Language Outcomes at 7

Years: Early Predictors and

Co-Occurring Difficulties
Cristina McKean, PhD,a,b Sheena Reilly, PhD,b,c Edith L. Bavin, PhD,d Lesley Bretherton, PhD,b,e,f,g Eileen Cini, BAppSci,b Laura
Conway, BA,b,f Fallon Cook, PhD,b Patricia Eadie, PhD,h Margot Prior, PhD,g Melissa Wake, MD,b,e,f Fiona Mensah, PhDb,e,f

OBJECTIVE: To examine at 7 years the language abilities of children, the salience of early abstract
life factors and language scores as predictors of language outcome, and co-occurring
difficulties
METHODS: A longitudinal cohort study of 1910 infants recruited at age 8 to 10 months.
Exposures included early life factors (sex, prematurity, birth weight/order, twin birth,
socioeconomic status, non–English speaking background,family history of speech/
language difficulties); maternal factors (mental health, vocabulary, education, and age);
and child language ability at 2 and 4 years. Outcomes were 7-year standardized receptive
or expressive language scores (low language: ≥1.25 SD below the mean), and co-occurring
difficulties (autism, literacy, social, emotional, and behavioral adjustment, and health-
related quality of life).
RESULTS: Almost 19% of children (22/1204;18.9%) met criteria for low language at 7 years.
Early life factors explained 9-13% of variation in language scores, increasing to 39-58%
when child language scores at ages 2 and 4 were included. Early life factors moderately
discriminated between children with and without low language (area under the curve:
0.68–0.72), strengthening to good discrimination with language scores at ages 2 and 4
(area under the curve: 0.85–0.94). Low language at age 7 was associated with concurrent
difficulties in literacy, social-emotional and behavioral difficulties, and limitations in school
and psychosocial functioning.
CONCLUSIONS: Child language ability at 4 years more accurately predicted low language at 7
than a range of early child, family, and environmental factors. Low language at 7 years was
associated with a higher prevalence of co-occurring difficulties.

aEducation, Communication and Language Sciences, Newcastle University, Newcastle-upon-Tyne, United

WHAT’S KNOWN ON THIS SUBJECT: Fluctuating
Kingdom; bMurdoch Childrens Research Institute, Melbourne, Australia; cMenzies Health Institute Queensland, preschool language abilities present challenges
Griffith University, Queensland, Australia; dSchool of Psychology and Public Health, La Trobe University, for identifying children at risk for later language
Melbourne, Australia; eThe Royal Children’s Hospital, Melbourne, Australia; and fDepartment of Pediatrics, impairment. Child, family, and environmental factors
gPsychological Sciences, and hGraduate School of Education, The University of Melbourne, Melbourne, Australia
explain a limited amount of variability in language
Drs McKean, Reilly, Cook, and Mensah designed the study and completed the data analyses; ability at 4 years, and prediction of low language
Drs Bavin, Bretherton, Eadie, Prior, and Wake designed the longitudinal cohort study and revised status is limited.
the manuscript; Ms Cini and Ms Conway designed data collection instruments, carried out
WHAT THIS STUDY ADDS: Language ability at 4
preliminary data analyses, coordinated and supervised data collection, and critically reviewed
years predicted language outcomes at 7 years more
the manuscript; and all authors approved the final manuscript as submitted.
saliently than child, family, and environmental factors.
DOI: 10.1542/peds.2016-1684 At 7, low language was associated with higher rates of
Accepted for publication Nov 22, 2016 social-emotional, behavioral, and literacy difficulties
Address correspondence to Sheena Reilly, PhD, Menzies Health Institute Queensland, Griffith and health-related quality of life limitations.
University, Queensland, QLD 4222, Australia. E-mail: s.reilly@griffith.edu.au
To cite: McKean C, Reilly S, Bavin EL, et al. Language Outcomes at 7 Years: Early
Predictors and Co-Occurring Difficulties. Pediatrics. 2017;139(3):e20161684

