Associations of adverse childhood experiences and depressive symptoms among children and adolescents: the mediating effect of resilience and sleep quality

Abstract

Adverse childhood experiences (ACEs) may precipitate depressive symptoms among children and adolescents, yet whether resilience and sleep quality jointly mediate this link remains unclear. Therefore, this study aimed to investigate these potential mediating pathways. From April to June 2023, a total of 3258 children and adolescents were included for this cross-sectional study from 11 primary and secondary schools in Chaohu City, Anhui Province, China. We used a series of scales to assess their depressive symptoms, ACEs, resilience, and sleep quality. To explore the potential mediating pathways between depressive symptoms, ACEs, resilience, and sleep quality, the chain mediation analysis was conducted using the PROCESS macro (Model 6). The prevalence of depressive symptoms was 30.1% among children and adolescents. The multivariate logistic regression analyses showed that females, ACEs total score, resilience, and poor sleep quality were independent correlates of depressive symptoms. When the dimensions of ACEs were involved in the regression model, emotional abuse, physical abuse, emotional neglect, physical neglect and mother suffered from domestic violence were also independent risk factors. Notably, resilience and sleep quality had a chain-mediated effect between ACEs and depressive symptoms. A high proportion of children and adolescents reported depressive symptoms. Moreover, resilience and sleep quality may have a chain-mediated effect between ACEs and depressive symptoms. Hence, the impact of ACEs, resilience and sleep quality on depressive symptoms among children and adolescents should be highlighted and targeted interventions should be implemented.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-25994-2.

Introduction

Depressive disorder, a mental condition typified by persistent low mood and anhedonia, is projected to be one of the leading contributors to the global disease burden by 2030¹. Notably, with the increase of social pressure and lifestyle changes, the incidence of depressive disorder among children and adolescents has shown a continuous climb globally². In the United States, the detection rate of adolescent depressive episodes rose from 8.1% in 2009 to 15.8% in 2019³. Comorbid depressive symptoms not only have a profound impact on cognitive development, academic achievement, and social functioning among children and adolescents, but also significantly elevate suicide risk^4,5. Therefore, investigating the factors influencing depressive symptoms in this population and their underlying mechanisms holds significant public health importance.

Family environmental risk factors play a critical role in the development of depressive symptoms, particularly adverse childhood experiences (ACEs). ACEs are defined as stressful or traumatic events occurring before age 18 years and include abuse (emotional, physical, or sexual), neglect (emotional or physical), and family dysfunction (e.g., family member’s substance abuse, mental illness, criminal behaviors, domestic violence, parental divorce or separation, etc.)⁶. Data from the 2021 China School Adolescent Health Survey showed that 47.3% of adolescents have experienced at least one ACE⁷. ACEs can have significant negative effects on the mental health of children and adolescents. For example, a national study from the United States has found that the more adolescents aged 12–17 years were exposed to ACEs, the higher their risk of developing mental health symptoms such as depression, anxiety, and behavioral problems⁸. However, the specific pathways through which ACEs influence depressive symptoms among children and adolescents, particularly their mediating mechanisms, are still not fully understood.

Resilience and sleep quality may serve as key mediating variables in the association between ACEs and depressive symptoms. Resilience refers to an individual’s capacity to recover and maintain psychological equilibrium in the face of adversity, trauma, stress, or other significant life challenges, and its development is characterized by considerable dynamism and plasticity⁹. According to psychosocial development theory, childhood is a critical window for the formation of resilience, with factors such as family environment quality and social support systems playing pivotal roles in shaping its trajectory¹⁰. A systematic review indicated that ACEs may diminish resilience among children and adolescents¹¹. And lower levels of resilience may, in turn, contribute to the onset of depressive symptoms among children and adolescents. For instance, a study conducted in Hong Kong, China, identified low resilience as a strong predictor of depressive symptoms among adolescents¹². Furthermore, Chen et al. proposed that resilience moderates the association between ACEs and depressive symptoms¹³.

Moreover, sleep quality, an important physiological indicator of mental health, may also act as a mediator in the association between ACEs and depressive symptoms. Numerous clinical studies have shown that children and adolescents exposed to ACEs face a significantly elevated risk of developing various types of sleep disorders^14–16. During adolescence, a critical period for the establishment of sleep-wake rhythms¹⁷, sleep problems are closely linked to several adverse mental health outcomes such as anxiety and depression^18,19. A recent study further found that childhood family dysfunction undermines resilience, which in turn precipitates poor sleep quality and subsequent depressive symptoms²⁰. These findings suggest that ACEs may affect depressive symptoms through the chain pathway of “impaired resilience → reduced sleep quality”. Yet, this mechanism remains underexplored among children and adolescents. Adolescence is a key sensitive period when psychological resilience is reshaped and sleep rhythms become consolidated, and both processes show marked plasticity and can be modified. Clarifying whether the proposed chain of effects operates in adolescents would therefore help to trace, moment by moment, how ACEs evolve into depressive symptoms and would point to new targets for preventive clinical interventions.

This study aimed to explore the epidemiological characteristics of depressive symptoms among children and adolescents, with a focus on the combined effects of socio-demographic characteristics, ACEs, resilience, and sleep quality. It also sought to validate the mediating mechanism of the chain “ACEs → Resilience → Sleep quality → Depressive symptoms”. Accordingly, we hypothesized that: (1) there is a significant burden of depressive symptoms in Chinese children and adolescents; (2) depressive symptoms are significantly associated with socio-demographic characteristics, ACEs, lower resilience, and poorer sleep quality; (3) resilience and sleep quality play a chain-mediated role between ACEs and depressive symptoms.

