Study protocol: families and childhood transitions study (FACTS) – a longitudinal investigation of the role of the family environment in brain development and risk for mental health disorders in community based children

Background Extant research has demonstrated that parenting behaviour can be a significant contributor to the development of brain structure and mental health during adolescence. Nonetheless, there is limited research examining these relationships during late childhood, and particularly in the critical period of brain development occurring between 8 and 10 years of age. The effects of the family environment on the brain during late childhood may have significant implications for later functioning, and particularly mental health. The Families and Childhood Transitions Study (FACTS) is a multidisciplinary longitudinal cohort study of brain development and mental health, with two waves of data collection currently funded, occurring 18-months apart, when child participants are aged approximately 8- and 10-years old. Methods/design Participants are 163 children (M age [SD] = 8.44 [0.34] years, 76 males) and their mothers (M age [SD] = 40.34 [5.43] years). Of the 163 families who consented to participate, 156 completed a video-recorded and observer-coded dyadic interaction task and 153 completed a child magnetic resonance imaging brain scan at baseline. Families were recruited from lower socioeconomic status (SES) areas to maximise rates of social disadvantage and variation in parenting behaviours. All experimental measures and tasks completed at baseline are repeated at an 18-month follow-up, excluding the observer coded family interaction tasks. The baseline assessment was completed in October 2015, and the 18-month follow up will be completed May 2017. Discussion This study, by examining the neurobiological and mental health consequences of variations in parenting, has the potential to significantly advance our understanding of child development and risk processes. Recruitment of lower SES families will also allow assessment of resilience factors given the poorer outcomes often associated with this population. Electronic supplementary material The online version of this article (doi:10.1186/s12887-017-0905-x) contains supplementary material, which is available to authorized users.


Measures associated with the Mock MRI procedure
Prior to completing the MRI, participants took part in a mock MRI procedure on a disused machine at the Royal Children's Hospital (RCH) in Melbourne. Mock MRI procedures improve both child MRI participation and the quality of structural/functional brain data collected during scans (de Bie et al., 2010;Carter, Greer, Gray & Ware, 2010). During the mock MRI, participants were shown photographs depicting each step of the MRI procedure which the researcher discussed with the child. After this, participants completed a visual analogue scale (VAS) to gauge whether they felt nervous, upset, interested and/or happy. For each emotion, participants were presented with a horizontal line bookended by the words 'Not at all' and 'A lot' and thus representing a continuum. Participants were then asked to draw an X at the point on the line which best represented their feelings toward completing the MRI.
The purpose of this scale was to assist the researcher in understanding the child's feeling towards the MRI procedure allowing any issues to be worked through during the mock MRI.
The researcher then showed the participant the mock MRI machine and encouraged them to practice lying in the machine while listening to a recording of sounds that an MRI machine typically makes while functioning. Furthermore, the participant tried wearing headphones and a helmet similar to those used in the actual MRI.

Image Acquisition
Neuroimaging data were acquired on the 3T Siemens TIM Trio scanner (Siemens, Erlangen, Germany) at the Murdoch Childrens Research Institute (MCRI), RCH, Melbourne, Australia.
Participants lay supine with their head supported in a 32-channel head coil.

Resting fMRI
A continuous functional gradient-recalled acquisition sequence is conducted at rest to acquire 154 whole-brain T2*-weighted echo-planar volumes (repetition time = 2400ms, echo time = 35ms, flip angle = 90°; field of view = 210 x 210mm 2 , 38 interleaved slices, voxel size of 3.3mm 3 ). Complex field maps are obtained in order to correct for distortion caused by magnetic field inhomogeneities. Total sequence duration 6.18 min.

