Objectives: To study the extent to which parents are able to serve as true proxies for their children with epilepsy using a more granular approach than has been found in any study to date.
Methods: Proxy resemblance to the child was based on discrepancy in z-centered child minus parent scores of matching quality-of-life (QOL) domains for 477 dyads. Latent class mixed models (LCMMs) were built, with child's age as the independent variable for epilepsy-specific and generic QOL. Data were obtained from the QUALITÉ Canadian cohort, which recruited children with epilepsy aged 8 to 14 years at baseline and their parents.
Results: Both epilepsy-specific and generic LCMMs produced latent classes representing proxies that were overly positive, overly negative, or in agreement relative to their children with posterior probabilities of 79% to 84%. The "agreement" classes had N = 411 and N = 349 in the epilepsy-specific and generic LCMMs, respectively. The epilepsy-specific LCMM had a small unique class of N = 5 with a posterior probability of 88% called "growing discrepancy."
Conclusions: Most parents of children with epilepsy can serve as valid proxies for their children on QOL scales. Poorer parental adaptation is more related to overly negative proxies, whereas low peer support from the child's perspective is more related to overly positive proxies.
Keywords: child self-report; epilepsy; latent variable model; parent proxy; quality of life.
Copyright © 2019 ISPOR–The Professional Society for Health Economics and Outcomes Research. Published by Elsevier Inc. All rights reserved.