Ethical approval
This study was approved by University of Reading Research Ethics Committee, the UK (SREC 2018/105) and Chiang Mai University Research Ethic Committee, Thailand (CMUREC 61/073). Informed written consent was obtained from all participants and from the parents of young people under 18 years of age based on standards prescribed by Chiang Mai University Research Ethic Committee.
Part 1 Psychometric properties of the Thai Mood and Feelings Questionnaire
Participants
Three public secondary schools in Thailand agreed to take part in the study; two from Phayao province and one from Chiang Mai province. A total of 1737 adolescents aged between 12 and 8 years old were invited to take part (male 44%, female 56%). Eighty percent (n = 1382) of those who were invited agreed to take part and provided written consent from their caregiver as well as written assent for themselves. Ninety-nine young people for whom consent was obtained were not included because they were absent from school on the day the research was conducted. Eight young people were excluded because more than 25% of their data were missing. Of the 1275 adolescents who took part 39% were male (n = 500) and 61% (n = 775) were female; significantly more females than males took part, χ2(1) = 10.72, p = 0.001. The flow chart of participant recruitment is presented in Fig. 1.
Measures
The Mood and Feelings Questionnaire (MFQ) [34] is a 33-item questionnaire used to assess depression symptoms in children and adolescents aged between 8 and 18 years old. Items are rated for the past 2 weeks. Each item is rated on a three-point Likert scale of “true” (“2”), “sometimes true” (“1”) or “not true” (“0”), yielding a maximum score of 66. The MFQ has high internal consistency (α = 0.94) [35] and can discriminate between depressed and non-depressed children and adolescents sampled from both clinical and community settings [37, 38]. A cut-off of 27 provided optimal sensitivity and specificity in identifying young people who met criteria for a diagnosis of major depressive disorder [36]. The MFQ has been adapted and translated into a number of languages and demonstrated an excellent internal consistency in translated versions e.g., Norwegian (α = 0.91) [39], Arabic (α = 0.92) [40], and Swedish version (α = 0.93) [41].
The Children’s Depression Inventory (CDI) [16] is a 27-item self-report questionnaire that measures of depression for children and adolescents ages 7 to 17 years. It measures current levels of depression symptoms on five factors: negative mood, interpersonal problems, negative self-esteem, ineffectiveness, and anhedonia. The CDI has good internal consistency and test–retest reliability [16]. The CDI has been translated into Thai [46] and had good internal consistency (α = 0.83).
The Strengths and Difficulties Questionnaire (SDQ) [47] is a 25-item screening questionnaire used to assess mental health in children and adolescents aged 4–16 years old. The SDQ has four subscales to measure emotional symptoms, peer problems, hyperactivity and inattention, and conduct problems and one subscale that assesses prosocial behaviour. The SDQ Thai version [48] has acceptable internal consistency, e.g., total difficulties subscales (α = 0.70) and emotional subscale (α = 0.63).
Translation of the MFQ
Permission was obtained from the developer for translation of the MFQ to the Thai language. The translation and adaptation of the MFQ followed guidance cross-cultural adaptation of psychological instruments [49]. First, the original version of the MFQ was translated into Thai by the first author (NF) a native Thai speaker. The Thai version was adapted for linguistic context and aimed to preserve all essential characteristics of the original version. Next, the Thai translation was back translated into English by a bilingual (Thai–English) translator from the Psychology Department at Chiang Mai University who did not have knowledge about the original instrument. The original version of the MFQ was then evaluated and compared with the back-translation. All differences were resolved by discussion. The consensus version was adequately adapted culturally and linguistically for the target population.
The Thai translation of the MFQ was then piloted with five young people in Thailand to check if they interpreted the questions as intended. Adolescents were asked to express their understanding of the measure and to suggest any changes they considered necessary. Adolescents understood all items on the MFQ. Once this was completed, the MFQ was released for use in this study (see Additional file 1).
Procedures
In part 1 of the study, following receipt of informed consent, adolescents (N = 1275) were given and completed a pack of questionnaires in their classroom during the school day at a time convenient to their schools. These were distributed and collected by the first author (NF). These data were used to examine internal consistency, descriptive statistics and construct validity of the MFQ. Participants were also asked to indicate if they were willing to take part in Part 2 of the study (diagnostic interview) and, if so, to provide contact details for the researcher.
Research procedures and risk management process were discussed with schools in advance. On the day of data collection, the researcher identified students whose response to the questionnaires suggested that they were at risk of suicide or self-harm. The researcher informed a member of the school safeguarding team on the same day as the risk was disclosed, following school’s safeguarding guidelines.
Statistical analysis
Data were analysed using SPSS version 25. Fewer than1% of items had missing values. Participants’ data were excluded from the analysis if more than 25% was missing (n = 8). Where fewer than 25% of item were missing, mean item substitution was used to impute missing data. An independent t-test was used to examine the mean difference between male and female. The association between age and depression symptoms was assessed by a Pearson’s correlation. Internal consistency of the MFQ was assessed with Cronbach’s alpha. Convergent validity of the MFQ was assessed through Pearson correlation coefficients between total score of the MFQ and the CDI total score and the SDQ Emotion symptoms subscale.