Methods

Participants

This cross-sectional survey was conducted from April to June 2023. Using convenience sampling, we selected 11 primary and secondary schools in Chaohu City, Anhui Province, that together cover urban, suburban, and rural areas; although not randomly drawn, the frame maximises regional diversity. The study population consisted of school students, with the following inclusion criteria: (1) age ≤ 18 years; and (2) voluntary participation in the study. The exclusion criteria were as follows: (1) those who are unable to understand the questionnaire; and (2) subjects or their guardians who refused to sign the informed consent form. A total of 3596 children and adolescents were initially recruited. Of these, 213 explicitly declined to participate, leaving 3383 who provided informed consent and completed the questionnaire. After quality control, 125 questionnaires with missing key variables were excluded. Consequently, 3258 valid questionnaires were retained for statistical analysis, yielding an effective response rate of 90.6% (Fig. 1). Among these children and adolescents, the mean age was 13.01 ± 1.61 years (range: 9–18 years), with 52.1% being male (Table 1).

Fig. 1.

Participants screening flow chart.

Table 1.

Comparison of socio-demographic and psychological health characteristics in the depressed and non-depressed groups.

Variables	Total sample (n = 3258)	Depressed group (n = 980)	Non-depressed group (n = 2278)	t/χ2	P
Gender				27.161	< 0.001
Males, n (%)	1696 (52.06)	442 (45.10)	1254 (55.05)
Females, n (%)	1562 (47.94)	538 (54.90)	1024 (44.95)
Age (years), mean (SD)	13.01 (1.61)	13.24 (1.54)	12.92 (1.63)	5.371	< 0.001
BMI (kg/m2), mean (SD)	19.32 (3.31)	19.39 (3.40)	19.29 (3.26)	0.781	0.435
Family economy was extremely hard, n (%)	140 (4.30)	84 (8.57)	56 (2.46)	62.266	< 0.001
ACEs total score, mean (SD)	0.95 (1.39)	1.74 (1.73)	0.60 (1.04)	19.180	< 0.001
Sexual abuse score, mean (SD)	5.50 (1.73)	5.84 (2.25)	5.35 (1.43)	6.269	< 0.001
Sexual abuse, n (%)	240 (7.37)	116 (11.84)	124 (5.44)	41.045	< 0.001
Emotional abuse score, mean (SD)	7.11 (3.06)	9.14 (3.94)	6.24 (2.05)	21.772	< 0.001
Emotional abuse, n (%)	230 (7.06)	184 (18.78)	46 (2.02)	293.223	< 0.001
Physical abuse score, mean (SD)	6.03 (2.21)	6.94 (2.93)	5.64 (1.66)	13.001	< 0.001
Physical abuse, n (%)	200 (6.14)	134 (13.67)	66 (2.90)	138.100	< 0.001
Emotional neglect score, mean (SD)	11.31 (4.77)	13.56 (5.04)	10.34 (4.29)	17.449	< 0.001
Emotional neglect, n (%)	719 (22.07)	393 (40.10)	326 (14.31)	265.023	< 0.001
Physical neglect score, mean (SD)	7.57 (2.75)	8.80 (3.02)	7.04 (2.44)	16.164	< 0.001
Physical neglect, n (%)	684 (20.99)	360 (36.73)	324 (14.22)	209.255	< 0.001
Family members were addicted to gambling, drugs or alcohol, n (%)	211 (6.48)	116 (11.84)	95 (4.17)	66.491	< 0.001
Parents had been separated or divorced, n (%)	390 (11.97)	168 (17.14)	222 (9.75)	35.584	< 0.001
Mother suffered from domestic violence, n (%)	264 (8.10)	158 (16.12)	106 (4.65)	121.044	< 0.001
Family members had mental disorders, n (%)	83 (2.55)	51 (5.20)	32 (1.40)	39.841	< 0.001
Family members jailed for crimes, n (%)	58 (1.78)	27 (2.76)	31 (1.36)	7.618	0.006
ACE accumulation category, n (%)				480.063	< 0.001
0	1774 (54.45)	301 (30.71)	1473 (64.66)
1	688 (21.12)	223 (22.76)	465 (20.41)
2	386 (11.85)	174 (17.76)	212 (9.31)
≥ 3	410 (12.58)	282 (28.77)	128 (5.62)
CYRM total score, mean (SD)	110.19 (20.64)	97.67 (20.20)	115.58 (18.37)	−23.842	< 0.001
PSQI total score, mean (SD)	4.60 (3.07)	6.68 (3.43)	3.70 (2.40)	24.675	< 0.001

BMI, Body Mass Index; CES-D, Center for Epidemiologic Studies Depression Scale; ACEs, Adverse Childhood Experiences; CYRM, Child and Youth Resilience Measure; PSQI, Pittsburgh Sleep Quality Index.

SD, Standard Deviation; Bolded P value: < 0.05.

This study’s data collection was conducted by uniformly trained investigators, utilizing a standardized questionnaire administration process. Before the formal survey, the researchers provided participants with a detailed explanation of the study’s purpose, the method of completing the questionnaire, and the principle of confidentiality. All participants and their legal guardians provided informed consent by signing the consent form. The study was approved by the Ethics Committee of the Fourth Affiliated Hospital of Anhui Medical University (KYXM-202210-010). It also followed the relevant ethical requirements of the Declaration of Helsinki (2013 version) (https://www.wma.net/policies-post/wma-declaration-of-helsinki/) and the STROBE checklist (see the attachment).

