Research design: We have already coded a wide range of behaviours in the video material obtained from the 60 families of children who developed psychiatric disorders, together with 120 control videos. The full dataset was sent to the ALSPAC team, who have added the data to their own database.

1. Clinical rating scale. We first plan to investigate the cases of inattentive ADHD and pervasive development disorder (PDD) / autism in much more depth. We have been provided with the identifiers for these cases and have been given permission for expert clinicians to perform detailed assessments of the videos. This should allow us to build up a diagnostic system for testing and for potential use by clinicians after testing in a longitudinal design. Analysis of sequencing and contingency in these videos will allow us to explore whether there are differences in the behavioural dynamics of mutual responsiveness in early face-to-face mother-infant interactions in infants who later go on to be diagnosed with ADHD or PDD. We shall first ask a group of clinicians (CG, CP, PW and Dr Helen Minnis) and a developmental psychologist (HM) to view the videos of the inattentive ADHD and PDD cases (and 33 sex-matched control videos) and identify behaviours that appear predictive of disorder in order to develop a diagnostic rating scale. A second group (5 General Practitioners & 5 Health Visitors) will view the 11 case videos and 33 control videos and be blind to the case and control status of the videos. They will use their own judgments to predict which infants will later go on to develop inattentive ADHD or PDD. Then this second group will look at the videos again, this time using the rating scale developed by the group 1 clinicians. The second group of clinicians will take part in this study as part of a CPD knowledge transfer protocol (a 2 day course). The findings from this study will have a direct potential impact for routine practice.

2. Adult motor activity. In view of the finding, using our holistic ratings, that adult hypoactivity predicts childhood disruptive behaviour disorders (ADHD and oppositional/conduct disorders), we wish to collect objective evidence on adult movement in the 180 videos in the same way that we have done for infant motor activity, using frame-by-frame analysis aided by SkillSpector software. Frame-by-frame locations (x-y coordinates, in pixels) of the adult's hands, elbows, shoulders and nose will be taken from all of the 180 videos. Methods of processing this information have been developed, including measures of total motion, average velocity and acceleration. We need to assess all 180 videos to give us robust comparisons and linkable data for adult and child. The relationship between motor activity, maternal depression and parenting behaviours will be investigated.

3. Sequencing. Once we have collected data on motor activity in the adults we shall investigate the sequencing of both infant and adult motor behaviours and of the infant and adult vocalisation behaviours we have already obtained with a view to establishing if there are particular patterns of disruption of social interaction which can be characterised objectively and which are associated with psychopathology. There is a strong a priori case that abnormalities will be found in cases of autism as well as some cases of ADHD.

Sex-adjusted logistic regression models will be used to estimate the odds ratios for the subsequent development of disorders associated with maternal and infant movement and their inter-relationship. Multivariate regression models will be used to determine the degree of confounding between measures and develop a subset of measures with independent predictive value for each group of cases. The predictive ability of the resultant scores will be summarised in terms of standard measures of diagnostic utility, including receiver operator characteristic curves.