Childhood Maltreatment Types As A Prognosticator Of Dysfunctional Attitudes In Major Depressive Disorder

Background:Depression（Major depressive disorder,MDD）is a heterogeneous disorder.Despite the high global prevalence,the mechanism underlying depression remains under investigation.The symptomatology,intensity,treatment response,and onset of depression vary from individual to individual.Therefore,it is critical to identify the risk factors that account for these variations to advance the understanding of depression.Childhood maltreatment（CM）,an established environmental risk factor of depression,influences the incidence of depression and its characteristics.Besides having a strong relationship with depression,childhood maltreatment is linked with earlier onset of depression,and affects the severity of the course of depression.There are five subtypes of CM: emotional abuse（EA）,physical abuse（PA）,sexual abuse（SA）,emotional neglect（EN）,and physical neglect（PN）.However,when considering the specific childhood maltreatment subtypes,physical abuse is associated with 1.5 times higher odds of developing depression and emotional abuse is associated three times higher risk of developing depression compared to those who never underwent CM.Under Beck’s views of depression,a child’s exposure to stressful life events,such as childhood maltreatment,can potentially develop cognitive vulnerabilities.The cognitive vulnerabilities are commonly referred to as dysfunctional attitudes.These dysfunctional attitudes are hallmark features of depression,and shall remain latent unless activated by a stressor.When activated,depression might result.However,only few studies investigated whether the dysfunctional attitudes are features exclusive to childhood maltreatment.There are eight forms of dysfunctional attitudes that have been described in medical literature.The association of the specific childhood maltreatment subtypes and the various forms of dysfunctional attitudes has never been investigated.Such knowledge can contribute to the medical community to understand the pathway from childhood maltreatment to depression.Objectives:1.To investigate the association between dysfunctional attitudes and childhood maltreatment.2.To investigate the predictability of the various childhood maltreatment subtypes on the different forms of dysfunctional attitudes.3.To investigate the predictability of cumulative childhood maltreatment types on the different types of dysfunctional attitudes.Methods:Part 1:1.In this study,two group of participants were enrolled;an MDD group including 171 MDD participants and a healthy control（HC）group including 156 healthy controls.2.Childhood maltreatment（CM）was retrospectively assessed using the Chinese version of the childhood trauma questionnaire short form（CCTQ-SF）.Dysfunctional attitudes were assessed using the Chinese version of the dysfunctional attitudes scale-form A（C-DAS-A）.Depression severity was assessed through a 24-items Hamilton depression rating scale(HAMD₂₄).3.Independent sample T-tests were used to examine group differences for continuous variables and Chi-square tests were used to analyze group differences for categorical variables.4.A series of two way analysis of covariance and post-hoc analyses,with unmatched demographic variables as covariates,were run to assess the effects of diagnosis（MDD and HC）,history of childhood maltreatment（CM+ and CM-）,and their interactions on the C-DAS-A total and subscale scores（vulnerability,attraction and repulsion,perfectionism,compulsion,seeking applause,dependence,self-determination attitude,and cognition philosophy）.Part 2:1.In this study,two groups of participants were enrolled;an MDD group including 168 MDD participants and a healthy control（HC）group including 153 healthy controls.2.Childhood maltreatment（CM）was retrospectively assessed using the Chinese version of the childhood trauma questionnaire short form（CCTQ-SF）.Dysfunctional attitudes were assessed using the Chinese version of the dysfunctional attitudes scale-form A（C-DAS-A）.Depression severity was assessed through a 24-items Hamilton depression rating scale(HAMD₂₄).Anxiety severity was assessed through a 14-items Hamilton anxiety rating scale(HAMA₁₄).3.Independent sample T-tests were used to compare group differences for continuous variables and Chi-square tests were used to analyze group differences for categorical variables.4.We used a hierarchical regression model to analyze the predictability of the various CM subtypes on C-DAS-A total score.A three levels