Measuring instruments

Socio-demographic characteristics

The socio-demographic characteristics of the children and adolescents, including their age, gender, body mass index (BMI) and family economic status, were collected using a self-designed questionnaire.

Depressive symptoms

The Center for Epidemiological Studies Depression Scale (CES-D) was utilized to assess subjects’ depressive symptoms²¹. This scale comprises 20 items rated on a 4-point scale, with total scores ranging from 0 to 60. A higher total score indicates more severe depressive symptoms. According to the original scale, the criteria for depressive symptoms are as follows: total score ≤ 15 indicates no depression, 16–19 indicates mild depressive symptoms, and ≥ 20 indicates moderate to severe depressive symptoms. Referring to previous research on Chinese children and adolescents²², this study defined the presence of depressive symptoms as a CES-D total score > 15. Currently, the Chinese version of the CES-D has demonstrated good reliability and validity and is widely used among Chinese children and adolescents²³. In our study, the Cronbach’s alpha coefficient for CES-D was 0.907. It is important to note that although the CES-D is a widely used screening tool for depressive symptoms, it does not provide a clinical diagnosis of depression. Our study measured the prevalence of depressive symptoms, not clinically diagnosed depression.

ACEs

In this study, ACEs were comprehensively assessed across 10 key dimensions²⁴. The Childhood Trauma Questionnaire (CTQ) was utilized to evaluate experiences of abuse and neglect among children and adolescents²⁵. The CTQ comprises 28 items rated on a 5-point scale and includes five subscales: emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect. Higher CTQ scores indicate a higher level of abuse or neglect experienced by the individual. For this study, each subscale was categorized into two groups (yes and no) based on established cut-off points: 13 for emotional abuse, 10 for physical abuse, 8 for sexual abuse, 15 for emotional neglect, and 10 for physical neglect²⁶. In addition to the CTQ, 5 other dimensions of family dysfunction were assessed, drawing on previous research: (1) family members were addicted to gambling, drugs or alcohol; (2) parents had been separated or divorced; (3) mother suffered from domestic violence; (4) family members had mental disorders; and (5) family members jailed for crimes²⁴. Each of these 10 dimensions was scored as 1 ACE if present. Consequently, the ACEs total score ranged from 0 to 10, with higher scores reflecting a greater accumulation of ACEs.

Resilience

The Child Youth Resilience Measurement Scale (CYRM) was used to assess the level of resilience²⁷. This scale comprises 28 items rated on a 5-point scale, with total scores ranging from 28 to 140. Higher scores indicate more resilient individuals, who are better able to demonstrate adaptability and resilience when facing life’s challenges and dilemmas. The reliability and validity of this scale have been thoroughly established in numerous studies^28,29. And the Cronbach’s alpha coefficient for CYRM was 0.961.

Sleep quality

In this study, the Pittsburgh Sleep Quality Index (PSQI) was utilized to evaluate sleep quality³⁰. The PSQI includes seven distinct components: sleep latency, sleep duration, sleep quality, habitual sleep efficiency, sleep disturbances, use of sleep medications, and daytime dysfunction. Each of these components is rated on a scale from 0 to 3, resulting in a total score that ranges from 0 to 21. A higher total score is indicative of worse sleep quality. And the Cronbach’s alpha coefficient for PSQI was 0.701.

Statistical analysis

This study employed SPSS 23.0 for data analyses. Categorical variables were presented as frequencies (%). The normality of continuous variables was evaluated using the Kolmogorov-Smirnov test. Variables that followed a normal distribution were expressed as mean ± standard deviation (SD). In the univariate analyses, independent samples t-tests or chi-square tests were utilized to compare the difference of demographic characteristics, ACEs, resilience, and sleep quality between the depressive symptoms group and the non-depressive symptoms group. Variables with P < 0.05 in the univariate analyses were included in a multivariate logistic regression model (Enter method) to identify independent predictors of depressive symptoms. And variance inflation factor (VIF) was employed to test the covariance problem of each variable. The results showed that the VIF values of all independent variables were below 5.0, indicating acceptable multicollinearity levels. This study used the Box-Tidwell method to test the linearity between continuous independent variables and the logit of the dependent variable. The results showed no significant nonlinear trends for any of the continuous predictors (all Box-Tidwell tests, P > 0.05), satisfying the linearity assumption of logistic regression. The model’s calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, and the area under the curve (AUC) was calculated to assess its discriminative ability. Additionally, Pearson correlation analyses were conducted to assess the correlations between ACEs, depressive symptoms, resilience, and sleep quality. Based on existing empirical evidence and the theoretical assumptions proposed in this study, ACEs may indirectly aggravate depressive symptoms through the pathway of “impaired resilience → decreased sleep quality”. To test this potential chain-mediation mechanism, the present study employed the PROCESS macro (Model 6). A 95% confidence interval (CI) was calculated using 5000 Bootstrap samples, and a mediation effect was deemed significant if the interval did not include zero. In the mediation analysis, potential confounding factors, including age, gender, and BMI, were controlled for. Statistical significance was set at a two-tailed P < 0.05.

