].In contrast, other researchers have suggested low self-esteem to be a product of an individual's experience of psychosis, for example, owing to hospitalization, the loss of their social role or rank, or stigmatization [43,44].
Self-esteem and psychotic experiences are often assessed using cross-sectional measures, with global scores [38][39][40][41].Another way of collecting data on these variables is through the experience sampling method (ESM).ESM assesses moment-to-moment variations in thoughts, feelings, and behaviors in daily life.It generates longitudinal data with a very limited recall bias and high ecological validity [45].

Objective
Overall, there is limited evidence that exposure to childhood trauma moderates the association between self-esteem and psychotic experiences, especially in daily life.Exposure to childhood trauma plays an important role in the development and maintenance of psychosis [17][18][19]46], but unfortunately, preventing childhood trauma remains challenging.By investigating whether self-esteem is a relevant underlying process in the association between childhood trauma and psychotic experiences, we might be able to identify individuals exposed to (high levels of) childhood trauma in the early stages and target self-esteem to prevent a full-blown psychotic disorder.Therefore, we aimed to investigate whether cross-sectional and temporal associations of momentary self-esteem and psychotic experiences in daily life were modified by prior exposure to childhood trauma (ie, physical, emotional, and sexual abuse and physical and emotional neglect) in patients with enduring psychotic disorders, their first-degree relatives, and controls.Specifically, using data from the Genetic Risk and Outcome in Psychosis (GROUP) study [47], a large multicenter study of patients with psychotic disorders, their first-degree relatives, and controls without a family history of psychotic disorders, we aimed to test the following hypotheses: 1.In patients, their first-degree relatives, and controls, the magnitude of associations of self-esteem and psychotic experiences in daily life (both measured with the ESM) is stronger in individuals exposed to high levels of each type of childhood trauma (ie, physical, emotional, and sexual abuse and physical and emotional neglect) than in those exposed to low levels of childhood trauma (measured using the Childhood Trauma Questionnaire [CTQ]; hypothesis 1). 2. The difference in the magnitude of associations of self-esteem and psychotic experiences in daily life between those exposed to high levels of childhood trauma and those exposed to low levels of childhood trauma is (1) greater in patients than in controls, (2) greater in relatives than in controls, and (3) greater in patients than in relatives (hypothesis 2).

XSL • FO
RenderX 3. In patients, their first-degree relatives, and controls, the temporal association (1) between self-esteem at t n-1 and psychotic experiences at t n and (2) between psychotic experiences at t n-1 and self-esteem at t n in daily life (both measured with the ESM) is stronger in individuals exposed to high levels of each type of childhood trauma than those exposed to low levels of childhood trauma (measured using the CTQ; hypothesis 3). 4. The difference in the magnitude of temporal associations of (1) self-esteem and psychotic experiences and (2) psychotic experiences and self-esteem in daily life between those exposed to high levels of childhood trauma and those exposed to low levels of childhood trauma is (1) greater in patients than in controls, (2) greater in relatives than in controls, and (3) greater in patients than in relatives (hypothesis 4).

Sample
A sample of patients with psychotic disorders, their first-degree relatives, and controls without a family history of psychotic disorders was recruited in the GROUP study [47], a large longitudinal multicenter study in the Netherlands and Belgium.Individuals with psychotic disorders were recruited from regional psychosis care facilities or academic centers in selected geographical areas.The patients' relatives were contacted after obtaining written informed consent.Participants in the control group were recruited by contacting random addresses in the same geographical areas as the patients.To be eligible, participants had to be between 16 and 50 years old and their command of Dutch language had to be sufficient.As an additional criterion, patients had to be diagnosed with a nonaffective psychotic disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria [48].In the control group, individuals with a family history of psychotic disorders were excluded.

Ethics Approval
Ethics approval was provided by the Ethical Review Board of the University Medical Centre Utrecht (METC: 0 4/003-O), and all participants gave written informed consent.For this analyses, only participants who completed the CTQ and, in line with previous ESM studies [49], at least one-third (33.3%) of the ESM assessments were included.

