Keywords
developmental language disorder - Spanish-speaking children - monolingual Spanish
speakers - prevalence
Learning Outcomes: As a result of this activity, the reader will be able to:
-
Understand the scope and magnitude of DLD in children aged 4;0 to 6;11 years in Mexico.
-
Emphasize the significance of employing rigorous and validated diagnostic tools to
accurately identify children with DLD.
-
Raise awareness of DLD to inform the health and educational authorities to establish
systems for early identification and diagnosis of children with DLD.
Developmental language disorder (DLD), formerly known as specific language impairment
(SLI), is a language condition characterized by deficits with grammar, phonology,
semantics, syntax, and/or pragmatics, in the absence of cognitive, motor, neurological,
or hearing problems (Leonard 2014; Bishop & Leonard 2000; Kapantzoglou et al. 2015).
Difficulties may change over time with age (Conti-Ramsden & Botting 1999; Heilmann
et al. 2010), as well as the severity and impact on daily life. The rate of prevalence
refers to the proportion of individuals within a specific sample who have a particular
condition such as DLD. The prevalence of DLD can vary across different countries and
populations due to various factors, including diagnostic criteria, cultural differences,
and access to healthcare services. In Mexico, it is estimated that 15% of people,
including children and adults, have a communication and/or language disability, according
to the Instituto Nacional de Estadística, Geografía e Informática (INEGI 2020). The
estimation of people with communication disorders has dramatically increased from
2010 (8.3%) to 2020 (15%). The majority of children in Mexico (93%) are reported to
be monolingual Spanish speakers. Among individuals who have a communication and/or
language disability, 3.2% are children between 3 and 4 years of age, and 14.8% are
children between 5 and 9 years of age.
The prevalence of communication disorders can vary depending on the questions asked
about language concerns. For instance, the INEGI (2020) included a single question
about language use in their surveys and census: “Do you have difficulties speaking,
communicating, or having a conversation in your everyday life?” This question encompasses
all types of speech and language deficits, regardless of their etiology or age of
onset. However, the prevalence of specific language disorders, such as DLD in monolingual
Spanish-speaking children in Mexico, has yet to be estimated. The present study, although
not an epidemiological study, conducted secondary analyses to obtain a first estimate
of the prevalence of diagnosed DLD in Mexico based on data from three different cities
of Mexico: Queretaro, Mexico, and Monterrey.
Prevalence of Developmental Language Disorder
Prevalence of Developmental Language Disorder
The estimated prevalence of diagnosed DLD varies across different countries and populations.
For instance, studies on monolingual English-speaking children from the United States
and England indicate that approximately 7.5% of children 4 to 6 years of age present
with the disorder (Norbury et al. 2016; Tomblin et al. 1997). In Australia, McLeod
and Harrison (2009) estimated a prevalence of 13 to 14.7% in 4- to 5-year-old children,
and Calder et al. (2022) found a prevalence of 6.4% in 10-year-old children. In Finland,
Hannus et al. (2009) reported that 2.5% of 0- to 6-year-old children presented with
DLD. In China, Wu et al. (2023) found a prevalence of 8.5% in children aged 5 to 6
years. Large variability in outcomes is also noted across studies that have been conducted
in developing countries. For example, Tchoungui Oyono et al. (2018) found that 4.3%
of 3- to 5-year-old French-speaking children in Cameroon had DLD; Melchiors Angst
and colleagues (2015) reported a similar prevalence of 4.58% in 4- to 6-year-old children
in Brazil; Oryadi-Zanjani et al. (2015) found a prevalence of 3.3% in 5-year-old Persian-speaking
children of Iran; and Gad-Allah et al. (2012) reported a prevalence of 19.7% in Egypt
in 3- to 5-year-old children.
There is a dearth of studies on the prevalence of DLD in Spanish-speaking countries
in Central or Latin America (De Barbieri et al. 1999; Villanueva et al. 2008). Chile
has been at the forefront of identifying children with DLD for many years. In the
study of Barbieri et al. (1999), 316 children aged between 1 and 7 years were evaluated
using an expressive communication developmental scale and specific diagnostic tools
for different language development levels. The authors found that 6.8% of children
presented with some type of language impairment, with a higher incidence in males
(66.4% of cases).
There is a variety of factors that affect the prevalence estimates including changes
in the definition of language disorders, variability in the characteristics of the
populations under study, differences in measurement of language abilities, and other
methodological characteristics of the studies conducted. Regarding the definition,
in recent years, the term “DLD” has been adopted based on a consensus study involving
a panel of experts (Bishop et al. 2017). These researchers emphasized the importance
of using the same label—DLD—for a complex and multifactorial disorder. Several years
ago, Bishop (2014) suggested maintaining the term “specific language impairment” with
the understanding that “specific” implies an unknown etiology rather than a pure impairment
that does not imply any other cognitive deficit (e.g., working memory). Moreover,
the DSM-V term “language disorder” remains confusing because it is wide in its definition,
and boundaries among disorders are diffuse. It is possible that a substantial misclassification
of children with expressive language disorders, including DLD, is taking place in
different countries, including Mexico, and variability in prevalence outcomes is partially
affected by the lack of consistency in the definition. Researchers working with Spanish-speaking
children have now adopted the term DLD to promote consistency and clarity in discussing
the condition (e.g., Andreu et al. 2021; Castilla-Earls et al. 2021), and for the
present study we also adopted the same term.
