Content uploaded by Harry J de Koning
Author content
All content in this area was uploaded by Harry J de Koning
Content may be subject to copyright.
DOI: 10.1542/peds.2006-3145
2007;120;1317Pediatrics
W. Verhoeven and Harry J. de Koning
Heleen M.E. van Agt, Heleen A. van der Stege, Hanneke de Ridder-Sluiter, Ludo T.
Effects on School Performance and Language Development at Age 8
A Cluster-Randomized Trial of Screening for Language Delay in Toddlers:
http://pediatrics.aappublications.org/content/120/6/1317.full.html
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2007 by the American Academy
published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point
publication, it has been published continuously since 1948. PEDIATRICS is owned,
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
ARTICLE
A Cluster-Randomized Trial of Screening for
Language Delay in Toddlers: Effects on School
Performance and Language Development at Age 8
Heleen M. E. van Agt, MA
a
, Heleen A. van der Stege, PhD
b
, Hanneke de Ridder-Sluiter, PhD
b
, Ludo T. W. Verhoeven, PhD
c
,
Harry J. de Koning, MD, PhD
a
a
Department of Public Health, Erasmus Medical Center, Rotterdam, Netherlands;
b
Dutch Foundation for the Deaf and Hard of Hearing Child, Amsterdam, Netherlands;
c
Pedagogical and Educational Sciences, Faculty of Social Sciences, Radboud University Nijmegen, Nijmegen, Netherlands
Financial Disclosure: Drs Van de Stege and de Ridder-Sluiter are employed by the Dutch foundation for the Deaf and Hard of Hearing Child, where the VTO language-screening instrument was created; Drs van
Agt, Verhoeven, and de Koning have indicated they have no financial relationships relevant to this article to disclose.
ABSTRACT
OBJECTIVE. The goal of this study was to assess the effects of screening and early
treatment of preschool children for language delay on language development and
school performance at age 8.
METHODS. A cluster-randomized, controlled trial and follow-up study of 55 child
health centers in 6 geographic regions were conducted from January 2002 to
September 2005. A total of 9419 children who were from the general population
and aged 15 months at entry were studied. School type end school progress was
known for 5406 (57.4%) children. In the intervention group, a structured screen-
ing instrument was conducted twice (at ages 15/18 and 24 months), and usual care
was applied in the control group. The screening instrument consisted of a uniform
set of questions for the parents and test elements for the child. A positive screen
result was followed by multidisciplinary assessments at speech and hearing centers
and subsequent early treatment if needed. Percentages of children who attended
a special school, repeated a class because of language problems, and scored low on
standardized language tests, in intention-to-screen analyses, were measured.
RESULTS. At age 8, 2.7% in the intervention group and 3.7% in the control group
attended a special school, 6.1% vs 4.9% had repeated a grade, 8.8% vs 9.7% had
deficient oral language performance, 4.7% vs 4.7% had deficient reading, and
2.8% vs 4.2% had deficient spelling.
CONCLUSIONS. Screening toddlers for language delays reduces the number of children
who require special education and leads to improved language performance at age
8. Nationwide implementation of the screening might be recommended.
www.pediatrics.org/cgi/doi/10.1542/
peds.2006-3145
doi:10.1542/peds.2006-3145
Key Words
screening, language delay, preschool
children, RCT
Abbreviations
RCT—randomized, controlled trial
VTO—VroegTijdige Onderkenning
Ontwikkelingsstoornissen
RR—relative risk
CI— confidence interval
Accepted for publication Jun 19, 2007
Address correspondence to Heleen M. E. van
Agt, MA, Department of Public Health,
Erasmus MC, PO Box 2040, 3000 CA
Rotterdam, Netherlands. E-mail: h.vanagt@
erasmusmc.nl
PEDIATRICS (ISSN Numbers: Print, 0031-4005;
Online, 1098-4275). Copyright © 2007 by the
American Academy of Pediatrics
PEDIATRICS Volume 120, Number 6, December 2007 1317
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
C
HILDREN’S GENERAL DEVELOPMENT is crucial. In
health care, there is much focus on the monitoring
of developmental steps in individual young children.