Basic Sample Characteristics
Using a sociodemographic schedule, data on age, gender, ethnicity, marital status, and educational level (adapted from the Dutch Standard Classification of Education) [50] were collected.

Childhood Trauma
The Dutch version of the 25-item CTQ-short version [2] was used to assess childhood trauma at the baseline.The 25 items inquire about 5 types of trauma in childhood (emotional, physical, and sexual abuse and emotional and physical neglect).
All 5 types of trauma are covered with 5 items rated on a 5-point Likert scale (1=never true to 5=very often true).The CTQ-short version has shown to be reliable and to provide adequate content coverage.There is also good evidence of criterion validity in both adolescent patients with a psychiatric disorder and individuals from a normative community sample [2].For the analyses, the mean score for each type of trauma was used.

ESM Procedure
The ESM was used at the 6-year follow-up assessment to collect data on self-esteem and psychotic experiences in daily life.

ESM Measures
To assess momentary self-esteem, the mean score of the following 2 ESM items was used: "I like myself" and "I doubt myself" (reversed) [55] (following the approach by Geldhof et al [56] to analyze multilevel reliability; within-person α=.22; between-person α=.76).To assess the intensity of psychotic experiences, the mean score of the following 8 ESM items was used: "My thoughts are influenced by others," "I can't get these thoughts out of my head," "I feel unreal," "My thoughts can't be expressed in words," "I feel suspicious," "I hear voices," "I see things that aren't really there," and "I am afraid I will lose control" [57,58] (within-person α=.64; between-person α=.85).
All the items were measured on a 7-point Likert scale, ranging from 1 ("not at all") to 7 ("very much").

Statistical Analysis
For this analyses, we used release 7.0 of the overall GROUP database and release 2.0 of the ESM data.The "mixed" command in Stata 13.0 (StataCorp) was used to fit linear mixed models.Because ESM data include multiple observations within each participant, they were treated as nested within participants and participants nested within families.Restricted maximum likelihood estimation was used to fit the models and estimated the associations between momentary self-esteem as the continuous independent variable and psychotic experiences as the outcome variable.We controlled for the potential confounders of age, gender, ethnicity (people of color or White), level of education, and marital status.
We then added two-way (self-esteem × abuse, self-esteem × group, and abuse × group) and three-way (self-esteem × abuse × group) interaction terms to test whether associations between self-esteem and psychotic experiences were modified by prior exposure to high (mean + 1 SD of continuous CTQ scores) versus low levels (mean − 1 SD of continuous CTQ scores) [59,60] of each type of childhood abuse to each group (patients, relatives, and controls).We standardized the continuous ESM and CTQ variables (mean 0, SD 1) to interpret significant three-way interaction terms [61].Wald tests were used to test the hypothesis that the interaction effects were equal to zero.
For the three-way interaction models, the significance levels of the Wald tests were adjusted to correct for type-I errors using family-wise error-corrected P value, which were computed by multiplying the unadjusted P value by the total number of tests.
The "lincom" command was used to compute linear combinations of coefficients to test hypotheses 1 and 2. Next, we conducted time-lagged analyses to test hypotheses 3 and 4.
To do so, we generated within-subject lagged variables of self-esteem and psychotic experiences (at t n-1 and t n ) and fitted linear mixed models.

Basic Characteristics
The third-wave (6-year follow-up) data of the GROUP study was completed by 486 participants (patients: n=194, 39.9%; relatives: n=169, 34.8%; and controls: n=123, 25.3%).Participants who did not complete the CTQ in the first wave (30/486, 6.2%) and a minimum of one-third of the ESM assessments in the third wave (88/486, 18.1%) were excluded from the analysis, resulting in a sample of 368 participants (patients: n=139, 37.8%; relatives: n=118, 32.1%; and controls: n=111, 30.2%).The basic characteristics of the excluded participants of the third-wave sample were broadly similar to those of the included participants (Multimedia Appendix 1).Overall, the basic characteristics of the included patients, their relatives, and controls were similar, except that included participants of all 3 groups were more often White compared with the excluded participants.Compared with the relatives and controls, the patients were younger and had a lower educational level.The patient group consisted of more men and were more often unmarried compared with the other 2 groups (Table 1 3).The levels of all types of childhood trauma were similar in controls and relatives.