Downloaded from by guest on February 8, 2017

PEDIATRICS Volume 139, number 3, March 2017:e20161684 ARTICLE
Robust language development associated with later language months living in 6 local government
is needed to meet the increasing outcomes to inform the design of areas of Melbourne, Australia, in
social-relational and academic primary preventive interventions16 2003, selected to represent high,
demands on children transitioning and targeted secondary prevention. medium, and low SES according to
to formal schooling.1 Yet 7% to 20% Early life factors measured in infancy the census-based Socioeconomic
of children experience low language explain only a small amount of Indexes for Areas (SEIFA). A total
in the preschool and early school variance in language abilities at 4 of 1910 infants were recruited
years2,3 and are at heightened risk years but more than is explained at via the Maternal and Child Health
of difficulties with literacy, academic 2 years. Socioeconomic status (SES) Service, with additional recruitment
attainments, and social-emotional has emerged as an increasingly at hearing screening appointments
and behavioral (SEB) adjustment.4–8 important predictor of language and and through press advertising; full
These children are also at elevated low language status between these 2 details are reported elsewhere.17
risk of experiencing poor mental ages,3,17 suggesting that the salience Parents unable to understand
health9 and periods of unemployment of early life factors may change over English sufficiently to respond to
in adulthood.10,11 a child’s developmental trajectory. the questionnaires designed for a
However, we lack a detailed Whether the effects of early life grade 6 reading level were excluded.
understanding of the natural history factors increase or decrease between Participant retention across the 8
of language development from the 4 and 7 years is unknown. waves of data collection to 7 years
preschool to the early school years, is shown in Fig 1. The in-scope
Little is known about the impact
and it remains unclear at what age sample for this analysis consists
of low language on the individual’s
co-occurring difficulties emerge of 1204 children with complete
functioning and health-related
for children with low language. direct receptive and expressive
quality of life (HRQL),18 the nature
Therefore, identifying when and language assessment at 7 years. This
and degree of HRQL limitations
how best to intervene remains article extends previously reported
experienced, or the incidence of
challenging. outcomes at 1, 2, and 4 years in this
co-occurring conditions. Yet this is
cohort.3,17
Because of the instability in child critical information needed to inform
language development in the appropriate preventive interventions. Measures
preschool years, neither screening The burden of these difficulties in
nor 1-time direct assessments the population of children with low Measures were collected via parent
are sufficiently reliable to identify language, as opposed to only those report and direct assessments
children at risk of persisting clinically referred, remains poorly undertaken by trained research
difficulties.12–14 At least half of quantified.19 assistants.
children with low expressive In a community-ascertained cohort3,17 Early Life Factors
language abilities at 2 years catch the aims of this study were (1) to
up with their peers by 4 or 5 years, quantify the contribution of early life Twelve early life factors identified by
and many 4-year-olds presenting factors to 7-year language outcomes, the US Preventive Services Taskforce
with low language have previously specifically receptive and expressive as potentially predictive of language
attained language scores falling language scores and low language development, and previously
within the typical range at 2 years.3 status; (2) to determine whether explored at 2 and 4 years, are
Furthermore, of those presenting these early life factors maintain included here. These included child
with low language at 7 years, in salience once 2- and 4-year language factors: male sex, twin birth, preterm
a community-based study 46% abilities are considered; and (3) birth, birth order and birth weight;
previously attained typical language document co-occurring diagnoses family factors: non–English speaking
scores at 4, and conversely 39% of autism and difficulties in literacy, background (NESB), family history
of children with low language at HRQL, and SEB adjustment in 7-year- of speech or language difficulties,
4 years had typical language by old children with low language. and SES; and maternal factors:
7 years.15 Therefore, a diagnostic age, mental health, education, and
approach, dichotomizing children vocabulary.17,20
into those with and without language METHODS NESB was defined where families
difficulties, risks both overserving reported that English was not the
and underserving even between 4 Sampling and Participants
main language spoken to the child at
and 7 years. The Early Language in Victoria Study home. The SEIFA Index of Relative
One approach has been to identify (ELVS) cohort was drawn from the Disadvantage is a census-derived
early risk and protective factors population of infants aged 7.5 to 10 measure of SES standardized for

Downloaded from by guest on February 8, 2017

2 MCKEAN et al
 of 44.21 Maternal mental health was
dichotomized to “no mental health
problem” (<4) or “likely mental
health problems” (4–24) on the
Kessler Psychological Distress Scale
(K6).22

Early Language Measures

At 2 years late talking was
determined via parental report of
children’s vocabulary knowledge via
the MacArthur–Bates Communicative
Development Inventory–Words and
Sentences.23 Based on sex-specific
norms, late talkers were children
with scores falling below the 10th
centile.
At 4 years we administered the
Clinical Evaluation of Language
Fundamentals–Preschool Second
Edition,24 yielding receptive and
expressive standard scores (mean =
100; SD = 15).

Outcomes
Language at 7 years was measured
with the Clinical Evaluation of
Language Fundamentals Fourth
Edition (CELF-4) Australian
Standardization.20 Low language was
defined as a receptive or expressive
standard score that fell ≥1.25 SDs
below the published normative
population means (ie, ≤81). This
follows precedent in previous
studies in ELVS allowing direct
comparison between earlier and
later data waves.3 Outcomes were
examined for difficulties known to
be associated with low language in
clinical populations and for which
robust measures were included in
data collection at 7 years, namely,
low nonverbal IQ, diagnosis of an
autism spectrum disorder, literacy
FIGURE 1 difficulties, SEB difficulties, and
Participant flowchart from wave 1 (8 months) to wave 8 (7 years). Denominator for percentages limitations in HRQL. Nonverbal
is number participating at baseline (N = 1910). Numbers participating, withdrawing and lost to skills were measured with the block
contact are given at each wave. Numbers change at each wave due to families actively withdrawing,
becoming lost to contacts or not participating in a particular wave but returning later.
design and matrices subtests of
the Wechsler Abbreviated Scale of
Intelligence,25 with low nonverbal
the population of Australia (mean = of the Mill Hill Vocabulary scale, a ability defined as a score falling
1000; SD = 100). Maternal vocabulary multiple-choice test that yields a ≥1.25 SDs below the published
was measured via a modified version summed raw score with a maximum normative population mean.25

