Speech and Language Disorders in Children: Implications for the Social Security Administration's Supplemental Security Income Program (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Comparison of Trends in Childhood Speech and Language Disorders in the General Population and the Supplemental Security Income Program Population One of the key aims of this report is to “identify trends in the preva- lence of speech and language disorders in children and assess factors that influence these trends” among both the general population and children receiving Supplemental Security Income (SSI) on the basis of speech and language disorders (see the committee’s statement of task in Box 1-1 in Chapter 1). To achieve this aim, this chapter builds on the discussion of speech and language disorders in children and their prevalence in Chapter 2, the persistence of these disorders in Chapter 3, and the structure of the SSI program for children with these disorders in Chapter 4. To develop an overview of trends in the prevalence of speech and language disorders, the committee reviewed available research and data- sets that include information on these disorders in children over the past decade. As is described in detail later in this chapter, available datasets have significant limitations, resulting in some limits on the extent to which trends in the prevalence of these disorders can be described with precision. Nevertheless, available research and datasets make it possible to draw evidence-based conclusions related to some basic issues concerning these trends—in particular, those issues that may directly impact the nature and size of the population of children receiving SSI benefits based on having these disorders. It was also possible to draw conclusions about gaps in knowledge that need to be addressed to provide more precise information about the prevalence of speech and language disorders among both the general population of children and those receiving SSI benefits on the basis of having these disorders. This chapter begins by reviewing the data sources used by the committee 159 

160 SPEECH AND LANGUAGE DISORDERS IN CHILDREN to gather information on trends in the prevalence of childhood speech and language disorders. It then presents this information (1) for the general population, (2) according to demographic and other characteristics, (3) for the SSI population, and (4) for Medicaid beneficiaries. Next is a discussion of the strengths and limitations of the data available for examining these prevalence trends. The final section contains the committee’s findings and conclusions. DATA SOURCES The committee used three different types of data to address the aim of this chapter: clinical data from small treatment studies, population data from large national surveys, and administrative or service data from large federal programs. Detailed information on the strengths and limitations of these data sources is provided later in the chapter. Clinical data come from published research studies conducted with clinical or treatment populations. They provide detailed information about each study subject; depending on the study, this may include information on diagnosis, the course of the disorder over time, and/or treatment. Clinical studies typically share the characteristic of the child’s speech or language disorder having been diagnosed by a professional (as opposed to reported by a parent or a guardian). Population data from large national surveys provide less detailed in- formation on each participant relative to clinical studies. However, infor- mation from national surveys is more representative of the populations of children from which respondents to the survey are drawn (e.g., all children living in the United States). Population data from national surveys typically are based on parent or guardian identification of speech and language dis- orders in children, and clinical corroboration of the condition or treatment received by the child is rare. Finally, administrative or service data describe the population of children who receive services through large federal programs such as the Individuals with Disabilities Education Act (IDEA) (special education ser- vices) or Medicaid (health insurance). By definition, information derived from administrative data is representative only of the children who qualify to receive services under these programs and therefore does not reflect the characteristics of children in the general U.S. population. For purposes of this chapter, trends are defined as changes in preva- lence over time, with prevalence being the proportion of a population with a speech or language disorder. To describe trends in speech and language disorders among children in the general U.S. population, the committee considered a number of national datasets, reviewed later in this section and described in greater detail in Appendix B. In addition, during its two 

COMPARISON OF TRENDS 161 public workshops, the committee solicited expert testimony that provided informed perspectives on the availability, content, and value of these na- tional data. The committee determined that no single dataset or small set of data sources can provide definitive population estimates of the prevalence of speech and language disorders among U.S. children. Where data that can be used for this purpose do exist, their utility is limited by the problems discussed later in this chapter. To examine trends in the number of speech and language disorders among children receiving SSI benefits, the committee used data from the Social Security Administration (SSA). Except where noted otherwise, these data were extracted for the years 2004 through 2014 from the SSA 831 file. Data related to poverty level and population size were retrieved from the U.S. Census Bureau for the years 2004 to 2013. Suspension and termi- nation data were retrieved from SSI annual statistical reports for the years 2004 to 2013. Trend data to which the committee had access, however, are best described as serial cross-sections because they do not follow individual children over time. Instead, these serial cross-sections provide snapshots over time. The committee pieced these snapshots together to provide rough estimates of changes in prevalence among populations of beneficiaries (in the case of administrative data) or respondents (in the case of population survey data) over successive years. Because children observed in 1 year may not be the same as those observed in another year, the composition of these populations changed over time. Consequently, these data cannot be used to describe the natural history of speech and language disorders over time. (This topic—the persistence of speech and language disorders following treatment—is covered in Chapter 3.) Not only may changes in population composition over time affect rates of observed prevalence, but changes in the SSA program procedures, changes in eligibility and determination guidelines (as described in Chapter 4), and economic conditions also may influence changes in the rate of identification of speech and language dis- orders within a given population of children at a given point in time. For these reasons, precise conclusions about trends cannot be drawn. The same caveats apply to changes in the gender distribution of children identified as having speech and language disorders. The committee derived information on comorbidities (i.e., other co- occurring disorders) among children with speech and language disorders from an analysis of Medicaid claims data from 20 states, performed by a team at Rutgers University under contract to the Institute of Medicine. These data provide a snapshot of comorbid disorders occurring among a subset of child Medicaid recipients with speech and language disorders who received treatment in the preceding year. The analysis examined rates of speech and language disorders among children who were Medicaid 

162 SPEECH AND LANGUAGE DISORDERS IN CHILDREN enrollees. It also yielded information on the subset of children who were eligible for Medicaid because of their participation in the SSI disability pro- gram. Medicaid claims are generated as a result of a treatment encounter between a child and his or her health care provider. To be captured in this Medicaid dataset, the child must have been eligible for Medicaid and to have obtained some sort of treatment or health care service. Furthermore, a child in the Medicaid dataset would be identified as having a speech or language disorder only if a diagnosis of such a disorder were listed on an insurance claim during the child’s Medicaid-covered encounter. Not all chil- dren with speech and language disorders are Medicaid beneficiaries; not all children with these disorders seek or obtain treatment; and not all children with these disorders will have their condition correctly coded during their health care visit. Just as the Medicaid population does not include all children with speech and language disorders, the SSI population with childhood disabili- ties does not include all children with speech and language disorders or even all children with disabilities. The eligibility and determination process required to qualify for the SSI program produces a beneficiary population that is very different from the general population of children with speech and language disorders in the United States (see Chapter 4). As discussed in Chapter 4, children with these disorders are eligible for SSI benefits only if their disorder is “marked” or “extreme” and they come from a family with low income. This profile reflects the intent of the SSI program—to support the most vulnerable and severely impaired children. For this reason, trends in the prevalence of speech and language disorders within the SSI popula- tion and the general U.S. population are not comparable. As noted earlier, the committee relied upon large national surveys and administrative or service data sources to identify trends in prevalence of childhood speech and language disorders. Generally, any nationally rep- resentative survey was considered if it used questions or other methods to ascertain whether a child had such a disorder. For example, data from the National Survey of Children with Special Health Care Needs and the National Survey of Children’s Health are reported here because both sur- veys include questions about speech and language disorders. On the other hand, some national surveys were excluded because they lack such ques- tions. Two prominent examples are the Medical Expenditure Panel Survey and the National Health and Nutrition Examination Survey. While these surveys collect extensive information on health and do include questions related to disabilities and chronic illnesses, neither asks any questions about speech and language disorders in children. Notably, the data sources the committee used to estimate trends in prevalence are heterogeneous in their approach to ascertaining speech and language disorders and in the populations they include. For example, the 

COMPARISON OF TRENDS 163 Survey of Income and Program Participation includes two salient ques- tions that can be used to identify children with these disorders: the first asks whether the child has any difficulty having speech understood, and the second asks whether “speech problems” were identified as the condi- tion that caused the child to have difficulty with activities of daily living. The National Survey of Children’s Health, by contrast, posed the question: “Has a doctor or other health care provider ever told you that [child’s name] had stuttering, stammering, or other speech problems?” (DRC, 2012b). Clearly, these are different ways of ascertaining whether a child has a speech disorder. When surveys rely solely on parental reports of a condi- tion, the responses may be inaccurate. Conversely, surveys that ask about a doctor’s opinion may underestimate the prevalence of speech and language disorders among children who do not come to the attention of a medical provider. All national surveys that do not include clinical corroboration (i.e., direct assessment of the child as an integral part of the study) present this potential problem. With respect to administrative or service data, such as Medicaid claims data, these data do not provide an accurate estimate of the prevalence of a given condition. Instead, these data can be used to understand the number and characteristics of children who receive services from a specific program (e.g., special education, Title V, SSI). Medicaid claims data, for example, typically are analyzed to understand utilization and costs of health care services among Medicaid participants. Table 5-1 provides an overview of the surveys and sources of admin- istrative/service data the committee used. The variety of approaches to determining speech and language disorders is evident in this table. For surveys, the second column of the table lists the question(s) used by each to identify these disorders. The table also shows that none of the survey data- sets can be used to determine the severity of the child’s condition, and very limited information is available for two of the datasets on what treatment an individual child received. Furthermore, as discussed earlier, individual children cannot be studied over time in any of these datasets. While the Survey of Income and Program Participation is longitudinal, and partici- pants can be studied over time, disability questions are asked only once. It is also important to emphasize that, with few exceptions (e.g., the National Survey of Children with Special Health Care Needs and the National Health Interview Survey—Voice, Speech, and Language Supplement), the surveys focus on speech problems and do not ask explicitly about language disorders, such as comprehension difficulties. This omission could result in an underestimate of the number of respondents with such disorders. Note that in the following discussion of data from these sources, use of the terms “speech,” “language,” and “speech and language” reflects the wording of the questions included in each survey. 

TABLE 5-1  Overview of Data Sources with Information on Children with Speech and Language Disorders 164 Is Condition Can Children Is Information Severity Be Studied Available on Population Data Indicator of Speech and Language Disorders Measured? Over Time? Treatment? National Health Interview During the past 12 months, has [child] had any of No No No Survey the following conditions? (Response list includes stuttering and stammering.) -AND- How long [has child] had a speech problem? National Health Interview Numerous survey questions related to “speech Yes No Yes Survey—Voice, Speech, problems,”including: and Language Supplement Did a health or education professional EVER tell you a diagnosis or reason for [child’s name] voice, swallowing, speech, or language problem? -AND- At what age did [child] FIRST begin to have any speech problems? National Survey of Compared to other children the same age, would No No Yes Children with Special you say [he/she] experiences a lot, a little, or Health Care Needs no difficulty speaking, communicating, or being understood? National Survey of Has a doctor or other health care provider No No No Children’s Health ever told you that [child] had . . . stuttering, stammering, or other speech problems? Survey of Income and Any difficulty having speech understood? No No (disability No Program Participation -AND- questions are asked “Speech problems” was identified as condition only in one topical causing child to have difficulty with activities of module) daily living. 