Affective faces fMRI task
Participants are administered (at the 18-month follow-up only) a modified version of the emotional face-matching task originally reported by Hariri et al. (2000). In this task participants must either match the gender of faces presented (face condition), or match shapes Last updated 10/04/2017 Page 4 of 42 (control condition). During each 4s "face trial", participants are presented with a target face (centre top) and two probe faces (bottom left and right) and are instructed to match the probe of the same gender to the target by pressing a button either on the left or right. During each 4s "shape trial" participants are presented with a target shape (centre top) and two probe shapes (bottom left and right) and are instructed to match the probe of the same shape to the target by pressing a button either on the left or right. Each block consists of six consecutive trials containing angry or fearful faces (face condition) or shapes (control condition). A total of three 24-s blocks of each emotional face condition (i.e. angry and fearful) and six 24-s blocks of the control condition (shapes) are presented interleaved in a pseudo-randomized order. A fixation cross lasting 10-s is interspersed between each block. The total task time is 7 minutes.
For each trial, response accuracy and response latency (reaction time) is obtained. Prior to the scan, participants complete a short practice version of the task with different emotional faces (happy and angry). Parameters include 136 whole-brain T2*-weighted echo-planar images (repetition time = 3000ms, echo time = 35ms, flip angle = 85°) within a field of view of 216x216mm 2 , with a voxel size of 3mm 3 . Forty interleaved slices are acquired. Total sequence duration 6:42min Hariri, A. R., Bookheimer, S. Y., & Mazziotta, J. C. (2000). Modulating emotional responses: effects of a neocortical network on the limbic system. Neuroreport, 11(1), 43-48.

Family Interaction Task: Event Planning Interaction & Problem Solving Interaction
The Family Interaction Task (FIT) included two 15-minute tasks, an Event Planning Interaction (EPI) followed by a Problem Solving Interaction (PSI), that mother-child dyads completed together. During the EPI, participants planed three fun events together such as 'taking a trip or vacation'. These activities were chosen from the Pleasant Events Checklist several training videos and a final reliability assessment for which trainees independently scored five videos using FIMS codes. Trainees were required to achieve 90% agreement with from a master document for each training video. Once this level of proficiency was achieved coding commenced. The coders worked systematically through the audiovisual material in a counterbalanced sequence; participant dyads were coded in batches of 20 with one coder starting at each end of the set. For the first 60 participants reliability was calculated between coders three times. At the completion of all coding reliability was calculated again with our coders achieving 95% agreement between codes.
FIMS items: excluded & additional items -The FIMS system includes general family measures designed for interactions involving more than two people (i.e. two parents and a child). Interactions for our study only involved mother-child dyads so most of these items were excluded excepting the item 'the family is able to reach agreement or resolution' which was included because none of the dyadic items covered this behaviour. Also, the FIMS has been used to assess parenting of children with chronic medical conditions with items such as 'active catering to the child' and 'parental behaviour that infantalises the child' being particularly relevant to such as focus. We excluded these items because the participants in our study represented a healthy population for whom these items were not relevant.
For the purposes of this study minor additional items were also added to the original