Downloaded from by guest on February 8, 2017

PEDIATRICS Volume 139, number 3, March 2017 3
Children with a diagnosis of autism scores only. The R2 and AUC values A total of 227 (18.9%) children were
spectrum disorder were identified of these models were examined to categorized as having either low
through parental report and quantify and compare the ability of receptive language (198; 16.4%) or
confirmed via follow-up telephone the models to predict 7-year language low expressive language (111; 9.2%),
interview by the authors (M.P. or or language status. The proportions but of these 82 children (6.8%) had
P.E.). SEB difficulties were identified with co-occurring difficulties (aim 3) low scores in both domains. This
via the Strengths and Difficulties were estimated with 95% confidence means 29 (2.4%) children had only
Questionnaire,26 and the clinical intervals (CIs) in 7-year-old children expressive language difficulties and
cutoff points for an “abnormal” score with and without low language, and 116 (9.6%) only receptive difficulties.
applied. Literacy difficulties were these proportions were compared via Therefore, it must be borne in mind
determined with an adapted version univariable logistic regression. when interpreting these data that
of the Wide Range Achievement Test many children in the receptive
To investigate whether sample
fourth Edition, where scores across and expressive groups had mixed
attrition had affected the inferences
reading and spelling subtests were impairments.
drawn, the sample of 1204 children
summed and scaled. In the absence
with complete language data at In the linear regression analyses
of population norms, literacy scores
7 years was reweighted to be examining associations between
falling ≥1.25 SDs below the sample
representative of the initial cohort early life factors and language ability
mean were classified as “impaired.”
with regard to the 12 early life at age 7 (Table 2, aim 1), mean
HRQL was measured on the parent-
factors. Weights were derived from CELF-4 standard scores were higher
reported Pediatric Quality of Life
the inverse predicted probabilities for girls than boys (4.3 points for
Scale,27 and “limitations” in HRQL
of participation estimated by a receptive and 2.4 for expressive
were defined as scores falling ≥1 SD
logistic regression model. Multiple language) and for children of
below the sample mean, a level found
imputation was used to account for mothers with a university degree
to be similar to that of children with
missing predictor variables.29,30 when compared with those who
severe or chronic health conditions
A series of 50 data sets were derived did not finish school (3.4 receptive,
such as rheumatic conditions and
from chained equations to implement 4.1 expressive). There was a clear
newly diagnosed cancer in a large
an imputation model including the effect of parity; firstborn children
population sample.28
explanatory and outcome variables had higher language scores, and
considered in the analyses. Because twin birth was associated with
Analyses results were similar, we present the lower expressive language scores
findings based on the complete and (5.6 points lower than singletons).
A total of 1204 children completed
unweighted data. Analyses were A family history of speech or
the receptive and expressive
implemented in Stata 13.0 (Stata language difficulties was associated
language assessment at 7 years.
Corp, Inc, College Station, TX). All with lower receptive language (2.7
To quantify the contribution of
analyses were also repeated with points), and lower SES and poorer
early life factors to 7-year language
the CELF-4 core language score as maternal vocabulary knowledge
outcomes (aim 1), multivariate
the outcome. Again results were were associated with lower language
linear regressions were fitted to the
similar, and so we present only the scores. For every 100-unit increase
CELF-4 outcomes and the coefficient
findings for receptive and expressive in SEIFA scores, receptive language
of determination (R2) derived.
language outcomes to enable direct scores increased by 2.2 points and
Multivariate logistic regression
comparison with previous research.3 expressive by 2.3.
models were fitted to the binary
outcome “low language status,” Together the early life factors
deriving the area under the curve accounted for 9% and 13% of
RESULTS
(AUC). To determine whether these the variation in receptive and
early factors have measurable effects Participant retention at 7 years expressive language standard scores,
once later language measures are is summarized in Fig 1 and the respectively, at 7 years (Table 3, aim 2).
considered (aim 2), these regression characteristics of the participants Together the addition of 2-year late-
analyses were repeated with, in turn, (N = 1204) and nonparticipants talker status and 4-year language
the following predictors: the 12 early (N = 706) are compared in Table 1. scores increased the variation
life factors described above; the 12 Participating families were more explained to 39% for receptive
factors plus 2-year-old late-talker likely to be socially advantaged and and 58% for expressive language.
status; the 12 factors, 2-year-old late- to have mothers who were more Language scores alone at 4 years
talker status plus 4-year-old language highly educated and less likely to (without the early life factors or
scores; and 4-year-old language speak a language other than English. late-talker status) produced almost

Downloaded from by guest on February 8, 2017

4 MCKEAN et al
identical R2 values, explaining 37% TABLE 1 Characteristics at Baseline and Wave 2 of Children Participating and Lost to Follow-up at 7 y
and 56% of 7-year receptive and Baseline and Wave 2 Characteristics Did Not Complete at Completed at 7 y P
expressive scores, respectively. 7 y (N = 706) (N = 1204)

