Academic underachievement among children with epilepsy: proportion exceeding psychometric criteria for learning disability and associated risk factors.(Report)

One percent of children will develop epilepsy by 20 years of age (Hauser, 1994). Children with epilepsy are at greater risk for academic difficulties compared to healthy children and compared to those with many other chronic illnesses of childhood (Austin, Huberty, Huster, & Dunn, 1998; Fowler, Johnson, & Atkinson, 1985; Westbrook, Silver, Coupey, & Shinnar, 1991). Furthermore, children with epilepsy are less successful than their peers without epilepsy at obtaining gainful employment as adults (Dodrill & Clemmons, 1984; Sillanpaa, Jalava, Kaleva, & Shinnar, 1998). Although many studies have documented that children with epilepsy tend to have difficulties in school, it is unclear how many of these children would meet criteria for learning disabilities (LD) or for obtaining services.

LD Screening

With the implementation of the Education for All Handicapped Children Act of 1975, the U.S. Office of Education provided an operational definition of LD that required a "severe discrepancy between achievement and intellectual ability in one or more ... areas" of achievement (Assistance to States for Education of Handicapped Children: Procedures for Evaluating Specific Learning Disabilities, 1977). The stipulation of a severe discrepancy between achievement and IQ was retained in the Individuals with Disabilities Education Act (IDEA) amendments of 1997. The majority of state definitions have included a discrepancy criterion for many years (Frankenberger & Fronzaglio, 1991; Mercer, Jordan, Allsopp, & Mercer, 1996; Mercer, King-Sears, & Mercer, 1990), and 96% of states were continuing to use IQ-achievement discrepancy in their classification criteria as recently as 2002 (Reschly & Hosp, 2004). More than half of the states reported using standard scores, and 90% of those specified a discrepancy cutoff of 1 SD or greater (Frankenberger & Fronzaglio, 1991). Criticism of the role and utility of IQ in defining LD has led some to advocate for defining LD using a "low-achievement" criterion (i.e., comparison to the mean for achievement tests, without regard to IQ). These definitions have been compared and debated (e.g., Epps, Ysseldyke, & Algozzine, 1983; Kavale, 1995), but both definitions persist in the literature.

Both the IQ-achievement discrepancy and low-achievement approaches for individual classification have come under scrutiny. Fletcher and his colleagues have challenged the assumption that comparisons to IQ are valid indicators of underachievement and have highlighted the statistical limitations of psychometric cutoffs for individual classification (e.g., Fletcher, Coulter, Reschly, & Vaughn, 2004; Francis, Fletcher, Stuebing, Lyon, Shaywitz, & Shaywitz, 2005). Instead, they have advocated for serial testing (i.e., documenting change over time) or response to intervention (RTI) as being more empirically defensible for LD identification. These alternatives are unencumbered by the statistical limitations attributed to the use of IQ in defining LD, and RTI carries the additional advantage of promoting earlier intervention. Whereas Fletcher et al. (2004) criticized the use of IQ in LD classification and extensive assessment to identify LD, they nonetheless acknowledged a role for focused achievement testing even with an RTI approach, stating that "screening and evaluation of academic skills in the service of intervention and to determine level of risk for LD is not time consuming" (p. 309). Therefore, low achievement on a focused battery of academic measures can help identify children who are at risk and who need intervention.

Others continue to advocate for IQ-achievement discrepancy definitions for screening purposes. In a comprehensive review and analysis of the issues and empirical data addressing the use of discrepancy models in LD, Kavale (2002) argued that an IQ-achievement discrepancy is useful for screening for LD and can help to document underachievement, which is a necessary (although not sufficient) element in the definition of LD. Therefore, an initial assessment comparing actual achievement to the child's potential (i.e., IQ) is a defensible approach to screening for LD, especially for group-level research. For screening purposes, a liberal discrepancy (e.g., 1 SD) would be appropriate, especially given error of measurement issues (e.g., Francis et al., 2005).

With the reauthorization of IDEA in 2004 (Individuals with Disabilities Education Improvement Act of 2004), local educational agencies are no longer required to consider discrepancy from IQ and are allowed to use empirically validated processes of RTI as alternatives. This change provides more latitude to local agencies. It remains unclear, however, whether states will change their criteria, how quickly they will change, and which definition they will adopt.

Limitations in Past Research on Children With Epilepsy