Is Condition Can Children Is Information Administrative/Service Severity Be Studied Available on Data Indicator of Speech and Language Disorder Measured? Over Time? Treatment? Individuals with Percentage of children reported by local education No No No Disabilities Education Act agencies as having a primary disability of “speech (IDEA) child count data or language impairment” Medicaid claims data ICD-9* diagnostic codes for developmental speech No severity codes Yes Yes disorders (315.3) or deafness (389), expressive for developmental language disorder (315.31), mixed receptive- speech disorders or expressive language disorder (315.32), speech deafness and language developmental delay due to hearing loss (315.34), childhood onset fluency disorder (315.35), other developmental speech or language disorder (315.39), speech and language deficit unspecified (438.10), other speech and language deficits (438.19), and other speech disturbance (784.59); speech and language disorders also must be coded based on their underlying cause, if any SSA data Impairment codes for “speech and language Yes No No impairment” (3153) or “loss of voice- communication impairment” (7840) on SSA Form 831 * ICD-9 refers to the International Classification of Diseases, Ninth Revision, Clinical Modification. SOURCES: CDC, 2015; DRC, 2012b,c. 165 

166 SPEECH AND LANGUAGE DISORDERS IN CHILDREN With respect to administrative/service datasets, Table 5-1 shows the indicator of speech and language disorders provided by each. As with the survey datasets, the table also shows whether the severity of the child’s con- dition and any treatment received can be determined, and whether children with these disorders can be studied over time. National Health Interview Survey The National Health Interview Survey is a cross-sectional survey based on personal household interviews. It serves as the principal source of in- formation on the health of the civilian noninstitutionalized population of the United States and is administered by the National Center for Health Statistics (CDC, 2012a). Its purposes include monitoring trends in illness and disability and tracking progress toward the achievement of national health objectives, and it enables analysis of these data by many demo- graphic and socioeconomic characteristics (CDC, 2012a). The National Health Interview Survey is an ongoing survey that has been conducted continuously since 1957, with substantial content revision every 10-15 years (CDC, 2012a). Apart from major revisions, its ques- tions remain largely unchanged from year to year, allowing trend analyses over multiple years (CDC, 2012a). The survey, with a sample size of ap- proximately 35,000 households containing about 87,500 persons, collects limited information on all individuals in a household (CDC, 2012a). For each household, one sample adult and one sample child (if present) are randomly selected, and additional information on health status is collected in the Sample Child Core and Sample Adult Core questionnaires (CDC, 2012a). This additional information encompasses numerous disabling and health conditions, health care services, and health behaviors (CDC, 2012a). The value of the National Health Interview Survey in the context of this study is that the sample is representative of all U.S. civilian households, and the data collected facilitate examining relationships among many health variables. A key shortcoming of this survey for purposes of this study is its limited ability to identify children with speech and language disorders. The salient questions ask about whether the child “had any stuttering or stam- mering” during the past 12 months or has a “speech problem [that] causes limitation” and if so, whether the speech problem was chronic (CDC, 2015). In addition, the data are reported by an adult living in the house- hold who is knowledgeable about the child, but include no corroborating information from clinical assessment. Because the sample varies from year to year, longitudinal monitoring is not possible, but trend data on stutter- ing/stammering or speech problems can be derived. 

COMPARISON OF TRENDS 167 National Health Interview Survey—Voice, Speech, and Language Supplement To gather additional information on speech and language impair- ments, the National Center for Health Statistics fielded a supplement to the National Health Interview Survey in 2012. The National Health Interview Survey—Voice, Speech, and Language Supplement is a one-time survey that was conducted using data collection methods similar to those used for the National Health Interview Survey to identify children aged 3-17 years with reported problems with voice, swallowing, speech, or language dur- ing the past 12 months (Black et al., 2015). The sample size was 13,275 children, 7.7 percent of whom had such problems according to parent or guardian reports—5.0 percent speech problems and 3.3 percent language problems1 (Black et al., 2015). Among younger children (aged 3-10) with voice, speech, language, or swallowing problems, one-third had multiple such problems, compared with about one-quarter among older children (aged 11-17) (Black et al., 2015). The Voice, Speech, and Language Supplement provides information on prevalence, causes, age of onset, problem severity, receipt of intervention services, and other (comorbid) conditions (Hoffman, 2015). Its greatest shortcoming for present purposes is its one-time occurrence; the relatively small sample size, which limits disaggregation in analyses; the inclusion of all levels of severity, from mild to severe; and the fact that all of the data are based on parent or guardian reports, with no direct clinical assessment or corroboration to confirm the presence of a speech or language disorder. National Survey of Children with Special Health Care Needs The National Survey of Children with Special Health Care Needs was a national cross-sectional telephone survey administered by the National Center for Health Statistics. Its purpose was to gather information on the health and functional status of children with special health care needs, in- cluding their physical, emotional, and behavioral health; their health care access; their service utilization and its quality; and the impact of chronic conditions on their family (DRC, 2012c). This survey was fielded three times, with data being collected on children with speech and language disorders in the 2005-2006 and 2009-2010 iterations; no questions about etiology or diagnoses were asked in the 2001 survey. The survey used a five-question special health care needs screener, de- veloped by the Maternal and Child Health Bureau, to identify a minimum 1  Some of the survey respondents reported that children had both speech and language disorders. 

168 SPEECH AND LANGUAGE DISORDERS IN CHILDREN of 750 children with special health care needs in each state (CDC, 2014a). The screener asked about (1) use of medication prescribed by a doctor; (2) more use of medical care, mental health, or educational services than is usual for children of the same age; (3) being limited or prevented in any way from doing things most children of the same age can do; (4) receipt of special therapy, such as physical, occupational, or speech therapy; and (5) any kind of emotional, developmental, or behavioral problem for which the child needed or received treatment or counseling (Blumberg, 2015). To qualify as a child with a special health care need, at least one of these questions had to be answered affirmatively because of a medical or other health condition that had or was expected to have a duration of at least 12 months (Blumberg, 2015). The data from this survey allow for analyses at the national, state, and regional levels. The 2005-2006 and 2009-2010 surveys each screened more than 370,000 households to identify more than 40,000 children with special health care needs (CDC, 2014a). The value of this survey for pres- ent purposes is that it provides information on a much larger sample of children with special health care needs relative to other surveys. The data are representative of noninstitutionalized children with special health care needs in each state. At the same time, however, these data have several shortcomings in the context of this study. First, the screener questions did not include a question about speech or language disorders; speech and language questions were asked only for children screened into the survey by the screener questions. As a result, data from this survey do not allow for a direct assessment of prevalence, although prevalence can be estimated. A study commissioned by the SSA in 2001-2002 found that about 10 percent of child SSI recipients had been missed by the special needs screener, and of those who had been missed, 53 percent had intellectual disability; 8 percent had speech distur- bances; 4 percent had other specific learning difficulties; 4 percent had at- tention deficit hyperactivity disorder/attention deficit disorder; and the rest had a wide range of physical, cognitive, and behavioral conditions (Ireys et al., 2004). Second, like all the other surveys described here, this survey provides no clinical corroboration of a child’s condition; all information is based on parent or guardian reports. Third, the survey did not collect detailed information about speech therapy or other service utilization, the etiology of the speech or language impairment, the child’s age at the time of diagnosis, receipt of social welfare benefits, or measures of poverty or family well-being. Fourth, all levels of severity, from mild to severe, were included and cannot be disaggregated. Finally, because the survey was cross-sectional, no information is available about individual children’s con- ditions over time. 

COMPARISON OF TRENDS 169 National Survey of Children’s Health The National Survey of Children’s Health was a cross-sectional tele- phone survey administered by the National Center for Health Statistics. Its purpose was to gather information on multiple, intersecting aspects of children’s lives, including physical and mental health status and function- ing; access to care; and the child’s family, neighborhood, and social context (DRC, 2012a). The survey, fielded in 2007 and 2011, collected data on children with and without disabilities or other special health care needs. Both the 2007 and 2011 surveys included more than 90,000 children who were representative of all noninstitutionalized U.S. children (CDC, 2013a). This survey has key shortcomings in the context of this study. First, it collected no information about receipt of SSI benefits, nor did it collect detailed information about speech therapy or other service utilization, etiology, age at time of child’s diagnosis, receipt of social welfare benefits, or measures of poverty or family well-being. Second, all the information collected is based on parent or guardian reports, with no clinical corrobora- tion of a child’s condition. Finally, the survey’s serial cross-sectional nature does not allow for measuring or monitoring specific children’s conditions or status over time. Survey of Income and Program Participation The Survey of Income and Program Participation is a longitudinal telephone panel survey administered by the U.S. Census Bureau. It col- lects data related to sources and amounts of various types of income and assets, labor force participation, participation in and eligibility for social welfare programs, and general demographic characteristics. Its purpose is to enable assessment of the effectiveness of existing federal, state, and local programs and to provide accurate statistics on the distribution of income and measures of economic well-being in the United States (U.S. Census Bureau, 2015b). This survey has been fielded annually since the late 1970s (U.S. Census Bureau, 2015b). Its design is a continuous series of national panels with sample sizes ranging from approximately 14,000 to 52,000 households (U.S. Census Bureau, 2015b). The 2008 survey included ap- proximately 15,000 children in 9,000 households (NBER, 2015). During the time frame of interest to this study (i.e., the past decade), sampled households participated in a panel for 2 years, and data were collected on each member of the household. Core questions were asked several times, allowing for some monitoring over time, and topical questions were asked in selected waves. This survey provides detailed data on income and assets, receipt of social welfare benefits, and material hardship (U.S. Census Bureau, 2015b). 