In the logistic regression analyses Female, % 46.7 51.1 .07

Twin birth, % 3.5 2.3 .12
(Table 2, aim 1) the same factors
Preterm birth (<36 wk), % 2.8 3.2 .62
as for the linear regression, with Birth wt (kg), mean ± SD 3.4 ± 0.6 3.5 ± 0.5 .04
the exception of twin birth, were Birth order, % .07
significant predictors of 7-year low- First 48.9 50.8
language status. The AUC values Second 38.2 33.7
Third 9.9 13.1
in Table 3 (aim 2) quantify the
Fourth or later 3.0 2.5
ability of the predictor variables to NESB, % 10.8 3.4 <.001
discriminate between 7-year-olds Maternal education level, %
with and without low language. An ≤12 y 28.6 20.5 <.001
AUC value of 0.5 indicates chance 13 y 41.3 39.5
Degree or postgraduate 30.1 40.1
discrimination, 0.7 to 0.8 moderate,
SEIFA score, mean ± SD 1028.2 ± 67.2 1040.6 ± 56.1 <.001
0.8 to 0.9 good, and 1 indicates Family history of speech or language 26.1 24.2 .36
perfect discrimination. difficulties, %
Maternal mental health symptoms, % 30.0 32.6 .28
The early life factors alone
Maternal vocabulary score, mean ± SD 26.2 ± 5.6 28.2 ± 4.7 <.001
reached only a moderate level of Maternal age at baseline, mean ± SD 31.4 ± 4.8 32.1 ± 4.4 <.001
discrimination (receptive 0.68, Participants completed direct assessment of both receptive and expressive language at 7 y. Child age at 7 y assessment,
expressive 0.72).31 Together the mean ± SD, 88.3 ± 2.2 mo. P values were derived through comparisons between those completing 7-y assessment and
addition of both 2-year late-talker those lost to follow-up via either χ2 tests for categorical variables or t tests for continuous variables.
status and 4-year language scores
improved the model’s ability to 48.6% expressive) and limitations in DISCUSSION
discriminate to a “good” level for their HRQL in the domains of school
receptive language (0.85) and to (32.1% receptive, 36.0% expressive) At 7 years a range of early child,
an “excellent” level for expressive and psychosocial functioning (27.6% family, and maternal factors
language (0.94). Notably, almost receptive, 31.3% expressive). It collectively explained 9% of
identical AUC values were found must be noted that only 29 children receptive and 13% of expressive
when only 4-year language measures had isolated expressive language language outcomes. Although it was
were used as predictors (receptive: difficulties. Children with only less than explained at 4 years (19%
0.84; expressive: 0.93), without the expressive difficulties experienced and 21%),3 it was higher than at
addition of the early life factors and lower levels of co-occurring 2 years (4% and 7%)17 in the same
late-talker status. difficulties than those with mixed cohort. Together, these predictors
expressive–receptive difficulties. moderately discriminated between
The prevalence of co-occurring children with typical language and
difficulties (autism, low nonverbal IQ, However, given the low numbers,
those with low language at 7 years.
literacy difficulties, SEB difficulties) estimates of co-occurrence for
Whereas late-talker status did not
and limitations in HRQL were these subgroups are not presented
substantially improve discrimination,
significantly higher for children with here because of concerns regarding
the addition of language scores at
low language at 7 years than for their reliability.
4 years improved discrimination
peers with typical language (Table 4). to “good” or “excellent” levels.
Two subtypes of socioemotional After multiple imputation to address
Similar levels of discrimination
problems as measured by missing predictors and weighting to were obtained when only 4-year
the Strengths and Difficulties reflect the sample characteristics at language scores were used to predict
Questionnaire (low prosocial baseline, no substantive differences 7-year-old language, suggesting that
scores and emotional difficulties) were identified in the estimates of language abilities are more stable
were not significantly different the prevalence of comorbidities to between 4 and 7 than between 2 and
between the 2 groups, although those with complete data. For the 4 years.
prevalence of low prosocial scores regression analyses, differences were
was significantly higher for children minimal; however, some effects were Children with low language at
with low expressive language but diluted when missing data were 7 years were more likely to have
not receptive. A high proportion of addressed (eg, SEIFA, family history low IQ, autism, SEB difficulties, and
children with low language also had of speech or language difficulties, and literacy difficulties than their peers
literacy difficulties (37.2% receptive, maternal vocabulary). with typical language, and rates of

Downloaded from by guest on February 8, 2017

PEDIATRICS Volume 139, number 3, March 2017 5
TABLE 2 Multiple Variable Linear Regression Analysis of CELF-4 Standard Scores and Logistic Regression Analysis of Language Status at 7 Years With
Respect to Early Life Factors
Language (N = 1132)a Low Language Status (N = 1132)a
Receptive Expressive Receptive Expressive
Mean 95% CI P Mean 95% CI P Odds 95% CI P Odds 95% CI P
Differenceb Difference Ratio Ratio
Child
Female 4.3 2.7 to 5.9 <.001 2.4 0.9 to 3.9 .002 0.57 0.41 to 0.80 .001 0.54 0.35 to 0.84 .007
Twin birth −3.0 –8.6 to 2.6 .29 −5.6 –10.9 to –0.2 .04 1.46 0.50 to 4.31 .49 1.68 0.50 to 5.65 .40
Preterm birth (<36 1.0 –3.9 to 5.9 .69 0.7 –4.0 to 5.4 .78 0.95 0.36 to 2.54 .92 1.26 0.42 to 3.79 .68
wk)
Birth wt (per kg) 1.2 –0.5 to 2.9 .16 1.5 –0.1 to 3.1 .07 0.84 0.59 to 1.20 .35 0.73 0.47 to 1.14 .17
Birth order
First Reference Reference Reference Reference
Second −1.6 –3.4 to 0.2 .07 −2.9 –4.6 to –1.2 .001 1.26 0.85 to 1.86 .25 1.36 0.81 to 2.26 .24
Third −4.1 –6.6 to –1.5 .002 −6.3 –8.7 to –3.9 <.001 1.65 0.99 to 2.74 .06 2.22 1.19 to 4.13 .01
Fourth or later −5.3 –10.6 to 0.0 .05 −8.9 –14.0 to –3.9 .001 2.11 0.82 to 5.44 .12 3.02 1.02 to 8.98 .05
Family
NESB 2.3 –2.4 to 7.0 .35 0.9 –3.6 to 5.4 .68 0.63 0.23 to 1.77 .38 0.95 0.3 to 3.04 .94
SEIFA score 2.2 0.7 to 3.6 .003 2.3 0.9 to 3.7 .001 0.61 0.46 to 0.80 .001 0.58 0.41 to 0.82 .002
(per 100 points)
Family history −2.7 –4.6 to –0.9 .004 −1.5 –3.3 to 0.2 .09 1.46 1.00 to 2.11 .05 1.09 0.66 to 1.78 .74
of speech
or language
difficulties
Mother
Maternal
education level
≤12 y Reference Reference Reference Reference
13 y 0.9 –1.2 to 3.1 .39 0.7 –1.3 to 2.8 .49 0.86 0.57 to 1.31 .49 0.89 0.53 to 1.48 .64
Degree or 3.4 1.2 to 5.6 .003 4.1 2.0 to 6.3 <.001 0.59 0.37 to 0.94 .03 0.41 0.22 to 0.78 .006
postgraduate
Maternal 0.3 –1.3 to 2.0 .69 0.4 –1.2 to 2.0 .63 0.94 0.66 to 1.34 .72 1.02 0.65 to 1.61 .92
mental health
symptoms
Maternal 0.3 0.1 to 0.5 .003 0.4 0.3 to 0.6 <.001 0.95 0.91 to 0.99 .01 0.95 0.90 to 1.0 .03
vocabulary
score (per
point)
Maternal age at 0.1 –0.1 to 0.3 .35 0.2 0.0 to 0.3 .11 0.99 0.95 to 1.03 .69 1.01 0.96 to 1.06 .71
baseline (per
year)
a Children with complete predictor and outcome data.
b Mean difference refers to the average difference between the reference group (eg, boys) and the comparator group (eg, girls) in the outcome (eg, receptive language score).