In addition, the studies conducted across different countries vary with respect to
the age, sex, and sociocultural characteristics of the participants, which contribute
to the variability in the findings. For example, Tchoungui Oyono et al. (2018) found
that the prevalence of DLD ranged from 2.6 to 7.1% for the three age groups included
in the study (3–5 years of age) with the highest prevalence observed in 5-year-old
children, whereas Barbieri et al. (1999) found higher frequency of language impairments
between 3 and 4 years of age (66.4% of the cases) than at 5 years of age (24.2% of
cases) and at younger than 3 (9.4% of cases). Furthermore, there is relative consistency
in findings across studies that the prevalence of the disorder is higher in male than
in female participants (e.g., Calder et al. 2022; De Barbieri et al. 1999; Norbury
et al. 2016; Tomblin et al. 1997). Regarding the sociocultural characteristics of
the groups examined, some studies used a nationally representative sample or stratified
sampling from urban, suburban, and rural areas to account for the socio-demographic
conditions in different residential areas (e.g., McLeod & Harrison 2009; Norbury et
al. 2016; Tomblin et al. 1997), whereas other studies used sample from a single city
(e.g., De Barbieri et al. 1999; Hannus et al. 2009; Tchoungui Oyono et al. 2018).
In addition, it is important to consider that the particular sociocultural characteristics
of the populations examined are likely to affect children's scores on the measures
used (e.g., Barragan et al. 2018); therefore, sociocultural differences across studies
are likely to affect differentially the scores and outcomes on the prevalence of the
disorder across the different populations (Kapantzoglou et al. 2016). In the present
study, the sample was not stratified but collected from three different urban areas
in Mexico, including children who attended public and private schools. Therefore,
besides estimating the prevalence of DLD for each age and sex group separately, the
study also examined the effects of the sociocultural characteristics of the sample—maternal
education and the type of school children attended—on the prevalence estimates. Finally,
identification of DLD in the present study was based on measures with validity evidence
for their use with the target population.
There is also a discussion in the literature regarding possible differences in the
prevalence of the disorder between developing and high-income countries. The rationale
behind this is that children living in socioeconomically disadvantaged situations
might be more prone to develop difficulties acquiring language (Black et al. 2008;
Hoff 2003; Jensen et al. 2017), which may be more pronounced in children with DLD
(Auza & Peñaloza 2019; Lara-Díaz et al. 2021; Norbury et al. 2021). For instance,
Norbury et al. (2021) found that the predicted probability of language disorder was
2.5 greater at the 10th percentile of Income Deprivation Affecting Children Index
(McLennan et al. 2010). The findings are mixed with studies suggesting that prevalence
rates vary across developing countries (e.g., Tchoungui Oyono et al. 2018; Gad-Allah
et al. 2012), and the present study will be the first to provide evidence from Mexico.
Prevalence studies also vary widely in their diagnostic approach for DLD and other
methodological characteristics. Specifically, some studies used direct assessment
that included a diagnostic battery with a variety of measures such as hearing screening,
parent interviews, standardized tests, and language sample analyses (e.g., Tomblin
et al. 1997), whereas in other studies, language assessment was based on a single
language test (e.g., Calder et al. 2022) or on indirect measures, such as parent and
teacher questionnaires (Gad-Allah et al. 2012). In some cases, the prevalence estimates
were based on retrospective data (Hannus et al. 2009). There is also variability in
the cut scores applied across studies for identifying DLD. Given lack of standard
cut scores, depending on the types of tests conducted and researchers' related decisions,
cut scores can range from 1 to 2 standard deviations below the mean (e.g., Calder
et al. 2022; Norbury et al. 2016; Tomblin et al. 1997). In some cases, a re-norming
process was followed, in which the study sample served as its own reference given
that the characteristics of the standardization sample in the test did not match the
characteristics of the target population (e.g., Tchoungui Oyono et al. 2018). Finally,
as expected, the sample size also varies widely across different studies and in some
studies the sample is population based, whereas in other cases it is clinically based
(e.g., De Barbieri et al. 1999; Hannus et al. 2009).
Screening and Diagnosing Developmental Language Disorder in Mexico
Screening and Diagnosing Developmental Language Disorder in Mexico
A main issue concerning the identification of DLD in Mexico is the diagnostic terminology
that has been used over time. For over three decades, expressive language disorders
in Mexico have been identified in several public institutions with a nomenclature
based on adult neurological deficits, such as “retardo anártrico” (anarthric delay)
and “retardo anártrico-afásico” (anarthric–aphasic delay) (Azcoaga et al. 1987). These
terms, as well as “developmental dysphasia” (Parise & Maillart 2009), are still used,
especially in clinical settings, although the international nomenclatures in the medical
classification system, such as the ICD-11 (F80.9) and the DSM-V, “language disorder,”
have become more common among clinicians and researchers globally. When obsolete terminology
is still used, it can potentially contribute to misdiagnosis because without clear
definitions, varied interpretations among healthcare professionals are possible. Many
clinicians still use older terms as synonyms and are less aware of the most recent
term “DLD” (Bishop et al. 2017). The accuracy in describing the condition of interest
(in this case DLD), as well as in the collection and analysis of data, are key components
in the outcomes of the prevalence of DLD (Hannus et al. 2009). For instance, the disorder
has been defined in various ways, such as language disorder with unspecified or of
unknown origin (Barbieri et al. 1999; McLeod & Harrison 2009; Norbury et al. 2016),
specific language disorder (Hannus et al. 2009; Tomblin et al. 1997), or DLD (Calder
et al. 2022). Changes in terminology within the field may lead to under-identification
of DLD if there is inconsistency in how terms are used or understood. If the term
has changed over the years and new terminology is introduced without a clear definition
and/or criteria, it may result in overlooked or incorrect diagnosis.
Low awareness of DLD in Mexico is another main issue for the accurate and early identification
of the disorder, even among the professional community of speech-language pathologists.