170 SPEECH AND LANGUAGE DISORDERS IN CHILDREN However, it has key shortcomings in the context of this study. First, it collects no information about speech therapy or other service utilization, etiology, or age of child’s diagnosis. Second, it provides no direct clinical as- sessment data to confirm a child’s condition; all information is provided by parents or guardians. Third, because many questions and the survey format have changed over the years, multiyear comparisons are often problematic. Fourth, even though the survey is longitudinal, questions about disability status are asked only once, and it is not possible to measure or monitor children’s disability status over time. Finally, the survey collects informa- tion about disabilities only for children who are 6 years of age and older. Individuals with Disabilities Education Act (IDEA) Child Count Data IDEA child count data are administrative data collected and reported annually. Each state reports the number of children receiving special edu- cation services funded in part by federal monies from IDEA Part B (ages 3-21) or Part C (ages birth through 2 years) (U.S. Department of Education, 2014). In 2012-2013, approximately 6.4 million children received special education services (U.S. Department of Education, 2015). The value of IDEA data is that the program is designed to serve all children aged 0-21 in the United States who are eligible to participate; eligibility for services is based on information derived from a variety of in- struments used to evaluate the child; and assessments should display no bias with respect to the student’s racial, cultural, or linguistic background or disabilities. The results of these assessments are used by a multidisciplinary team to determine whether children have disabilities. While all children with disabilities are legally entitled to receive IDEA services, the child count data may underrepresent the true prevalence of conditions because the data are based only on those children for whom parents agreed to receive publicly funded services, and the data quality may be affected by the multiple reporting entities. Variability in the number of recipients across states and school districts suggests that eligibility criteria and service delivery processes vary considerably. A significant shortcoming in the context of this study is that the condition reported in the child count data is only the primary disability used in making the service eligibility determination. Therefore, a child may have a speech or language disorder but be counted under a different eligibility category (e.g., autism, learning disability, hearing impairment), which can result in an underestimate of the prevalence of speech and language disorders. Nonetheless, these data do provide valuable information about trends in the percentage of children whose primary disability is determined to be a speech or language impair- ment, obtained from a multifaceted assessment that was evaluated by a multidisciplinary team. 

COMPARISON OF TRENDS 171 The next section provides information about the prevalence of speech and language disorders in the general population of U.S. children derived from national survey data, including the National Survey of Children’s Health, the National Survey of Children with Special Health Care Needs, and the National Health Interview Survey—Voice, Speech, and Language Supplement. Further details on all of these surveys is provided in Appendix B. TRENDS IN THE PREVALENCE OF CHILDHOOD SPEECH AND LANGUAGE DISORDERS IN THE GENERAL POPULATION Chapter 2 includes clinical data on the overall prevalence of childhood speech and language disorders in the general U.S. population. In this sec- tion, population data are used to estimate trends—or changes over time—in the prevalence of these disorders in the general population. A subsequent section presents estimates of trends in the child SSI program population based on administrative or service data. Estimates of the prevalence of childhood speech and language disorders in the general population vary depending on the survey data that are used. Figures 5-1 and 5-2 show the proportion of noninstitutionalized children with speech and language disorders in the general population derived from two datasets: the 2005-2006 and 2009-2010 National Survey of Children with Special Health Care Needs (see Figure 5-1) and the 2007 and 2011 National Survey of Children’s Health (see Figure 5-2). In an effort to have comparable data points (similar periods in time from two different surveys) the committee used data from the two National Surveys of Children with Special Health Care Needs that included speech and language disorders and the two closest years in which the National Survey of Children’s Health was fielded. To derive estimates of trends in prevalence in the general U.S. popula- tion, the committee used weighted percentages from the National Survey of Children’s Health and the National Survey of Children with Special Health Care Needs. (See Appendix D for a detailed description of the methods used to calculate estimates of trends in prevalence from the national survey data.) For the National Survey of Children’s Health, the result was an estimated 26 percent increase from 2007 to 2011 (from 3.8 to 4.8 percent or from 2,697 to 3,916), with relatively equivalent increases noted in children of different racial or ethnic identities (CDC, 2013b,c) (see Figure 5-2). For the 2005-2006 and 2009-2010 National Survey of Children with Special Health Care Needs, the committee multiplied the weighted percentage of children with speech problems in each year (using the survey weights rec- ommended by the National Center for Health Statistics) by the percentage 

172 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 6.0% 56% Proportion of Children with Speech and Language Disorders increase 5.0% 5 5.0% 4.0% Compared to other 3 2% 3.2% children the same age, would you say [he/she] 3.0% experiences a lot, a little, or no difficulty speaking, communicating, or being 2.0% understood? 1.0% 0.0% 2005/6 2009/10 Year FIGURE 5-1 Proportion of children with speech and language disorders in the general U.S. population based on the 2005-2006 and 2009-2010 National Survey of Children with Special Health Care Needs. NOTE: The sample size in 2005-2006 is 38,296. The sample size in 2009-2010 is 39,747. of children with special health care needs in the larger population.2 The per- centages of children with speech problems for 2005-2006 and 2009-2010 were 23 percent and 33 percent, respectively (CDC, 2007, 2012b, 2014b,c). In 2005-2006 and 2009-2010, the percentages of children with special health care needs were 13.9 percent and 15.1 percent, respectively (HHS, 2013; HHS et al., 2008). The result was an estimated 56 percent increase in prevalence from 3.2 to 5.0 percent (from 8,435 to 11,936) from 2005- 2006 to 2009-2010 (see Figure 5-1). Together, these prevalence estimates indicate that the proportion of children in the general U.S. population with speech and language disorders increased from 3.2 percent in 2005-2006 to 4.8 percent in 2011 (CDC, 2007, 2013c, 2014b; HHS et al., 2008). The 2  In 2009-2010, the percentage of children with special health care needs in the larger population was 15.1 percent (HHS, 2013). 

COMPARISON OF TRENDS 173 committee notes that this is an approximate estimate, based on parent and guardian reports, that includes all levels of severity of speech and language disorders as measured by the questions used in the respective surveys. Not shown in Figures 5-1 and 5-2 is the National Health Interview Survey’s single question on stuttering. Responses to this question suggest a slight but gradual increase from 2000 through 2012, particularly among the younger age groups (3-11), and more stable rates for children aged 12-17. To derive an estimated range for the overall prevalence of reported speech and language problems of any severity, the committee used estimates based on clinical data (see Chapter 2) and the population data described above. The result was an estimated overall prevalence of speech and lan- guage disorders of 3-16 percent of the general U.S. population aged birth through 21 years. Administrative data from IDEA suggest that, relative to their propor- tion of the general U.S. population (2-15 percent for those aged birth-21, as 6.0% 26% 6% Proportion of Children with Speech and Language Disorders rease increase 5.0% 4.8% 48 4.0% 3 8% 3.8% Has a doctor or other health care provider ever told you that [child] had . . . stuttering, 3.0% stammering, or other speech problems? 2.0% 1.0% 0.0% 2007 2011 Year FIGURE 5-2 Proportion of children with speech and language disorders in the general U.S. population based on the 2007 and 2011 National Survey of Children’s Health. NOTE: The sample size in 2007 is 81,955. The sample size in 2011 is 85,581. 

174 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 5 Percentage of Children Served Under IDEA 4.5 4 3.5 3 Learning disabiliƟes Part B 2.5 Other 2 Speech or language Intellectual disabiliƟes 1.5 1 0.5 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Year FIGURE 5-3  Percentage of the general U.S. population of children aged 6-21 years served under IDEA Part B, by year and disability category, fall 2003 through fall 2012. NOTE: According to the U.S. Department of Education, percentage was calculated by dividing the number of students ages 6 through 21 served under IDEA, Part B, in the disability category in the year by the estimated U.S. resident population ages 6 through 21 for that year, then multiplying the result by 100. The numbers of students served under IDEA Part B are included in Appendix C. SOURCE: U.S. Department of Education, 2014. reported above), a substantially smaller percentage of children who receive special education services have a primary condition of speech and/or lan- guage impairment (1.5-1.7 percent of those aged 6-21) (see Figure 5-3) (U.S. Department of Education, 2014). The smaller percentages for the IDEA da- taset are also consistent with other sources indicating that only about two- thirds of children reported to have speech and language problems received intervention services during the preceding year (Bainbridge, 2015; Black et al., 2015). Some of the children with speech and language impairments in IDEA data also likely have more serious impairments than children in the survey datasets described above. With respect to trends in prevalence, IDEA data stand in contrast to the survey data reported above, indicating that the prevalence of speech and language disorders among children receiving IDEA services remained steady from 2003 to 2012. Why the prevalence of speech and language disorders as measured by the National Survey of Children’s Health and the National Survey of 

COMPARISON OF TRENDS 175 Children with Special Health Care Needs has increased over time is unclear. However, analyses of similar increases among children with autism spec- trum disorder and attention deficit hyperactivity disorder may provide some insight. The available epidemiologic literature shows a consistent increase in the prevalence of these latter two conditions over time (Newschaffer et al., 2005; Visser et al., 2010). Factors identified as causes for these increases include increased awareness of developmental disorders due to media attention and advocacy, increased availability of early intervention and special education services, changes in access and practices that are leading to children being identified as having a developmental disorder at a higher frequency, changes in the definition of certain conditions (e.g., au- tism spectrum disorder), and an increase in the assignment of co-occurring diagnoses among children with developmental disorders (NASEM, 2015). Although there have been no similar studies of the epidemiology of speech and language disorders, the committee believes the available evidence sug- gests that the same factors have contributed to the increasing prevalence of these disorders. VARIATIONS IN THE PREVALENCE OF SPEECH AND LANGUAGE DISORDERS BY DEMOGRAPHIC AND OTHER CHARACTERISTICS This section summarizes the available evidence on variations in the prevalence of childhood speech and language disorders by the follow- ing characteristics: age, gender, race/ethnicity, socioeconomic status, and comorbidities. Age The available evidence indicates that speech and language disorders are highly variable with age. Age of Children with Speech and Language Disorders Across national population surveys, the percentage of children with reported speech and language disorders is generally higher among younger children; these disorders are most common among children aged 3-6 years (11.0 percent), followed by those aged 7-10 years (9.3 percent), and their reported prevalence is lowest among children in their teens (4.9 percent) (see Table 5-2 and the discussion of persistence in Chapter 3) (Bainbridge, 2015; Black et al., 2015). IDEA child count data show that the percentage of children with speech or language impairments is considerably higher in the younger age group 

176 SPEECH AND LANGUAGE DISORDERS IN CHILDREN TABLE 5-2  National Health Interview Survey—Voice, Speech, Language Supplement: Prevalence of Speech and Language Disorders, by Age Age Group Prevalence (%) 3-6 years 11.0 7-10 years  9.3 11-17 years  4.9 NOTE: The prevalence estimates for children aged 3-6 and 7-10 are significantly different from that for children aged 11-17 years (p < 0.05). SOURCE: Black et al., 2015. of 3-5 years (44.7 percent) than in the older age group of 6-21 years (18.2 percent) (see Figure 5-4) (U.S. Department of Education, 2014). Age of Onset As noted in Chapter 1, age of onset of speech and language disorders is difficult to pinpoint, although it is likely to be early in the child’s life. Available data on these disorders are more likely to reflect age of identifi- cation than age of onset. Identification of speech and language disorders requires that children be of an age when speech is typically present, and disruptions or delays in speech become evident. Data from the National Health Interview Survey—Voice, Speech, and Language Supplement indi- cate that these disorders are identified in the overwhelming majority of chil- dren by age 6. More than half of speech problems are identified in the first 2 years of the child’s life, and almost one-third between 3 and 5 years of age. Gender Data sources consistently document sex differences in speech and lan- guage disorders, with boys demonstrating speech and language problems approximately twice as frequently as girls (see also Chapter 2). Data from the National Health Interview Survey (2000-2012) indicate that about 69 percent of children identified with stuttering and stammering problems are boys, compared with 31 percent for girls (Bainbridge, 2015). The National Health Interview Survey—Voice, Speech, and Language Supplement shows the prevalence of voice, speech, swallowing, or language problems to be 4.3 percent of all boys versus 2.2 percent of all girls (Hoffman, 2015). Finally, data from the National Survey of Children with Special Health Care Needs indicate that 67 percent of all children with a great deal of difficulty with speech are boys (Blumberg, 2015). 