Results

Comparison of socio-demographic and psychological health data in the depressed and non-depressed groups

In our study, the prevalence of depressive symptoms was 30.1% among children and adolescents. Univariate analyses revealed that compared to the group without depressive symptoms, children and adolescents with depressive symptoms were more likely to be females (χ² = 27.161, P < 0.001) and older (t = 5.371, P < 0.001). They were more likely to experience ACEs (t = 19.180, P < 0.001), including sexual abuse (χ² = 41.045, P < 0.001), emotional abuse (χ² = 293.223, P < 0.001), physical abuse (χ² = 138.100, P < 0.001), emotional neglect (χ² = 265.023, P < 0.001), physical neglect (χ² = 209.255, P < 0.001), family members were addicted to gambling, drugs or alcohol (χ² = 66.491, P < 0.001), parents had been separated or divorced (χ² = 35.584, P < 0.001), mother suffered from domestic violence (χ² = 121.044, P < 0.001), family members had mental disorders (χ² = 39.841, P < 0.001), family members jailed for crimes (χ² = 7.618, P = 0.006), and family economy was extremely hard (χ² = 62.266, P < 0.001). In addition, they had lower levels of resilience (t = -23.842, P < 0.001) and poorer sleep quality (t = 24.675, P < 0.001) (Table 1). Additionally, regarding the risk of cumulative ACEs levels, we found that the risk of depressive symptoms gradually increases with the accumulation of ACEs scores.

Independent factors associated with depressive symptoms

When ACEs total score involved in the multivariate logistic regression model (Model 1), females (OR = 1.322, 95% CI 1.103–1.586, P = 0.003), ACEs total score (OR = 1.453, 95% CI 1.351–1.562, P < 0.001), resilience (OR = 0.973, 95% CI 0.969–0.978, P < 0.001), and poor sleep quality (OR = 1.322, 95% CI 1.276–1.369, P < 0.001) were independent correlates of depressive symptoms (Table 2). When the dimensions of ACEs involved in the regression model (each CTQ dimension treated as a continuous variable) (Model 2), females (OR = 1.256, 95% CI 1.040–1.516, P = 0.018), age (OR = 1.066, 95% CI 1.002–1.133, P = 0.041), emotional abuse score (OR = 1.209, 95% CI 1.160–1.259, P < 0.001), emotional neglect score (OR = 1.034, 95% CI 1.008–1.061, P = 0.011), physical neglect score (OR = 1.070, 95% CI 1.025–1.118, P = 0.002), family members jailed for crimes (OR = 0.467, 95% CI 0.222–0.982, P = 0.045), resilience (OR = 0.976, 95% CI 0.972–0.981, P < 0.001), and poor sleep quality (OR = 1.279, 95% CI 1.233–1.326, P < 0.001) were independent correlates (Table 2). When the dimensions of ACEs involved in the regression model (each CTQ dimension treated as a binary variable) (Model 3), females (OR = 1.313, 95% CI 1.092–1.580, P = 0.004), emotional abuse (OR = 2.998, 95% CI 1.989–4.518, P < 0.001), physical abuse (OR = 1.702, 95% CI 1.129–2.567, P = 0.011), emotional neglect (OR = 1.463, 95% CI 1.145–1.869, P = 0.002), physical neglect (OR = 1.455, 95% CI 1.140–1.857, P = 0.003), mother suffered from domestic violence (OR = 1.496, 95% CI 1.068–2.095, P = 0.019), family members jailed for crimes (OR = 0.445, 95% CI 0.214–0.925, P = 0.030), resilience (OR = 0.974, 95% CI 0.969–0.979, P < 0.001), and poor sleep quality (OR = 1.316, 95% CI 1.270–1.364, P < 0.001) were independent correlates (Table 2). When ACEs accumulation category involved in the multivariate logistic regression model (Model 4), ACEs accumulation category also emerged as an independent risk factor for depressive symptoms (P < 0.001) (Table 2). Compared to individuals with no ACEs, those with 1, 2, and 3 or more ACEs had 1.623 times, 2.431 times, and 4.461 times higher odds of depressive symptoms, respectively.

Table 2.

Independent factors associated with depressive symptoms by multivariate logistic regression analyses (total sample n = 3258; depressive symptoms group n = 980).