model was designed.The predictors for level 1 were: age,gender,and education.Level 2 predictors were: HAMA₁₄,HAMD₂₄,episodes count,and duration of the current episode.As for level 3,the following predictors were added: emotional abuse（EA）,physical abuse（PA）,sexual abuse（SA）,emotional neglect（EN）,and physical neglect（PN）.For the HC group models,the following two items were removed from level 2: episode counts and duration of the current episode.5.The C-DAS-A total score was afterward replaced with the eight different subscales of C-DAS-A（vulnerability,attraction and repulsion,perfectionism,compulsion,seeking applause,dependence,selfdetermination attitude,and cognition philosophy）.The procedure was performed in the MDD group and the HC group separately.6.To test whether cumulative childhood trauma counts（CTC）predict more dysfunctional attitudes,a similar procedure was used.A hierarchical regression analysis of CTC on the C-DAS-A total score was run.7.A three levels model was designed.The predictors for level 1 were: age,gender,and education.Level 2 predictors were: HAMA₁₄,HAMD₂₄,episodes count,and duration of the current episode.As for level 3,CTC was added as the predictor.For the HC group models,the following two items were removed from level 2: episode counts and duration of the current episode.8.The C-DAS-A total score was afterward replaced with the eight different subscales of C-DAS-A（vulnerability,attraction and repulsion,perfectionism,compulsion,seeking applause,dependence,selfdetermination attitude,and cognition philosophy）.The procedure was performed in the MDD group and the HC group separately.Results:Part 1:1.Within the MDD group,the mean age of depression onset was 35.06 years,while the mean episodes of MDD were 2.03.In the MDD group,the youngest participant was 18 years old and the oldest participant was 58 years old;the mean age of the participants in the MDD group was 34.85 years.Within the MDD group,participants in the MDD/CM+ subgroup had a lower mean level of education（9.55 ± 3.29 years）compared with those in the MDD/CM-subgroup（11.26 ± 3.52 years）;the difference was statistically significant（p = 0.001）.Within the HC group,the participants of the HC/CM+ subgroup had a lower mean level of education（10.26 ± 3.37 years）compared to those in the HC/CM-subgroup（11.80 ± 3.68 years）;the difference was statistically significant（p = 0.008）.In the MDD group,participants of the MDD/CM-subgroup had a lower mean HAMD₂₄ score（31.00 ± 6.80）compared to those in the MDD/CM+ subgroup（32.01 ± 7.67）;the difference was not statistically significant（p > 0.05）.In the HC group,participants in the HC/CM-subgroup had a lower mean HAMD₂₄ score（1.06 ± 1.61）compared to those in the HC/CM+ subgroup（1.79 ± 1.89）;the difference was statistically significant（p=0.011）.In the MDD group,participants in the MDD/CM+ subgroup had a higher mean C-CTQ-SF scores（48.12 ± 9.20）compared to participants in the MDD/CM-subgroup（32.16 ± 4.46）;the difference was statistically significant（p < 0.001）.In the HC group,participants in the HC/CM+ subgroup had higher mean C-CTQ-SF scores（44.94 ± 8.81）compared to those in the HC/CM-subgroup（30.64 ± 4.10）;the difference was statistically significant（p<0.001）.In the MDD group,participants in the MDD/CM+ subgroup had a higher mean C-DAS-A total score rating（159.11 ± 25.50）compared to those in the MDD/CM-subgroup（149.37 ± 29.13）;the difference was statistically significant（p=0.022）.In the HC group,participants in the HC/CM+ subgroup had a higher mean total CDAS-A score rating（133.13 ± 22.83）compared to those in the HC/CMsubgroup（117.40 ± 22.72）;the difference was statistically significant（p<0.001）.The MDD/CM+ subgroup had the highest mean C-DAS-A total score（159.11）compared to HC/CM-subgroup which had the lowest mean C-DAS-A total score（117.4）.In the MDD group,C-DAS-A total score was not significantly correlated（p>0.05）with age,education level,HAMD₂₄,and HAMA₁₄.In the HC group,C-DAS-A total score was significantly correlated with HAMD₂₄（r=0.299,p<0.001）,and a significant negative correlation with education level（r=-0.236,p=0.003）;C-DAS-A total score has no significant correlation with age（p>0.05）and gender（p>0.05）.60.2% of participants in the MDD group experienced CM（MDD/CM+）while 44.2% of participants experienced CM in the HC group（HC/CM+）.2.The 2x2 analysis of covariance showed no statistically significant two-way interaction between Diagnosis and CM on C-DAS-A total score while controlling for age,gender,and education（F（1,320）=1.20,p=0.275,partial η2=0.004）.Therefore,main