comorbid literacy difficulties were Using limitations in HRQL in children explanation for these diminished
particularly high (37% and 49%). with chronic and severe health effects may be that universal access
Children with low language were conditions as a benchmark to define to schooling partly compensates
2 to 3 times more likely than their a significant limitation in HRQL, it is for early life factors; for example,
peers to have SEB difficulties. Given of significant concern that children children from an NESB catch up
the relationship reported between with low language were experiencing with their monolingual peers when
low language in childhood and later limitations at more than twice the given access to English-speaking
mental health outcomes in adulthood, frequency reported in their peers environments.32 In addition, as young
this concurrent finding at 7 years with typical language.28 mothers mature, perhaps they are
suggests that this relationship able to provide more supportive
A number of factors that were
begins early.9,10 To our knowledge home learning environments.33,34
predictive of language outcomes
this is the first time an association at 4 years (NESB, maternal age, Between 2 and 4 years3 we found
between low language abilities family history of speech or language an increase in the contribution of
and HRQL has been reported in problems, and birth weight) were social and environmental factors
childhood in a community sample. not at 7 years in this cohort. One to children’s language outcomes.

Downloaded from by guest on February 8, 2017

6 MCKEAN et al
TABLE 3 Proportion of Variance in CELF-4 Standard Scores Explained and Accuracy of Predictors in entirely fixed, and factors such
Discriminating Between Children With and Without Low Language Status at 7 y as the frequency of being read
Languagea Low Language Status b to and the number of children’s
Nc Receptive Expressive Receptive Expressive books in the home are associated
Model R2 R2 AUCd 95% CI AUC 95% CI
with change in relative ability
from 4 to 7 years, albeit with small
(1) Predictors (n = 12) in infancy 1132 0.09 0.13 0.68 0.64 to 0.72 0.72 0.67 to 0.77
only
effects.16
(2) As (1) plus late talking status 1084 0.14 0.20 0.72 0.68 to 0.76 0.82 0.78 to 0.86
(2-y) The strengths of the ELVS include
(3) As (2) plus receptive and 1026 0.39 0.58 0.85 0.82 to 0.89 0.94 0.92 to 0.96 the cohort of children that reflects a
expressive language scores
broad spectrum of the community,
(4-y)
(4) Receptive and expressive 1132 0.37 0.56 0.84 0.81 to 0.88 0.93 0.91 to 0.95 the prospective, longitudinal
language scores only (age 4) data collection, repeated robust
a Linear regression. language measures, and concurrent
b Logistic regression. measures of other aspects of child
c Children with complete predictor and outcome data.
d AUC value of 0.5 indicates chance discrimination, 0.7–0.8 indicates moderate, 0.8–0.9 indicates good, and 1 indicates
development. Inevitably sample
perfect discrimination. attrition has occurred, and there
are missing data. However, the
similarity of findings when inverse
However from 4 to 7 years the explained by the early life factors is probability weighting and multiple
salience of these factors (eg, mediated by a measure of language imputation were implemented
SES, maternal education) was at 4 years. Therefore, it appears provides assurance that the
substantively unchanged. Age 4 that the associations between the results presented reflect the
language scores alone yielded early life factors measured here and original cohort. The inclusion of a
similar levels of prediction as a child’s relative language abilities child’s access to speech pathology
models including late-talker status have played out by age 4. However, interventions and its relationship
and early life factors. This finding it is important to note that a child’s to child outcome would have
suggests that most of the variability relative language ability at 4 is not strengthened the study. The ELVS