There are no established procedures in place in the country, in schools or medical
settings, to screen children for DLD. Parents typically seek help when they are concerned
about their children's development, particularly when they observe speech delays rather
than difficulties with grammar use. The Encuesta Nacional de Salud y Nutrición (National
Survey of Health and Nutrition—ENSANUT—) indicated that 30.8% of boys and 27.8% of
girls aged between 2;0 and 9;0 years were identified as being at risk of experiencing
a developmental disorder. This risk is defined as the presence of a disorder in any
of the neurodevelopmental domains, including motor skills, cognition, language, and
personal-social development (Romero-Martínez et al. 2012). The parents who participated
in the survey observed problems in the cognitive and communicative domains. Specifically,
parents reported difficulties with talking, understanding (classified in this survey
as cognitive deficits), or maintaining a conversation to be the main issues in 13.3%
of children between 2;0 and 5;0 years of age, and in 11.5% of children between 6;0
and 9;0 years of age (Romero-Martínez et al. 2012).
Finally, another main challenge in identifying DLD is that the instruments for screening
and diagnosing monolingual Spanish-speaking children in Mexico, as in other developing
countries, are scarce. Nevertheless, recently there has been some improvement in identification
methods due to the emergence of certain tests. Specifically, for children less than
5 years of age, the Evaluación del Desarrollo Infantil/Child Development Assessment
(EDI; Rizzoli-Córdoba et al. 2014) can be used for the identification of neurodevelopmental
delays, including language delays. For children older than 4 years, there are also
some options. Recently, in 2018, the first screening tool for identifying monolingual
Spanish-speaking children with a risk of presenting grammatical disorders was published
in Mexico (Auza Benavides et al. 2018b, 2018c). Also, other tests such as the Bilingual
English-Spanish Language Test (BESA; Peña et al. 2014), and the Spanish Clinical Evaluation
of Language Fundamentals – Fourth Edition, Spanish (CELF-4; Semel et al. 2006) that
have been standardized on Spanish-English speaking populations in the United States,
are also used to diagnose DLD. However, the efficacy of the detection system could
be enhanced through the implementation of a systematic methodology for language screening
and assessment. Moreover, further investigation is needed regarding the diagnostic
standards that are appropriate for the characteristics of local populations in Latin
America, to identify children with DLD accurately and efficiently. Currently, the
gold standard for language assessments is the use of converging evidence from various
methods of assessment (Auza Benavides et al. 2018b; Barragan et al. 2018; Castilla-Earls
et al. 2020) and this is the approach followed in the present study as well.
Studying the prevalence of children with DLD in Mexico is crucial not only for understanding
the rate of affected individuals but also for evaluating the consequences of our assessment
protocols and establishing a system to early identify and diagnose DLD, as well as
for gathering further evidence regarding possible differences in the prevalence of
DLD between developing and high-income countries. Given so, the purpose of this study
was to obtain a first estimate of the prevalence of DLD in Mexico based on secondary
analyses. Specifically, the study aimed to answer the following research questions:
-
What is the overall estimate of prevalence of children presenting with DLD?
-
What is the estimate of prevalence of children presenting with DLD at different ages?
-
What is the estimate of prevalence of boys and girls presenting with DLD?
Method
A cross-sectional retrospective study was conducted to investigate the prevalence
of children with DLD in a sample of monolingual Spanish-speaking children aged between
4;0 and 6;11 years. The study recruited children from 137 public and private schools
across three cities in Mexico: Queretaro, Mexico, and Monterrey. Parents of the children
were invited to participate in the study through open and free talks delivered in
schools and public health centers in the three cities. During these talks, information
was provided about language development and disorders, and it was emphasized that
anyone could participate, whether they were concerned about their child's language
development or not.
Participants
The initial sample for this study consisted of 872 children who were eligible for
inclusion in a larger study that validated a screener instrument (Auza Benavides et
al. 2018b). Four children were excluded from the sample because they were native speakers
of languages other than Spanish. Of the remaining 868 children, parental consent was
not provided for 118 children, who either withdrew from the study or did not complete
testing. In some cases, children did not attend school on the days of testing or were
taken out of school. Additionally, some parents did not provide the necessary biological
and sociocultural information on the initial questionnaire (Auza et al. 2023), which
was crucial for sample selection (Auza et al. 2019). A total of 38 children were excluded
during testing due to other conditions such as neurological impairment, hearing loss,
intellectual disability, autism spectrum disorder, or a psychiatric condition. This
left us with a sample of 712 children. However, since this sample came from a larger
one, we did not include the 115 children who were tested in hospitals or clinics,
as this may have led to potential referral bias and inflated prevalence rates in the
community. Therefore, only children attending public or private schools were included
in this sample. Another 54 children were ruled out of the sample because they did
not complete the whole language assessment. In total, 543 monolingual Spanish-speaking
children aged 4 to 6 years participated in this study (typical language development
[TLD]: N = 497, M
age = 63, SD = 9.22; DLD: N = 46, M
age = 62, SD = 9.81). At the time of testing, all the 46 children diagnosed with DLD
had undergone language services for duration of less than 2 months, even children
aged 6 who were late diagnosed. None of the children had any contact with indigenous
languages. Additionally, none of the participants had a history of hearing loss, sensorimotor
or neurological problems, severe psychological disorders, or health problems, as reported
by their parents. In cases where there was a concern about any of these issues, children
were referred for evaluation, and if the problem was clinically confirmed, they were
excluded from the study.