1
Serious problems in cognitive and/or socioemotional de-
velopment at school age or adolescence often originate
from developmental disorders in childhood, of which
language delays are the most prevalent.
2–5
In a large
Dutch sample from the open population, the prevalence
of language delays was estimated between 2.4% and
5.3% in 3-year-olds.
6
Although up to 60% of language
delays at the age of 2 to 3 years probably resolve spon-
taneously,
5
some indicate severe and long-lasting lan-
guage impairment with detrimental effects at later age.
7
Although effective treatment exists for young children
with several underlying causes of language delay,
8
it is
unclear whether systematic screening of language delay
at an early age is effective.
9
Whether screening leads to
better language performance as compared with usual
practice can be investigated only in a randomized, con-
trolled trial (RCT).
Several studies
10–14
have evaluated test characteristics
of specific screening instruments to detect language de-
lays in preschool children. Only 1 study was set up as an
RCT to evaluate the accuracy of a structured test and a
parent-led method for language screening among 582
3-year-old children
11
; however, none of these studies
evaluated the effects of screening on language perfor-
mance at later age.
In the Netherlands, on average, 85% of all 0- to
4-year-old children and their parents visit child health
centers (for free) at regular times for assessing the child’s
general development by physicians, including language
development. This article describes the effects of a spe-
cific screening instrument in a cluster-RCT among 9419
children on school performance and linguistic skills at
age 8 in the Netherlands. At this age, children in Dutch
schools should normally be in grade 2 of primary school,
having had 1 year of reading education. If children are
not capable of attending a regular primary school be-
cause of learning, behavioral, or health problems, then
special education services are offered. Children with se-
vere language delays run a high risk of being placed in
special schools or having to repeat a grade.
12
We hypoth-
esized that the screening would result in a reduction in
the proportion of children who need to attend a special
school, repeat a year in regular school, or have scores in
the lowest percentile of several standardized language
tests. We reported previously
6
that the screening did lead
to more earlier diagnoses and treatments in the first 3
years of life, as compared with a control group.
METHODS
Methods of this cluster-RCT have been published be-
fore.
6
Individual randomization is the ideal design, but
we used a cluster trial design to avoid biased results
(induced by the alternating use of the specific screening
instrument for the intervention children and standard
monitoring for the control subjects by 1 physician).
Child health care physicians were the units for random-
ization, and children were the units for analysis.
Randomization
We asked the child heath care physicians to identify low-
and high-socioeconomic neighborhoods within their re-
gion. Within the identified socioeconomic strata, each
physician was then allocated a number and randomly
classified by rolling dice by the trial’s manager as alter-
nately intervention or control physician. Physicians in
the control group performed the usual monitoring sys-
tem, which is based on physicians’ observation and on
questioning the parents in a limited manner without
clearcut referral criteria.
15
The child health center phy-
sicians in the intervention group were trained to use the
specific screening instrument.
Screening and Diagnosis
The VroegTijdige Onderkenning Ontwikkelingsstoornis-
sen (VTO; early detection of developmental disorders)
Language Screening instrument consisted of questions
about the language production, language reception, and
interaction of children in the age group 12 to 29 months
(Appendixes 1 and 2).
6,16,17
With this instrument, the
child health center physician in the intervention group
interviewed the parents who routinely visited the child
health center with their child, which took ⬃5 minutes.
The complete screening procedure embraced a screening
interview at 15/18 months as well as at 24 months. The
final score was obtained by adding the scores on both
screenings, which ranged between 0 and 7. When chil-
dren had a final score of ⱕ2, they were referred to the
general practitioner for additional assessment at a speech
and hearing center to confirm language delay and, if so,
to assess the underlying causes.
6
This was done by a
uniform protocol of multidisciplinary diagnostic proce-
dures in all regions, which included assessment of lan-
guage production, language reception, hearing, cogni-
tive development, and socioemotional development.