Variables	B	SE	Wald c2	OR e	95% CI e	OR f	95% CI f	P
Model 1 a
Females	0.279	0.093	9.074	1.322	1.103–1.586	1.322	1.103–1.586	0.003
Age	0.056	0.030	3.453	1.057	0.997–1.121	1.094	0.995–1.203	0.063
Family economy was extremely hard	0.136	0.229	0.352	1.145	0.732–1.793	1.145	0.732–1.793	0.553
ACEs total score	0.373	0.037	101.322	1.453	1.351–1.562	1.681	1.519–1.860	< 0.001
CYRM total score	−0.027	0.002	119.666	0.973	0.969–0.978	0.574	0.519–0.634	< 0.001
PSQI total score	0.279	0.018	239.450	1.322	1.276–1.369	2.353	2.111–2.622	< 0.001
Model 2 b
Females	0.228	0.096	5.600	1.256	1.040–1.516	1.256	1.040–1.516	0.018
Age	0.064	0.031	4.156	1.066	1.002–1.133	1.108	1.004–1.223	0.041
Family economy was extremely hard	0.301	0.240	1.571	1.351	0.844–2.161	1.351	0.844–2.161	0.210
Sexual abuse score	−0.029	0.031	0.904	0.971	0.915–1.031	0.951	0.857–1.055	0.342
Emotional abuse score	0.189	0.021	83.147	1.209	1.160–1.259	1.786	1.577–2.024	< 0.001
Physical abuse score	0.023	0.028	0.709	1.023	0.970–1.080	1.053	0.934–1.186	0.400
Emotional neglect score	0.034	0.013	6.526	1.034	1.008–1.061	1.174	1.038–1.329	0.011
Physical neglect score	0.068	0.022	9.320	1.070	1.025–1.118	1.206	1.069–1.360	0.002
Family members were addicted to gambling, drugs or alcohol	0.216	0.189	1.303	1.240	0.857–1.796	1.240	0.857–1.796	0.254
Parents had been separated or divorced	0.034	0.144	0.057	1.035	0.780–1.373	1.035	0.780–1.373	0.812
Mother suffered from domestic violence	0.266	0.177	2.263	1.305	0.923–1.845	1.305	0.923–1.845	0.133
Family members had mental disorders	0.388	0.308	1.585	1.474	0.806–2.698	1.474	0.806–2.698	0.208
Family members jailed for crimes	−0.761	0.379	4.033	0.467	0.222–0.982	0.467	0.222–0.982	0.045
CYRM total score	−0.024	0.003	83.864	0.976	0.972–0.981	0.612	0.551–0.680	< 0.001
PSQI total score	0.246	0.019	172.810	1.279	1.233–1.326	2.126	1.900–2.379	< 0.001
Model 3 c
Females	0.273	0.094	8.370	1.313	1.092–1.580	1.313	1.092–1.580	0.004
Age	0.052	0.030	2.896	1.053	0.992–1.118	1.087	0.987–1.197	0.089
Family economy was extremely hard	0.333	0.237	1.967	1.395	0.876–2.222	1.395	0.876–2.222	0.161
Sexual abuse	0.199	0.182	1.189	1.220	0.853–1.745	1.220	0.853–1.745	0.275
Emotional abuse	1.098	0.209	27.513	2.998	1.989–4.518	2.998	1.989–4.518	< 0.001
Physical abuse	0.532	0.210	6.438	1.702	1.129–2.567	1.702	1.129–2.567	0.011
Emotional neglect	0.381	0.125	9.277	1.463	1.145–1.869	1.463	1.145–1.869	0.002
Physical neglect	0.375	0.125	9.080	1.455	1.140–1.857	1.455	1.140–1.857	0.003
Family members were addicted to gambling, drugs or alcohol	0.339	0.185	3.348	1.403	0.976–2.017	1.403	0.976–2.017	0.067
Parents had been separated or divorced	0.096	0.142	0.461	1.101	0.834–1.453	1.101	0.834–1.453	0.497
Mother suffered from domestic violence	0.403	0.172	5.486	1.496	1.068–2.095	1.496	1.068–2.095	0.019
Family members had mental disorders	0.414	0.307	1.820	1.514	0.829–2.764	1.514	0.829–2.764	0.177
Family members jailed for crimes	−0.811	0.374	4.708	0.445	0.214–0.925	0.445	0.214–0.925	0.030
CYRM total score	−0.026	0.003	107.519	0.974	0.969–0.979	0.579	0.522–0.642	< 0.001
PSQI total score	0.275	0.018	226.930	1.316	1.270–1.364	2.324	2.082–2.593	< 0.001
Model 4 d
Females	0.286	0.093	9.483	1.331	1.110–1.597	1.331	1.110–1.597	0.002
Age	0.062	0.030	4.196	1.064	1.003–1.128	1.105	1.004–1.215	0.041
Family economy was extremely hard	0.204	0.225	0.820	1.226	0.789–1.907	1.226	0.789–1.907	0.365
ACEs accumulation category
0	Ref	–	–	–	–	–	–	–
1	0.484	0.116	17.326	1.623	1.292–2.039	1.623	1.292–2.039	< 0.001
2	0.888	0.140	40.110	2.431	1.847–3.200	2.431	1.847–3.200	< 0.001
≥ 3	1.495	0.150	98.923	4.461	3.323–5.990	4.461	3.323–5.990	< 0.001
CYRM total score	−0.026	0.002	112.151	0.974	0.969–0.979	0.582	0.526–0.643	< 0.001
PSQI total score	0.277	0.018	235.481	1.319	1.273–1.367	2.340	2.099–2.608	< 0.001

ACEs, Adverse Childhood Experiences; CYRM, Child and Youth Resilience Measure; PSQI, Pittsburgh Sleep Quality Index.

SE, Standard Error; OR, Odds Ratio; CI, Confidence Interval; Bolded P value: < 0.05.

^a, ACEs total score were involved in the regression model; ^b, Various types of ACEs were involved in the regression model, with each CTQ dimension treated as a continuous variable; ^c, Various types of ACEs were involved in the regression model, with each CTQ dimension treated as a binary variable; ^d, ACEs accumulation category were involved in the regression model; ^e, For continuous variables, the OR corresponds to a per-unit increase; ^e, for continuous variables, the OR corresponds to a per-standard-deviation increase.

Model evaluation

The calibration of the four models was evaluated using the Hosmer-Lemeshow goodness-of-fit test. The χ² values for Models 1, 2, 3 and 4 were 8.800, 5.390, 11.502 and 2.516, respectively, with corresponding p-values of 0.359, 0.715, 0.175 and 0.961 (all P > 0.05) (Table 3). These results indicate that no systematic discrepancy exists between the predicted probabilities and the observed events for any model, suggesting that all four models exhibit good calibration. And the AUCs for Models 1, 2, 3 and 4 were 0.827, 0.842, 0.830 and 0.827, respectively (all P < 0.001).

Table 3.

Model evaluation.

Model	AUC	95% CI	P	Hosmer-Lemeshow c2	Hosmer-Lemeshow P
Model 1 a	0.827	0.811–0.842	< 0.001	8.800	0.359
Model 2 b	0.842	0.827–0.857	< 0.001	5.390	0.715
Model 3 c	0.830	0.815–0.846	< 0.001	11.502	0.175
Model 4 d	0.827	0.811–0.842	< 0.001	2.516	0.961

AUC, Area Under the Curve; CI, Confidence Interval; Bolded P value: < 0.05.