The study included children with TLD and children with DLD. For all children, their
non-verbal IQ score was 80 or above on the Kaufman Assessment Battery for Children—2nd
edition (KABC-2; Kaufman & Kaufman 2004). Children with TLD met two of the following
three criteria: (1) the number of grammatical errors per T-unit in their language
sample was less than 20% (Restrepo 1998); (2) the mean length of utterance in words
(MLUw) was age appropriate (Simón-Cereijido & Gutiérrez-Clellen 2007); and (3) children
scored at or above the cut score on the Tamiz de Problemas del Lenguaje (TPL), which
included a morphology subtest and a sentence repetition subtest (Auza et al. 2018c).
Children with DLD met two of the following three criteria: (1) the percentage of grammatical
errors per T-unit in their language sample was 20% or more; (2) the MLUw was low for
their age (Simón-Cereijido & Gutiérrez-Clellen 2007); and (3) children scored below
the cut score on the TPL which included a morphology subtest and a sentence repetition
subtest (Auza et al. 2018c).
General Procedures
Initially, an information session was organized for parents and teachers to apprise
them of the significance of our study. During this session, we provided information
on TLD and discussed potential red flags that indicate the likelihood of a language
disorder in children. Parents willing to participate in the study completed the parent
consent form. The KABC-2 non-verbal subtests were administered to rule out cognitive
impairment. As a third step, the TPL (Auza et al. 2018c) and a language-sample were
administered as a comprehensive assessment. The language-sample was used to obtain
the MLUw and the percentage of grammaticality. All measures were administered by trained
graduate students in linguistics or experienced SLPs in two or three distinct sessions
during the same week at school. In each session, all tasks were administered in a
randomized order. The test administrators were blind to the children's language status.
The trained examiners conducted all the sessions in a quiet area at the children's
school. Approximately 60% of all tests were independently scored by a second rater
with a target agreement of 90%. In the case of any disagreements, the first author
and the graduate students resolved them by consensus.
Measures
The KABC-2
The nonverbal subtests of the KABC-2 were administered to rule out cognitive impairment.
The subtests used in the study varied according to the age of the participants and
encompassed a combination of several tasks, such as the following: conceptual thinking,
face recognition, non-verbal story completion with images, triangles, block counting,
pattern reasoning, and hand movements.
Grammaticality in Language Samples
The children completed a story-retell task in Spanish using picture and a script support,
utilizing the books “Si le das una galletita a un ratón/If You Give a Mouse a Cookie”
(Numeroff 1985) and “Con una rana es suficiente/One Frog Too Many” (Mayer & Mayer
1975). The books are similar in length, they are wordless, have multiple episodes,
and they are appropriate for the target age group. The children retold one of the
stories after listening to the script, supported by accompanying pictures. In terms
of story equivalence, previous reports have indicated no significant difference in
the measures obtained from these two narratives, specifically in terms of MLUw and
the percentage of ungrammaticality (Auza et al. 2018a). Consequently, 33.2% of the
entire sample of children retold “If You Give a Mouse a Cookie,” while 66.8% retold
“One Frog Too Many.” Within this total, 39.2% of children with DLD retold “If You
Give a Mouse a Cookie,” and 60.8% retold “One Frog Too Many.” The narratives were
transcribed, and the MLUw and grammatical errors were coded using the Systematic Analysis
of Language Transcripts (SALT; Miller & Iglesias 2010). Grammatical errors included
omissions, substitutions, additions, and word order errors. For instance, an example
of an omission (*) of an obligatory preposition is: “Corrieron *con-prep zapatos/(They ran *with-prep shoes”). An example of an article substitution [Subs] is “El niño se enojó con la[Subs] (Feminine article) sapo/The boy got angry with the (Feminine article) toad.” An example of an addition [Adic] is: “la (Feminine clitic) estaba buscándola- (Feminine clitic) [Adic]/(he) was looking
for her-(Feminine clitic) [Adic].” Semantic, phonological, or cohesive errors were excluded from the coding process.
The percentage of ungrammatical sentences (number of sentences with grammatical errors
divided by the total number of complete and intelligible sentences/terminal units
[TU]) was computed. The MLUw was automatically obtained using SALT and was based on
the complete sentences, children produced in the sample. All transcribers were graduate
students in linguistics and received training in linguistic analysis to code grammatical
errors. The inter-rater agreement for TU was 91%, and for percentage of ungrammatical
sentences it was 90%.
TPL
This test comprises two grammatical tasks: a morphology cloze task and a sentence
repetition task (Auza Benavides et al. 2018b; Auza et al. 2018c). These tasks are
designed to target vulnerable grammatical elements that have been identified as strong
indicators of DLD in monolingual Spanish-speaking children (e.g., Bedore & Leonard
2001, 2005; Morgan et al. 2009, 2013; Simón-Cereijido & Gutiérrez-Clellen 2007). The
technical manual provides cut scores for children between 3;0 and 6;11 years of age.
Sensitivity ranges between 74.6 and 88.9%, and specificity ranges between 92.1 and
95.0% across age groups. The TPL test also provides information for classifying children
with a high probability (“red light”) or a low probability (“yellow light”) of having
DLD. Children who score below the 16th percentile are considered to have a DLD.
Statistical Analysis
In this study, we initially characterized the socio-demographic and language development
profiles of children with DLD and TLD across three age groups—3, 4, and 5 years old.
We then compared these variables between the DLD and TLD groups. For continuous variables,
we utilized Welch's t-test for comparison (Delacre et al. 2017; Skovlund & Fenstad 2001; Ruxton 2006),
and Cohen's d was employed to determine effect sizes. Categorical variables were examined using
the Chi-square test (Kroonenberg & Verbeek 2018), with the Phi coefficient for effect
size assessment.