6
The cutoff score of the VTO Language Screening instru-
ment was obtained in a pilot study by using the Reynell
language comprehension test as gold standard. A cutoff
score of ⱕ2 was found to be the most optimal point,
allocating 80% of the children as having either true-
positive or true-negative results. More details on the
validity of the VTO, which was proved to be satisfactory,
have been published before.
6
Follow-up
Follow-up was aimed at all children in both intervention
and control groups who according to their date of birth
should normally now have been in grade 2 of primary
school, in the school years 2001–2002 and 2002–2003,
respectively. In the Netherlands, there are separate spe-
1318 VAN AGT et al
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
cial schools for children with learning problems, for chil-
dren with a visual disability, for children who are deaf
and hard of hearing and for children with severe speech
difficulties, for children with mental and/or physical dis-
abilities, and for children with behavioral difficulties. In
January of each school year, we informed all primary
schools and special schools in the regions of the study
population about the follow-up project. Then we in-
formed the parents by mail and asked them for their
written consent to obtain data on their child’s linguistic
abilities from the school and teacher (plus name and
address details of the school and teacher). Two months
later, the parents received a questionnaire and, if neces-
sary, a reminder for the informed consent. The parents
were asked a number of detailed questions about the
history of language problems (age and type of problem)
and related treatment. The parent questionnaire con-
tained some questions about background characteristics
(number of older brothers/sisters, foreign language spo-
ken at home, educational level of parents, and whether
the child had a physical or mental disability/illness),
which are known predictors for language develop-
ment.
18,19
Schools received a list with the names of the chil-
dren for which we obtained informed consent from
the parents. At the end of the school year, the teachers
of these children were asked to fill out a questionnaire
and to supply the scores on a set of widely known
specified standardized language assessment tests. In
case these specified tests were not (yet) applied in a
particular school, we included the relevant test mate-
rial in the mail parcel and asked the teacher whether
he or she was willing to administer the test(s) to the
child. We also included a book for the classroom and
a theater voucher as presents for all teachers (inde-
pendent of the response). Reminders were sent to all
parents and teachers who did not respond in the pre-
vious periods.
Participants
In 6 regions in the Netherlands, 4 regions in the south, 1
in the midsouth, and in 1 large city in the west, 55
physicians of child health centers were randomly as-
signed. Inclusion started in May 1995 in the 4 regions in
the south, in March 1996 in the midsouth region, and in
August 1996 in the city in the west. The participating
children were those who were between the age of 15 to
24 months in the given inclusion period and were living
within the area of the intervention physicians’ health
care location and those who were living within the area
of the control physician (n ⫽ 11 440).
Primary Outcome Measures
The primary outcome measures (at the individual level)
were school performance and linguistic ability at age 8.
School performance embraced school career, defined by
the type of school the child was currently attending, and
functioning at school, assessed by determining whether
the child had repeated a grade. We assessed both oral
and written linguistic abilities of the children by means
of standardized assessment tests and the independent
(blinded toward arms) judgment of teachers.
School Type and Grade
Data on school type and grade were derived from the
parent report on school address details. School type was
determined by linking name and address details of the
school to the Dutch central registry system, in which
each school is allocated a unique number and catego-
rized by school type. Children were assumed to have
repeated a grade when their grade was below grade 2
(group 4 in the Netherlands) of primary school, which
was the expected grade given the age of the included
children.
Standardized Assessment Measures
Teachers were asked to supply the scores of the individ-
ual children with respect to receptive and productive
oral and written language usage, consisting of a vocab-
ulary test
20
; spelling
21
and reading comprehension
22
tests,
which can be administered at group level; and sentence
construction
23
and technical reading
24
tests, which are to
be administered for each child individually. These out-
comes are part of the national pupil monitoring system,
which is widely used by teachers in Dutch schools to
follow the school progress of individual children in pri-
mary school. Each test comprises separate units designed
for specific measuring moments in the school year.
