26474
Predicting Attrition in a Longitudinal Study of Children with Autism Spectrum Disorder

Oral Presentation
Friday, May 11, 2018: 2:52 PM
Willem Burger Zaal (de Doelen ICC Rotterdam)
A. E. Richard1, I. M. Smith2, E. Duku3, M. G. Hayes4, T. Vaillancourt5, C. Waddell6, J. Volden7, P. Mirenda8, L. Zwaigenbaum7, T. Bennett9, S. Georgiades3, M. Elsabbagh10, W. J. Ungar11 and P. Szatmari12, (1)Autism Research Centre, IWK, Halifax, NS, Canada, (2)Dalhousie University / IWK Health Centre, Halifax, NS, CANADA, (3)McMaster University, Hamilton, ON, Canada, (4)Dalhousie University, Halifax, NS, Canada, (5)University of Ottawa, Ottawa, ON, Canada, (6)Simon Fraser University, Vancouver, BC, Canada, (7)University of Alberta, Edmonton, AB, Canada, (8)University of British Columbia, Vancouver, BC, Canada, (9)Offord Centre for Child Studies, McMaster University, Hamilton, ON, CANADA, (10)McGill University, Montreal, PQ, Canada, (11)University of Toronto / The Hospital for Sick Children, Toronto, ON, Canada, (12)Centre for Addiction and Mental Health, Toronto, ON, Canada
Background: Attrition is considered an inescapable reality of longitudinal studies. Nonetheless, little research has carefully examined predictors of attrition in studies of children, in which an interplay of child and parent variables may influence participation. Notably, attrition has not been systematically explored in longitudinal observational studies of children with neurodevelopmental disorders such as autism spectrum disorder (ASD). Identifying predictors of attrition in such studies may improve quality of research as well as families’ experiences of participation.

Objectives: We examined how the rate of attrition varied across assessment points in an inception cohort of preschoolers with ASD followed over a nine-year period, and investigated which child and parent characteristics predicted attrition.

Methods: Data were drawn from the multi-site Canadian Pathways in ASD study, which followed 421 children with ASD and their caregivers. This study was divided into two phases. Phase I followed participants from diagnosis at age 2 to 4 years (Mage = 38.2 months [SD = 8.7]; MIQ = 50.9 [SD = 28.7]; M:F = 356:65) until age 6 years, and Phase II, from the age of 7.5 to 11 years. Participation involved psychometric assessment of children and completion of interview and questionnaires by parents and teachers to identify child, family, and community predictors of child outcome. We used discrete-time survival analysis to identify baseline predictors of attrition.

Results:

One hundred and eighteen families (28%) ceased participation over nine years in this longitudinal study. The attrition rate was relatively stable in Phase I, with 2-6% of participants ceasing participation at each assessment point. Participant loss peaked at the start of Phase II, with 10% of participants not consenting to this second phase. Interestingly, of those who consented to Phase II, only 3% left the study over the subsequent three-year period. Competing discrete-time survival analysis models were fit and compared based on AIC and BIC. The final model included age of the primary caregiver, family income, general distress of the parent, a measure of severity of child’s restricted and repetitive behavior, and child IQ, as predictors of attrition (χ2model(18) =103.1, p <0.001). Age of the caregiver (Wald statistic(1) =8.24, p =0.004), family income (Wald statistic(1) =7.02, p =0.008), and severity of restricted and repetitive behaviour (Wald statistic(1) =7.19, p =0.007) were the most important predictors. Whereas lower family income, younger age of the parent, and lower IQ of the child increased the risk of attrition, self-reported general distress of the parent and more severe ASD symptoms were associated with a reduced risk.

Conclusions:

The rate of attrition in this study of children with ASD was commensurate with observational studies of children at risk for, or with, a medical condition, and influenced by both child and parent variables. These findings suggest that addressing parent-perceived challenges to participation and emphasising relevance of participation may improve retention rates and thereby increase the representativeness of longitudinal data. In particular, reviewing the study aims and renewing participants’ commitment may offer a compelling approach to increasing participant retention.