COMPARISON OF TRENDS 177 Percentage of Children Served Under IDEA 100% 90% 80% 70% 55.30% 60% 81.70% Other disabiliƟes combined Part B 50% 40% Speech or language impairments 30% 20% 44.70% 10% 18.20% 0% 3 to 5 years 6 to 21 years Age Group FIGURE 5-4  Percentage of children served under IDEA Part B by age group: Fall 2012. NOTE: In the age group 3-5 years, the “other disabilities combined” category includes developmental delay, autism, deafness-blindness, emotional disturbance, hearing impairments, intellectual disabilities, multiple disabilities, orthopedic im- pairments, other health impairments, specific learning disabilities, traumatic brain injury, and visual impairments. In the age group 6-21 years, the “other disabilities combined” category includes specific learning disabilities, other health impairments, autism, intellectual disabilities, emotional disturbance, deafness-blindness, develop- mental delay, hearing impairments, multiple disabilities, orthopedic impairments, traumatic brain injury, and visual impairments. SOURCE: U.S. Department of Education, 2014. Race/Ethnicity Figure 5-5 shows the weighted racial and ethnic composition of chil- dren with speech and language disorders, using data from the 2007 and 2011 National Survey of Children’s Health (CDC, 2013b,c). In addition to depicting the proportion of children with speech and language disorders by race and Hispanic ethnicity, the figure shows the reported increase in each group over time. As is evident, the percentage of children reported to have speech and language disorders increased over time within each racial or ethnic group. Figure 5-6 shows the prevalence of speech and language disorders in children by race/ethnicity based on the National Health Interview Survey— Voice, Speech, and Language Supplement. Although not statistically 

178 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 6% 5.1% Percent of Children with Speech and 4.9% 4.8% 5% 4.5% 4.4% Language Disorders 4% 3.7% 3.7% 2.9% 3% 2007 2% 2011 1% 0% Non-Hispanic Non-Hispanic Other Non- Hispanic White Black Hispanic Race/Ethnicity FIGURE 5-5  Prevalence of speech and language disorders by race/ethnicity among children aged 0-17 years based on the 2007 and 2011 National Survey of Children’s Health. NOTE: The sample size is 81,955 in 2007 and 85,581 in 2011. Figure 5-5 4.0% 3.7% 3.7% Speech or Language Problems Prevalence 3.5% 3.0% 3.0% 2.7% 2.6% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Non-Hispanic Hispanic Non-Hispanic Asian Other Children Black White Race/Ethnicity FIGURE 5-6 Prevalence of speech and language disorders among children aged 3-17 years based on the National Health Interview Survey—Voice, Speech, and Language Supplement. 

COMPARISON OF TRENDS 179 significant, this survey showed higher rates for non-Hispanic black (3.7 percent) and Hispanic (3.7 percent) children compared with non-Hispanic white (3.0 percent), Asian (2.6 percent), and other (2.7 percent) children (Hoffman, 2015). In summary, the committee found no apparent racial or ethnic dif- ferences in the prevalence of speech and language disorders based on the surveys it reviewed. The surveys show slight variability across racial/ethnic groups, but the findings are not statistically significant. Socioeconomic Status A greater proportion of children with speech and language disorders than of children without such disorders are in families that live in poverty or have low incomes. Low income here means total income below 200 percent of the federal poverty level (FPL) (Boushey et al., 2001; National Center for Children in Poverty, 2013). The FPL varies by household size. In 2015, for a family of four living in the 48 contiguous United States, it was $24,250 (HHS, 2015a). Thus, for example, a family of four with income below 200 percent of that number ($48,500) would be defined as having low income. In 2013, approximately 44 percent of U.S. children lived in low-income households (National Center for Children in Poverty, 2013). Figures 5-7 and 5-8 show data from the Survey of Income and Program Participation indicating that children with speech and language disorders are more likely than other children to live in poor or low-income households (26 versus 21 percent and 28 versus 23 percent, respectively) (NBER, 2015). The National Survey of Children with Special Health Care Needs, 2009-2010, yields similar findings, showing that relative to all children with special health care needs, those with any speech and language prob- lems were more likely to live in poverty (30.8 versus 22.2 percent) or in low-income households (24.0 versus 21.8 percent) and less likely to live at 400 percent of the FPL or above (19.5 versus 27.5 percent) (Blumberg, 2015). This finding was even more marked for children with the most severe speech and language disorders who lived in poverty (33.4 versus 22.2 percent) or in low-income households (23.0 versus 21.8 percent); these children also were less likely to live well above the poverty level (18.3 versus 27.5 percent) (Blumberg, 2015). In data from the National Health Interview Survey—Voice, Speech, and Language Supplement (2012), the prevalence of communication problems decreased from 5.4 percent in children living below the FPL to 1.9 percent for children in families at five times the poverty level (Hoffman, 2015). Families with low incomes often face material deprivation. Figure 5-9 shows 2010 data from the Survey of Income and Program Participation in- dicating that households that experience various types of material hardship 

180 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 26% < 100% FPL 46% 100–199% FPL ≥ 200% FPL 28% FIGURE 5-7 Children aged 6-17 with speech and language disorders, by family income relative to the federal poverty level (FPL). NOTE: The sample size is 277. SOURCE: 2010 Survey of Income and Program Participation. 21% < 100% FPL 56% 100–199% FPL 23% ≥ 200% FPL FIGURE 5-8  Children aged 6-17 without speech and language disorders, by family income relative to the federal poverty level (FPL). NOTE: The sample size is 14,507. SOURCE: 2010 Survey of Income and Program Participation. 

COMPARISON OF TRENDS 181 40% 34.83% 35% 30% Households with 6- to 17- Percent of Material Hardship year-olds with speech and 25% 23.05% language disorders 22.85% Households with 6- to 17- 20% year-olds without speech problems 14.40% 15% 10% 8.12% 6.61% 5% 0% Unmet EssenƟal Low Food Security Child Food Insecurity Expenses Types of Hardship FIGURE 5-9 Material hardship among families with and without children with speech and language disorders. NOTE: The sample size of children with speech and language disorders is 277 and without speech and language disorders is 14,507. SOURCE: 2010 Survey of Income and Program Participation. are more likely than households that do not experience such hardship to have children with speech and language disorders (NBER, 2015). Food insecurity is defined as insufficient access to adequate food, which occurs as a result of limited financial and other resources (Coleman-Jensen et al., 2013). In the case of household food insecurity, many parents forego their own meals so they can feed their children. Child food insecurity (il- lustrated in Figure 5-9), a situation in which children have insufficient food, is rarer. Figure 5-9 shows that both household food insecurity and child food insecurity are higher among households with children with speech and language disorders than among other households (e.g., 8 percent versus 7 percent for child food insecurity) (NBER, 2015). The third indicator of hardship is the inability to meet necessary household expenses (e.g., rent, utilities), which shows a similar pattern. With respect to health service utilization of children with speech and language disorders, having insurance is generally considered a prerequisite for receiving appropriate care when it is needed (HHS, n.d.). According to data from the 2011 National Survey of Children’s Health, just 4 percent of children with speech problems were uninsured (CDC, 2013c). Of the 

182 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 60% 57% 50% 50% Percent of Unmet Need 40% 30% 2007 2011 20% 18% 13% 14% 8% 10% 0% No prevenƟve care ≥1 unmet No help with care in last year health need coordinaƟon Indicators of Receiving Health Care FIGURE 5-10  Unmet health care needs for children with speech problems. NOTE: In 2007, the sample sizes for no preventive care in past year, ≥ 1 unmet health need, and no help with care coordination are 2,469, 2,487, and 980, respec- tively. In 2011, the respective sample sizes are 3,529, 3,548, and 1,410. Figure 5-10 SOURCE: 2007 and 2011 National Survey of Children’s Health. remainder, 44 percent had public insurance (Medicaid, TriCare, or the State Children’s Health Insurance Program), and 53 percent had private insur- ance (CDC, 2013c). Yet despite having nearly universal levels of insurance, children with speech or language disorders experienced significant barriers to receiving needed health care services. Figure 5-10 illustrates the relatively high rates of unmet need for these children in terms of three common in- dicators of receiving health care (CDC, 2013b,c). The rates of unmet need for care were relatively stable from 2007 to 2011. Comorbidities As discussed in Chapter 2, children with speech and language disorders have a high likelihood of experiencing other problems as well. Based on the 2000-2012 National Health Interview Survey, more than 40 percent of children with speech and language disorders experience such comorbidities as developmental delay3 (32 percent), autism (12 percent), or intellectual 3 Developmental delay refers to physical, cognitive, communication, social or emotional, or adaptive developmental milestones that are not reached by a child at the expected times. 

