Measures
Social Adjustment. Social adjustment was measured using the Withdrawal subscale of the Behavioral Assessment System for Children, Third Edition, Parent Rating Scales (BASC-3).29 The BASC-3 is a standardized parent-report measure of behavioral and emotional functioning in children and adolescents. The Withdrawal subscale is comprised of 10 items scored on a 5-point scale (1=never, 5=almost always) which are summed to provide a total subscale score. Higher scores reflect higher social withdrawal. The BASC-3 has evidence of validity for use with pediatric cancer survivors and is a reliable measure to detect differences between survivors and healthy controls.30BASC-3 scores are standardised for age. During statistical analyses, we used the standard t-scores of participants for analyses which have a mean of 50 and a standard deviation of 10. In addition, we considered the frequency of participants who scored above the clinical cut-off (>60) on the BASC-3 Social Withdrawal scale as a measure of those experiencing clinically significant social difficulties.29 Alpha was .87 for the child scale (participants age 8-11) and .84 for the adolescent scale (participants age 12-18) in the current sample.Executive Functioning. Executive functioning was measured using the total score of the Behavior Rating Inventory of Executive Functions, Second Edition (BRIEF-2) Parent Form.31The BRIEF-2 is a 59-item parent-report measure of executive function. Items are scored on a 5-point scale (1=never, 5=almost always) which are summed to provide a total score. Higher scores indicate greater deficits in executive function. BRIEF-2 scores are standardised for age. During statistical analyses, we used the standard t- scores of participants for analyses which have a mean of 50 and a standard deviation of 10. Alpha was .97 in the current sample.Demographic information. Participating parents completed a demographic questionnaire which collected the following information: child(s) sex, age, and ethnicity, annual household income, and highest level of parental and maternal education. Clinical information was collected for all survivors of ALL from their medical records including: age at diagnosis, time since diagnosis, time off treatment, and diagnostic risk level (low, standard, intermediate).Statistical AnalysesThe sample for this analysis (N =96) was comprised of 38 survivors, 20 survivor siblings, and 38 age- and sex-matched controls. Data were analyzed using SPSS 26 using a familywise alpha rate of .05. Descriptive statistics were calculated for the full sample.
For aim 1, multilevel modeling was used to evaluate group differences (survivor, survivor sibling, control) in BASC-3 Social Withdrawal. The repeated measures formulation within SPSS Mixed Models was used to handle potential systematic variation (i.e., dependence) within survivor families.32 This dependence was considered ‘nuisance’ variability to be disentangled from the fixed parameters (i.e., mean BASC-3 differences between groups), which were of substantive interest. This method models nonindependence at the level of the family as a covariance by allowing for error terms from each child within the same family to be correlated. For aim 2, a main effects regression model was tested with age at diagnosis, time off treatment, and BRIEF-2 score as predictors of BASC-3 Social Withdrawal in survivors only. For aim 3, a mediation model was tested with the significant group pairwise comparisons from aim 1 as the antecedent variable, BRIEF-2 score as the mediator, and BASC-3 Social Withdrawal as the outcome. See Fig. 1 for a description of the tested model. For aim 4 a mediation model was tested with the significant disease-related predictor(s) from aim 2 as the antecedent variable(s), BRIEF-2 score as the mediator, and BASC-3 Social Withdrawal as the outcome in survivors only. The PROCESS macro33 for SPSS was used to test the mediation models; the statistical significance of the indirect effect was tested using 5,000 bootstrapped samples.