Are There Adverse Mental Health Effects of Being a Sibling of a Person with a Disability? A Propensity Score Analysis

Background and Purpose

Although many scholars in the disability field have questioned whether there is an advantage or disadvantage to having a brother or sister with a disability (Cuskelly & Gunn, 2006; Hastings, 2007; Orsmond & Seltzer, 2007; Ross & Cuskelly, 2006; Pilwksky, et al., 2004), the answer to this question remains elusive. Most sibling studies to date have been based on small, convenience samples. This study uses a national data set with propensity score analysis to examine whether there are adverse effects to being a sibling of a child with a disability.

Methods

This study is a secondary analysis of the 2006 National Health Interview Survey (NHIS) data. From the 9837 children who participated in the Sample Child file, we included only those who were siblings and did not have disabilities themselves. 3859 sample children (siblings) participated. 338 had at least one brother or sister with a disability, and 3521 had only brothers or sisters without disabilities. Three outcomes were examined: the Child Mental Health Brief (CMB), a three item mental health service use measure, and a child difficulties measure.

Data were analyzed using propensity score analysis. Propensity scores were estimated using logistic regression with sibling group (with or without brother/sister with a disability) as the dependent variable. Then, two procedures were used. In the first procedure, siblings were stratified into quintiles based on propensity score followed by a two-way (two conditions, five strata) MANOVA. In the second procedure, a one-to-one greedy matching strategy was used followed by Guo and Fraser's (2010) stratification after greedy matching technique for estimating average treatment effect (ATE).

Results

Prior to the propensity score analysis, siblings with brothers or sisters with disabilities had significantly higher CMB scores (t = -3.46), more child difficulties (t = -3.26), and used more mental health services (t = -2.80). After the propensity score analysis, most comparisons were non-significant. In the two-way MANOVA, The overall multivariate F test was non-significant, F(3, 3847) = 2.38, p = .068. The univariate F test for mental health services use was significant, F (4, 3849) = 6.66, p = .012. Siblings of children with disabilities used more services. However, the univariate F tests for CMB, F (4, 3849) = 3.09, p = .079 and child difficulties, F (4, 3849) = 2.15, p = .143 were non-significant. In the stratification after greedy matching procedure, all ATEs were non-significant (1.25 for the CMB, 1.42 for mental health services, and 1.31 for child difficulties).

Implications and Conclusions

Few sibling studies have used national datasets (Dunn, Deater-Deckard, Pickering, & Golding, 1999; Hodapp & Urbano, 2007) and we know of no other studies that use propensity score analysis. This study supports a growing body of evidence that being a sibling of a child with a disability has little to no negative effect, and some positive effects (e.g., Rossiter & Sharpe, 2001). Findings suggest that professionals should be alert to the needs of siblings of children with disabilities but they should not assume that siblings will have poorer outcomes.