COMPARISON OF TRENDS 183 disability (10 percent) (Bainbridge, 2015). Data from the National Health Interview Survey—Voice, Speech, and Language Supplement indicate that intellectual disability is the most common comorbidity, representing 52 per- cent of all comorbidities, followed by developmental delay (27.6 percent), seizures (26.4 percent), and attention deficit hyperactivity disorder (10.4 percent) (Hoffman, 2015). What cannot be ascertained from these data is the frequency with which children experience speech and language disor- ders as comorbidities of other disorders or primary diagnoses. TRENDS IN CHILDHOOD SPEECH AND LANGUAGE DISORDERS IN THE SUPPLEMENTAL SECURITY INCOME POPULATION In analyzing trends in childhood speech and language disorders in the SSI population, the committee included any child who was allowed based on code 3153 (speech and language impairment) or 7840 (loss of voice [communication impairment]) from SSA Form 831. From this point for- ward, unless otherwise noted, the discussion of children with speech and language disorders who receive SSI benefits encompasses those with these two primary impairments. It is also important to remember that the SSA’s eligibility and determi- nation process requires that children have a “marked” or “extreme” im- pairment, as described in Chapter 4. Because the datasets used in estimating the prevalence of speech and language disorders in the general U.S. popula- tion impose no such severity restrictions, definitive comparisons between the two populations are not possible. Impairments determined through the SSA process can be either primary—where the child’s chief presenting problem is a speech or language impairment—or secondary—where the child has been determined to have another impairment for which a speech or language impairment is a comorbidity or a secondary impairment. As described in detail in Chapter 2, speech and language impairments are com- monly associated with a variety of other developmental disorders, including disorders that may themselves result in a child’s qualifying for SSI benefits (e.g., autism spectrum disorder, attention deficit hyperactivity disorder, intellectual disability). The remainder of this section presents information regarding • the number of children qualifying for (i.e., allowances) and receiv- ing (i.e., recipients) SSI benefits for speech and language disorders as a primary condition; • the frequency with which other impairments (i.e., conditions) iden- tified in the SSA eligibility and determination process are comorbid with primary speech or language disorders; 

184 SPEECH AND LANGUAGE DISORDERS IN CHILDREN • the number of children for whom speech and language disorders are secondary impairments, and the comorbidity of other primary impairments among these children; • the reasons that children who receive SSI benefits exit from the program (i.e., are suspended or terminated); and • the concentration of the lowest-income children within the SSI program resulting from suspensions and terminations related to income and resources. Note that in the figures that follow, the number of recipients represents the total number of children receiving SSI benefits in a given federal fiscal year, and includes children under 18 years of age who qualified in previ- ous years and were still eligible. Allowances, on the other hand, represent the new cases of children who became eligible for SSI benefits in that year. Because the number of newly eligible children is smaller than the total number of current beneficiaries, the number of allowances for a given year will always be lower than the number of recipients. As discussed previ- ously, recipients of SSI benefits for speech and language disorders must have “marked” or “extreme” disorders to qualify for SSI benefits, and the vast majority of these children are not expected to recover from or be cured of those disorders. Thus, the majority of these children will continue to qualify for SSI benefits until they are age 18 unless their eligibility status changes for another reason (e.g., family income and/or resources increase to the point that the child no longer qualifies; the child dies; or some other event occurs that affects a relatively small number of children, such as moving out of the United States, failing to complete paperwork, or moving into a residential institution). Trends in Initial Allowances and Total Number of Recipients for Primary Speech and Language Impairments The committee worked to identify and understand the trends over time in the total number of SSI recipients for primary speech and language im- pairments. To do so, the committee examined trends in initial allowances and trends in the overall number of SSI child recipients for speech and lan- guage disorders, and examined these trends in the context of the cumulative effect of new children being entered into the program after establishment of this diagnostic eligibility in 1994 and the relatively few departures from the program. These findings are presented below. The numbers of initial determinations, denials, and initial allowances for speech and language disorders are shown in Figure 5-11. An initial de- termination refers to a decision made by Disability Determination Services as to whether a child is allowed or denied SSI benefits based on disability. 

COMPARISON OF TRENDS 185 60,000 Number of Determinations, Allowances, and 50,000 40,000 DeterminaƟons Denials 30,000 Allowances Denials 20,000 10,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year FIGURE 5-11  Number of determinations, allowances, and denials for speech and language disorders (includes primary impairment codes 3153 and 7840) at the ini- tial level, fiscal years 2004-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. Therefore, determinations equal the sum of initial allowances and denials. The committee’s analysis showed that initial allowances for children with speech and language disorders exceed the number of denials. The data sug- gest that more children who meet the requirements for SSI disability benefits than children who do not meet the requirements are applying for benefits and being allowed into the SSI program. The numbers of allowances for primary speech and language impair- ment over an 11-year period are shown in Figure 5-12. The figure also shows increases in allowances within the SSI program during years that cor- respond to the years of administration of the National Survey of Children with Special Health Care Needs and the National Survey of Children’s Health. The number of children who met the eligibility criteria each year for a marked or extreme speech or language impairment increased from 18,299 in 2004 to 22,727 in 2014. During that same period, the population of children aged 0-17 in the United States increased by just 0.39 percent (U.S. Census Bureau, 2015c). It is noteworthy that during the economic recession that began in 2007, the number of initial allowances increased sharply, and the number then decreased in 2013 and 2014 as the economy improved. 

186 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 30,000 25,000 20,000 Does child qualify for SSI benefits because Allowances 15,000 of a speech or language disorder? 10,000 Increase from 5,000 2004 to 2014: 24% 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year FIGURE 5-12  Initial allowances for speech and language impairment as a primary condition, fiscal years 2004-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. The 27 percent increase in SSI allowances for children with speech and language impairments is markedly lower than the 56 percent increase in prevalence of these disorders reported by the National Survey of Children with Special Health Care Needs between 2005-2006 and 2009-2010. On the other hand, the 40 percent increase in SSI allowances is larger than the 26 percent increase in prevalence reported by the National Survey of Children’s Health between 2007 and 2011 (compare Figures 5-2 [presented earlier] and 5-12). Figure 5-13 illustrates the total number of children receiving SSI ben- efits for speech and language impairment over the period 2004 to 2014, showing a 171 percent increase from 78,827 to 213,688. However, the number of initial allowances per year (i.e., newly eligible beneficiaries) increased by only 24.2 percent during this same period. Figure 5-14 illustrates why the annual number of initial allowances increased by 24.2 percent during this 11-year period, while the total num- ber of recipients increased by 171 percent. As elaborated below, because the impairment code for speech and language impairment (3153) was not 

COMPARISON OF TRENDS 187 250,000 200,000 150,000 Recipients 100,000 50,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year FIGURE 5-13  Numbers of child recipients of SSI benefits for speech and language impairment, fiscal years 2004-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. established until 1994, and children qualifying for SSI benefits have severe disorders that are unlikely to remit, the growth in the number of overall recipients reflects the aggregation of new allowances over the 18 preceding years. In other words, the total number of participants in a particular year is an accumulation of new cases added in that year plus all of those deemed eligible in prior years (less those who age out or are otherwise no longer eligible). In most cases, children who are deemed eligible will continue to have a severe speech or language disorder, and unless their eligibility changes for some other reason, they will not exit the program until they reach 18 years of age. Thus, a 24.2 percent increase in the number of an- nual initial allowances leads to a much larger 171 percent increase in the total number of beneficiaries over time because once qualified for SSI, few children with severe speech and language disorders subsequently become ineligible for the program. In 1994, the first year in which the new impairment code (3153) for speech and language impairment existed, only 1,585 children met the eli- gibility criteria for this new code. In 1995, an additional 4,109 children were allowed benefits under this new code; in 1996, an additional 4,119 children were allowed; and so forth for each subsequent year. Given that children with severe speech and language impairments are likely to continue 

188 

COMPARISON OF TRENDS 189 # of Initial Allowances Each Year FIGURE 5-14  Illustration of how the total number of children receiving SSI benefits grew over time as a result of the addition of a new category, fiscal years 1994-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. 

190 SPEECH AND LANGUAGE DISORDERS IN CHILDREN to have severe impairments throughout adolescence and into adulthood, the total number of children (aged 0-18) who received benefits under this new code in any given year approaches the total number of children who became eligible in each of the preceding 18 years (those years are shaded in Figure 5-14). Thus, the total number of children receiving SSI benefits for speech and language impairments in 2013 will include almost all of the children who became eligible in the years from 1996 through 2013, but will not include children who became eligible in 1994 or 1995. In fact, when the numbers in Figure 5-14 are adjusted for the approximately 4 percent4 of children whose benefits are terminated each year (see the later discussion of suspensions and terminations), the graph of the number of recipients each year in Figure 5-14 is similar to the graph in Figure 5-13. In other words, the increase in the number of recipients is almost completely explained by the fact that a new category for speech and language impairment was created in 1994; that virtually all children in this category have marked or extreme forms of impairment (as required by the SSA eligibility criteria); and that very few of these children will cease to have marked or extreme impairments as a result of receiving the available treatments. Comorbidities Among Allowances for Primary Speech and Language Disorders Figure 5-15 shows the number of allowances for children with primary speech and language impairments who also had secondary impairments (co- morbidities). The top two secondary impairment categories are none estab- lished, no medical evidence in file, and none/no secondary impairment for children with primary speech and language impairments. However, these data should be viewed with caution. In its review, the U.S. Government Accountability Office (GAO, 2012) found “significant inconsistencies” in how secondary impairments were coded nationwide. Furthermore, the absence of a recorded secondary impairment does not mean that the child did not actually have another impairment, only that it was not recorded as part of the eligibility determination process. Only about one-third of children with speech and language impair- ments had a secondary impairment recorded. Among these children, autistic disorder and other developmental disorders was the secondary impairment category most commonly listed. The number of children with secondary autism spectrum disorder increased between 2004 and 2014 from 643 to 2,488. The second most commonly occurring secondary impairment, espe- cially between 2007 and 2010, was attention deficit hyperactivity disorder. 4  Based on unpublished data provided by the SSA, the 4 percent figure was calculated by dividing total child terminations by total child SSI recipients in 2008. 

COMPARISON OF TRENDS 191 12,000 None established/no medical evidence in file None/no secondary 10,000 impairment AuƟsƟc disorder and other Secondary Impairment Allowances developmental disorders 8,000 AƩenƟon deficit hyperacƟvity disorder Learning disorder 6,000 Hearing loss Intellectual disability Asthma 4,000 Epilepsy Organic mental disorders 2,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year FIGURE 5-15  Top secondary impairments for initial allowances for primary speech and language disorder (includes impairment codes 3153 and 7840), fiscal years 2004-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. Figure 5-15 Between 2008 and 2014, autism spectrum disorder, attention deficit hyper- activity disorder, and learning disability were the most frequent comorbid conditions associated with an allowance based on a primary speech or language impairment. Primary Impairments for Which Speech and Language Impairments Were Secondary Impairments Figure 5-16 shows the most frequent primary impairments among children for whom speech and language impairments were reported as a secondary condition. The most commonly occurring of these primary impairments was autism spectrum disorder, for which allowances ranged from 2,402 to 6,238 between 2008 and 2014. Between 2004 and 2007, the most commonly occurring primary impairment was intellectual disability. Commonly occurring in addition to autism spectrum disorder and intel- lectual disability were attention deficit hyperactivity disorder and learning disabilities. 