Norm scores consist of 5 levels, which are based on the
scores of these tests on the specified measuring moments
in a national sample (A: 25% highest scores; B: 25% just
above the average score; C: 25% just below the average
score; D: 15% far below the average score; and E: 10%
lowest scores).
Teacher and Parent Questionnaires
Apart from the standardized tests, the teacher and par-
ent questionnaires also included questions with respect
to oral and written linguistic abilities and learning. In
addition, the teacher was asked about the child’s future
development (“Do you think that in the future the child
would develop in a normal way?”).
Secondary Outcome Measures
The secondary outcome measure (at the individual
level) was the frequency of occurrence of (past) treat-
ment to spur the child’s language development, as re-
ported by the parent. Interventions for language diffi-
culties may take many forms because of the broad range
of problems as well as the broad range of underlying
causes.
5
To assess the number of children treated for
language problems in both intervention and control
PEDIATRICS Volume 120, Number 6, December 2007 1319
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
group, we therefore asked the parents the following
question: “Has there ever been anything done to spur on
your child’s language development?” Children were
identified as treated for a language problem when par-
ents confirmed the question by answering ⱖ1ofthe
following: “Yes, treatment by speech and language ther-
apist/ear, nose, and throat specialist/remedial teacher/
physiotherapist/psychologist/social worker.” In addition,
parents were asked to report the age around which their
child had been treated.
Sample Size
Under conditions of usual care, we estimated that
⬃2.5% of children would be having serious language
problems at 8 years (as defined by attending special
education). On the basis of the estimated treatment ef-
fect sizes from Law et al,
5
the observed proportions of
different types of language delays among the children
whose screening was positive at age 2, and the number
of children who as a result of the screening would be
treated for language delay,
6
we estimated that we could
reduce the percentage of serious language delays at age
8 with 20% in the intervention arm, for a significance
level of 5% (1-sided), a power of 80%, and equal allo-
cation. For a trial with randomization of individual chil-
dren to be able to detect such a reduction, a minimum of
2925 children in total would need to be recruited. Be-
cause we used cluster randomization, we required a
larger sample size to compensate for this design effect.
The formula 1 ⫹ [(m ⫺ 1) ⫻ R], where m ⫽ the number
ot children per cluster, R ⫽ s
2
b/(s
2
b ⫹ s
2
w), the intra-
cluster correlation coefficient, is used where s
2
b is the
variance between clusters and s
2
w is the variance within
clusters.
25
On the basis of the additional assumptions of
an estimated intracluster correlation coefficient of 0.005
and an average of 110 eligible and participating children
in each cluster [1 ⫹ (110 ⫺ 1) ⫻ 0.005 ⫽ 1.55], we
would need 1.55 ⫻ 2925 ⫽ 4534 children in total, which
is at least 20 clusters in each trial arm.
Analyses
Comparisons were made between intervention and con-
trol groups (intention-to-screen analysis) and between
children who completed the full screening procedure
(screened at age 15/18 months as well as 24 months)
and children who had never been screened by VTO (in
this comparison, we excluded children who were
screened at 1 age only). The primary outcomes were put
to binary variables: regular education/special education,
repeating a grade (yes/no), oral and written linguistic
abilities according to standardized tests (E level/higher
than E level), and normal future development (yes/no)
according to the teacher. We analyzed the primary out-
come variables by multilevel analysis with 2 levels (clus-
ter and child) by using EGRET 2.0.1
26
for logistic regres-
sion with distinguishable binomial random effect. To
adjust for possible regional differences, we subsequently
included region in the model. Although the trial was
designed with a 1-sided hypothesis,
27
we report the re-
sults for 2-sided 5% tests for the primary outcomes as
well to follow statistical convention.
