Participants

In this study, 38 adolescents (aged between 12 and 18 years old) were recruited in the study between February and December 2019. Recruitment was based on patients’ admission to the Children and Adolescents Outpatient Department of Mental Health, the First Hospital of Shanxi Medical University. The inclusion criteria consisted of: (1) age between 12 and 18 years old; (2) functional assessment of school refusal as suggested by Kearney and Albano,2004 [34]; (3) more than 50% of school absence or absence in the 4 weeks prior to the visit. Exclusion criteria consists of an anxiety and depression diagnosis which is assessed by the psychiatrist of the study using the Mini-International Neuropsychiatric Interview (M.I.N.I.)(compatible with the Diagnostic and Statistical Manual of Mental Disorders, fifth Edition (DSM-5)) [35]. We recruited healthy controls (n = 35) locally from advertisement. The inclusion for the HC including aged 12–18 years old, without gender limitation, the flyers for including healthy controls (aged 12–18 years old, without gender limitation, Han nationality) were posted in middle schools in Taiyuan City. All the informed consent were obtained from the participants themselves and their parents/caregivers. (NO. of ethics approval: KYLL-2023-080).

MeasuresGeneral demographic data

We collected demographic data from all participants that included age, gender, and education level.

Eysenck Personality Questionnaire

We used the Chinese version of Eysenck Personality Questionnaire (EPQ) that consists of four subscales: extraversion (E), neuroticism (N), psychoticism (P) and lying (L) [36]. Binary answers were provided. The Chinese version of Eysenck Personality Questionnaire has high reliability and validity [37, 38].

Zung Self-rating Depression Scale

The Chinese version of Zung Self-rating Depression Scale (SDS) [39] is used for assessment of depressive symptoms which includes 20 items and each item is graded 1 to 4. Some items 2, 5, 6, 11, 12, 14, 16, 17, 18 and 20 are graded in reverse. Previous studies have found the scale to be appropriate and commonly used by Chinese people [40,41,42]. Mild depression is considered when the total score is between 50 and 59. A moderate depression is considered when the total score is between 60 and 69, and a severe depression is associated to a score above 69. The Chinese version of Zung Self-rating Depression Scale has high reliability and validity [43, 44].

Zung Self-Rating anxiety scale

There are 20 items in the Chinese version of Zung Self-Rating Anxiety Scale (SAS) [45], and each item is graded on 1 to 4 levels. Among them, items 5, 9, 13, 17 and 19 are graded in reverse. Studies demonstrate that the scale can be widely used to screen for anxiety in the Chinese population [40, 42, 46]. A score between 50 and 59 is associated with mild anxiety, a score between 60 and 69 is associated with moderate anxiety, and a score above 69 scores is associated with severe anxiety. The Cronbach’s α coefficient was 0.913, and has high constructive validity [47] .

Symptom Checklist 90

We used the Chinese version of Symptom Checklist 90 (SCL-90),which consists of 90 items with each item graded 1 to 5. SCL-90 includes 10 factors that reflect somatization, obsessive symptoms, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, psychoticism, and the other aspects of psychological symptoms(measurement of individual sleep and diet) [48]. If the total score is over 160, or the number of positive items is more than 43, or the mean of factor score is ≥ 2, this will be an indication of mild or above psychological problems. If the mean of each factor score is ≥ 2.5, this indicates that the psychological pain has reached a moderate level or above. If the individual factor score is ≥ 3, this indicates that the pain level has reached a mild or above severe level, indicating the possibility of psychological problems. The Chinese version of SCL-90 has high reliability and validity. The reliability of the general scale was 0.97, and was over 0.67 for all the subscales. Test-retested correlation was over 0.70. SCL-90 had high content validity and constructive validity [49].

Data collection

The hemodynamic responses in the prefrontal cortices and superior temporal cortices was measured by a 52-channel fNIRS system (ETG-4100. Hitachi Medical Co., Tokyo, Japan) with 2 NIR light wavelengths (695 and 830 nm). The fNIRS system contains 16 light detectors and 17 light emitters, all of which were arranged in a 3 × 11 array to form 52 measurement channels. All the participants were asked to perform a Verbal Fluency Task (VFT) in a quiet environment. Participants were asked to seat with eyes open, avoiding excessive body and head movements, and focusing on a cross on the screen. The VFT test comprised a 30-s pre-task period, a 60-s task period, and a 70-s post-task period. During the pre- and post-task periods, the participants were asked to constantly say “one, two, three, four, five” repeatedly. During the task period, the participants were asked to think as many four-character idioms or phrases as possible, which begin with big, white, and sky [29].

fNIRS analysisData preprocessing

The near-infrared spectroscopy signals were preprocessed using the NIRS-SPM toolbox, which is a MATLAB-based software package (MATLAB 2013b). The preprocessing steps included: transforming all .csv files into NIRS-SPM available .mat files; checking for participants’ available channels.

Calculate the β-value

The NIRS-SPM toolbox mainly uses the general linear model (GLM) method in data analysis, The GLM is formulated as follows: Y = βX + ε. In this study, β is represents the level of cortical activation during the VFT.

First, low-frequency drift generated by breathing, heartbeat, or other factors was conducted using the discrete cosine transform (DCT). Physiological noise was filtered using a low-pass filter that is based on the hemodynamic response function (HRF).

Second, a GLM was constructed using the time series associated with rest and task performance as the independent variables, the oxyhemoglobin concentration as the dependent variables. The first-order derivative and second-order derivative of the time series were used as covariates in this process.

Third, the value of β was calculated.

Index extraction

The values of β were extracted for 52 channels for participants. The degree of activation of the brain cortex during the VFT task was assessed by δβ value of oxy-hemoglobin (VFT β value minus baseline β value).

Statistical analysis

SPSS 22.0 was used for the data analysis (SPSS Inc, Chicago, IL, USA).

The categorical data (gender) were analyzed with the chi-square test. The numerical data was analyzed using two independent sample test, including age, educational age, duration of SR, total scores and each subscale scores of EPQ, SAS, SDS, SCI-90 (with age, gender and educational years as covariates). A one-way ANOVA was used, with group as the between-group factor (SR group and HC group), and age and gender as covariates. This was used to compare the differences in frontal and temporal cortex activation levels between the SR group and HC group. The p-value was corrected by false discovery rate correction(FDR, the FDR correction method ranks multiple hypotheses according to the magnitude of the p-value and then the significance level of each hypothesis is determined according to the ranked order) [50, 51]. And less than 0.05 was considered statistically significant. Bonferroni test was used for post-hoc analysis to identify the sources of differences (Bonferroni corrections to minimize type I errors, specifically raw p value*number of t tests. Values less than 0.05 reached by the Bonferroni correction were considered statistically significant).

Mean δβ values were extracted for channels with statistically significant results, and correlation between δβ values and clinical symptoms (EPQ subscale scores, SAS total score, SDS total score, and SCL-90 total score) using Pearson’s correlation analysis (with age, gender and educational years as covariates). The p-value was also corrected by FDR.