Given the sampling design, the proportion of DLD observed in our sample should not
be construed as representative of the Mexican infant population; however, it offers
a preliminary estimate. We present this proportion with 95% confidence intervals.
Statistical significance was established at a p-value below 0.05. Effect sizes were interpreted using Cohen's benchmarks: a Cohen's
d of 0.8, 0.5, and 0.2 (absolute values) signified large, medium, and small effects,
respectively. The Phi index followed a similar scale, with 0.5, 0.3, and 0.1 (only
positive values) representing large, medium, and small effects, respectively (Cohen
1988).
To answer the first and second research questions, we calculated the overall estimate
of prevalence of having DLD and the estimate of prevalence for each of the three age
groups (4, 5, and 6 years), using a logistic regression model. Also, logistic regressions
were used to calculate the estimate of prevalence of boys and girls presenting with
DLD. Odd ratios with 95% confidence intervals and a significance level of 0.05 were
used. All statistical analyses were performed using R Project for Statistical Computing.
Results
[Table 1] presents the characterization of children with DLD and TLD by their socio-demographic
and language development profiles, along with a comparison of these group profiles.
The table also provides the same analysis for each of the three age groups.
Table 1
Demographic and sociocultural characteristics of children with and without DLD
|
Total, n = 543
|
DLD
n = 46
|
TLD
n = 497
|
ES (95% CI)
|
p
|
|
Socio-demographic characteristics
|
|
Age [month], mean (SD)
|
60.5 (9.1)
|
62.5 (9.3)
|
−0.22 (−0.52, 0.08)
|
0.15
|
|
Sex [male], n (%)
|
33 (72%)
|
253 (51%)
|
0.12 (0.05, 0.20)
|
0.007
|
|
Maternal education [year], mean (SD)
|
10.7 (4.6)
|
12.1 (4.2)
|
−0.32 (−0.65, 0.01)
|
0.06
|
|
Types of school [public], n (%)
|
24 (52%)
|
192 (39%)
|
0.08 (0.00, 0.17)
|
0.08
|
|
Diagnostic tools
|
|
K-ABC [score], mean (SD)
|
100.5 (10.2)
|
104.0 (11.3)
|
−0.31 (−0.58, −0.03)
|
0.01
|
|
Language samples
|
|
MLU [by no. of words], mean (SD)
|
6.67 (3.81)
|
8.14 (2.99)
|
−0.49 (−0.87, −0.11)
|
0.03
|
|
PU [percentage], mean (SD)
|
52.0 (26.0)
|
12.9 (11.5)
|
3.39 (2.70, 4.08)
|
<0.001
|
|
Screening task
|
|
TPL [percentage], mean (SD)
|
47.1 (19.7)
|
87.9 (10.4)
|
−3.92 (−4.52, −3.31)
|
<0.001
|
|
TPL [red], n (%)
|
42 (91%)
|
9 (2%)
|
0.85 (0.77, 0.94)
|
<0.001
|
|
4-year-olds, n = 224
|
n = 24
|
n = 200
|
|
|
|
Socio-demographic characteristics
|
|
Age [month], mean (SD)
|
53.3 (3.5)
|
53.5 (3.4)
|
−0.09 (−0.52, 0.35)
|
0.39
|
|
Sex [male], n (%)
|
17 (71%)
|
92 (46%)
|
0.15 (0.07, 0.29)
|
0.022
|
|
Maternal education [year], mean (SD)
|
11.6 (4.1)
|
12.4 (4.1)
|
−0.19 (−0.61, 0.24)
|
0.39
|
|
Types of school [public], n (%)
|
11 (46%)
|
59 (30%)
|
0.11 (0.00, 0.25)
|
0.10
|
|
Diagnostic tools
|
|
K-ABC [score], mean (SD)
|
99.6 (10.4)
|
104.0 (11.0)
|
−0.40 (−0.81, 0.01)
|
0.061
|
|
Language samples
|
|
MLU [by no. words], mean (SD)
|
6.06 (4.17)
|
7.29 (2.56)
|
−0.48 (−1.15, 0.19)
|
0.17
|
|
PU [percentage], mean (SD)
|
53.9 (25.5)
|
12.8 (12.0)
|
3.44 (2.51, 4.36)
|
<0.001
|
|
Screening task
|
|
TPL [percentage], mean (SD)
|
41.2 (19.9)
|
83.7 (12.3)
|
−3.44 (−4.18, −2.70)
|
<0.001
|
|
TPL [red], n (%)
|
22 (92%)
|
5 (3%)
|
0.85 (0.72, 0.98)
|
<0.001
|
|
5-year-olds, n = 209
|
n = 13
|
n = 196
|
|
|
|
Socio-demographic characteristics
|
|
Age [month], mean (SD)
|
63.7 (3.1)
|
64.3 (3.3)
|
−0.19 (−0.72, 0.35)
|
0.51
|
|
Sex [male], n (%)
|
8 (62%)
|
109 (56%)
|
0.03 (0.00, 0.18)
|
0.68
|
|
Maternal education [year], mean (SD)
|
9.4 (5.8)
|
12.6 (4.1)
|
−0.79 (−1.57, 0.00)
|
0.035
|
|
Types of school [public], n (%)
|
7 (54%)
|
63 (32%)
|
0.11 (0.00, 0.26)
|
0.11
|
|
Diagnostic tools
|
|
K-ABC [score], mean (SD)
|
103.0 (10.7)
|
103.5 (12.0)
|
−0.04 (−0.55, 0.47)
|
0.87
|
|
Language samples
|
|
MLU [by no. words], mean (SD)
|
7.53 (3.83)
|
8.51 (3.06)
|
−0.32 (−1.01, 0.38)
|
0.38
|
|
PU [percentage], mean (SD)
|
42.1 (28.3)
|
11.8 (10.6)
|
2.87 (1.38, 4.36)
|
0.002
|
|
Screening task
|
|
TPL [percentage], mean (SD)
|
56.2 (14.1)
|
90.1 (8.2)
|
−4.13 (−5.16, −3.10)
|
<0.001
|
|
TPR [red], n (%)
|
12 (92%)
|
2 (1%)
|
0.88 (0.75, 1.00)
|
<0.001
|
|
6-year-olds, n = 110
|
n = 9
|
n = 101