^a, ACEs total score were involved in the regression model; ^b, Various types of ACEs were involved in the regression model, with each CTQ dimension treated as a continuous variable; ^c, Various types of ACEs were involved in the regression model, with each CTQ dimension treated as a binary variable; ^d, ACEs accumulation category were involved in the regression model.

Correlations between depressive symptoms, ACEs, resilience and sleep quality

Correlation analyses showed that depressive symptoms were significantly correlated with ACEs total score, resilience, and sleep quality, respectively (all P < 0.05) (Table 4). As well, ACEs total score was also significantly correlated with resilience and sleep quality, respectively (all P < 0.05) (Table 4).

Table 4.

Correlations between depressive symptoms, ACEs, resilience and sleep quality.

Variables	1	2	3	4	5	6	7
1. Gender	1.000
2. Age	−0.034 (95% CI −0.068–0.001)	1.000
3. BMI	−0.105 *** (95% CI −0.139 − −0.068)	0.219 *** (95% CI 0.186–0.252)	1.000
4. CES-D total score	0.080 *** (95% CI 0.048–0.114)	0.088 *** (95% CI 0.054–0.120)	0.005 (95% CI −0.030–0.040)	1.000
5. ACEs total score	0.014 (95% CI −0.019–0.049)	−0.061 *** (95% CI −0.095 to −0.024)	−0.005 (95% CI −0.042–0.030)	0.448 *** (95% CI 0.414–0.481)	1.000
6. CYRM total score	−0.088 *** (95% CI −0.123 to −0.052)	0.001 (95% CI −0.034–0.037)	0.005 (95% CI −0.031–0.043)	−0.482 *** (95% CI −0.514 to −0.450)	−0.419 *** (95% CI −0.453 to −0.386)	1.000
7. PSQI total score	0.104 *** (95% CI 0.068–0.139)	0.237 *** (95% CI 0.204–0.268)	0.066 *** (95% CI 0.033–0.102)	0.550 *** (95% CI 0.521–0.578)	0.300 *** (95% CI 0.263–0.337)	−0.365 *** (95% CI −0.399 to −0.331)	1.000

BMI, Body Mass Index; CES-D, Center for Epidemiologic Studies Depression Scale; ACEs, Adverse Childhood Experiences; CYRM, Child and Youth Resilience Measure; PSQI, Pittsburgh Sleep Quality Index.

^*, P < 0.05; ^**, P < 0.01; ^***, P < 0.001.

Mediating roles of resilience and sleep quality in the association between aces and depressive symptoms

The results of mediation analyses showed that resilience and sleep quality played a chain mediating role between ACEs and depressive symptoms. Specifically, the direct effect value was 1.747 (95% CI 1.512–1.981), while the total indirect effect value was 1.666 (95% CI 1.480–1.865). The specific indirect effects were as follows: “ACEs - Resilience - Depressive symptoms” had an indirect effect of 0.781 (95% CI 0.647–0.927); “ACEs - Sleep quality - Depressive symptoms” had an indirect effect of 0.544 (95% CI 0.416–0.676); and “ACEs - Resilience - Sleep quality - Depressive symptoms” had an indirect effect of 0.341 (95% CI 0.278–0.413) (Table 5; Fig. 2).

Table 5.

Mediating roles of resilience and sleep quality in the associations between aces and depressive symptoms.

Effect	Unstandardized β	SE	95% CI	Effect size (%)
Total effect	3.413	0.127	3.164–3.661	–
Direct effect	1.747	0.120	1.512–1.981	51.19
Indirect effect	1.666	0.099	1.480–1.865	48.81
ACEs - Resilience - Depressive symptoms	0.781	0.071	0.647–0.927	22.88
ACEs - Sleep quality - Depressive symptoms	0.544	0.067	0.416–0.676	15.94
ACEs - Resilience - Sleep quality - Depressive symptoms	0.341	0.034	0.278–0.413	9.99

ACEs, Adverse Childhood Experiences.

SE, Standard Error; CI, Confidence Interval; Bolded P value: < 0.05.

Fig. 2.

Direct and indirect relationship models of ACEs, resilience and sleep quality on depressive symptoms. ACEs, Adverse Childhood Experiences. SE, Standard Error.

Discussion

In our study, depressive symptoms were detected in 30.1% of children and adolescents, consistent with previous findings^31–33. A survey of Chinese school adolescents revealed that about 34% of adolescents had depressive symptoms³². Another study using the CES-D scale among 32,389 Chinese children and adolescents also found that the overall prevalence of depressive symptoms was as high as 39.9%³³. A meta-analysis further revealed that the prevalence of depressive symptoms in this population has gradually increased over time in recent years, reaching a current prevalence of 26.3%³¹. Notably, the data-collection window (April–June) overlapped with the peak examination period in the latter half of the semester; academic stress may therefore have inflated the observed prevalence of emotional problems. Future studies can systematically incorporate “exam season” as a seasonal stressor to clarify its true contribution to fluctuations in depressive symptoms. Additionally, this study found that the risk of depressive symptoms increased with age, a conclusion that previous research had repeatedly confirmed^34,35. On one hand, the sharp fluctuations of sex hormones and cortisol after puberty directly disrupt emotion regulation pathways, making adolescents (especially girls) more vulnerable to emotional lows^36,37. On the other, the mounting academic workload, peer comparisons, entrance exam competition, and family expectations that come with each additional year create chronic psychological stress, further magnifying the probability that depressive symptoms will surface^38–40. Also, our study indicated that girls were significantly more likely than boys to report depressive symptoms. Sun et al. found that female adolescents exhibited a higher prevalence of depression and more severe depressive symptoms at all ages compared to males⁴¹. This gender difference may be related to the following reasons. First, the onset of the menstrual cycle and fluctuations in hormone levels can significantly affect the mood of female children and adolescents^42,43. For example, during the premenstrual period, female adolescents may experience emotional problems such as irritability and depression, which may gradually develop into depressive symptoms if they are not addressed and alleviated in a timely manner^44,45. Moreover, female children and adolescents tend to exhibit higher levels of rumination and anxiety when faced with negative events, and these unique psychological developmental characteristics make them more vulnerable to depressive symptoms^46,47. Taken together, these findings suggest that a significant proportion of Chinese children and adolescents (especially girls) exhibit depressive symptoms, a problem that cannot be ignored. Therefore, parents, school staff, and the community should closely monitor the mental health status of this population and proactively explore the risk factors associated with depressive symptoms, so that effective interventions can be implemented in a timely manner.

