The overarching aim of this study is to assess the extent to which individual differences in a variety of early psychosocial measures taken in the ALSPAC cohort predict individual differences in outcome 10 to 15 years later. To achieve this aim, this study will use a case-control design, where a group of individuals with autism (cases) will be compared with a group of individuals without autism (controls). Nonautistic adolescents matched for childhood IQ, sex, birth order, and parental social class will be selected as control individuals. To begin, we will determine the developmental outcomes of case and control individuals, including current autistic symptomatology, social and communicative competence, cognitive functioning, academic attainment, and mental health status. We will then examine whether early psychosocial predictors are independently associated with multiple outcomes, and whether the same variables are predictive of outcomes (and to the same extent) for adolescents with and without autism.

Suitable families will be contacted and invited to take part in the follow-up study. Only families who have not actively declined participation previously will be eligible for participation in this study. The extent of attrition is expected to be reasonably high given the amount of time for which the longitudinal study has been running; therefore, we expect to see approximately 40-50 individuals with autism and 40-50 matched controls. Initially, we will interview case and control parents via telephone to obtain information about developmental history (Social Communication Questionnaire; Rutter et ao., 2003) and basic demographic factors. The interviewer(s) will be blind to the case/control status of the participants. We will then invite participants and their primary caregiver to come in to the University for interviews and assessments. Outcome measures will include (1) extent and nature of autistic symptomatology (as indexed by the Autism Diagnostic Interview - Revised [Lord et al., 1994] and the Autism Diagnostic Observation Schedules [Lord et al., 2000]); (2) specific cognitive skills (as measured by a battery of tasks tapping social cognition, executive function, and visuospatial ability); (3) social adaptation (as measured by the subject and informant versions of the Socio-emotional Functioning Intervew; Rutter et al., 1988); (4) IQ (as assessed by the full form of the Wechsler Adult Scales of Intelligence - Third Edition; Wechsler, 1999); (5) academic functioning (as indexed by measures of maths, reading, and phonological skills); and (6) psychiatric functioning (particularly anxiety and depression). Outcome variables will be assessed blind to predictor variables.

Initial analyses will be performed by comparing cases and controls on key outcome variables. Data will be linked with retrospective data on key psychosocial variables (deprivation, family size, temperament, attachment behaviour, parenting style, behaviour/emotions, social relationships) as well as other child variables (autistic symptoms, childhood IQ, language ability) for individual participants. Multiple regression models will be used to predict outcome, with 'caseness' (i.e., autism, nonautism) entered as an additional independent variable. For each outcome variable, we will determine (1) the amount of total variance explained, (2) the independent and separate contributions of each predictor variable, (3) and the interaction term (caseness by predictor variable). If the interaction term turns out to be significant, then this would suggest that the link between the predictor and outcome variable differs according to whether or not the individual has an autism spectrum condition.

This study should advance our knowledge of the sorts of factors predictive of individual children's developmental outcomes. This knowledge is not only critical for achieving a complete aetiological understanding of the condition, but also will highlight which factors play a role in maximise children's potential as they 'move on up' into adulthood (National Autistic Society, 2007).

A list of specific ALSPAC variables is attached.