|
|
|
|
Socio-demographic characteristics
|
|
Age [month], mean (SD)
|
75.0 (3.0)
|
76.8 (3.7)
|
−0.49 (−1.06, 0.07)
|
0.11
|
|
Sex [male], n (%)
|
8 (89%)
|
52 (51%)
|
0.21 (0.00, 0.40)
|
0.031
|
|
Maternal education [year], mean (SD)
|
10.2 (3.9)
|
10.4 (4.4)
|
−0.04 (−0.66, 0.58)
|
0.90
|
|
Types of school [public], n (%)
|
6 (67%)
|
70 (69%)
|
0.02 (0.00, 0.20)
|
0.87
|
|
Diagnostic tools
|
|
K-ABC [score], mean (SD)
|
99.0 (9.1)
|
104.5 (10.8)
|
−0.53 (−1.11, −0.06)
|
0.11
|
|
Language samples
|
|
MLU [by no. words], mean (SD)
|
7.07 (2.68)
|
9.12 (3.21)
|
−0.64 (−1.22, −0.06)
|
0.056
|
|
PU [percentage], mean (SD)
|
61.3 (21.3)
|
15.3 (12.2)
|
3.77 (2.49, 5.04)
|
<0.001
|
|
Screening task
|
|
TPL [percentage], mean (SD)
|
49.6 (22.4)
|
91.9 (6.5)
|
−6.53 (−8.96, −4.07)
|
< 0.001
|
|
TPL [red], n (%)
|
8 (89%)
|
2 (2%)
|
0.83 (0.65, 1.00)
|
< 0.001
|
Abbreviations: K-ABC, Kauffman cognitive subtests; MLU, mean length of utterances;
PU, percentage of ungrammatical utterances; TPL, Tamiz de Problemas del lenguaje;
ES, effect of size; for quantitative variables: Glass's Δ (interpretation <0.2: null;
0.2–0.5, small; 0.5–0.8: medium; >0.8 large). For qualitative variables: ϕ index (interpretation
<0.1: null; 0.1–0.3, small; 0.3–0.5: medium; >0.5 large).
The socio-demographic variables reveal modest differences: the DLD group was slightly
younger than the TLD group (60.5 vs. 62.5 months, Cohen's d = −0.22, p = 0.15), had a higher proportion of girls (72 vs. 51%, Phi = 0.12, p = 0.007), and had mothers with fewer years of education (10.7 vs. 12.1 years, Cohen's
d = −0.32, p = 0.06). Furthermore, a larger percentage of children with DLD attended public schools
(52 vs. 39%, Phi = 0.12, p = 0.08). In terms of non-verbal cognitive abilities, differences in K-ABC scores,
albeit small, were statistically significant (100.5 vs. 104.0, Cohen's d = −0.31, p = 0.01). In the analysis of the language development profile, the DLD group demonstrated
lower performance across all measured variables compared to the TLD group. Specifically,
the quantification of ungrammatical sentences as a percentage and the total percentage
of language proficiency (TPL) exhibited markedly vast disparities, at 52.0 versus
12.9% and 47.1 versus 87.9%, respectively. The trends in the distribution of these
variables were consistent when analyzed by age group. A detailed inspection indicates
that, for 5-year-olds, the discrepancy in the proportion of girls between the DLD
and TLD groups was less pronounced (62 vs. 56%), while the difference in maternal
education levels was more pronounced (9.4 vs. 12.6 years). In contrast, the K-ABC
scores showed negligible differences (103.0 vs. 103.5). There was a modest divergence
in MLUw (7.53 vs. 8.51), and for 6-year-olds, there was a slight difference in maternal
education levels (10.2 vs. 10.4 years), and the percentage of children attending public
schools (67 vs. 69%).
Regarding our first research question about the overall estimate of prevalence of
children presenting with DLD, results suggested an 8.5%. Regarding our second research
question about the estimate of prevalence of children presenting with DLD at different
ages, we found 10.7% for 4-year-old children, 6.2% for 5-year-old children, and 8.2%
for 6-year-old children. This showed no trend, such as a monotonic increase or decrease,
across the ages. These results are summarized in [Table 2].
Table 2
Global and age-stratified relative frequency of children presenting with DLD
|
Age group
|
DLD frequency [95% CI] (%)
|
ES (95% CI)
|
p
|
|
All
|
Male
|
Female
|
|
Global
|
8.5 [6.4, 11.1]
|
11.5 [8.3, 15.8]
|
5.1 [3.0, 8.5]
|
0.12 (0.05, 0.20)
|
0.007
|
|
4-y-olds
|
10.7 [7.3, 15.4]
|
15.6 [10.0, 23.6]
|
6.1 [3.0, 12.0]
|
0.15 (0.07, 0.29)
|
0.022
|
|
5-y-olds
|
6.2 [3.7, 10.3]
|
6.8 [3.5, 12.9]
|
5.4 [2.3, 12.1]
|
0.03 (0.00, 0.18)
|
0.68
|
|
6-y-olds
|
8.2 [4.4, 14.8]
|
13.3 [6.9, 24.2]
|
2.0 [0.1, 10.5]
|
0.21 (0.00, 0.40)
|
0.031
|
For the third question, concerning the estimates of DLD prevalence by sex, a higher
prevalence was observed among boys overall, and specifically among 4- and 6-year-olds,
with differences from girls reflecting small effect sizes that were statistically
significant. In the sample of 5-year-olds, sex differences were seldom found.