TABLE 4 Prevalence of Co-Occurring Difficulties in Children With Low Language at 7 y Versus Those With Typical Language
Receptive Expressive
Typical Language Low Language Typical Language Low Language
(N = 1006)a (N = 198)a (N = 1093)a (N = 111)a
Total Nb % 95% CI % 95% CI P % 95% CI % 95% CI P
Autism 1204 1.6 1.0 to 2.6 6.1 3.5 to 10.4 <.001 1.6 1.0 to 2.6 9.0 4.9 to 16.0 <.001
Low nonverbal IQ 1197 0.5 0.2 to 1.2 10.8 7.1 to 16.0 <.001 1.2 0.7 to 2.0 12.0 7.1 to 19.7 <.001
Literacy difficulties 1199 9.5 7.8 to 11.4 37.2 30.7 to 44.2 <.001 10.6 8.9 to 12.5 48.6 39.3 to 58.0 <.001
Social–emotional and
behavioral difficulties
Emotional symptoms 1070 7.6 6.0 to 9.5 12.3 8.0 to 18.3 .05 7.8 6.3 to 9.7 13.5 8.0 to 22.0 .06
Conduct problems 1070 7.3 5.8 to 9.2 17.2 12.1 to 23.8 .001 7.9 6.4 to 9.8 17.7 11.3 to 26.7 .002
Hyperactivity or 1070 7.4 5.8 to 9.3 20.9 15.3 to 27.8 <.001 8.3 6.7 to 10.2 20.8 13.8 to 30.2 <.001
inattention
Peer problems 1069 5.1 3.8 to 6.7 12.3 8.0 to 18.3 .001 5.7 4.4 to 7.3 11.5 6.4 to 19.6 .03
Low prosocial 1070 2.2 1.4 to 3.4 4.9 2.5 to 9.5 .05 2.2 1.4 to 3.3 7.3 3.5 to 14.6 .005
behavior
Total difficulties 1070 3.9 2.8 to 5.3 14.7 10.1 to 21.1 <.001 4.4 3.3 to 5.9 16.7 10.4 to 25.6 <.001
Quality of life
Physical health 1068 10.4 8.6 to 12.6 16.7 11.6 to 23.1 .02 10.4 8.6 to 12.5 20.8 13.8 to 30.2 .003
Emotional functioning 1069 12.4 10.4 to 14.7 21.5 15.8 to 28.5 .002 12.8 10.9 to 15.1 22.9 15.6 to 32.4 .007
Social functioning 1069 14.0 11.9 to 16.4 23.9 18.0 to 31.1 .002 14.4 12.3 to 16.7 27.0 19.1 to 36.9 .001
School functioning 1034 12.9 10.8 to 15.3 32.1 25.2 to 39.8 <.001 13.9 11.8 to 16.2 36.0 26.6 to 46.5 <.001
Psychosocial 1069 13.2 11.2 to 15.6 27.6 21.3 to 35.0 <.001 13.8 11.8 to 16.2 31.3 22.7 to 41.2 <.001
functioningc
Total score 1069 12.4 10.4 to 14.7 27.0 20.7 to 34.4 <.001 13.1 11.1 to 15.3 30.2 21.8 to 40.2 <.001
a In sample of 1204 children with complete expressive and receptive language data at 7 y.
b Numbers differ because of missing outcome variables and test scoring guidelines with respect to handling missing responses.
c This is a composite derived from the emotional, social, and school functioning subscales.

Downloaded from by guest on February 8, 2017

PEDIATRICS Volume 139, number 3, March 2017 7
data allowed identification of the Future research should be they be equipped and supported
proportion of children with low directed toward developing and to promote robust oral language
language at 4 and 7 years who robustly evaluating child language development for all children.
had received some form of speech surveillance and intervention
pathology intervention (4 years: pathways over the early school
38% receptive, 40% expressive; years. Such pathways should ACKNOWLEDGMENTS
7 years: 28% receptive, 39% include reliable and low-burden We sincerely thank the
expressive). However, insufficient language assessment tools within participating parents and children
information were available to a holistic surveillance approach, and acknowledge the contribution
draw reliable conclusions as to the monitoring language development of the Victorian Maternal and Child
amount, type, and quality of the alongside key domains such Health nurses who supported
therapy received, precluding its as literacy and psychosocial recruitment to this cohort. Ethical
inclusion in analyses. functioning. Tiered interventions approval was obtained from
must be developed and evaluated the Royal Children’s Hospital
that are tailored to the severity of (#23018 and #27078) and La
CONCLUSIONS language and associated difficulties Trobe University, Human Ethics
Low language at 7 years is prevalent and risks experienced by the child. Committee (#03-32). All
and associated with poor literacy Evaluation of such pathways must parents provided written,
and SEB adjustment. Furthermore, weigh the relative burden, costs, informed consent.
these 7-year-olds experience and benefits of overserving and
significant limitations in school and underserving children, given the
psychosocial functioning at twice instability in children’s language ABBREVIATIONS
the rate reported in peers with status that continues from 4 to
AUC: area under the curve
typical language.28 Although the 7 years.
CELF-4: Clinical Evaluation of
results reported here suggest that
Language Fundamentals
education and health services could We recommend monitoring the
Fourth Edition
reasonably estimate a child’s risk progress of children with low
CI: confidence interval
for persisting low language abilities language at 4 years through the
ELVS: Early Language in Victoria
at 7 years based on assessment of early school years with respect to
Study
his or her language abilities at 4, language, literacy, and psychosocial
HRQL: health-related quality of
1-time language assessments at functioning. Additionally, where
life
this age that categorize children as children present with literacy or
NESB: non–English speaking
with or without low language have psychosocial difficulties over this
background
proven insufficiently reliable as the period, their language abilities should
SEB: social-emotional and
sole means of targeting language be investigated. It is essential that
behavioral
interventions.35 In this cohort, 6.2% schools recognize the vital role oral
SEIFA: Socioeconomic Indexes
had low language at 4 but not 7 years, language skills play in children’s
for Areas
8.4% at 7 but not at 4 years, and social and emotional adjustment
SES: socioeconomic status
9.8% at both time points.15 and acquisition of literacy36 and that

