Description of work:
Overview of Epidemiology Study of SLI

Children with Specific Language Impairment (SLI) have unexpected and unexplained difficulties learning and using spoken language. Although these difficulties are most apparent during the preschool and early school years, evidence now exists that these problems are usually present well into adulthood and are probably present throughout the person's life. Associated with difficulties in spoken language, first evident during early childhood, are impairments in reading and writing. As a result, those individuals presenting this condition are at high risk for academic difficulties. Despite life long academic, social and occupational consequences on those affected, there is little information about the prevalence of this condition or factors that place persons at risk for SLI. The few studies that do exist suffer from small sample sizes, poor sampling design and inadequate diagnostic procedures; furthermore none of these studies have been performed in the United States. The NIDCD, through a contract with The University of Iowa, has conducted an epidemiologic study of SLI in 5-year-old children based upon a sample of more than 7,256 children in the states of Iowa and Illinois. This study was designed to address the following issues:

1. Review existing literature on the epidemiology of SLI and related disorders.

2. Establish a definition of specific language disorder that is consistent with current research and clinical practice. In accordance with this definition, develop an explicit criterion for the diagnosis of specific language disorder.

3. Estimate, based upon this definition, the preva1ence of specific language impairment in five-year-old children each of three community environments in the United States: urban, suburban, and rural. Furthermore, estimate this prevalence rate in both males and females within these strata.

4. Characterize for each stratum the percentage of children with SLI who had primarily receptive primarily expressive, and receptive-expressive SLI. Also, determine the proportion of children with SLI within these strata who had concomitant phonological disorder.

5. Evaluate potential risk factors for specific language disorder.

6. Determine the history of clinical intervention for those children within each stratum.

Method
Three regions of the state of Iowa and one from Illinois, each centering upon a metropolitan area constituting an SMA (Standard Metropolitan Area) were used to obtain subjects from each of these strata.

Specifically these metropolitan areas were Waterloo/Cedar Falls/Cedar Rapids (IA), the Iowa-Illinois Quad Cities (Davenport/Bettendorf, IA, Moline, Rock Island, IL, and Des Moines, IA.). Within each of the strata and sampling region, all children who were age eligible for kindergarten during the study year (between 5 and 6 years old on September 15) were equally likely to be sampled. A stratified-cluster sampling method was used where the cluster unit was the attendance zone for a public school within one of the three community strata. The schools within each sampling region and strata were randomly selected and all kindergarten children attending the school were included in the study kindergarten children attending the school were included in the study. Using this sampling scheme over 7,000 children were sampled for participation in this study.

A two-stage assessment procedure was used to obtain the data needed to address the questions concerned with prevalence and risk for SLI. In the first stage, a brief language screening test was administered to all kindergarten children attending the schools sampled. The second stage of the assessment consisted of a diagnostic examination for SLI, administered to each child who failed. The screening test, as well as any child who had passed the screen and was randomly sampled from the same strata and community as the screening failure. The diagnostic examination consisted of a set of language measures, phonological measures, two subtests of the Wechsler Preschool and Primary Scale of Intelligence-R (Wechsler, 1989), and a pure tone hearing test. Additional information about the child's language background, history of speech and language intervention, and performance on pre-reading tasks was also obtained. In order to obtain information bearing on risk factors placing children at risk for SLI, the parents of each child who was found to present SLI, as well as 4 normal-control subjects per SLI child, were then contacted by Ames Statistical Laboratory. A 45-minute telephone questionnaire was administered to obtain background information pertaining to potential risk factors for SLI.

RESULTS
Summary & Introduction
The results in this summary will be collapsed over the residential stratification variable.

Screening Phase
Over the duration of the study, 7,844 children were eligible to participate, 626 of these children were not screened for one of three reasons: (1) they were non-English speaking or were from homes in which English was a second language, (2) they presented with unequivocal exclusionary conditions such as blindness or deafness, or (3) their parents did not wish them to participate. Of the 7,218 children who were screened, 1,933 failed the screening and 5,285 passed the screening. Of the 1,933 children who failed the screening, 1,835 were monolingual English speakers. Of the 5,285 children who passed the screening, 5,159 were monolingual English speakers. Thus, among monolingual English speakers 73.9% passed and 26.1% failed the screen.

Diagnostic Phase
Of the 3,877 children selected for the diagnostic (1,933 screening failures and 1,944 screening passes), 2,084 were given permission by their parent's to participate in the study. Of these children, 75 were reported by their parents to speak a second language. The rest of the 2,009 children constituted the final monolingual English speaker sample. The screening status for 901 of these monolingual English speakers was fail and for the rest (1,108) it was pass. However, 49 (39 of them had failed screening and the rest, 10, had passed) of these 2,009 monolingual English speakers were reported to have exclusionary conditions such as autism, head injury, mental retardation, etc. Although these children were not administered the diagnostic protocol, they were included in the denominator for calculating the prevalence of SLI. The diagnostic results of the remaining 1,960 children who were administered the diagnostic protocol showed that 1,479 had normal language status, 31 failed the hearing criteria, 215 presented language impairment but also failed the non-verbal cognitive criterion and 216 qualified as SLI by presenting a language impairment in the context of normal non-verbal IQ, normal hearing and no other exclusionary conditions.

Prevalence of SLI
The data above provided the raw frequencies necessary to compute the prevalence of SLI. Due to the nature of the study design in which all screening failures, and a partial sample of screening passes were diagnosed, the sample of children who received the diagnostic protocol were not a representative sample of all the monolingual English speaking children sampled and screened. Therefore, the prevalence could not be determined by simply dividing the number of SLI children by the number diagnosed. Instead, it was necessary to determine the prevalence of SLI in the screening failure group and the screen pass group separately and then to combine these based upon the differential rate of screening pass and failure. Among the screening pass children, 2.8% (31 of 108) were found to be SLI, whereas 20.5 % (185 of 901) of the screening failures were found to be SLI. The prevalence rate of SLI in the total screened population was then obtained by weighting 2.8% by the proportion of children passing the screen (.738) and likewise weighting the 23.4% by the proportion of children failing the screen (.262). The prevalence rate of SLI in the total population was thus estimated to be 7.42%.

Risk Factors
A 180 item questionnaire was administered to 228 parents of children with SLI and the parents of 860 children with normal hearing, intellectual, and language status. This questionnaire provided information on: parental background; maternal prenatal and perinatal health, parental exposure to tobacco, alcohol, and other drugs; family history of speech, language and learning problems; and several features of the child rearing environment. In addition, there were several items that provided extended insight into certain behavioral co-morbidities of SLI.
The principal risk factors for SLI were concentrated the background of the parents. In comparison with the parents of the control children, the parents of children with SLI, especially those in the urban setting, were less educated, younger, and had greater rates of learning problems, especially learning disorders in the mothers and speech and learning disorder in the fathers. The parents of the children with SLI were also more likely to smoke tobacco during and after the study child's birth than the parents of the control children. Finally, children who were read to infrequently, and/or had little opportunity to verbally share experiences or feelings were also at greater risk for SLI. When these risk factors were analyzed by strata, it was found that these risk factors were more often associated with SLI in the urban setting.