In-depth analyses found that ACEs were central to the development of depressive symptoms in youth. A series of previous clinical studies have elucidated the strong and complex association between ACEs and depressive symptoms^8,48,49. For instance, a 12-year longitudinal study demonstrated that ACEs, including childhood abuse and physical violence victimization, were significantly and positively linked to increased depression from adolescence to early adulthood, with females showing higher vulnerability⁴⁹. The link between ACEs and depressive symptoms may involve multiple mechanisms. At the biological level, ACEs may increase the likelihood of depression among children and adolescents through specific epigenetic changes (e.g., in NR3C1, SLC6A4, BDNF, FKBP5), dysregulated cortisol secretion rhythms, elevated levels of inflammatory markers (such as IL-6, CRP, and TNF-α), and by disrupting brain structures and functional connectivity involved in emotion regulation^50–54. At the psychosocial level, a cross-sectional study found that ACEs can contribute to depressive symptoms by interfering with an individual’s normal ability to regulate emotions⁵⁵. Another study from China found that ACEs can promote the development of borderline personality traits and reduced self-control in adolescents, which are strongly associated with depression⁵⁶. Furthermore, the present study also found that emotional abuse, emotional neglect, physical neglect, and mothers suffering from domestic violence appeared to be more strongly associated with depressive symptoms among children and adolescents. This may suggest that these specific subtypes of ACEs have a more pronounced impact on the emergence of depressive symptoms in this population. However, further longitudinal and foundational research is essential to elucidate how these ACE types specifically influence depressive symptoms among children and adolescents. This study further revealed that, in terms of cumulative effects, compared to those who had not experienced ACEs, children and adolescents who had experienced one, two, and three or more ACEs were at 1.623, 2.431, and 4.461 times higher risk of developing depressive symptoms, respectively. It indicated that the more ACEs children and adolescents accumulated, the more severe the impact on their mental health. Therefore, schools and social workers were advised to identify ACEs exposure in that population as early as possible and provide early intervention.

In mediation analyses, resilience mediated the association between ACEs and depressive symptoms among children and adolescents. Indeed, the formation of resilience can be traced back to individuals’ childhood stages, and early encounters with ACEs can disrupt the normal trajectory of psychological development, consequently increasing the risk of low resilience¹⁰. A study of patients with mood disorders revealed that individuals presenting with childhood trauma (especially emotional abuse and emotional neglect) generally exhibit low resilience⁵⁷. Furthermore, prior research has confirmed the association between resilience and depressive symptoms^58,59. A cross-sectional study focusing on the adolescent population showed that reduced resilience was significantly and positively associated with current depressive symptoms⁵⁸. Elmore et al. noted that children with ≥ 4 ACEs and low resilience were 8.75 times more likely to be depressed⁵⁹. On the one hand, individuals with low resilience tend to have significant deficits in their ability to cope with stress, which can be manifested in difficulties in seeking support from others and coping effectively with emotions, and this can significantly increase individuals’ risk of falling into poor mental health states, including depression⁶⁰. On the other hand, children and adolescents with low resilience tend to have difficulty viewing problems and setbacks in a positive light and are more likely to develop negative cognitive patterns, which can increase susceptibility to depression^61,62. Similarly, such mediating effects have been tentatively confirmed in previous studies. A survey of children from Slovakia also showed that resilience mediated the associations between ACEs and emotional and behavioural problems⁶³. Another meta-analysis further showed that childhood trauma, resilience, and depression were significantly associated, with resilience playing a mediating role between the other two⁶⁴. In summary, this study underscored the need to consider resilience as an underlying factor of interventions for depressive symptoms among children and adolescents, and actively developing more targeted psychological interventions. For example, a randomised controlled trial demonstrated that a resilience-themed curriculum significantly improved children’s resilience and mental health⁶⁵.

Also, the study identified sleep quality as an additional mediator linking ACEs to depressive symptoms among children and adolescents. The robust associations between ACEs and sleep disturbances have been widely studied. A prospective cohort study found that ACEs experienced during youth significantly predict poor sleep outcomes later on¹⁴. Such associations involve multiple mechanisms. Studies have shown that adolescents experiencing ACEs often face significant psychological stress (e.g., hypervigilance, anxiety), which can markedly disrupt the sleep-wake cycle^66,67. In terms of neurobiochemistry, ACEs also affect the balance of neurotransmitters associated with sleep regulation, such as by reducing serotonin and dopamine levels^68,69. Furthermore, our data further showed that poor sleep quality was independently linked to depressive symptoms, aligning with previous research findings^70–72. For example, a prospective study spanning nearly 30 years demonstrated a significant bidirectional association between sleep disorders and depressive symptoms in childhood and adolescence through cross-lagged modelling⁷⁰. Another human experimental study found that sleep deprivation heightens inflammatory markers, thereby increasing susceptibility to depressive symptoms⁷¹. As well, neuroimaging studies have further revealed that sleep disorders may induce depressive-like manifestations in adolescence by affecting the integrity of brain structures such as the cingulate bundle⁷².