effects analysis and a Bonferroni post hoc test were performed for CM and diagnosis.Main effect analysis and Bonferroni post-hoc tests of CM was statistically significant（p<0.001）.3.The 2x2 analysis of covariance of the C-DAS-A subscale scores showed no significant interaction effect between CM and Diagnosis on CDAS-A subscale scores（p>0.05）,except for C-DAS-A dependence score（p=0.034）.Main effect analysis and Bonferroni post hoc tests were performed for all the models,except C-DAS-A dependence score.Main effects analysis of CM was statistically significant in the following CDAS-A subscales: vulnerability（p<0.001）,attraction and repulsion（p<0.001）,perfectionism（p=0.029）,seeking applause（p<0.001）,and selfdetermination attitude（p=0.018）.The main effect of CM on C-DAS-A compulsion and C-DAS-A cognition philosophy were not statistically significant（p>0.05）.As for C-DAS-A dependence subscale,an analysis of simple main effects for CM was performed using a Bonferroni adjustment.The analysis showed no statistical significance（p>0.05）.Part 2:1.The mean age of the MDD group（34.88±9.61 years）participants was similar to that of the HC group（34.51±9.21 years）participants with no statistically significant difference（p>0.05）.The mean level of education was higher in the HC group（11.13 ± 3.63 years）than in the MDD group（10.31±3.46 years）;the difference was statistically significant（p=0.039）.The HC group participants had a lower mean HAMD₂₄ score（1.38±1.77）compared to those in the MDD group（31.60±7.38）;the difference was statistically significant（p<0.001）.The HC group participants had a lower mean HAMA₁₄ score（1.25±1.92）compared to those in the MDD group（18.31±6.25）;the difference was statistically significant（p<0.001）.The HC group participants had a lower mean C-DAS-A total score（125.75±23.90）compared to those in the MDD group（155.50 ± 27.51）;the difference was statistically significant（p<0.001）.The HC group participants had a lower mean C-CTQ-SF total score（37.05±9.78）compared to participants in the MDD group（41.76±10.75）;the difference was statistically significant（p=0.002）.The HC group participants had a lower mean CTC score（0.72±0.99）compared to participants in the MDD group（1.08±1.10）;the difference was statistically significant（p=0.002）.The prevalence of CM in the sample（MDD+HC）was 52.3%,with EA（6.2%）being the least reported and PN（43.0）was the most reported type of CM.The prevalence of the other CM subtypes were: PA（7.5%）;SA（7.5%）and EN（25.9%）.Within the whole sample（MDD+HC）,24.6% of the participants experienced any one subtype of CM;20.9% of the participants experienced any two subtypes of CM;4.4% of the participants experienced any three subtypes of CM;2.5% of the participants experienced any four subtypes of CM and 0.3% experience all the five subtypes of CM.A higher proportion of the MDD group reported having experienced any form of CM subtype compared to the HC group（MDD: 60.1%,HC: 3.8%,p=0.003）.The prevalence of CM in the MDD group was 60.1%,and EA and SA had the lowest prevalence（8.9%）,and PN（49.4）had the highest prevalence.The prevalence of the other CM subtypes were: PA（9.5%）and EN（21.5%）.In the MDD group,24.4% of the participants experienced any one form of CM subtypes;26.8% of the participants experienced any two forms of CM subtypes.The prevalence of CM in the HC group was 43.8%,EA（3.3%）had the lowest incidence and PN（35.9%）had the highest prevalence.The prevalence of the other CM subtypes were: PA（5.2%）;SA（5.9%）and EN（19.6%）.In the HC group,24.8% of the participants experienced any one subtype of CM;14.4% of the participants experienced any two subtypes of CM.2.The hierarchical regression analysis to predict C-DAS-A total score in the MDD group was statistically significant with ΔR² of 7.9%,F=2.915,p=0.015 from the second to the third level;among the level 3 predictors,only EA had a statistically significant standard coefficient of 0.247（p= 0.011）.The hierarchical regression analysis to predict C-DASA attraction and repulsion subscale score in the MDD group was also statistically significant with a ΔR² of 8.2%,F=2.926,p=0.015 from the second to the third level;among the level 3 predictors,only PN had a statistically significant standard coefficient of 0.276（p=0.009）.The hierarchical regression analysis to predict C-DAS-A self-determination type subscale score in the MDD group was statistically significant with a ΔR² of 6.9%,F=2.602,p=0.027 from the second level to the third level;among the level 3 predictors,only EA had a statistically significant standard