Participants
This study used data from the Tokyo Teen Cohort study (TTC, http://ttcp.umin.jp/), a population-based longitudinal survey focusing on children’s health from biopsychosocial multidisciplinary viewpoints. The TTC was conducted from October 2012, and the participants were recruited from three municipalities in Tokyo (Setagaya, Mitaka, and Chofu) using the Basic Resident Register. The candidate participants included 14,553 children born between September 1, 2002, and August 31, 2004. Invitation letters were sent to the primary parents of those children around their tenth birthday. Of these children, 10,234 could be contacted, and the children were invited to participate in the cohort study. Of the 10,234 children, 4,478 children participated in the baseline survey named the Tokyo Early Adolescence Survey (T-EAS) (20-24). This baseline survey was conducted from October 2012 to January 2015 when participants were approximately 10 years old (time 1, T1). Then, we carried out an adjustment of the social economic status for the participants in this baseline survey, and 3,171 participants were extracted as a target of TTC. The second wave in TTC was carried out from August 2014 to December 2016 at the time the participants were approximately 12 years old (time 2, T2), and the follow-up rate was 94.8%.
Ethical approval
Ethical approval for this study was obtained from the research ethics committees of Tokyo Metropolitan Institute of Medical Science (Approval number: 12-35), SOKENDAI (The Graduate University for Advanced Studies) (2012002), and the Graduate School of Medicine and Faculty of Medicine, The University of Tokyo (10057). We obtained informed assent from children and written informed consent from primary parents (mostly mothers). In each wave of data collection, trained interviewers visited participants’ homes. They distributed questionnaires to the children and primary parents and conducted psychological tests for the children.
Measures
Tics
We evaluated tics at T1 and T2. Participants’ primary parents answered the questionnaire about children’s tics, which had been used in a previous study (15). The questionnaire included a section with five questions about specific motor and vocal tics in the past year: Q1: Has your child had any repeated movements of parts of the face and head (e.g., eye blinking, grimacing, sticking tongue out, licking lips, spitting)?; Q2: Has your child had repeated movements of the neck, shoulder or trunk (e.g., twisting around, shoulder shrugging, bending over, nodding)?; Q3: Has your child had repeated movements of arms, hands, legs, feet?; Q4: Has your child had repeated noises and sounds (e.g., coughing, clearing throat, grunting, gurgling, hissing)?; Q5: Has your child had repeated words or phrases?. Each question was answered as “definitely”, “probably” or “not at all” present. Furthermore, we asked the following question about the frequency of these repetitive behaviors: "Q6: About how often does/did this happen in the last year?". This question was answered on a 5-Likert scale: “1: Less than once a month, 2: 1-3 times a month, 3: About once a week, 4: More than once a week, 5: Every day.”
We evaluated presence of tics by binary valuables; with tics or without tics. We defined participants who responded “Definitely” or “Probably” to any of Q1, Q2, and Q4 as having tics. Participants who only endorsed repeated movements of the arms, hands, legs or feet (Q3) or repeated words or phrases (Q5) in the absence of a positive response to other questions about types of tics (Q1, Q2, Q4) were excluded from all case definitions to remove nontic movements such as stereotypy or isolated echolalia. We defined all responses of “definitely” or “Probably” to motor and/or vocal tic as tics regardless of the frequency because there is no condition of the frequency in diagnostic criteria of tic disorders (25) and because we aimed to exhaustively find tics in the general population. In post hoc analyses, we used a narrower definition of tics that required a frequency of the motor and/or vocal tic of twice a week or more.
Maternal depressive/anxious symptoms
We employed the Kessler Psychological Distress Scale (K6) (26-28) for T1 and the General Health Questionnaire-28 (GHQ-28) (29, 30) for T2. The K6 and the GHQ-28 are both widely used self-reported questionnaires developed to evaluate depressive/anxious symptoms. If primary parents other than mothers answered the K6 or GHQ-28, we regarded those responses as missing values. In this study, we used raw values of K6 and GHQ-28 as continuous scales not as screening scales for the purpose of evaluating the severity of depressive/anxious symptoms including normal range in general population. Cronbach’s alpha was .84 for K6, and .88 for GHQ-28. We found the distributions of K6 and GHQ-28 were similar by the graphing cumulative distribution of the Z score of K6 and GHQ-28.
Other variables
Sex (7, 15, 31), age (9, 10, 32), maternal age (17, 19, 33, 34), socioeconomic status (35), and maternal alcohol use during pregnancy (36) were included in the analyses since previous studies reported that these factors influenced the occurrence of TS/CT. To assess socioeconomic status, family income was evaluated on a 10-point scale, which ranged from "0-990,000 yen” to “more than 10,000,000 yen”. Information on maternal alcohol use during pregnancy was obtained from maternity record books that were provided for almost all mothers by local public organizations in Japan.
Statistical analysis
Longitudinal relationships between maternal depressive/anxious symptoms and children’s tics were studied with structural equation modeling (Figure 1). We used SPSS® (Statistical Package for Social Science; IBM Corp., Armonk, N.Y. USA) version 21.0 for characteristics of the study participants and Amos ver. 22.0 (IBM Corp, New York) for structural equation modeling. We used the following three cross-lagged design models. The first model analyzed the longitudinal relationships between maternal depressive/anxious symptoms and children’s tics without adjusting covariates (unadjusted model). The second model adjusted for sex, age in months, family income, maternal age, and maternal alcohol use during pregnancy (adjusted model). Finally, we took into account the nonnormal distribution of particular dependent variables (maternal depressive/anxious symptoms and children’s tics at T2), logarithmically converted these measures and conducted the analysis (logarithmically transformed model).
We also conducted post hoc analysis by using a narrower definition of tics, which had been used in a previous study about the prevalence of TS/CT (15). The aims of the post hoc analysis were to compare the prevalence of tics in this study with the previous study and to examine the stability of the results on the relationships between maternal depressive/anxious symptoms and children’s tics.
Missing values in tics, maternal depressive/anxious symptoms, and covariates were accounted for by full information maximum likelihood procedures available in Amos. This method estimates model parameters and standard errors using all available data while adjusting for the uncertainty associated with missing data (37).
A threshold for statistical significance was set to p < .05 (two-sided) for all analyses. We evaluated the fit of our models by using the comparative fit index (CFI) and the root mean square error of approximation (RMSEA). A good model fit was indicated by an RMSEA value smaller than .05 and a CFI value larger than .95 (38, 39).