192 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 7,000 AuƟsƟc disorder and other developmental disorders 6,000 Intellectual disability AƩenƟon deficit hyperacƟvity disorder 5,000 Hearing loss Primary Impairment Allowances Cerebral palsy 4,000 Organic mental disorders Learning disorder 3,000 Epilepsy Borderline intellectual funcƟoning 2,000 Development and emoƟonal disorders/infants 1,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Year FIGURE 5-16  Top primary impairments for which speech and language disorders were secondary impairments (includes impairment codes 3153 and 7840), fiscal years 2004-2014. SOURCE: Unpublished dataset provided by the Social Security Administration. Figure 5-16 Exits from the Supplemental Security Income Program Recipients may exit from the SSI program because of suspension, ter- mination, age-18 transitions, or death. A suspension occurs when a recipi- ent’s payments are stopped because of ineligibility. Suspensions cannot last longer than 1 year. If a recipient does not become eligible again within 12 months, benefits are terminated. A termination is a permanent loss of SSI eligibility. A terminated recipient must file a new application to reinstate his or her benefits. (See also the discussion of the SSI program in Chapter 4.) Figure 5-17 shows suspensions from the SSI program from 2004 through 2013. Notably, these data are for all children receiving SSI ben- efits, not just those with speech and language disorders as has been the case previously in this section. Depending on the year, suspensions represent 12-15 percent of the total number of SSI recipients under 18 years of age. Most suspensions occurred because the child recipient’s family in- come or assets exceeded SSI program limits. The number of such income- related suspensions increased from 84,900 in 2004 to 95,138 in 2013. Approximately 3.7 to 18.3 percent of the suspensions occurred because the child no longer qualified as having a marked or extreme impairment. 

COMPARISON OF TRENDS 193 250,000 200,000 Recipient Suspensions 150,000 No longer disabled Other 100,000 Excess income and resources 50,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year FIGURE 5-17  Number of all child SSI recipients suspended, fiscal years 2004-2013. SOURCE: Unpublished dataset provided by the Social Security Administration. This trend is consistent with information presented in Chapter 3 indicating that the severity of speech and language disorders required for SSI eligibil- ity is unlikely to resolve over time, even when treatment is provided. Some children were suspended for “other” reasons, which include admission into a residential facility, whereabouts unknown, no representative payee, in public institution, failed to furnish report, moved outside United States, in transition, and other or unknown reasons. Figure 5-18 shows the number of children (again not limited to those with speech and language disorders) whose SSI benefits were terminated between 2004 and 2013. As in Figure 5‑17, the largest number of termina- tions occurred when recipients’ families had income and/or resources that exceeded program limits—28,270 terminations in 2004, rising to 32,825 in 2013 (which represents 48.9 percent of all terminations in that year). Terminations due to no longer being disabled decreased by more than 70 percent between 2004 and 2008 before increasing to 27.5 percent in 2013—approximately the same level as in 2004. Concentration of Lowest-Income Children Within the Supplemental Security Income Program Resulting from Suspensions and Terminations The net result of suspending and terminating higher-income children from the SSI program is the concentration of children from the poorest 

194 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 80,000 70,000 60,000 Recipient Terminations 50,000 Death Other 40,000 No longer disabled 30,000 Excess income and resources 20,000 10,000 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year FIGURE 5-18  Number of SSI child recipients terminated, fiscal years 2004-2013. SOURCE: Unpublished dataset provided by the Social Security Administration. families remaining in the program (see Figure 5-19). The largest number of child SSI recipients are those with the lowest family incomes (at or below 100 percent of the FPL). Over time, a larger proportion of the lowest- income children thus are retained within the SSI program. TRENDS IN CHILDHOOD SPEECH AND LANGUAGE DISORDERS AMONG MEDICAID BENEFICIARIES Medicaid data are administrative and can be used to examine comorbid conditions and services among children receiving treatment for a medical diagnosis. As described below, entry into the SSI program establishes eligi- bility for Medicaid in most states. Consequently, most children eligible for SSI are also eligible for Medicaid. The Medicaid data on such children can then be tracked over time to examine trends in the number of Medicaid enrollees who are diagnosed with a speech or language disorder and their resulting utilization of treatment services, if any. This section first describes the relationship of children with speech and language disorders enrolled in SSI to the Medicaid program, with specific reference to their eligibility for, and entry into, Medicaid. It then illustrates the prevalence of speech and language disorders among a sample of Medicaid enrollees. The section ends by describing some of the limitations in the use of Medicaid data to 

COMPARISON OF TRENDS 195 1,800 1,600 1,400 Rate per 100,000 1,200 children at/below 100% FPL 1,000 Rate per 100,000 Rate children at/below 800 200% FPL 600 Rate per 100,000 children 400 200 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Year FIGURE 5-19  Rate of children receiving SSI benefits on the basis of speech and lan- guage disorders (includes impairment codes 3153 and 7840) by the federal poverty Figure 5-19 level (FPL), fiscal years 2004-2013. NOTE: The Current Population Survey table creator was used to generate numbers of children below 200 percent of the FPL. Parameters used to generate the numbers include get count of: persons in poverty universe (everyone except unrelated indi- viduals under 15); years: 2004 to 2013; Census 2010 weights; row variable: age; column variable: income-to-poverty ratio; and customized formatting: income-to- poverty ratio cutoff of 200 percent. SOURCES: U.S. Census Bureau, 2015a,c, and unpublished dataset provided by the Social Security Administration. establish trends in the prevalence of speech and language disorders among children with these disorders enrolled in the SSI program. Established in July 1965 as Title 19 of the Social Security Act, Medicaid is the dominant insurer of low-income adults and children in the United States today. Jointly funded by the federal and state governments, Medicaid is administered by the states, which have latitude in determining program eligibility, available benefits, health care provider payment rates, drug for- mularies, and a range of program characteristics, within federal parameters. As a result, there is wide variability among states in program characteristics and beneficiary populations. (Detailed discussion of the origins and struc- ture of Medicaid is available elsewhere [see, e.g., Engel, 2006, and Olson, 2010].) 

196 SPEECH AND LANGUAGE DISORDERS IN CHILDREN As the description above implies, most children are eligible for Medicaid because their families have low incomes. However, states deter- mine income limits for their Medicaid programs, as long as those limits meet or exceed federal minimums. The family income above which an infant below 1 year of age is no longer eligible for Medicaid, for example, varies from 139 percent of the FPL in Utah to 375 percent of the FPL in Iowa (HHS, 2015b). Income is not the sole criterion by which children qualify for Medicaid. Children who are in certain highly vulnerable categories (e.g., disability status) and meet citizenship and state residency criteria are eligible for Medicaid in certain states under several other eligibility categories; being a recipient of SSI is one such category. In 33 so-called 1634 states and the District of Columbia, SSI eligibility automatically qualifies an individual for Medicaid coverage (SSA, 2014). There are 10 “209(b) states” that use criteria more restrictive than those of the SSI program to determine Medicaid eligibility for their residents; in these states, meeting SSI eligibility criteria does not guarantee Medicaid eligibility (SSA, 2014). Additionally, there are 7 “SSI criteria states” that make their own determinations regarding Medicaid eligibility once an individual has met SSI eligibility criteria (SSA, 2014). The net result of all this variation is that the Medicaid and SSI beneficiary populations display substantial overlaps but are by no means identical. Once a child SSI recipient is eligible for Medicaid, the child’s parent or guardian must successfully seek services from a provider who bills Medicaid for those services. A diagnosis is necessary for such a bill (a “claim”) to be paid. Medicaid claims are generated as a result of such a billing process, and represent units of services that are paid for by state Medicaid agencies. Medicaid data, therefore, reflect the receipt of health care services or treat- ment. These are data of a particular type: they reflect those services received by a child who is eligible for Medicaid, whose provider accepts Medicaid reimbursement, and whose treatment was paid for by the Medicaid agency in the state in which the child lives. Children who are unable to access services—for example, because no providers in their region will accept Medicaid reimbursement—will not appear on Medicaid rolls even if they are Medicaid beneficiaries or otherwise eligible for Medicaid. Furthermore, Medicaid claims data do not include claims of children who are in Medicaid managed care. This exclusion is substantial, as up to half of all children reflected in Medicaid data are in managed care and thus are not reported to state Medicaid agencies in most states. Hence their diagnoses or use of services cannot be established. Consequently, Medicaid data cannot be used to determine the prevalence of speech and language disorders in the general population—that is, they are not population data. They can be used only to quantify the rate of diagnoses of these disorders occurring among 

COMPARISON OF TRENDS 197 a treated population whose care was paid for by Medicaid under certain payment arrangements—that is, they are a type of service data. These data cannot be interpreted to include all low-income children with speech and language disorders, or even all children who receive Medicaid funding for their health insurance. An analysis of Medicaid claims data from 20 states conducted by a team at Rutgers University sheds light on speech and language disorders among participants in the SSI and Medicaid programs. The Rutgers team examined claims data for children aged 3-17 with 11+ months of continu- ous Medicaid eligibility and no dual eligibility in a given year. Data were ob- tained for 2001-2010 from Alabama, Alaska, Arkansas, California, Florida, Idaho, Illinois, Indiana, Louisiana, Michigan, Mississippi, Montana, North Carolina, North Dakota, New Hampshire, New Mexico, South Dakota, Vermont, Virginia, and Wyoming.5 Importantly, these are not longitudinal or panel data, as discussed earlier. Hence, these data should be interpreted as comparing diagnostic codes for speech and language disorders observed over time among correlated but possibly independent groups of children receiving services paid for by Medicaid. Figure 5-20 displays the proportion of children in these 20 states whose Medicaid claims indicate a diagnosis of speech and language disorders (blue bars). It also shows the proportion of children who were eligible for Medicaid during 2001-2010 because they were SSI beneficiaries whose claims had the same diagnosis during the same time period (gray bars). As seen in the figure, a greater proportion of those children who were eligible for Medicaid because of participation in the SSI program had observed diagnoses of speech and language disorders compared with other children on Medicaid. In calendar year 2010, for example, 11.9 percent of the children on Medicaid because of SSI program participation (n = 56,980) were diagnosed as having speech and language disorders, compared with 2.7 percent of all children on Medicaid (n = 221,630) diagnosed as having these disorders. This pattern reveals that, compared with non-SSI children on Medicaid, more children with speech and language disorders are concentrated within the SSI/Medicaid population. This pattern may reflect increased entry into the SSI program of children with speech and language disorders over time, as discussed earlier in this chapter. Some of these children on SSI, then, may have found their way into the Medicaid program through the eligibility 5  Thissubset of 20 states was selected because they (1) had predominantly fee-for-service youth Medicaid populations or (2) had been identified as having relatively complete and usable managed care encounter data for that population. See https://www.cms.gov/ Research-Statistics-Data-and-Systems/Computer-Data-and-Systems/MedicaidDataSources GenInfo/Downloads/MAX_IB_15_AssessingUsability.pdf and http://mathematicampr.com/ publications/pdfs/health/MAX_IB14.pdf (accessed September 30, 2015). 