Saliva Samples
Short-term circulating levels of cortisol, testosterone, DHEA and DHEA-S were obtained through the measurement of average salivary levels across two mornings. DHEA and DHEA-S are hormonal markers of adrenarcheal development, testosterone of adrenarcheal development in girls and both adrenarcheal and gonadarcheal development in boys, and cortisol an important corollary of HPA axis functioning. With the help of a parent or guardian, children collected two saliva samples: one was collected the day before their appointment and the second was collected on the day of their appointment. All equipment was provided. Participants were instructed to collect the sample immediately after waking (prior to tooth brushing, eating, drinking or taking medication). Saliva was collected via the passive drool of whole saliva through a straw into the test tube. Parents were asked to time the Last updated 10/04/2017 Page 8 of 42 duration of the 2ml saliva collection using a stopwatch. Samples were then frozen with Techni-Ice TM in the participant's freezer within a sealed container. This was then transported to RCH on the day of the appointment in a cooler bag. During the assessment the sample was stored in a larger Styrofoam cooler box packed with Techni-Ice TM , this was later transferred to the MCRI freezer and stored at -30°C. Families were asked to minimize the time that the samples spent out of the freezer. At time of assay (at the conclusion of T1 data collection), samples were defrosted and centrifuged, with the supernatant assayed for levels of testosterone, cortisol, DHEA and DHEA-S. T1 assays were conducted at MCRI, using Salimetrics ELISA kits of the same lot number, and with in-house controls. The inter-assay coefficients of variation (CVs) for the baseline assessment were: DHEA = 11.76%; DHEA-S = 13.77%; testosterone = 10.47%; cortisol = 5.32%. The intra-assay CVs were: DHEA = 9.03%; DHEA-S = 7.82%; testosterone = 8.17%; cortisol = 3.47%.
Remaining supernatant is stored in 1ml aliquots (typically x3) in a -80°C freezer for future assays when funding allows, including other hormones (e.g., oestradiol) and immune system biomarkers (e.g., C-reactive protein and secretory immunoglobulin A). Salivary assays of biomarkers testosterone, cortisol, DHEA, and DHEA-S, are well-accepted substitutes for measuring serum levels (Byrne et al., 2013;Granger et al., 2007) Anthropometry Body composition of the child was obtained through measurements of height, weight and waist circumference as described in Simmons (Simmons et al., 2014). Two measurements of each were taken, however if they were not within a specified range (0.5cm for height, 0.1 kg for weight, 0.5cm for waist) a third measurement was collected. The mean value was calculated for two measurements; the median value if three measurements were obtained.
Raw figures (either the mean or median) were transformed into z scores according to age and gender related charts-these values were used to indicate body composition, a validated indicator for obesity across childhood and adolescence (Cole, Bellizzi, Flegal, & Dietz, 2000).
Waist circumference was used to indicate intra-abdominal fat-this has been validated as a Last updated 10/04/2017 Page 11 of 42 reliable indicator (Taylor, Jones, Williams, & Goulding, 2000). Measurements were taken using the following materials: a portable rigid Invicta stadiometer was used to measure height; calibrated Tanita THD 382 digital scales measured weight; and waist circumference was measured by non-stretch anthropometric tape (according to the International Society for the Advancement of Kinanthropometry (ISAK) protocols). To improve accuracy of these measurements children were instructed to remove bulky or heavy items such as shoes and coats.

Demographics and Health Information
Detailed demographic information was collected including parental age, language spoken at home, race, ethnicity, adoption status, and country of birth for the maternal and paternal grandparents, mother, father, and child. Also collected was socioeconomic data such as residential neighbourhood, parental education, occupation and annual household income.
Information about family structure was collected including significant caregivers and siblings (both biological and non-biological) living inside as well as outside the home. A brief mental health history of the primary caregivers was taken using the maternal-reported lifetime diagnosis of psychiatric symptomsa brief interview using the dedicated subsection of the Kiddie-Schedule for Affective Disorders and Schizophrenia-Present and Lifetime Version (K-SADS-PL; Kaufman et al., 1997). During this interview, the parent was asked to recall whether they had been diagnosed, or experienced symptoms relating to, the following presentations: depression, anxiety, mania/hypomania, schizophrenia, psychotic symptoms, conduct or antisocial disorders, and substance use. If mental health diagnosis/symptoms were  (7), 980-988.

Parent Questionnaires
Questionnaires completed by parents/guardians comprised self-report and reports about the child. The broad domains covered by questionnaires were; parenting and family environment, symptoms and behaviours relating to the parent, and symptoms, traumatic events, adjustment and development relating to the child. the frequency of behaviour, 1= never to 5 = always (Shelton, Frick & Wootton, 1996). Factor analyses have substantiated the five factor subscales (Essau, Sasagawa, & Frick, 2006), and the questionnaire has been utilised and published widely with a range of populations. The APQ has normative data in the relevant age range, as well good evidence for reliability and validity (e.g., Dadds, Maujean, & Fraser, 2003). The ASSIST (version 3.1) is a questionnaire originally developed by the World Health Organization (WHO) as a tool for early screening of psychoactive substance use in primary health care setting. The questionnaire investigates the use of substances, not for medical purpose, across the life span and in the three months preceding the assessment. The substances listed are alcohol, tobacco, cannabis, cocaine, amphetamine type stimulants, inhalants, sedative or sleeping pills, hallucinogens, opioids and other -as specified by the respondent. There are 8 questions in total, which assess use, frequency, desire and urge to use (retrospective report), and functional impacts. Response options for Question 2 to 5 are never (not used in the last 3 month), once or twice (1 -2 times in the last 3 month), monthly (1 -3 times in each month, of the last 3 month), weekly (1 -4 times per week, over the last 3 months), daily or almost daily (5 -7 days per week, over the last 3 months). Questions 6 and 7 refer back to substances endorsed in Question 1 (lifetime use) asking how often a friend, relative or anyone else has ever expressed concern about the use of the substance, and any failure tentative to control, cut down or stop using the substance. Response options for Questions 6 and 7 are Nonever, yesin the past three months, yesbut not in the past three months. Question 8 asks about any use of drug by injection, not for medical purpose.

