The ARCHER study is a multidisciplinary, three year, longitudinal cohort study, using a convenience sample. The research hypothesis of the study is that onset and/or tempo of the rise in puberty hormones play a significant role in adolescent health and wellbeing. The ARCHER study commenced recruitment in June 2011 and has enrolled 200 adolescent participants (with parent/guardian) as of March 2012.
A feasibility study for all aspects of baseline data collection was completed in 28 students in years 5&6 at a rural primary school. Sample collection, labeling, transport and storage were successful. The plasma testosterone and oestradiol levels demonstrated that the participants (except 2 -1 M, 1 F) had levels consistent with pre- or very early puberty. On-line questionnaires were revised and successfully re-piloted in August 2009 in a rural primary school with 26 students and their parents. In 2010 focus group discussions (single and mixed gender with 10–12 year olds and 13–15 year olds) were held with 58 participants. Topics included what would positively influence recruitment and retention, how peers might participate in the study and the collection of biological samples. The young people expressed strong preferences in relation to the collection process and the focus group outcomes are now published
. The methodology was further tested in 2010 in a funded pilot study on puberty, sleep and depression, with included the use of the Actiwatch. This is a digitally integrated recording of wrist activity which provides more detailed information on sleep wake/times
Setting and study sample
The study is being conducted in Dubbo and Orange, two large towns in regional Australia in the State of New South Wales and surrounding areas. Enrollment is in school Grades 5, 6 and 7 (10–12 years of age) to ensure that both the transition into early puberty and the end of puberty are captured in adequate numbers. Each student will require a parent or guardian to also participate in the study. Lack of parental competence in English or adolescent intellectual disability are exclusion criteria. The least common primary outcome that we measure is depression, which we estimate will affect 5% of children in mid- to late-puberty
. We estimate that 160 boys and 160 girls at Tanner Stage 3 (the pubertal stage where clinical onset of depression is most likely) are required to detect a one standard deviation (SD) difference in testosterone nmol/l (SD = 5)
 or oestradiol pmol/l (SD = 250)
 levels between those with and without depression. The anticipated loss to follow up is 5% per annum in this setting. Assuming a mean age of puberty onset of 11.4 years
, 400 subjects will ensure adequate numbers of adolescents at Tanner Stage 3 in the study, as a few younger subjects may not reach this Stage at the end of the study (age 14 years) and some older children may have already entered Stage 3 at the beginning of the study (age 12 years).
Recruitment procedures and follow up rates
Recruitment is primarily through schools, as well as through the media (targeted at parents) and through established links with local community groups. We have permission from the state Department of Education and Communities - Western Region and the Catholic Education Office – Bathurst Diocese to recruit through their schools. Data collection does not take place in the schools. Information sessions are held in the community and at consenting schools. Parents are asked to provide two additional adult contact names with the written consent, to reduce loss to follow up. Participants who agree are sent birthday cards, updates and study newsletters as reminders by their preferred mode of contact (email, letter or Short Message Service (SMS)). These practices have been shown to increase retention rates in school age children
. The young people requested ‘get togethers’ in the focus groups to celebrate their involvement in the study and these will be held annually.
Adolescents will complete questionnaires at baseline and annually for the following three years. We will use a computerized on-line questionnaire with branched algorithmic structure, to enhance confidentiality and accuracy of responses and to reduce exposure to sensitive questions (particularly related to sexual activity). This has been created and trialed. The total number of questions for adolescents is 256. These are completed in approximately one hour (shorter for the older participants and closer to 75 minutes for younger participants). Adolescents complete the questionnaire under supervision of research staff. Questionnaires will not be performed in the fasted state and, as with all other data and biological sample collection, are done outside school hours. Physical examination occurs at baseline and annually for three years. Blood is collected at baseline and annually for three years, in the fasting state. A first morning urine sample is collected at baseline and every three months for three years. A single indicator question providing information on mood fluctuation is sent to all participants with access to mobiles phones every 3 months and functions as a reminder to complete urine collection.