Discussion
This study is the first in Mexico to investigate the prevalence of DLD in monolingual
Spanish-speaking children between 4;0 and 6;11 years old. The study's findings indicated
an overall estimated prevalence of 8.5%, with specific estimates of 10.7% for 4-year-old
children, 6.2% for 5-year-old children, and 8.2% for 6-year-old children. This overall
prevalence in our study, while somewhat higher, closely aligns with the prevalence
of the disorder in monolingual English-speaking children from the United States and
England, which is approximately 7.5%, as established in prior research conducted by
Norbury et al. (2016) and Tomblin et al. (1997). Importantly, all three studies encompassed
samples of children aged 4 to 6 years, included methodologies to account for socio-demographic
variables in various residential areas, and employed a range of assessment tools for
identifying DLD. Furthermore, all three studies consistently observed a higher prevalence
of the disorder among male participants, a trend consistent with the broader literature.
However, it must be clarified that this study is a retrospective study based on data
collected in three different Mexican cities—Queretaro, Mexico, and Monterrey—whereas
studies of Norbury et al. (2016) and Tomblin et al. (1997) were based on probabilistic
sampling from a population. The overall prevalence estimate of children with DLD in
our current study stands significantly higher than figures reported in various other
regions. For instance, in Finland, a sample of children aged 0 to 6 years yielded
a prevalence of 2.5% (Hannus et al., 2009), and in Cameroon, a sample of French-speaking
children aged 3 to 5 years yielded a prevalence of 4.3% (Tchoungui Oyono et al. 2018).
In Chile, research involving children aged 1 to 7 years indicated a prevalence rate
of 6.8% (De Barbieri et al. 1999), and in Iran, a study involving 5-year-old Persian-speaking
children reported a prevalence of 3.3% (Oryadi-Zanjani et al., 2018). Notably, these
studies involved on average younger age groups than the current study. There is limited
research that supported lower prevalence figures in older children (e.g., Calder et
al. 2022; 6.4%). While the numeric estimates vary considerably in these studies with
younger groups than our participants, they might be collectively suggesting potential
challenges in identifying DLD in younger ages, likely due to the less advanced stage
of language development at such ages.
Regarding within study comparisons across ages, in the present study, the highest
prevalence estimate was found among the 4-year-old children (10.7%) and the lowest
in the 5-year-old group (6.2%). These results are consistent with Barbieri et al.
(1999) who also found higher frequency of language impairments, in Chile, between
3 and 4 years of age than at 5 years of age and the lowest frequency was identified
at ages younger than 3, which our study did not include. Results are also consistent
with Tchoungui Oyono et al. (2018) who found that the prevalence of DLD in French-speaking
children in Cameroon was similar to that in our study (7.1%) for 5-year-old children;
however, prevalence in their 3- and 4-year-old groups was significantly lower (3.3
and 4.2%). Our results were also similar to a study with 5- and 6-year-old Mandarin-speaking
children, although their methodology was different from ours (Wu et al. 2023). Higher
prevalence of 4-year-olds with DLD in our study could be due to differences in the
identification measures across studies.
Drawing comparisons between estimates to gauge the proportion of DLD across different
studies presents several inherent challenges. These challenges arise from a lack of
consensus in clinical criteria, variations in assessment methods, differences in the
age of children studied, limited knowledge about the disorder itself, diverse nomenclature,
and the scarcity of standardized tests, among other complexities. In the present study,
we addressed these challenges by employing a combination of assessment tools specifically
tailored to our context. We utilized a grammatically standardized norm-referenced
test developed in Mexico, along with two commonly accepted measures based on language
sample analyses, namely, MLUw and the percentage of ungrammatical sentences. It is
recognized that combining standardized measures of morphosyntax with MLUw and the
assessment of ungrammatical utterances can provide valuable diagnostic information
(Bedore et al. 2010; Eisenberg & Guo 2016; Gutiérrez-Clellen & Simon-Cereijido 2009).
The results of the present study differ substantially from those reported by the National
Survey of Health and Nutrition in Mexico. Specifically, the National Survey of Health
and Nutrition in Mexico has reported a range of 27.8 to 30.8% for communication disorders
among children aged 2 to 9 years (Romero-Martínez et al. 2012). However, this data
did not include specific information on the proportion of children diagnosed with
DLD. This disparity can be attributed to the broad classification criteria for language
disorders and the wide age range considered in their analysis. In the present study,
we focused on children between the ages of 4 and 6 years, a developmental stage where
DLD is typically diagnosed. This narrower age range allowed us to provide more specific
insights into the prevalence of DLD, which might account for the variation in our
findings compared to the broader language disorder estimates in the National Survey.
In the present study, it was observed that maternal education was significantly lower
in the DLD group than in the TLD group among 5-year-olds. It is worth noting that
low maternal education, often indicative of a lower socioeconomic status, is potential
risk factor for children with DLD, as mentioned in previous studies (Rudolph 2017;
Tomblin et al. 1997; Valade et al. 2023; Wu et al. 2023). Maternal education's influence
may arise from inadequate parent–child interaction practices and an increased likelihood
of receiving a DLD diagnosis ( Lara-Díaz et al. 2021). However, it is essential to
clarify that these findings do not seek to establish a causal link between maternal
education, socioeconomic status, and the occurrence of DLD. This distinction is vital
because DLD is a complex condition characterized by diverse and dynamic manifestations.