Cross-sectional Association Between Self-esteem and Psychotic Experiences by Childhood Trauma and Group
As shown in Table 4, after controlling for age, gender, ethnicity, education, and marital status, we found strong evidence that the association between self-esteem and psychotic experiences was modified by exposure to childhood trauma, as indicated by the statistically significant interaction effect of self-esteem × CTQ total score × group (χ 2  2 =9.0, P=.01).We found statistically significant interaction effects (all P≤.001) of self-esteem × physical abuse × group, self-esteem × sexual abuse × group, and self-esteem × physical neglect × group.However, there was no evidence of interaction effects of self-esteem × emotional abuse × group (χ 2 2 =5.4,P=.07) and self-esteem × emotional neglect × group (χ 2  2 =1.8, P=.41).

Within-Group Comparison-Hypothesis 1
Lower levels of momentary self-esteem were associated with more intense psychotic experiences in relatives exposed to high levels of childhood trauma in general compared with those exposed to low levels of childhood trauma (adjusted β high vs low =−.04;P=.03).There was no evidence that this association was stronger in patients or controls exposed to high versus low levels of childhood trauma in general.Specifically, in relatives, we found a stronger association between lower self-esteem and more intense psychotic experiences for those exposed to high versus low levels of physical abuse (adjusted β high vs low =−.12;P<.001); however, we did not find evidence that this association was modified in patients or controls (Figure S1 in Multimedia Appendix 2).Furthermore, lower self-esteem was associated with more intense psychotic experiences in relatives and controls exposed to high versus low levels of sexual abuse (relatives: adjusted β high vs low =−.05;P<.001 and controls: adjusted β high vs low =.05; P=.049) but not in patients (Figure S2 in Multimedia Appendix 2).Finally, in patients, we found a stronger association between lower self-esteem and more intense psychotic experiences in those exposed to high versus low levels of physical neglect (adjusted β high vs low =−.13;P<.001; Figure S3 in Multimedia Appendix 2).However, there was no evidence that this association was modified in relatives and controls.

Between-Group Comparison-Hypothesis 2
Next, we examined the differences in the magnitude of associations of self-esteem and psychotic experiences between those exposed to high versus low levels of childhood trauma in general, physical and sexual abuse, and physical neglect across groups.These differences in magnitude were only examined if both groups showed significant (P<.05) within-group associations (Table 5).When comparing relatives with controls, we found differences in the magnitude of associations between self-esteem and psychotic experiences between those exposed to high versus low levels of sexual abuse (adjusted β high vs low =−.10;P<.001), with differences in magnitude of association being greater in relatives.

Temporal Association Between Self-esteem and Psychotic Experiences by Childhood Trauma and Group
Findings regarding the temporal associations between momentary self-esteem and psychotic experiences in patients, relatives, and controls are shown in Tables 6 and 7.After controlling for age, gender, ethnicity, education, and marital status, we found no evidence that associations between self-esteem at t n-1 and psychotic experiences at t n or between psychotic experiences at t n-1 and self-esteem at t n were modified by exposure to any of the types of childhood trauma (ie, emotional, physical, and sexual abuse and emotional and physical neglect) within (hypothesis 3) and across (hypothesis 4) groups (ie, patients, relatives, and controls).b Adjusted β, standardized regression coefficients (continuous independent variables were standardized [mean 0, SD 1] for interpreting significant three-way interaction terms and examining the difference in associations between high [mean + 1 SD], average [mean], and low [mean − 1 SD] levels of childhood trauma within and across groups [patients, relatives, and controls]).c Family-wise error-corrected P values were computed by multiplying the unadjusted P value by the total number of tests to adjust significance levels of Wald tests for three-way interactions.b Adjusted β, standardized regression coefficients (continuous independent variables were standardized [mean 0, SD 1] for interpreting significant three-way interaction terms and examining the difference in associations between high [mean + 1 SD], average [mean], and low [mean−1 SD] levels of childhood trauma within and across groups [patients, relatives, and controls]).c Family-wise error-corrected P values were computed by multiplying the unadjusted P value by the total number of tests to adjust significance levels of Wald tests for three-way interactions.