Part 2: Diagnostic interview
Participants
One thousand and twenty-one (80%) adolescents agreed to take part in a follow up interview. Based on the MFQ score in Part 1, participants were divided into two groups (elevated, i.e., 27 and above, and sub-threshold, i.e., below 27 [36]). Random samples were then selected from each group. MadCalc version 19.7 was used to estimate a required sample size for a Receiver Operating Characteristics (ROC) curve. The power analysis determined a minimum sample size of 31 participants in each group (i.e. elevated and sub-threshold group) included in ROC analyses to achieve a sufficient power of 0.80 with an Area Under the Curve (AUC) of 0.70 and α = 0.5. To ensure that there were sufficient participants with elevated symptoms of depression, young people with elevated MFQ scores were over-sampled in a ratio of 1.8 (elevated) to 1.0 (sub-threshold). Of the 184 young people invited to the interview, 138 young people and their guardians consented to take part (75%). Twenty-five young people could not subsequently be contacted, and ten young people were not at school on the day of the interview. The remaining 103 young people were interviewed (see Fig. 1). Based on their MFQ scores on the second administration participants were classified as having an elevated Thai MFQ (n = 44) or sub-threshold Thai MFQ (n = 59).
Measure
The Kiddie Schedule for Affective Disorders and Schizophrenia for School-Age Children; Present and Lifetime Version; Depressive disorder (K-SADS-PL DSM-5) [50] is a semi-structured interview with well-established psychometric properties that generates a reliable and valid diagnosis of depression in children and adolescents [42]. The K-SADS-PL has been widely used in epidemiological and treatment research [45, 51] and has also been translated and adapted into many languages [44, 45, 52] In the present study, the depressive disorder section of the K-SADS-PL, revised to be compatible with DSM-5 diagnoses, was used to determine the presence/absence of depressive disorder in Thai adolescents.
Translation of the K-SADS-PL; Depressive Disorder
The translation and adaptation of the K-SADS-PL followed the procedure in Part 1. The K-SADS-PL; Depressive disorder was translated into Thai by the first author (NF). The Thai version was adapted for linguistic context and aimed to preserve all essential characteristics of the original version. The K-SADS-PL; Depressive disorder Thai version was back-translated into English by a bilingual child and adolescent psychologist (Thai-English) translator from the Psychology Department at Chiang Mai University. The back-translation version was reviewed and compared with the original version. The final translation was fixed by consensus. The completed Thai version of the K-SADS-PL; Depressive disorder was administered to five young people in Thailand to check if they understood the questions of the measure. Once this was completed, the K-SADS-PL was used in this study.
Procedures
Part 2 of the study took place between 11 and 30 days after participants completed the self-report questionnaires. Participants took part in the K-SADS-PL interviews, which were conducted in a quiet room at school by the first author (NF), and completed the MFQ after the interviews. The interviews were audio-taped and detailed assessment notes were taken. Subsequently each interview recoding (n = 103) was coded according to K-SADS-PL diagnostic criteria by the first author (NF). NF has enhanced DBS and was trained to deliver and score the K-SADS-PL through training which included verbal instruction, watching training videos, and participating in diagnostic consensus supervision meetings where each individual symptom was discussed for parent and child, and a consensus agreed for each symptom before reaching an overall diagnosis decision. To check reliability of the diagnosis, 10% of the samples were double-rated by an experienced K-SADS-PL assessor (FO). Inter-rater reliability for the presence of depression diagnoses on the K-SADS-PL was κ = 0.80, and on individual symptoms was κ = 0.75.
The researcher discussed the diagnostic interview and safeguarding procedures with the schools before the interview. The school’s safeguarding guidelines were followed on the day of interview. Therefore, the researcher alerted the school safeguarding team on the same day as the risk was disclosed (i.e., experienced or thoughts of self-harm or suicide). Research procedures and safeguarding processes were also explained in information sheets to students and their parents.
Statistical analysis
In Part 2, the MFQ data from the second administration were used and analysed using SPSS version 25. Diagnoses were assigned according to the K-SADS-PL. There was no missing data on the MFQ or demographic information for the sub-sample of 103. Chi-square test was used to determine association of categorical variables. Mean differences of the MFQ were examined for young people who were given a diagnosis of MDD versus those who did not receive a diagnosis of MDD. Test–retest reliability was assessed by using Intraclass correlation coefficients (ICCs) absolute agreement model [53]. Because the interval between completing the MFQ at first and second administration varied, participants were allocated to two groups. Group one completed the MFQ approximately 2 weeks apart (11–19 days, n = 40); group two completed the MFQ approximately 4 weeks apart (20—30 days, n = 63). Receiver Operating Characteristic (ROC) analyses [54] were conducted to examine the ability of the MFQ to discriminate between individuals with and without a diagnosis of Major Depressive Disorder.
The ROC curve analysis provided information regarding sensitivity (the probability that the test correctly classifies subject with condition as positive), specificity (the probability that the test correctly classifies subject without condition as negative), positive predictive value (the probability of the presence of disease in those with a positive test result), and negative predictive values (the probability of the absence of disease in those with a negative test results) [55] of the MFQ for candidate cut-off points. As the MFQ is designed to be used primarily as a screening instrument, the optimal cut-off was determined by favouring sensitivity over specificity; therefore, minimum sensitivity was set as 80% as minimum specificity was set at 70% [56]. The accuracy of the MFQ in detecting depression was evaluated by the area under the curves (AUC). AUC measures the ability of screening measures to correctly classify those with and without the disease. AUC of 1.0 represents perfect diagnostic accuracy, AUC greater than 0.9 reflects high accuracy, 0.7–0.9 moderate accuracy and below 0.7 indicates low diagnostic accuracy [54].