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2017 by the American Academy of Pediatrics
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by project grants 237106, 9436958, and 1041947 from the Australian National Health and Medical Research Council and small grants from
the Murdoch Childrens Research Institute and the Faculty of Health Sciences, La Trobe University. Dr McKean’s postdoctoral position was funded by Australian
National Health and Medical Research Council (NHMRC) Centre of Research Excellence in Child Language (1023493). Professor Wake was supported by NHMRC
Senior Research Fellowship 1046518; Dr Mensah by NHMRC Early Career Fellowship 1037449 and Career Development Fellowship 1111160; and Professor Reilly
by NHMRC Practitioner Fellowship 491210. Research at the Murdoch Childrens Research Institute is supported by the Victorian Government’s Operational
Infrastructure Support Program. The funding organizations are independent of all researchers and had no role in the design and conduct of the study; in
the collection, analysis, and interpretation of the data; in the decision to submit the article for publication; or in the preparation, review, or approval of the
manuscript.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

Downloaded from by guest on February 8, 2017

8 MCKEAN et al
REFERENCES
1. Duncan GJ, Dowsett CJ, Claessens 10. Law J, Rush R, Schoon I, Parsons S. Washington, DC: The National
A, et al. School readiness and Modeling developmental language Academies Press; 1994
later achievement. Dev Psychol. difficulties from school entry into 19. Feeney R, Desha L, Ziviani J, Nicholson
2007;43(6):1428–1446 adulthood: literacy, mental health, and JM. Health-related quality-of-life of
employment outcomes. J Speech Lang children with speech and language
2. Tomblin JB, Records NL, Buckwalter P,
Hear Res. 2009;52(6):1401–1416 difficulties: a review of the literature.
Zhang X, Smith E, O’Brien M. Prevalence
of specific language impairment in 11. Beitchman JH, Wilson B, Johnson Int J Speech-Language Pathol.
kindergarten children. J Speech Lang CJ, et al. Fourteen-year follow-up 2012;14(1):59–72
Hear Res. 1997;40(6):1245–1260 of speech/language-impaired and 20. Semel E, Wiig EH, Secord WA. Clinical
control children: psychiatric outcome. Evaluation of Language Fundamentals:
3. Reilly S, Wake M, Ukoumunne OC,
J Am Acad Child Adolesc Psychiatry. Australian Standardised Edition.
et al. Predicting language outcomes
2001;40(1):75–82 4th ed. Sydney, NSW: Psychological
at 4 years of age: findings from Early
Language in Victoria Study. Pediatrics. 12. Nelson HD, Nygren P, Walker M, Corporation; 2006
2010;126(6). Available at: www. Panoscha R. Screening for speech and 21. Raven JC. Mill Hill Vocabulary Scale.
pediatrics.org/cgi/content/full/126/6/ language delay in preschool children: Oxford, UK: Oxford Psycholinguistics
e1530 systematic evidence review for the Press; 1965
US Preventive Services Task Force. 22. Kessler R, Mroczek D. Final Version of
4. Catts HW, Fey ME, Tomblin JB, Zhang X.
Pediatrics. 2006;117(2). Available at: Non-Specific Psychological Distress
A longitudinal investigation of reading
www.pediatrics.org/cgi/content/full/ Scale. Ann Arbor, MI: Survey Research
outcomes in children with language
117/2/e298 Center of the Institute for Social
impairments. J Speech Lang Hear Res.
2002;45(6):1142–1157 13. Siu AL; US Preventive Services Task Research, Univserity of Michigan; 1994
Force. Screening for speech and 23. Fenson L, Dale PS, Reznick JS,
5. Johnson CJ, Beitchman JH, Brownlie
language delay and disorders in et al. MacArthur Communicative
EB. Twenty-year follow-up of children
children aged 5 years or younger: Development Inventories: User’s Guide
with and without speech-language
US Preventive Services Task Force and Technical Manual. San Diego, CA:
impairments: family, educational,
recommendation statement. Singular Publishing Group; 1993
occupational, and quality of life
Pediatrics. 2015;136(2) Available at:
outcomes. Am J Speech Lang Pathol. 24. Wiig EH, Secord WA, Semel E.
www.pediatrics.org/cgi/content/full/
2010;19(1):51–65 Clinical Evaluation of Language
136/2/e474
Fundamentals—Preschool: Australian
6. Tomblin B. Validating diagnostic
14. Wallace IF, Berkman ND, Watson Standardised Edition. 2nd ed. Sydney,
standards for specific language
LR, et al. Screening for speech and Australia: Harcourt Assessment; 2006
impairment using adolescent
language delay in children 5 years 25. Wechsler D.Wechsler Abbreviated
outcomes. In: Norbury CF, Tomblin
old and younger: a systematic review. Scale of Intelligence.San Antonio, TX:
B, Bishop DVM, eds. Understanding
Pediatrics. 2015;136(2). Available at: The Psychological Corporation; 1999
Developmental Language Disorders:
www.pediatrics.org/cgi/content/full/
From Theory to Practice. Hove, UK: 26. Goodman R. The Strengths and
136/2/e448
Psychology Press; 2008:93–114 Difficulties Questionnaire: a research
15. Reilly S, Tomblin B, Law J, et al. Specific note. J Child Psychol Psychiatry.
7. Voci SC, Beitchman JH, Brownlie
language impairment: a convenient 1997;38(5):581–586
EB, Wilson B. Social anxiety in late
label for whom? Int J Lang Commun
adolescence: the importance of early 27. Varni JW, Sied M, Rode CA. The PedsQL:
Disord. 2014;49(4):416–451
childhood language impairment. Measurement Model for the Pediatric
J Anxiety Disord. 2006;20(7):915–930 16. McKean C, Mensah FK, Eadie P, Quality of Life Inventory. Philadelphia,
et al. Levers for language growth: PA: Lippincott Williams & Wilkins; 1999
8. Durkin K, Conti-Ramsden G. Young
characteristics and predictors of 28. Varni JW, Burwinkle TM, Seid M,
people with specific language
language trajectories between 4 and 7 Skarr D. The PedsQL 4.0 as a pediatric
impairment: a review of social
years. PLoS One. 2015;10(8):e0134251 population health measure: feasibility,
and emotional functioning in
adolescence. Child Lang Teach Ther. 17. Reilly S, Wake M, Bavin EL, et al. reliability, and validity. Ambul Pediatr.
2010;26(2):105–121 Predicting language at 2 years of 2003;3(6):329–341
age: a prospective community study. 29. Sterne JA, White IR, Carlin JB,
9. Schoon I, Parsons S, Rush R, Law
Pediatrics. 2007;120(6). Available at: et al. Multiple imputation for missing
J. Children’s language ability
www.pediatrics.org/cgi/content/full/ data in epidemiological and clinical
and psychosocial development: a
120/6/e1441 research: potential and pitfalls. BMJ.
29-year follow-up study. Pediatrics.
2010;126(1). Available at: www. 18. Mrazek P, Haggerty R.Reducing the 2009;338:b2393
pediatrics.org/cgi/content/full/126/1/ Risks for Mental Disorders: Frontiers 30. Lee KJ, Simpson JA. Introduction
e73 for Preventive Intervention Research. to multiple imputation for dealing