Self-Report Questionnaires
Response options for question 8 are Nonever, yesin the past three months, yesbut not in the past three months. consistency (Cronbach's alpha) that was .89 for the total score (i.e. the sum of response across all substances) and ranged from .77 to .94 considering each substance separately. The questionnaire demonstrated also to be able to discriminate between "use" (low risk), "abuse" (moderate risk) and "dependence" (high risk) groups. Therefore cut-off scores were created resulting is the current version The BAI scale is a self-report measure of severity of anxiety. It was originally develop with psychiatric population to overcome the lack of measures able to discriminate between anxiety and depression. It consists of 21 items that represent common symptoms of anxiety and tap on two main components (factors): somatic symptoms and subjective anxiety/panic symptoms.
The respondent is required to rate how much they have been bothered by each symptom in the week preceding (including the day of) the assessment. The scale ranges from 0 to 3, where 0 corresponds to not at all, 1 to mildly ("it did not bother me much"), 2 to moderately ("it was very unpleasant but I could stand it") and 3 to severely ("I could barely stand it"). The scale takes about 5-10 minutes to complete.
The BAI was able to discriminate between patients with anxiety disorders (either primary or secondary) from depressed patients (without anxiety disorders) and controls. It are asked to rate "how often this past week did you…", response options range from 0 to 3 for each item (0 = rarely or not at all, 1 = sometimes, 2 = occasionally, 3 = most of the time). The total score ranges from 0-60 and a higher score indicates a greater symptomatology. It is widely adopted as clinical significance cut-off a total score equal to or above 16. Validation studies were conducted both on the general population and on two clinical subsamples. The internal consistency (Cronbach's alpha) ranged between .84 and .85 in general population and .90 in the clinical subsample. Test-retest reliability on general population ranged from .51 to .67 and from .32 to .54 depending on the method of re-test (self-administered mail-backs or re-interviews respectively). It is worth to consider that the scale is designed to measure the current level of symptomatology (the past week) therefore variability is expected. In the clinical sample correlation of the score at the admission and after 4 weeks of treatment was .53. The scale was also able to discriminate well between patients and general population and Items are presented in a six point format requiring respondents to answer "never", "only once", "several times", "monthly", "weekly" or "daily" in a twelve month period. The four following

Conflict Behaviour Questionnaire (CBQ)
The CBQ is a 20-item true/false self-report measure that assesses general conflict between parents and their children over a two week period. Items includes statements such as, 'My child is easy to get along with' and 'My child often doesn't do what I ask'. It has demonstrated sound discriminant validity (Grace, Kelley, & McCain, 1993) and adequate internal consistency (Robin & Foster, 1989). In addition, the CBQ has been found to discriminate between distressed and non-distressed families (Robin & Foster, 1989). The shortened 20-item version is highly correlated with the original 75-item questionnaire (.96;Prinz, Foster, Kent & O'Leary, 1979;Robin & Foster, 1989). likelihood of adopting each response on a Likert scale from 1 (very unlikely) to 7 (very likely). Mothers of children between age 4 and 7 constituted the original sample (Ladouceur, Reid & Jacques, 2002). A further study adapted the original version to an older sample of early adolescents 11-13 years old by modifying 5 of the 12 scenarios (Yap, Allen & Ladouceur, 2008). In this study correlational analyses were conducted and three subscales (socialisation, encouragement and discomfort) appeared to be positively correlated between each other and negatively correlated with the subscale encouragement. Therefore the three subscales were grouped together into one composite scale defined "invalidating reactions" opposite to encouragement that it supposes to reflect non-restrictive (validating) responses to positive affectivity. In the original sample the internal consistency of the subscales was good The RDAS is a 14-item scale designed to measure relationship satisfaction (Busby, Christensen, Crane, & Larson, 1995). Items one to six rate the approximate extent of agreement or disagreement between the participant and their partner on issues such as, religious matters and career decisions. Six responses are available, ranging from 'always agree' to 'always disagree'. The language of the response options changes for items seven to ten to 'all of the time' to 'never'. For item 11, 'Do you and your mates engage in outside interests together?', five responses are presented ranging from 'everyday' to never'. The final three items revert back to six responses, scored from 'Never' to 'More Often'. The RDAS is a revised version of the original 32-item Dyadic Adjustment Scale (Spanier, 1976). The revised version offers improved psychometric properties, is shorter, and includes only 3 of the original 4 subscales: (1) Dyadic Consensusdegree to which respondent agrees with partner; (2) Dyadic Satisfaction --degree to which respondent feels satisfied with partner; (3)