coefficient of 0.262（p= 0.007）.The hierarchical regression analysis to predict the other C-DAS-A subscales scores in the MDD group did not have a statistically significant from level 2 to level 3: C-DAS-A vulnerability score（ΔR²=5.4%,F=1.878,p=0.101）;C-DAS-A perfectionism score（ΔR²=1.9%,F=0.619,p=0.685）;C-DAS-A compulsion score（ΔR²=3.2%,F=1.154,p=0.334）;C-DAS-A seeking applause score（ΔR²=4.9%,F=1.752,p=0.126）;CDAS-A dependence score（ΔR² = 3.1%,F = 1.034,p=0.400）and C-DASA cognition philosophy score（ΔR²=5.2%,F=1.806,p=0.115）.3.The hierarchical regression analysis to predict C-DAS-A seeking applause subscale score in the HC group was statistically significant with a ΔR² of 7.7%,F=2.504,p=0.033 from the second to the third level;among the level 3 predictors,only PN had a statistically significant standard coefficient of 0.216（p= 0.046）.The hierarchical regression analysis to predict C-DAS-A total score and C-DAS-A attraction and repulsion score in the HC group was statistically significant from level 2 to level 3.However,none of the level 3 predictors had statistically significant standard coefficient（C-DAS-A total score: ΔR²=8.8%,F=0.360,p=0.007;C-DAS-A attraction and repulsion score: ΔR²=9.0%,F=3.091,p=0.011）.The hierarchical regression analysis to predict the other C-DAS-A subscale score in the HC group was not significant from level 2 to level 3: C-DAS-A vulnerability score（ΔR²=6.1%,F=1.949,p=0.090）;C-DAS-A perfectionism score（ΔR²=6.0%,F=1.952,p=0.089）;C-DAS-A compulsion score（ΔR²=1.6%,F=0.536,p=0.749）;C-DAS-A dependence score（ΔR²=5.4%,F=0.054,p=0.098）;C-DAS-A self-determination attitude score（ΔR²=3.9%,F=1.296,p=0.269）and C-DAS-A cognitive philosophy score（ΔR²=5.1%,F=1.768,p=0.123）.4.The hierarchical regression analysis to predict C-DAS-A total score in the MDD group was statistically significant with ΔR² of 3.8%,F=6.845,p=0.010 from the second level to the third level;the standard coefficient of CTC as a predictor was 0.213（p=0.010）.The hierarchical regression analysis to predict C-DAS-A vulnerability subscale score in the MDD group was statistically significant with a ΔR² of 2.5%,F=4.208,p=0.042 from the second level to the third level;the standard coefficient of CTC was 0.171（p=0.042）.The hierarchical regression analysis to predict CDAS-A attraction and repulsion subscale score in the MDD group was statistically significant with a ΔR² of 5.4%,F=9.513,p=0.002 from the second level to the third level;the standard coefficient of CTC was 0.253（p=0.002）.The hierarchical regression analysis of C-DAS-A seeking applause subscale score in the MDD group was statistically significant with a ΔR² of 3.4%,F=6.145,p=0.014 from the second level to the third level;the standard coefficient of CTC was 0.202（p=0.014）.The hierarchical regression analysis to predict the other C-DAS-A subscale score in the MDD group was not statistically significant from level 2 to level 3: C-DAS-A perfectionism score（ΔR²=1.0%,F=1.615,p= 0.206）;C-DAS-A compulsion score（ΔR²=1.2%,F=2.224,p=0.138）;CDAS-A dependence score（ΔR²=0.1%,F=0.170,p=0.681）;C-DAS-A self-determination attitude score（ΔR²=1.9%,F=3.468,p=0.064）and C-DASA cognitive philosophy score（ΔR²= 0.5%,F= 0.831,p=0.363）.5.The hierarchical regression analysis to predict C-DAS-A attraction and repulsion subscale score in the HC group was statistically significant with ΔR² of 2.7%,F=4.480,p=0.036 from the second level to the third level;the standard coefficient of CTC as a predictor was 0.167（p=0.036）.The hierarchical regression analysis to predict the other C-DAS-A subscale scores in the HC group were not statistically significant from level 2 to level 3: C-DAS-A total score（ΔR²=1.1%,F=2.030,p=0.156）;C-DASA vulnerability score（ΔR²=0.2%,F=0.350,p=0.555）;C-DAS-A perfectionism score（ΔR²=0.2%,F=0.235,p=0.629）;C-DAS-A compulsion score（ΔR²=0.0%,F=0.047,p=0.829）;C-DAS-A seeking applause score（ΔR²=1.4%,F=2.239,p=0.137）;C-DAS-A dependence score（ΔR²=0.7%,F=1.175,p=0.280）;C-DAS-A self-determination attitude score（ΔR²=0.8%,F=1.304,p=0.255）and C-DAS-A cognitive philosophy score（ΔR²=0.0%,F=0.025,p=0.874）.Conclusion:1.Childhood maltreatment and dysfunctional attitudes are associated.2.Emotional abuse and physical neglect are of particular importance in predicting dysfunctional attitudes.3.Emotional abuse predicts specific types of dysfunctional attitudes among the depressed.Physical Neglect predicts specific types of dysfunctional attitudes among the depressed and the never-depressed.4.Cumulative childhood trauma counts are predictive of more forms of dysfunctional attitudes.