Principal Findings
Using an experience sampling design, the results of this study showed strong evidence that associations between momentary low self-esteem and increased intensity of psychotic experiences in daily life were modified by several types of childhood trauma, that is, physical and sexual abuse and physical neglect.For physical abuse, this was only the case for relatives of patients with psychotic disorders, and for physical neglect, this was only the case for patients.Sexual abuse modified the association in relatives and controls, but not in patients.Emotional abuse and XSL • FO RenderX emotional neglect did not modify the association between self-esteem and psychotic experiences in any group.When investigating temporal order, we found no evidence that childhood trauma modified the temporal associations between self-esteem at t n-1 and psychotic experiences at t n or those between psychotic experiences at t n-1 and self-esteem at t n .

Methodological Considerations
Several limitations of this study should be considered when interpreting its findings.First, only a selection of the baseline sample made it through the third wave.For our analyses, we had to exclude 118 individuals because they did not complete either a sufficient number of ESM assessments or the CTQ.Possibly, these assessments were too burdensome and, therefore, might have led to selection bias.However, when comparing the excluded individuals with those in the analytical sample, who participated in the third wave of GROUP assessments, in terms of basic sample characteristics, the included and excluded participants were comparable to a great extent, except for ethnicity (included participants were more often White in all 3 groups).Second, the CTQ, which was used to measure exposure to several types of trauma during childhood, is a retrospective self-report measure.It has been argued that the CTQ is prone to recall bias and that the manifestation of psychotic symptoms might affect the ratings of this measure [62,63].However, Gayer-Anderson et al [64] demonstrated the accuracy, strength of agreement, and convergent validity to be broadly similar between patients with first-episode psychosis and controls.Notably, although measured differently, in our control group, the prevalence of several childhood trauma types was higher than that in the general population in the Netherlands [65].However, if we only look at the categories "severe to extreme" and "moderate to severe" in our sample, the percentages are quite similar, except for sexual abuse.Possibly related to this, we found that lower levels of self-esteem were associated with more intense psychotic experiences in controls exposed to high versus low levels of sexual abuse; this was not the case for any of the other types of childhood trauma (in controls).This might indicate that higher levels of sexual abuse in this group might have affected the outcome.In addition, all ESM assessments of self-esteem and psychotic experiences were based on subjective self-reports, which might have led to bias.However, the ESM has been found to be a feasible, reliable, and valid assessment method in various populations [45,51,66], including patients with psychosis [51,52,58].
Moreover, momentary self-esteem was measured using only 2 items.Using fewer items in ESM research is quite common because it minimizes reactivity to the assessment method [67].The construct of self-esteem involves both positive and negative self-esteem [68,69].Therefore, we used 1 item that measured the positive dimension and 1 that measured the negative dimension of self-esteem, which is in line with previous ESM studies investigating self-esteem [55,70].Because of this heterogeneity of the self-esteem construct and, hence, ESM items, the internal consistency of the 2 momentary self-esteem items that we used was as expected (within-person α=.22; between-person α=.76) [71].In previous analyses, we investigated the extent to which the 2 ESM items adequately tapped into the self-esteem construct in a subsample of the current sample.We found that the convergent validity between the 2 momentary self-esteem items and the Rosenberg Self-Esteem Scale [69] was fair, with the intraclass correlation coefficient being 0.41 [33].It would have been interesting, however, to triangulate momentary ESM measures with explicit (Rosenberg Self-Esteem Scale) and implicit measures [72] of self-esteem to corroborate our findings.In addition, for future research, we would recommend using multiple items to measure both the positive and negative dimensions of momentary self-esteem.
Next, we controlled for potential a priori confounders, such as age, gender, ethnicity, education level, and marital status.We did not include medication use as a confounder because 98% (98/100) of the patients were using medication.Nevertheless, unmeasured confounders, such as other childhood adversities, comorbidities, the impact of illness chronicity, and molecular genetic measures, were not considered and may have, therefore, influenced the findings.
Finally, we standardized the continuous childhood trauma variables because it allowed us to interpret associations at higher versus lower levels of childhood trauma [60,61].Notably, this implies that by using continuous and standardized variables for childhood trauma, no discrete distinction was made between those exposed and not exposed, rather between those exposed to higher and lower levels of each type of childhood trauma.Therefore, these results could possibly be an underestimation of the reality.