Parent-Report on Child Questionnaires
Children's Depression Inventory 2 -Parent Report (CDI-2-PR) The CDI-PR is a parent-report of the child's depressive symptoms. It has 17 items in total, with 4 reverse-scored items to avoid negative bias. Similar to the CDI-C, questions aim to identify the presence of internalising pathology across a number of areas including: emotionally, behaviorally, academically and socially. Each item is a statement about the parent's observations of their child across the past two weeks. Items are rated on a 4-point scale from "not at all (0)" "some of the time (1)", "often (2)", "much or most of the time (3)".
The use of the CDI-C and CDI-P in combination may provide a clearer understanding of the child's emotional and behavioral experience. A multi-informant approach is considered a superior way of obtaining information about child psychiatric disorders (Fristad et al., 1991). For instance, parents are superior at reporting child externalizing behaviours and peer relationships while the child is better at reporting their internal experience of emotions and completion. Questions are presented in nine sections, such as child's everyday activities, child's pain and child's self-esteem. Questions about the impact of the child's health status on the parent are also included. Except for three general questions, the questions refer to the 4 weeks preceding the assessment. Response options are on a Likert formats with 4, 5 or 6 options.
The reliability of the short version has been investigated more recently (Raat, Botterweck, Landgraf, Hoogeveen & Essink-Bot, 2005). The physical and psychosocial summary scores showed adequate internal consistency (Cronbach's alpha > .80), while considering the single scales it ranged between .34 and .85. Test-retest reliability (ICC) was higher for the psychosocial summary measure than the physical (.78 and .44 respectively) while for the single scales it ranged from .14 to .78. The CHQ score correlated with the VAS rating of the parent about how good/bad they judge their child health state and it showed to be able to discriminate between children with a chronic condition such as asthma and frequent headache and children without. Children's Report of Parental Behaviour Inventory-Parent Report (CRPBI-PR) The CRPBI (Schaefer, 1965) was developed on the theoretical assumption that children's perceptions of parental behaviour directed towards them influences their social and personality development. The CRPBI-PR was created through minor changes to the stem and adjustment of the questions of the CRPBI, and the 56-item version was employed (Margolies & Weintraub, 1977). The 56-item version has six subscales (i.e., Acceptance, Control through Guilt, Nonenforcement, Lax Discipline, Child-centredness, and Instilling Persistent Anxiety), but all versions consistently load on three parenting factors: a) acceptance versus rejection (e.g., "I cheer my child up when he/she is sad"; b) psychological control versus psychological autonomy (e.g., "I say if my child really cared for me, he/she would not do things that make me worry"); and c) firm control versus lax control (e.g., "I let my child get away with a lot of things"). Mothers were asked to respond twice to each statementonce for their own behaviour, and then again for any current partner, on a 3-point scale (i.e., 2-very true, 1somewhat true, 0-not at all true). A sample of sixth graders, and their mothers and fathers, yielded satisfactory 6-month test-retest reliability (.75 for fathers and .76 for mothers) and good internal consistency (alphas ranging from .91 to .95 for mothers, fathers, and adolescents; Almeida & Galambos, 1991). Interrater agreement and concurrent validity has also been demonstrated (Schwartz et al., 1985). Higher scores reflect greater amounts of acceptance, psychological control, and firmness.