Finally, this study provided preliminary evidence that resilience and sleep quality may have a chain-mediated effect between ACEs and depressive symptoms. This sequential pathway was consistent with stress-coping theory⁷³. Specifically, ACEs act as chronic stressors that persistently impact individuals’ psychological and physiological well-being. According to stress-coping theory, individuals employ various coping strategies to regulate the impact of stress⁷⁴. Resilience plays a key mediating role in this process, serving as a positive psychological resource that helps individuals cope more effectively with stress and mitigates its negative impacts on sleep quality and mental health^24,75. Conversely, if individuals have lower resilience, stress may be more associated with poor sleep quality and, in turn, exacerbate depressive symptoms⁷⁶. Chang et al. found that ACEs significantly undermine individuals’ resilience, which in turn disrupts sleep quality²⁰, thereby supporting our findings. These findings provided a new perspective for understanding the complex association between ACEs and depressive symptoms. They also offered a theoretical basis for developing interventions that enhance resilience to improve sleep quality and alleviate depressive symptoms. Certainly, the above pathways are only hypothetical paths supported by cross-sectional mediation analysis, and future longitudinal studies are still needed for validation. Moreover, subsequent research should also consider the possibility of reverse pathways—for example, depressive symptoms may in turn affect sleep quality and resilience.

Undeniably, the present study has some limitations: (1) This study was cross-sectional; thus, the mediation findings rest on the strong assumption of sequential ignorability, and the results should therefore be interpreted as statistical mediation rather than causal mediation. Future research should adopt a longitudinal follow-up design with multiple time points to verify the mediation model of this study. (2) This study was conducted only in Chaohu, Anhui Province, China, which may limit the external validity of the findings. It is recommended that subsequent studies expand the sampling scope by adopting a multicentre, multiregional stratified sampling strategy and consider the inclusion of samples from different cultural backgrounds to improve the generalizability of the findings. Additionally, because students were nested within schools, standard errors were likely to be underestimated, increasing the risk of Type I error. Interpretations of these findings should thus be made with caution and, whenever possible, validated in future studies using clustering-aware methods such as multilevel linear models or cluster-robust standard errors. (3) The main variables of this study were collected using self-report scales, which may introduce recall bias. Additionally, the reliability of some scales was relatively low in this study; for example, the Cronbach’s α for the PSQI was only 0.701, which could affect the stability and reliability of the findings. Therefore, it is necessary to introduce objective measures (e.g., sleep monitoring devices) in future studies to improve the reliability of the findings. (4) Depression was identified solely based on the CES-D cutoff score, without confirmation through standard clinical diagnosis. Consequently, the reported prevalence may not accurately reflect the true rate of clinical depression and could be either overestimated or underestimated. (5) The present study did not measure or control for potential moderators (e.g., cultural background) or confounders (e.g., social support, parent-child relationship quality, current stress levels), which may limit the reliability of the findings. Future research should therefore incorporate a broader set of variables to fully examine the associations among depressive symptoms, ACEs, resilience, and sleep quality among children and adolescents.

Conclusion

In this study, a high proportion of school children and adolescents exhibited depressive symptoms, which were significantly associated with ACEs, low levels of resilience, and poor sleep quality. Further analyses indicated that resilience and sleep quality may have a chain-mediated effect between ACEs and depressive symptoms. Therefore, parents and school staff should pay attention to ACEs among children and adolescents and implement timely interventions to improve resilience, prevent sleep disorders, which may be associated with a reduced risk of depression. Concrete steps include designing and delivering school-based programs aimed at strengthening youth resilience and developing evidence-based interventions to improve sleep quality as complementary strategies for alleviating depressive symptoms.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ACEs

Adverse Childhood Experiences

DSM-5

The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition

BMI

Body Mass Index

CES-D

The Center for Epidemiological Studies Depression Scale

CTQ

The Childhood Trauma Questionnaire

CYRM

The Child Youth Resilience Measurement Scale

PSQI

The Pittsburgh Sleep Quality Index

SD

Standard Deviation

VIF

Variance Inflation Factor

AUC

Area Under the Curve

CI

Confidence Interval

Author contributions

YS and HL: the study’s design. WJ, LL, PT, LZ, ZW, GQ and YC: collection, analyses and interpretation of the data. JW: drafting of the manuscript. YS and HL: revision of the manuscript. The final version of the publication was approved by all authors.

Funding

This study was supported by Anhui Health Research Project (Grant Number. AHWJ2022c057), Key Project of Natural Science Research of Educational Department of Anhui Province (Grant Number: 2022AH050664) and The “Undergraduate Open Research Project” Fund of the Key Laboratory of Birth Population Health Education, Ministry of Education (Grant Number: JKBK20245).

Data availability

The data used for this study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics statement

All participants and their legal guardians provided informed consent by signing the consent form. The study strictly adhered to the code of ethics for medical research and was approved by the Ethics Committee of the Fourth Affiliated Hospital of Anhui Medical University (Approval No. KYXM-202210-010). It also followed the relevant ethical requirements of the Declaration of Helsinki (2013 version) (https://www.wma.net/policies-post/wma-declaration-of-helsinki/).

Consent for publication

Not Applicable.

Clinical trial number

Not applicable.