Downloaded from by guest on February 8, 2017

PEDIATRICS Volume 139, number 3, March 2017 9
 with missing data. Respirology. 33. Lewin A, Mitchell SJ, Ronzio CR. 35. Wake M, Levickis P, Tobin S, et al. Two-
2014;19(2):162–167 Developmental differences in year outcomes of a population-based
31. Kirkwood BJ, Sterne JAC. parenting behavior: comparing intervention for preschool language
Essential Medical Statistics. 2nd ed. adolescent, emerging adult, and delay: an RCT. Pediatrics. 2015;136(4).
Oxford, England: Blackwell Sciences; adult mothers. Merrill-Palmer Q. Available at: www.pediatrics.org/cgi/
2003 2013;59(1):23–49 content/full/136/4/e838

32. Cattani A, Abbot-Smith K, Farag R, 34. King T, McKean C, Rush R, et al. 36. Justice LM, Mashburn A, Hamre B,
et al. How much exposure to English The impact of increasing maternal Pianta R. Quality of language and
is necessary for a bilingual toddler education on single word reading in literacy instruction in preschool
to perform like a monolingual peer in middle childhood: An analysis of the classrooms serving at-risk pupils.
language tests? Int J Lang Commun Millennium Cohort Study. Merrill- Early Child Res Q. 2008;23(1):
Disord. 2014;49(6):649–671 Palmer Quarterly. 2017. In press 51–68

Downloaded from by guest on February 8, 2017

10 MCKEAN et al
Language Outcomes at 7 Years: Early Predictors and Co-Occurring Difficulties
Cristina McKean, Sheena Reilly, Edith L. Bavin, Lesley Bretherton, Eileen Cini,
Laura Conway, Fallon Cook, Patricia Eadie, Margot Prior, Melissa Wake and Fiona
Mensah
Pediatrics; originally published online February 8, 2017;
DOI: 10.1542/peds.2016-1684
Updated Information & including high resolution figures, can be found at:
Services /content/early/2017/02/06/peds.2016-1684.full.html
References This article cites 25 articles, 7 of which can be accessed free
at:
/content/early/2017/02/06/peds.2016-1684.full.html#ref-list-1

Subspecialty Collections This article, along with others on similar topics, appears in
the following collection(s):
Developmental/Behavioral Pediatrics
/cgi/collection/development:behavioral_issues_sub
Cognition/Language/Learning Disorders
/cgi/collection/cognition:language:learning_disorders_sub
Permissions & Licensing Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
/site/misc/Permissions.xhtml
Reprints Information about ordering reprints can be found online:
/site/misc/reprints.xhtml

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
Grove Village, Illinois, 60007. Copyright © 2017 by the American Academy of Pediatrics. All
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from by guest on February 8, 2017

Language Outcomes at 7 Years: Early Predictors and Co-Occurring Difficulties
Cristina McKean, Sheena Reilly, Edith L. Bavin, Lesley Bretherton, Eileen Cini,
Laura Conway, Fallon Cook, Patricia Eadie, Margot Prior, Melissa Wake and Fiona
Mensah
Pediatrics; originally published online February 8, 2017;
DOI: 10.1542/peds.2016-1684

The online version of this article, along with updated information and services, is
located on the World Wide Web at:
/content/early/2017/02/06/peds.2016-1684.full.html

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned,
published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point
Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2017 by the American Academy
of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from by guest on February 8, 2017