RESULTS
Figure 1 shows the flow of children in clusters through
the trial. In all, 28 child health center physicians were
allocated to the intervention group and 27 physicians to
the control group. Thirty-six physicians were not en-
rolled because of either the very small numbers of chil-
dren in their care or their inability to meet requirements
for participation. During the follow-up period, 15% of
the cohort was not reachable as a result of a change of
address. The parents of a total of 9419 children were
asked for consent, 5424 (57.6%) of whom agreed; the
parents of 5406 children supplied usable information on
school type, and the teachers of 4329 children and the
parents of 4281 children completed detailed question-
naires. The response (written consent) in low-socioeco-
nomic neighborhoods was 53.7% (1447 of 2695),
whereas the response in middle- and high-socioeco-
nomic neighborhoods was 58.9% (3763 of 6388) and
63.9% (214 of 335; P ⫽ .000), respectively. In low-
socioeconomic neighborhoods, the response in the in-
tervention and control groups was 55.6% and 51.6%
(P ⫽ .015), in middle-socioeconomic neighborhoods was
59.4% and 58.3% (P ⫽ .537), and in high-socioeco-
nomic neighborhoods was 71.4% and 56.9% (P ⫽ .006),
respectively. Baseline and follow-up characteristics of
clusters and children were similar between arms (Table
1). Only between regions were there some differences in
the number of clusters and children.
Before the age of 2 (start of screening), there were no
differences in the cumulative percentages of reported
treatments between the intervention and control groups
(Table 2). Before age 3, 3.5% of the children in the
intervention group and 2.4% in the control group had
been treated to spur language development (P ⫽ .069).
Before age 5, the percentage of children who were ever
treated was significantly higher in the intervention
group than in the control group: 10.8% vs 8.6% (P ⫽
.024). Before the age of 9, 26.5% of the children in the
intervention group and 23.7% in the control group had
been treated to spur language development (P ⫽ .054).
The intention-to-treat analyses revealed that, in children
who were allocated to the intervention arm, the relative
risk (RR) for special school attendance was 0.71 and the
RR for the lowest level of the spelling test was 0.68,
calculated according to logistic regression with distin-
guishable binomial random effect, which takes cluster
randomization into account (Table 3). After adjustments
for regional differences, the RRs were 0.70 (95% confi-
dence interval [CI]: 0.49 –1.02; P ⫽ .063, P ⫽ .032 for
1-sided testing) and 0.66 (95% CI: 0.43–1.01; P ⫽ .054,
1320 VAN AGT et al
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
P ⫽ .027 for 1-sided testing), respectively. For the other
outcomes, no significant differences were found. In the
group of children who underwent the complete screen-
ing procedure, the RR for special school attendance was
0.75 (95% CI: 0.62– 0.91; P ⫽ .003) and the RR for
lowest level of oral language performance was 0.74
(95% CI: 0.62– 0.90; P ⫽ .002, adjusted for regional
differences; Table 4). For the other outcomes, no signif-
icant differences were found.
DISCUSSION
We have shown that early detection of language delays
in toddlers at child health care centers by means of a
specific screening instrument followed by early treat-
ment can reduce the percentage of children who attend
a special school at 8 years by 30%. At the same time, the
number of children with spelling problems was reduced
by 33%. Screened children seemed to have fewer prob-
lems with oral linguistic skills. The screening led to more
treatments and support in the preschool period.
Parent report on school type and grade proved to be
reliable; only ⬍1% of the school addresses were found to
be incorrect. With the help of the teachers, the children’s
linguistic abilities were measured by standardized tests,
which were validated and proved to be reliable in pre-
vious research.
28
Only a small proportion of teachers
were not familiar with the tests; however, this propor-
tion did not differ between the intervention and control
groups, so this could not have been a potential bias.
Importantly, none of the teachers knew whether the
child belonged to the intervention or control group.
