Identifying Latent Subgroups of Parent-Child Relationships in Youth with Type 1 Diabetes

Management of type 1 diabetes in youth places added demands on the parent-child relationship, particularly across the transition from childhood to adolescence. The purpose of this study was to identify latent subgroups of parent-child relationships in youth with type 1 diabetes and examine associations of class membership with demographic characteristics and diabetes outcomes.

Data were from the baseline assessment of youth (N=390, mean A1c 8.4±1.2, 33.8% pump regimen) aged 9 to 14.9 years and their parents enrolled in a multi-center behavioral intervention targeting diabetes management. Participants completed measures of diabetes-related parent-child conflict (parent and child report), parent involvement in diabetes management (parent and child report) and collaborative diabetes-related parenting (child report). These variables were dichotomized using a median split. To identify latent subgroups and determine associations with class membership, latent class analysis (LCA) and multinomial logistic regression were conducted using PROC LCA in SAS 9.3. Associations of class membership with age, income, diabetes management adherence, glycated hemoglobin (a biomarker of glycemic control), child quality of life (QOL), and child depression were examined. LCAs were conducted with 2, 3, and 4 classes specified.

Based on the lowest model fitting scores (AIC, BIC, ABIC) and class interpretability, a 3-class model was optimal. Class profiles were as follows: Class 1: Harmonious–low probabilities of conflict and high probabilities of parent involvement and collaborative parenting; Class 2: Indifferent–low probabilities of conflict, parent involvement, and collaborative parenting; and Class 3:  Discordant–high probabilities of conflict and parent involvement but low probability of collaborative parenting. Setting the Harmonious class as a referent group, greater income was associated with lower odds (p< 0.001) of being in Class 2 (Odds Ratio [OR]=0.88) and Class 3 (OR=0.64), while being older was associated with greater odds (p< 0.001) of being in Class 2 (OR=4.11) and Class 3 (OR=1.59). Greater adherence was associated with lower odds (p< 0.001) of being in Class 2 (OR=0.93) and Class 3 (OR=0.70); Poorer glycemic control was associated with greater odds (p< 0.001) of being in Class 2 (OR=1.68) and Class 3 (OR=3.09). Greater QOL was associated with lower odds (p< 0.001) of being in Class 3 (OR=0.84); greater depression was associated with greater odds (p< 0.001) of being in Class 2 (OR=1.28) and Class 3 (OR=1.65).

Findings suggest that parent involvement facilities positive outcomes only if low conflict and collaborative parenting are maintained.  Preventive strategies targeting youth with type 1 diabetes could be tailored to these unique family profiles.