Comparison With Previous Research
There is well-established evidence that exposure to childhood trauma is a risk factor for developing a psychotic disorder [3][4][5][6][7][8][10][11][12], and self-esteem has been proposed to be involved in the pathway from childhood trauma to psychosis [38][39][40][41][42].In addition, mediation models linking childhood trauma and self-esteem in pathways to psychosis have been previously proposed and tested [41,73].However, evidence regarding whether prior exposure to childhood trauma (emotional, physical, and sexual abuse and emotional and physical neglect) moderates the association between momentary self-esteem and psychotic experiences in daily life remains limited.
As hypothesized (hypothesis 1), and extending previous findings [33], we found evidence that exposure to physical and sexual abuse and physical neglect modified the association between momentary self-esteem and psychotic experiences in daily life.However, the results also suggested that exposure to emotional abuse and emotional neglect did not modify the association between momentary self-esteem and psychotic experiences in daily life.This is unexpected, as previous studies demonstrated that when comparing the 5 types of childhood trauma, emotional abuse and neglect are the types that affect self-esteem the most [74,75].Therefore, we would expect that especially these types of trauma would have a particular bearing on the association between self-esteem and psychosis.However, the timing, chronicity, and severity of emotional abuse and neglect are related to the extent to which maltreatment affects developmental trajectories [76].For example, when maltreatment is less severe or ends early in childhood, it is XSL • FO RenderX possible that its impact might fade with time in some individuals [77,78].Indeed, our data show that the vast majority of the participants experienced none or mild to moderate levels of emotional abuse and neglect.Another explanation for these unexpected results might be that cross-sectional modeling of associations between momentary self-esteem and psychotic experiences does not consider moment-to-moment variation in these measures.However, when we evaluated these associations using time lags, we observed that exposure to childhood trauma did not modify the temporal association between self-esteem and psychotic experiences and vice versa.
On the basis of previous literature [38][39][40][41][42], we hypothesized that the effect of exposure to childhood trauma would modify the association between low self-esteem and increased intensity of psychotic experiences in daily life most substantially in patients, followed by their relatives.We found that physical neglect modified the association between self-esteem and psychotic experiences in patients; physical abuse modified this association only in relatives; and sexual abuse modified this association in relatives and controls, with the impact being stronger in relatives than in controls.It is possible that the controls exposed to childhood trauma may have better coping strategies [79] compared with first-degree relatives of patients with psychotic disorders who have a familial liability to psychosis.Therefore, controls might be more resilient to lower levels of self-esteem and psychotic experiences in daily life [79].In line with this, there is also evidence that controls exposed to high levels of childhood trauma are more resilient to daily life stress, compared with patients with first-episode psychosis, individuals with at-risk mental state, and help-seeking service users [22,23].Moreover, previous literature showed that resilience at baseline was lower in individuals who developed psychosis at follow-up than in those who did not [80], suggesting that resilience is a protective factor in the formation of psychosis [81].
In addition, relatives have a familial liability to psychosis [16,18], and they share a degree of exposure to childhood trauma with their siblings with psychotic disorders [20], which may explain the stronger associations in relatives than in controls.Patients were recruited via treating clinicians, which implies that they all received a form of (standard) mental health care, including medication.Treatments such as psychoeducation, cognitive behavioral therapy, and antipsychotic medication have been shown to reduce symptoms and prevent relapse [82][83][84].It has been demonstrated that the therapies target distorted beliefs about delusions and hallucinations, thereby decreasing the negative consequences of psychotic symptoms [83][84][85].Moreover, the results of a meta-analysis showed that the interpretation of these beliefs is addressed during treatment by considering psychological mechanisms and processes that might contribute to the formation and maintenance of psychotic symptoms, such as emotions, arousal, attachment, interpersonal issues, trauma, and self-esteem [86].For example, the results of a randomized clinical trial showed that treatment, such as eye movement desensitization and reprocessing and prolonged exposure, in patients with psychotic disorders reduced trauma symptoms and psychotic symptoms [87], also at 6-month follow-up.This potentially implies that the influence of trauma on patients might become less impactful over time owing to the effects of the treatment they had received.Furthermore, 98% (98/100) of patients were using antipsychotic medication.Emotional flattening might be a negative symptom of schizophrenia, but it is also a side effect of medication.Emotional flattening interferes with expressiveness, and this often leads to problems in interpersonal interactions, which in turn leads to more withdrawal from (social) activities [88].Selective perception and selective memory are important features of low self-esteem, and as a consequence, negative convictions about one's self are repeatedly confirmed in new situations [89].However, this will not occur often if patients who are using antipsychotics socially withdraw and therefore experience fewer social interactions.Potentially, directly or indirectly, medication use and the effects of illness chronicity might have a flattening effect on self-esteem.Therapy and medication use might limit the impact of exposure to traumatic experiences on the association between self-esteem and psychotic experiences in daily life for patients with enduring psychotic disorders.
Overall, the cross-sectional findings indicate that it might possibly be relevant to develop and implement early and low-level interventions that directly target self-esteem, especially for individuals exposed to physical and sexual abuse and physical neglect.Thus, the intensity of psychotic experiences in daily life can be reduced.We are currently evaluating the ecological momentary intervention (EMI) "SELFIE" [90].The next step would be to implement these types of EMIs in routine mental health care.
Previous research has investigated the temporal order of self-esteem and psychotic experiences in daily life and found that self-esteem preceded psychotic symptoms only in controls and psychotic experiences had a temporal effect on self-esteem in patients with psychotic disorders, their first-degree relatives, and controls [33].In this study, we investigated whether prior exposure to childhood trauma modified the temporal associations between self-esteem at t n-1 and psychotic experiences at t n and between psychotic experiences at t n-1 and self-esteem at t n .However, the results showed no evidence of these associations.Nevertheless, we would not rule out the hypothesis of a temporal order entirely because, in this study, we investigated the interaction between prior exposure to childhood trauma, psychotic experiences at t n , and momentary self-esteem at the previous time point (t n-1 ) and vice versa.Assessments took place within 90-minute time blocks.It is possible that longer time lags would have yielded different results, as the occurrence of psychosis might be preceded by weeks, months, or even years of psychological and behavioral abnormalities [91].Perhaps self-esteem is a process that needs more time to unfold to be succeeded by psychotic experiences.