Special school attendance can be considered to be a valid
91 eligible clusters (child health center
physicians) N = 91
36 clusters not enrolled
(too-small clusters or
inability to participate)
Lost to follow-up
Moved (n = 971 [15.0%])
Refused (n = 190 [2.9%])
Lost to follow-up
Moved (n = 752 [15.2%])
Refused (n = 108 [2.2%])
55 clusters randomized with
children aged 15 mo at entry
Analyzed
Mean (SD) age (y): 8.1 (0.4)
School type (n = 3118)
Class (n = 3084)
Oral language tests (n = 1270)
Reading tests (n = 1844)
Spelling tests (n = 1728)
Invited to participate at age 8 y (n = 5324) Invited to participate at age 8 y (n = 4095)
Analyzed
Mean (SD) age (y): 8.1 (0.4)
School type (n = 2288)
Class (n = 2250)
Oral language tests (n = 925)
Reading tests (n = 1328)
Spelling tests (n = 1225)
Intervention group (28 clusters with
6485 children)
Specific screening at 15/18 and 24
mo (n = 3776)
Specific screening at 15/18 or 24
mo (n = 560)
No specific screening/usual care
(n = 2149) (for logistic reasons, holidays,
etc
)
Control group (27 clusters with 4955
children)
No specific screening/usual care
(n = 4955)
Written consent (n = 3127 [58.8%])
Teacher questionnaires (n = 2491)
Parent questionnaires (n = 2474)
Written consent (n = 2297 [56.1%])
Teacher questionnaires (n = 1838)
Parent questionnaires (n = 1807)
FIGURE 1
Flowchart of child health physicians (clusters) and children
through the trial.
PEDIATRICS Volume 120, Number 6, December 2007 1321
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
measure of school performance, because only children
with severe educational problems attend such schools in
the Netherlands. It should be noted, however, that prob-
lems other than language might be the only reason for
special school entrance (eg, behavioral problems); there-
fore, we took account of possible differences in the re-
ferral policy for special education by adjusting for re-
gional differences.
At 3 years of age, the VTO screening had been found
to result in larger proportions of children with diagnosed
and/or treated language problems.
6
In this study, parents
in the intervention group reported significantly more
TABLE 1 Characteristics of Child Health Care Physicians (Clusters) and Children: Initially Recruited and
Follow-up at Age 8
Characteristic Intervention
Group
Control
Group
Total P
Initially recruited
Clusters
Total No. of clusters 28 27 55 —
No. of children per cluster, mean (SD) 224 (173) 184 (133) 204 (155) —
Region, No. of clusters (children)
South part (south) 4 (1153) 4 (1141) 8 (2294) —
South part (mid) 5 (1537) 4 (957) 9 (2494) —
South part (southwest) 3 (824) 2 (466) 5 (1290) —
South part (southeast) 3 (1166) 4 (1084) 7 (2250) —
Midsouth 5 (1409) 8 (1210) 13 (2619) —
Large city in west part 8 (396) 5 (97) 13 (493) —
Socioeconomic neighborhood, No. of clusters
Low 5 6 11 —
Middle 21 20 41 —
High 2 2 4 —
Children
Total 6485 4755 11440 —
Male, % 50.1 52.0 50.9 .045
Follow-up age 8
Total 3127 2297 5424 —
Age, mean (SD), y 8.1 (0.4) 8.1 (0.4) 8.1 (0.4) .862
Male, % 49.9 50.1 50.0 .926
Parental questionnaire
Physical disability/illness, % 6.8 7.8 7.2 .226
Maternal education, % .196
Low 14.6 16.6 15.4 —
Intermediate 59.9 59.2 59.6 —
High 25.5 24.3 25.0 —
Paternal education, % .921
Low 18.7 18.8 18.8 —
Intermediate 45.5 44.8 45.2 —
High 35.8 36.3 36.0 —
Foreign language in family, % 9.5 10.7 10.0 .208
Children in family, n (SD) 2.31 (0.84) 2.26 (0.82) 2.28 (0.83) .053
TABLE 2 Children Being Treated Per Age (Parent Questionnaire)
Age at Which Child Was Treated
to Spur Language
Development, y
a
Intervention (n ⫽ 2192),
Cumulative n (%)
Control (n ⫽ 1601),
Cumulative n (%)
Total (N ⫽ 3793),
Cumulative n (%)
P
2 26 (1.2) 20 (1.2) 46 (1.2) .881
3 76 (3.5) 39 (2.4) 115 (3.0) .069
4 141 (6.4) 85 (5.3) 226 (6.0) .165
5 237 (10.8) 137 (8.6) 374 (9.9) .024
6 331 (15.1) 214 (13.4) 545 (14.4) .134
7 452 (20.6) 296 (18.5) 748 (19.7) .107
8 557 (25.4) 364 (22.7) 921 (24.3) .060
9 581 (26.5) 380 (23.7) 961 (25.3) .054
a
Has there ever been anything done to spur on your child’s language development (by speech language therapist; ear, nose, and throat specialist, remedial teacher, physiotherapist, psychologist,
or social worker)?