For instance, some children may indeed manifest DLD, but their grammatical challenges
could be relatively mild and may not be readily apparent through standard clinical
assessments.
Another finding of this study was that the prevalence of DLD differed according to
sex. In our sample, the overall estimated prevalence of DLD was 11.5% in boys and
5.1% in girls. According to the provided prevalence rates, approximately 11 boys and
5 girls per 100 individuals were identified with DLD. The data were further stratified
into three age groups; however, age-stratification resulted in a substantial reduction
in sample size within each group. This necessitates caution in interpreting sex-difference
findings. Consequently, prioritizing the overall data for analysis is advisable due
to its more robust statistical interpretation. These sex-based trends align with global
research, which consistently indicates a higher incidence of DLD in males compared
to females (e.g., Calder et al. 2022; De Barbieri et al. 1999; Norbury et al. 2016;
Tomblin et al. 1997). Historical research has also contributed to our understanding,
as a classical study conducted on 3-year-old children found a male-to-female ratio
of 2:1 (Stevenson & Richman 1976). In a recent study in Iran, the estimated prevalence
of DLD in males and females was reported as 4.2 and 2.4%, respectively (Oryadi-Zanjani
et al. 2015). It is worth noting that the variable of sex may exhibit some variability
in certain studies, with some indicating male sex as a risk factor for language disorders
(Chilosi et al. 2023; Hannus et al. 2009; Harrison & McLeod 2010), while others do
not find a significant sex difference (Beitchman et al. 1986; Calder et al. 2022;
Tomblin et al. 1997). In general, language disorders have been regarded as having
a genetic component, supported by epidemiological studies, twin studies, family aggregation
research, and investigations into sex chromosome trisomy, although genetics alone
do not determine outcomes (Chilosi et al. 2023). Therefore, the predisposition of
males to DLD may be related to neurobiological factors on one hand, while environmental
factors might also influence the manifestation of DLD.
Clinical Implications
The National Survey of Health and Nutrition in Mexico did not include specific data
on the prevalence of DLD. This underscores the importance of recognizing and addressing
DLD in Mexico and thus, the findings of the present study. Increasing awareness of
this disorder is crucial, as it can lead to the development of strategies within the
healthcare and educational systems aimed at early identification and diagnosis of
DLD in children. Early recognition and diagnosis can significantly benefit children
by positively impacting their language development and enhancing their educational
and social skills. Identifying children with DLD allows for the provision of tailored
interventions and support to address their unique needs.
Furthermore, the results of our current study hold critical significance as the initial
reference for evaluating the consequences of our assessment protocols. If these protocols
lead to under- or over-identification of DLD, it is essential to review and refine
them to gradually establish an improved system for the early identification and diagnosis
of DLD.
Limitations of the Study
One significant limitation of our current study is its retrospective nature, as it
is based on a non-probabilistic sample derived from data collected in three distinct
Mexican cities: Queretaro, Mexico, and Monterrey. The prevalence estimate of 8.5%
in our study closely aligns with the reported rates for English-speaking children
of similar ages in England and the United States, approximately 7.5%, as established
in prior research by Norbury et al. (2016) and Tomblin et al. (1997), which employed
population samples. Our result is also similar to another non-English study, which
reported 8.5% in a large-scale population-based survey (Wu et al. 2023). However,
it is important to note that our estimate may be subject to bias due to the non-probabilistic
sampling method employed.
Another potential source of bias affecting the estimated percentages of children with
DLD and those with TLD is the influence of parental concern. When parents do not believe
their children are at risk of developing a language disorder, they may be less inclined
to participate in a study focused on developmental language concerns. Consequently,
a greater number of typically developing children attending school, whose parents
do not have language-related concerns, may have chosen not to take part in the study.
This, in turn, could result in an underestimation of the percentage of typically developing
children in our sample.
Furthermore, it is important to consider the evolving nature of parental concerns
as children grow, as such changes in parental concerns across different age groups
may impact the estimated prevalence of DLD differently. Specifically, younger children
might be less likely to have their language deficits recognized, as parents may not
readily identify the warning signs of a language disorder. In fact, in the present
study, the percentage of parental concern in children with DLD was nearly three times
higher than in children with TLD at 4 years of age. However, this difference increased
to five times at ages 5 and 6 years. This finding suggests that language disorders
are not generally perceived as a significant issue in younger Spanish-speaking children
in Mexico, even though a diagnosis can be obtained by the age of 4 years.
Additionally, the study's limited sample size led to a wider confidence interval.
As a result, there were no statistically significant differences in prevalence across
age groups or between genders. It is also important to note that the statistical associations
between maternal education and DLD in the 5-year-old cohort do not imply causation
due to the study's inherent design limitations. To gain a more comprehensive understanding
of this association, future research could entail more extensive investigations employing
methodologies that enable the assessment of causality.
Conclusion
The prevalence of DLD in a Mexican sample of children was 8.5% between 4 and 6 years
and 11 months of age. This study is the first in Mexico to estimate the prevalence
of DLD and the risk of DLD. Identifying a prevalence of children with a clinically
confirmed DLD in preschool and the first years of elementary school is crucial. We
also found significant risk factors that may predict the diagnosis of DLD, such as
being male. Early identification of these children is critical to reducing the impact
throughout life. Using sensitive clinical assessment tools is essential for this process.
The prevalence of DLD in Mexico should inform education and public health authorities
to raise awareness of this long-lasting disorder and establish a system to identify
and diagnose it effectively and deliver effective interventions.