198 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 14.0% 12.0% 10.0% Percentage of Enrollees Speech and Language 8.0% Disorders as a Percentage of All Medicaid Enrollees 6.0% Speech and Language Disorders as a Percentage 4.0% of SSI/Medicaid Enrollees 2.0% 0.0% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Year FIGURE 5-20  Speech and language disorders among Medicaid enrollees and SSI/ Medicaid enrollees. SOURCE: Medicaid Analytic eXtract [MAX] data. mechanisms discussed earlier. The SSI program, therefore, may act as a gateway into Medicaid for some low-income children with severe speech and language disorders who do not already have Medicaid coverage. The alternative pathway may be equally operant, in which low-income children on Medicaid are diagnosed with speech and language disorders and then seek SSI program participation. Both of these trends support the increas- ing prevalence of speech and language disorders among children receiving SSI discussed earlier in this chapter. Irrespective of mode of entry, the two programs are highly complementary, with Medicaid resourcing the health care needs of children and SSI resourcing the added costs associated with raising children with speech and language disorders. The concentration of children with speech and language disorders within the SSI program is also evident in the heights of the blue and red bars in Figure 5-20 over time. Between 2001 and 2010, the proportion of children in the SSI/Medicaid population with this diagnosis increased from 6.3 percent to nearly 12 percent; the rate of increase among the Medicaid population—from 1.6 percent to 2.7 percent—was far more moderate. The trends in the occurrence of speech and language disorder diagnoses over time within the SSI/Medicaid population, then, parallel the increase in the occurrence of these disorders over time among children receiving SSI benefits. 

COMPARISON OF TRENDS 199 The increases in speech and language disorders in the SSI/Medicaid and Medicaid populations differ from the prevalence of speech and lan- guage disorders among children in the general U.S. population discussed in Chapter 2. The simplest explanation for this difference is that the estimated prevalence in the general population (up to 16 percent) includes children with conditions ranging from mild to severe. The 6.3-12 percent prevalence is for a treated (Medicaid) population, and children receiving SSI must have severe speech and language disorders to qualify for the program. Children receiving Medicaid-funded services by definition must have the condition for which services are necessary, and must also successfully procure the ser- vices. Even if children in the general population have speech and language disorders, they are not necessarily able to procure services successfully. Other explanations for this difference in percentages are cited in the discus- sion of data limitations below; they include such factors as ascertainment challenges inherent in speech and language disorders, differences in the severity of the ascertained disorders, and limitations in the nature of the data used to arrive at diagnoses of speech and language disorders. Medicaid data can also be used to identify comorbid conditions. One approach is to examine additional diagnostic codes within the claims of children with a diagnosis of speech or language disorder. Using this strategy, the Rutgers University team determined that the most commonly occurring comorbid diagnosis in the claims of children with speech and language disorders is attention deficit hyperactivity disorder. Figure 5-21 displays the proportion of the SSI/Medicaid population with comorbid speech and language disorders and attention deficit hyperactivity disorder. 14% SSI/Medicaid Enrollees with ADHD and 11.5% Speech and Language Disorder 12% 10.1% 10% 9.5% 8.7% 8.8% 8.1% 8.1% 8% 6.9% 7.3% 6.7% 6% 4% 2% 0% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Year FIGURE 5-21  SSI/Medicaid enrollees with attention deficit hyperactivity disorder and a comorbid diagnosis of speech or language disorder. NOTE: ADHD = attention deficit hyperactivity disorder. SOURCE: MAX data. 

200 SPEECH AND LANGUAGE DISORDERS IN CHILDREN Among children who were eligible for Medicaid because of SSI program participation, 6.7 percent (n = 38,466) had comorbid attention deficit hy- peractivity disorder and speech or language disorder in calendar year 2001. This rate of comorbidity rose over time to reach 11.5 percent (n = 84,519) in calendar year 2010—roughly consonant with the 11.9 percent rate of speech and language disorders among the SSI/Medicaid population shown in Figure 5-20. These two rates differ because 11.5 percent is the rate of comorbidity, and not all children with speech and language disorders also have attention deficit hyperactivity disorder. While the committee had access to Medicaid enrollment information, it lacked data on the number and nature of services received by Medicaid ben- eficiaries with speech and language disorders. Consequently, the committee was unable to answer questions related to treatment and service utilization among those eligible for Medicaid because of SSI program participation. DATA STRENGTHS AND LIMITATIONS This chapter has relied on three different types of data: clinical data drawn from small treatment studies; population data drawn from large national surveys; and service (administrative) data drawn from the SSA, IDEA, or Medicaid. Each of these three types of data has unique strengths and limitations, as described below. The primary strength of clinical data is the high validity with which a diagnosis of speech and language disorders can be established. Because these clinical data often are derived from treatment studies, they typi- cally yield high-quality and highly reliable information on the nature of a disorder; its antecedents, concomitants, and consequences; the nature of treatment received; and the outcomes resulting from treatment. Clinical data, however, cannot be used to establish the population prevalence of a disorder unless the study sample is nationally representative. These studies usually are conducted within a narrow geographic area and are, therefore, susceptible to local influences on the identification, treatment, course, and outcome of a disorder. If clinical data are narrow and “deep dives” into a condition, popula- tion data are broad and shallow. The principal advantage of survey data is the ability to establish the rate of occurrence of a disorder in a population (the prevalence of the disorder). Because survey data cover broad geo- graphic areas, even the entire United States, they are less susceptible than clinical data to local or regional variations in identification or treatment. This breadth comes with a few drawbacks, however. Because the surveys elicit information from large numbers of respondents and the time allotted for each respondent is very limited, they usually do not probe deeply into the condition a child may have. In many instances, the data are not specific 

COMPARISON OF TRENDS 201 enough to permit reliable ascertainment of speech and language disorders, and different surveys use different ways of identifying such disorders. In contrast with clinical studies, in which the investigators follow up with patients over time, surveys are difficult and expensive to implement on a larger scale. Consequently, most national health surveys obtain information at a single point in time, and cannot be used to monitor the evolution of a disorder within an individual over time. In addition, as noted previously, these data often are based on parent or guardian reports without clinical corroboration of a child’s condition. Administrative or service data capture the entire universe of children who receive services. Like clinical data, they are good sources of knowledge regarding the nature and types of services a child receives or consumes and the providers who treat them. Like survey data, they cover large geographic areas. But service data pose some unique problems. For IDEA data, many children who receive services for speech and language disorders do not have a primary speech or language category in special education. This situation creates problems in disaggregating which children are receiving treatment specifically for speech and language disorders. The diagnoses contained in Medicaid data may not be highly reliable, and data on services can be prob- lematic as well (Bright et al., 1989). Moreover, all administrative data are sensitive to changes in rules, regulations, and funding, which may cause the number of children in the dataset to fluctuate even when there is no change in the population prevalence of the underlying disorder. The three types of data also capture speech and language disorders very differently, as illustrated earlier in Table 5-1. The National Survey of Children with Special Health Care Needs elicits from respondents impres- sions of problems with speaking, communicating, and being understood, while the National Survey of Children’s Health elicits whether a health care provider has diagnosed a fluency disorder. Other surveys focus on “speech problems” (e.g., National Health Interview Survey) without querying about language impairments. There is, consequently, great heterogeneity in the disorders that national surveys seek to identify and in the severity of condi- tions that are captured. These problems limit the extent to which national surveys can be used to answer questions about the prevalence of speech and language disorders at a national level. In addition, the questions used to ascertain speech and language disorders in these national surveys result in the identification and inclusion of children with relatively milder forms of the disorders. In contrast, children who are eligible for Medicaid because of SSI program participation are likely to have the most severe limitations. As discussed earlier, this is the case because the SSI program serves only children with “marked” and “extreme” disorders, and their Medicaid claims reflect this high level of impairment. Clinical data offer the greatest leeway in ascertaining levels of impairment. Because clinician-investigators 

202 SPEECH AND LANGUAGE DISORDERS IN CHILDREN can choose to focus on the entire range of severity of speech and language disorders, these data reflect the greatest variations in how the disorders are manifested. Given these limitations to existing data, the committee’s task was to obtain and utilize the best available data for the intended purposes of this study—minimizing the constraints of particular types of data while maxi- mizing their particular utility in answering specific questions. As a result, this chapter relies on a mix of data sources. The overall purpose was to use clinical, national survey, and administrative data in ways that would build a valid picture of what childhood speech and language disorders look like in the general population and in the SSI population. FINDINGS AND CONCLUSIONS The committee’s review of clinical, national survey, and administrative data and careful consideration of expert testimony yielded a picture of what is known about past and current trends in the prevalence of speech and language disorders among children (under age 18) in the general U.S. popu- lation and in the SSI population. Based on its review of the best available evidence, the committee reached the following findings and conclusions. Findings 5-1. Multiple data sources provide estimates of speech and language disorders in children (under age 18) in the general U.S. population. 5-2. Administrative and service data provide estimates of children who receive benefits or services for speech and language disorders through large federal programs. The SSA established the impair- ment code for speech and language disorders (3153) in 1994. 5-3. The overall prevalence of reported speech and language disorders of any severity ranges from 2 to 16 percent of the general popula- tion of children and young adults aged birth through 21 years. 5-4. Two national surveys show that boys manifest speech and lan- guage disorders approximately twice as frequently as girls. 5-5. Two national surveys show increases in the prevalence of speech and language disorders of between 26 and 56 percent over 4-year periods within the past decade. 5-6. The number of children receiving SSI benefits for a primary speech or language impairment increased by 171 percent between 2004 and 2014. However, the number of initial allowances per year (i.e., newly eligible beneficiaries) increased by only 24 percent during this time period, with the exception of a more significant 

COMPARISON OF TRENDS 203 increase (28 percent) during the recession (December 2007 to June 2009). 5-7. More than 40 percent of children with speech and language disor- ders in the general U.S. population experience comorbidities such as neurodevelopmental conditions, autism spectrum disorder, and behavioral problems. 5-8. The most commonly occurring comorbid diagnosis in the Medicaid claims of children with speech and language disorders is attention deficit hyperactivity disorder, seen in between 6.7 and 11.5 percent of children on Medicaid. 5-9. Only about one-third of children with speech and language impairments in the SSI program have a secondary impairment recorded. Among these children, autistic disorder and other devel- opmental disorders was the most commonly occurring secondary impairment category listed. 5-10. Between 8.3 and 57.1 percent of children with speech problems in the general U.S. population are reported to have unmet needs for at least one indicator of health care access or service utilization. 5-11. Administrative data from IDEA suggest that between 1.5 and 1.7 percent of all U.S. children receive special education services as a result of a primary condition of speech and/or language impairment. 5-12. Children with speech and language disorders in the general U.S. population are more likely than children without such disorders to live in poverty (26 versus 21 percent) or low-income house- holds (28 versus 23 percent). 5-13. The greatest number of child SSI recipients are those with the lowest family incomes (at or below 100 percent of the FPL). Conclusions 5-1. No single dataset or even a small set of data sources can provide definitive population estimates of the prevalence of speech and language disorders over time among U.S. children. 5-2. Children with speech and language disorders in the general U.S. population are not comparable to the population of children who receive SSI benefits for speech and language disorders. 5-3. Changes in the SSA’s program procedures, changes in eligibility and determination guidelines, cohort composition, and macroeco- nomic conditions all influence changes in the rate of identifica- tion of speech and language disorders within a given cohort of children at any point in time. 