A parent or guardian completes a questionnaire at baseline and annually for three years, and in the same month as their child. The questionnaires are available online or as hard copy and contain approximately 180 questions. This questionnaire reduces reporter burden on adolescents for demographic data, as well as providing other relevant family and environmental data.
Ethical approval has been granted for the pilot questionnaires, anthropometry, biological sample collection (HREC 10612), the pilot SMS and sleep studies (HREC 12502) and for the full study described in this protocol (HREC 13094) from the Human Research Ethics Committee, University of Sydney.
Questionnaires - adolescent
The Child Behaviour Check List (CBCL) in the version validated for ages 11–18 years as the Youth Self Report (YSR)
 is the main questionnaire measurement instrument. It contains 20 social competence items that measure participation in hobbies, games, sports, jobs, chores, friendship, and activities and 8 sub-scales, measuring internalizing and externalizing behaviour. In addition the subscales provide data on social competence, learning and engagement in education, sleep, unintentional injury, drug and alcohol use, conduct disorders, disruptive behaviour disorders, depression and anxiety, and self harm. The YSR also identifies the presence of physical illness and disability.
The YSR is supplemented by the following questionnaires.
Selected measures of self competency taken from the Raine cohort
, including Cowen’s Self Efficacy
 and Adolescent Self Perception Profile
12 selected items from Motivation and Engagement; Enjoyment of School, Academic Buoyancy and Class Participation scales
[48, 49]. These results can be linked to the Australian National Assessment Program Literacy and Numeracy scores
A questionnaire is used to document sleep/wake patterns, and sleep disorders in the adolescent’s natural sleeping environment
. In addition, a convenience subset of adolescents will use the Actiwatch home monitor system annually.
Direct questions on unintentional injury type in the preceding 12 months provided by RI, selected questions from The Australian School Students Drug & Alcohol Survey
, branched age appropriate sexuality questions on romantic relationships, sexual feelings, history of STI and pregnancy
 and the 40 item validated Australian self-reported delinquency scale by Mak
 in those adolescents who score highly on the YSR subscale for externalizing problems.
The Short Moods and Feeling Questionnaire (13 item) in all subjects
. The short form has been validated as a self-reported unidimensional measure of symptom severity of childhood depression in community samples.
The 16 item Deliberate Self Harm inventory
 for those who respond with a 1 or 2 for YSR item 18 or 91 (relating to self-harm or suicide).
Selected questions from Health Behaviors in School-aged Children for physical inactivity, nutrition and tobacco
One SMS question which is a linear analogue self-assessment scale question on mood providing information on mood fluctuation
Tanner Stage of puberty by self report using standardized line drawings which allow determination of accuracy of this method of pubertal staging, which is the main method used in epidemiological and non-clinical studies.
Questionnaires – parent/guardian
Demographic data for family and adolescent.
Child Behavior Checklist (CBCL) validated questionnaire for parents of children aged 6–18 years to corroborate the adolescent’s YSR
The validated Macmaster’s Family Assessment Device
 to obtain information on family and local environmental factors, which are potential confounders.
After the final 3-year questionnaire all adolescents will be asked whether they believed that the speed of their puberty was faster, the same as or slower than their age peers and whether the onset of their puberty was earlier, the same as or later compared to their age peers. Their parent/guardian will be asked the same questions in relation to their child.
Physical examination – adolescents only
Weight and body composition is measured in light clothing using Tanita TBF-300 Pro Body Composition Analyzer
. Height, and foot length as an indicator of growth stage
 are measured on a portable stadiometer (to 0.1 cm) and with a metal ruler (to 0.5 cm) respectively, and waist
 with a tape (to 0.1 cm) using standard techniques. Body Mass Index, as a measure of overweight (kg/m2) and waist circumference are expressed as individual z-scores based upon age and sex related reference charts
. Blood pressure (BP) and pulse rate are measured using an automated BP monitor, under standardised conditions.