Conclusions
Taken together, our findings suggest that the association between self-esteem and psychotic experiences in daily life is stronger in those exposed to high versus low levels of several types of childhood trauma.Hence, self-esteem might be a psychological process that links childhood trauma to psychosis.The results showed that this is especially the case in first-degree relatives and, to a lesser extent, in patients with psychotic disorders.Although we did not find evidence of temporal associations, the cross-sectional results indicate that improving self-esteem may potentially reduce the intensity of psychotic experiences in daily life.This underlines the importance of developing and evaluating early and low-level EMIs that directly target self-esteem to reduce the intensity of psychotic experiences in daily life.We currently evaluate such an intervention in the SELFIE study [90].

Table 2 .
Categorical Childhood Trauma Questionnaire severity scores by group.

Table 3 .
Differences in scores of Childhood Trauma Questionnaire severity between groups.
a CI adjusted for age, gender, ethnicity, education, and marital status.

Table 4 .
Associations between momentary self-esteem and psychotic experiences by childhood trauma in patients, relatives, and controls a .

Table 5 .
Difference in associations in those exposed to high versus low levels of childhood trauma across groups (Δ high vs low).

Table 6 .
Effect of self-esteem and psychotic symptoms at t n-1 on psychotic experiences at t n by group and modified by childhood trauma a .Adjusted for age, sex, ethnicity, level of education, and marital status. a

Table 7 .
Effect of self-esteem and psychotic symptoms at t n-1 on self-esteem at t n by group and modified by childhood trauma a .