204 SPEECH AND LANGUAGE DISORDERS IN CHILDREN 5-4. The information available on national trends in speech and lan- guage disorders in children—especially those disorders that are severe—is extremely limited. Available data suggest an increase in the overall prevalence of speech and language disorders in the general child population over the past decade. 5-5. The sharp increase in the number of SSI recipients eligible because of speech and language disorders observed between 2004 and 2014 is explained almost entirely by (1) the introduction in 1994 of a new impairment code for speech and language impairment, and (2) the marked and extreme levels of impairment among chil- dren who receive SSI benefits for speech and language disorders that are unlikely to be resolved by the time a child reaches age 18. 5-6. Children with speech and language disorders have a high likeli- hood of experiencing other health conditions as well. 5-7. The modest rates of comorbidities in the SSI data are due to inconsistencies in how secondary impairments are coded; the ab- sence of a recorded secondary impairment does not mean that the child did not actually have another impairment, only that such a secondary impairment was not recorded as part of the eligibility determination process. 5-8. Children with speech and language disorders experience signifi- cant barriers to receiving needed health care services. 5-9. The net result of suspending and terminating higher-income children from the SSI program is the concentration of children from the poorest and most vulnerable families remaining in the program. 5-10. Data from national surveys indicate that speech and language dis- orders are more common among children from families living in poverty than among children from families not living in poverty. REFERENCES Bainbridge, K. 2015. Speech problems among US children (age 3-17 years). Workshop pre- sentation to the Committee on the Evaluation of the Supplemental Security Income (SSI) Disability Program for Children with Speech Disorders and Language Disorders, May 18, Washington, DC. Black, L. I., A. Vahratian, and H. J. Hoffman. 2015. Communication disorders and use of intervention services among children aged 3-17 years: United States, 2012. NCHS data brief, no. 205. Hyattsville, MD: National Center for Health Statistics. Blumberg, S. J. 2015. Understanding speech and language data in the National Survey of Children with Special Health Care Needs. Workshop presentation to the Committee on the Evaluation of the Supplemental Security Income (SSI) Disability Program for Children with Speech Disorders and Language Disorders, May 18, Washington, DC. Boushey, H., C. Brocht, J. Bernstein, and B. Gundersen. 2001. Hardships in America: The real story of working families. Washington, DC: Economic Policy Institute. 

COMPARISON OF TRENDS 205 Bright, R. A., J. Avorn, and D. E. Everitt. 1989. Medicaid data as a resource for epidemiologic studies: Strengths and limitations. Journal of Clinical Epidemiology 42(10):937-945. CDC (Centers for Disease Control and Prevention). 2007. 2005-2006 National Survey of Children with Special Health Care Needs Datasets. http://www.cdc.gov/nchs/slaits/cshcn. htm (accessed September 30, 2015). CDC. 2012a. About the National Health Interview Survey. www.cdc.gov/nchs/nhis/about_ nhis.htm (accessed August 18, 2015). CDC. 2012b. 2009-2010 National Survey of Children with Special Health Care Needs Datasets. http://www.cdc.gov/nchs/slaits/cshcn.htm (accessed September 30, 2015). CDC. 2013a. National Survey of Children’s Health. www.cdc.gov/nchs/slaits/nsch.htm (ac- cessed August 18, 2015). CDC. 2013b. 2007 National Survey of Children’s Health. http://www.cdc.gov/nchs/slaits/nsch. htm#nsch2007 (accessed September 30, 2015). CDC. 2013c. 2011-2012 National Survey of Children’s Health. http://www.cdc.gov/nchs/ slaits/nsch.htm#nsch2011 (accessed September 30, 2015). CDC. 2014a. National Survey of Children with Special Health Care Needs. www.cdc.gov/ nchs/slaits/cshcn.htm (accessed August 18, 2015). CDC. 2014b. 2005-2006 National Survey of Children with Special Health Care Needs. http:// www.cdc.gov/nchs/slaits/cshcn.htm#2005CSHCN (accessed September 30, 2015). CDC. 2014c. 2009-2010 National Survey of Children with Special Health Care Needs. http:// www.cdc.gov/nchs/slaits/cshcn.htm#09-10 (accessed September 30, 2015). CDC. 2015. Questionnaires, datasets, and related documentation 1997 to the present. http:// www.cdc.gov/nchs/nhis/quest_data_related_1997_forward.htm (September 17, 2015). Coleman-Jensen, A., M. Nord, and A. Singh. 2013. Household food security in the United States in 2012. Washington, DC: U.S. Department of Agriculture, Economic Research Service. www.ers.usda.gov/media/1183208/err-155.pdf (accessed July 13, 2015). DRC (Data Resource Center for Child and Adolescent Health). 2012a. Frequently asked ques- tions. http://www.childhealthdata.org/learn/faq#what is NSCH (accessed April 1, 2015). DRC. 2012b. Guide to topics and questions asked: National Survey of Children’s Health 2011/12. http://www.childhealthdata.org/learn/topics_questions/2011-12-nsch (accessed September 17, 2015). DRC. 2012c. Guide to topics and questions asked: National Survey of Children with Special Health Care Needs, 2009/10. http://www.childhealthdata.org/learn/topics_ questions/3.3.0-2009-10-national-survey-of-cshcn--topics-questions (accessed September 17, 2015). Engel, J. 2006. Poor people’s medicine: Medicaid and American charity care since 1965. Durham, NC: Duke University Press. GAO (U.S. Government Accountability Office). 2012. Supplemental Security Income: Better management oversight needed for children’s benefits. GAO-12-497. http://purl.fdlp.gov/ GPO/gpo25551 (accessed July 7, 2015). HHS (U.S. Department of Health and Human Services). 2013. Child Health USA 2012. http:// mchb.hrsa.gov/publications/pdfs/childhealth2012.pdf (accessed September 2, 2015). HHS. 2015a. Annual update of the HHS poverty guidelines. Federal Register 80(14):3236-3237. HHS. 2015b. State Medicaid and CHIP income eligibility standards. www.medicaid.gov/ medicaid-chip-program-information/program-information/downloads/medicaid-and- chip-eligibility-levels-table.pdf (accessed September 2, 2015). HHS. n.d. Access to health services. www.healthypeople.gov/2020/topics-objectives/topic/ Access-to-Health-Services (accessed on September 2, 2015). HHS, HRSA (Health Resources and Services Administration), and MCHB (Maternal and Child Health Bureau). 2008. The National Survey of Children with Special Health Care Needs chartbook 2005-2006. Rockville, MD: HHS. 

206 SPEECH AND LANGUAGE DISORDERS IN CHILDREN Hoffman, H. 2015. Speech and language impairments in the U.S. pediatric population: Findings from the National Health Interview Survey (NHIS) voice, speech, and lan- guage supplement. Workshop presentation to the Committee on the Evaluation of the Supplemental Security Income (SSI) Disability Program for Children with Speech Disorders and Language Disorders, May 18, Washington, DC. Ireys, H., D. Kasprzyk, A. Takyi, and J. Gillcrist. 2004. Estimating the size and characteristics of the SSI child population: A comparison between the NSCF and three national surveys. www.ssa.gov/disabilityresearch/documents/nscf/finalcomparison/nscf-comparison-final. pdf (accessed September 2, 2015). NASEM (National Academies of Sciences, Engineering, and Medicine). 2015. Mental disorders and disabilities among low-income children. Washington, DC: The National Academies Press. National Center for Children in Poverty. 2013. Demographics of low-income children: United States. New York: Columbia University, Mailman School of Public Health. www.nccp. org/profiles/US_profile_6.html (accessed September 1, 2015). NBER (National Bureau of Economic Research). 2015. SIPP data—reading Survey of Income and Program Participation (SIPP) data, using SAS, SPSS, or Stata. http://www.nber.org/ data/survey-of-income-and-program-participation-sipp-data.html (accessed September 30, 2015). Newschaffer, C. J., M. D. Falb, and J. G. Gurney. 2005. National autism prevalence trends from United States special education data. Pediatrics 115(3):e277-e282. Olson, L. K. 2010. The politics of Medicaid. New York: Columbia University Press. SSA (Social Security Administration). 2014. Program operations manual system: Medicaid and the Supplemental Security Income (SSI) program. https://secure.ssa.gov/poms.nsf/ lnx/0501715010 (accessed July 6, 2015). U.S. Census Bureau. 2015a. CPS table creator. www.census.gov/cps/data/cpstablecrea tor.html (accessed July 13, 2015). U.S. Census Bureau. 2015b. Survey of income and program participation. http://www.census. gov/programs-surveys/sipp/about/sipp-introduction-history.html (accessed September 30, 2015). U.S. Census Bureau. 2015c. Total population by child and adult populations. http://data- center.kidscount.org/data/tables/99-total-population-by-child-and-adult#detailed/1/any/ false/869,15/39/416,417 (accessed September 23, 2015). U.S. Department of Education. 2014. 36th annual report to congress on the implementation of the Individuals with Disabilities Education Act, 2014. http://www2.ed.gov/about/reports/ annual/osep/2014/parts-b-c/36th-idea-arc.pdf (accessed June 12, 2015). U.S. Department of Education. 2015. Children and youth with disabilities. http://nces.ed.gov/ programs/coe/indicator_cgg.asp (accessed September 2, 2015). Visser, S. N., R. H. Bitsko, M. L. Danielson, R. Perou, and S. J. Blumberg. 2010. Increasing prevalence of parent-reported attention-deficit/hyperactivity disorder among children— United States, 2003 and 2007. Morbidity and Mortality Weekly Report 59(44):1439-1443.