1322 VAN AGT et al
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
treatments related to language development in their
child’s preschool period, in particular around the time of
school entrance. Given the time period, parents might be
remembering their child’s being treated for language
problems rather than the occurrence of (past) language
problems; however, these results must be evaluated with
caution, because we did not collect detailed data on
language problems from specialists as we did previously.
6
The parents of 57% of all eligible children participated in
the follow-up study. Given that we addressed an open
population and asked to sign for consent, the response is
moderate but acceptable. The response (proportion of writ-
ten consent) was only slightly lower in low-socioeconomic
neighborhoods than in middle- and high-socioeconomic
neighborhoods, so we think that the sample is still repre-
sentative of the total population. The response of the
schools and teachers was very high (82%). Although small
but significant differences in response between interven-
tion and control groups within low- and high-socioeco-
nomic neighborhoods were found, this could not have
been a potential bias of the results because we did not find
any significant differences in the distribution of educational
level of the parents between the intervention and control
groups. In addition, there were no substantial differences in
loss to follow-up and nonresponse between the study arms
and hence could not have biased the observed differences
in outcomes between the intervention and control groups
either.
The population visiting the child health centers in the
Netherlands is not a selected group: ⬃95% of all parents
visit these centers during the first year of their child’s life.
6
Only children who have severe disabilities diagnosed at
birth and need specialized care during their first year are
seen by pediatricians and rarely visit a child health center.
Most studies
11–14,29,30
on early detection of language
problems concluded that it is possible to identify children
with language problems at an early stage in the preschool
period, sometimes by 2-step screening methods. These
studies focused on the test characteristics of the screening
instrument. Comparisons across studies are difficult be-
cause there is no generally accepted definition of language
problems or gold standard, and methods of assessment
differ.
5
The most valid method would be clinical examina-
tion; however, this is not feasible in population-based stud-
ies. Apart from the sensitivity and specificity of the instru-
ment, it is important for economic reasons to consider the
proportion of children who have positive screening results
and require additional assessment. Previously, we
6
found
that the sensitivity of the VTO instrument, resulting in
2.4% of positive screenings, was between 25% and 52%,
depending on the assumed prevalence of language prob-
lems, which was based on either specialist or parent report.