Laboratory – adolescents only
Fasting bloods are collected using local anaesthetic cream and separated serum and spot urine samples frozen at −80°C, as well as buffy coat in one sample. Total blood draw is approximately 30 ml. Menstrual cycle stage will be recorded in post-menarchal females, with the perimenstrual week avoided.
Testosterone and oestradiol will be measured using AP5000 LC Tandem MS in the Andrology laboratory, ANZAC Research Institute
 (DHEA, DHT and oestrone will also be measured). There will be adequate blood collected to measure other potential biological variables of interest, which will form the basis of future studies but which are not integral to this study. These variables include LH, FSH, SHBG and IGF-1, inhibin, anti-Mullerian hormone, ACTH, cortisol, growth hormone, prolactin, oxytocin, TSH, thyroxine, triiodothyronine and Vitamin D (all of which are hormones relevant to puberty)
, full blood count, ferritin, glucose, liver function, urea, electrolytes, full lipid profile, creatinine, and insulin, fatty acid profile, leptin, adiponectin, resistin, TNF alpha, CRP and Interleukin 6 (as indicators of cardiovascular risk and also to derive eGFR)
. There is the potential to be able to study genes relevant to puberty and sex steroid action; KiSS-1, KAL1, FGFR1, GnRHR, GRPR-54, TAC3, neurokininB, androgen and oestrogen receptor, steroidogenic enzymes such as CYP, StAR, P450scc, P450 (17alpha) and 17beta-HSD
, and genes relevant to body composition and insulin resistance, such as leptin gene and product, MC4R, FTO, IL-6, and UCP-2
A first morning, fasting urine sample will be collected at home. Urinary testosterone and oestradiol will be measured, also using AP5000 LC Tandem MS. Urinary hormones, although less well standardized than blood, are essential to define hormone trajectories which can never be accomplished by a single annual sample. Sperm (indicating spermarche) will be measured in male urine samples
. There will be adequate urine to measure creatinine and microalbumin, but these measures are not integral to the current study.
Residential address will be geocoded
 to obtain additional information such as neighbourhood socioeconomic status, access to services and amenities and other environmental exposures, which may be study confounders for which adjustment will be required.
The main independent variables are tempo of puberty and age of onset at puberty. Both will be measured by testosterone and oestradiol, in blood and urine. Data analysis will have three main components.
The first component is testing the reliability and validity of measures under focus. These analyses will centre on: (a) descriptive, reliability and item functioning; (b) factor analysis to test factor structure and validity of measures; (c) ANOVAs for preliminary tests of main and interaction effects of background socio-demographics; and (d) tests of (in)variance in factor structure to ensure congruence in measurement properties across sub-groups to justify pooling sample data for subsequent modeling throughout the project
The second component is a correlational one and assesses the impact of tempo and onset of puberty over time. Specifically, for example, by assessing in the one analytic model (eg. in linear regression models for continuous variables such as YSR scores for sensation seeking and in logistic regression models for dichotomous outcomes such as depression or sexual debut) the effects of Time 1 onset and tempo of puberty on Time 2 outcomes after controlling for Time 1 outcomes, it is possible to get a sense of the relative salience of the onset, and tempo of puberty in one time period over factors in a following time period
. Also in these longitudinal analyses, key socio-demographic (and other) factors such as gender, school grade and socioeconomic factors can be included to get a further sense of puberty by controlling for other potential influences. The three monthly urine samples for the biological markers of puberty can be used to regress time on time measures, with tempo being defined by the residuals. Thus the larger and the more positive a residual over time, the more rapid the tempo of change.
The third component is based on mean-level differences (e.g. repeated measures) and explores for significant upward and downward shifts in means on target outcomes as a function of main effects (e.g. early and late onset; rapid and slow tempo) and interactions (e.g. early onset/rapid tempo; late onset/rapid tempo; early onset/slow tempo and late onset/slow tempo) of puberty effects that also control for key socio-demographic (and other) factors as covariates
. Taken together, these analyses (and adaptations of them) allow for integrative tests of the substantive and methodological issues at hand.
Analyses will be conducted using SAS, Stata and WinBUGS (for repeated measures analyses).