Some studies
31
found higher sensitivity measures at the
expense of high referral rates. One RCT reported that the
sensitivity of a structured test and a parent-led method
was, respectively 66% and 56%
11
; however, the applica-
TABLE 3 Primary Outcome Measures at Age 8: Intervention and Control Groups (Intention-to-Screen Analysis)
Intervention Group,
N (n 关%兴)
Control Group,
N (n 关%兴)
Total, N (n 关%兴) RR 95% CI P Intracluster
Correlation
2-Sided
Testing
1-Sided
Testing
In special school 3118 (83 关2.7兴) 2288 (85 关3.7兴) 5406 (168 关3.1兴) 0.71
a
0.48–1.04 .076 .038 0.0031
0.70
b
0.49–1.02 .063 .032 0.0028
Repeating a grade 3084 (443 关14.4兴) 2250 (318 关14.1兴) 5334 (761 关14.3兴) 0.99
a
0.81–1.21 .905 .453 0.0000
0.99
b
0.81–1.22 .935 .468 0.0000
Repeating a grade because of language problems in 2401 (146 关6.1兴) 1721 (84 关4.9兴) 4122 (230 关5.6兴) 1.26
a
0.89–1.80 .196 .098 0.0070
regular primary school (parent questionnaire) 1.28
b
0.89–1.84 .189 .095 0.0072
Below 10th percentile of oral language tests 1270 (112 关8.8兴) 925 (90 关9.7兴) 2195 (202 关9.2兴) 0.88
a
0.63–1.23 .464 .232 0.0043
0.89
b
0.64–1.24 .248 .124 0.0000
Below 10th percentile of reading tests in grade 2 1844 (86 关4.7兴) 1328 (62 关4.7兴) 3172 (148 关4.7兴) 1.00
a
0.72–1.40 .994 .497 0.0000
1.00
b
0.71–1.40 .944 .497 0.0000
Below 10th percentile of spelling tests in grade 2 1728 (48 关2.8兴) 1225 (52 关4.2兴) 2953 (100 关3.4兴) 0.68
a
0.41–1.13 .138 .069 0.0064
0.66
b
0.43–1.01 .054 .027 0.0000
Do you think that in the future the child would 1769 (201 关11.4兴) 1311 (175 关13.3兴) 3080 (376 关12.2兴) 0.83
a
0.67–1.03 .096 .048 0.0000
develop in a normal way (teacher “no”) 0.83
b
0.67–1.03 .094 .047 0.0000
a
Calculated according to logistic regression with distinguishable binomial random effect, which takes cluster randomization into account.
b
Calculated according to logistic regression with distinguishable binomial random effect, which takes cluster randomization into account, adjusted for region.
PEDIATRICS Volume 120, Number 6, December 2007 1323
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
bility of these results to the general population is question-
able because the sample came from a deprived area. More-
over, according to Laing et al,
11
the low uptake and high
attrition rate had probably biased the results toward over-
estimating the performance of the screening test. Laing et
al did not recommend formal screening on language prob-
lems, because they considered the sensitivity of the struc-
tured test not to be substantially higher than that of the
parent-led method; however, we think that the sensitivity
of the VTO screening instrument pertains to a realistic and
acceptable figure, given the low referral rate and the young
age of the children.
6
Most important, this is the first study
to evaluate the effects of an early language screening pro-
gram in an RCT, to assess whether the specific screening
procedure leads to extra diagnosed and/or treated language
problems and, most important, to extra “health” benefits at
later age as compared with usual practice.
9
After all, the
results of an RCT allow inferences about causal relation-
ships between the screening and the effects.
CONCLUSIONS
We have shown that an early language screening pro-
gram including a protocol of multidisciplinary diagnostic
procedures can reduce special school entrance and lin-
guistic problems. Nationwide implementation of the in-
tervention as part of routine monitoring of children’s
general development can be recommended.
ACKNOWLEDGMENTS
This study was funded by the Health Care Insurance
Board (College voor zorgverzekeringen).
We are grateful to the schools, teachers, parents,
and children who participated in this study; Anna
Brandt, Sandra Laurier, Toon Juttmann, Marianne
Quak, and Dick Slof for assistance with the data col-
lection and computer input of the data; Jan Alberts for
designing the procedures for data management and
data processing; and Carin Reep-van den Bergh, MA,
and Paul van der Maas, MD, PhD, for general and
methodologic support.
APPENDIX 1 Items of the VTO Language Screening Instrument
17
Age, mo Items
15 1. Word production
Animals, people, toys, food/drink
2. Language comprehension
Going out, eating, where is the ball, put the
doll to bed, get a spoon, let the doll drink
3. Understanding each other
Can clearly express her/his need for food/drink,
can clearly express her/his need for help
4. Playing (parent-child interaction)
How often playing together, what is favorite
play, can clearly express that she/he wants to
play, playing alone
18 1. Word production
Animals, people, toys, food/drink
2. Playing (parent-child interaction)
How often playing together, what is favorite
play, can clearly express that she/he wants to
play, playing alone
3. Language comprehension
Getting 3 objects: sock, spoon, small block
24 1. Word production
Animals, people, toys, food/drink
2. Playing (parent-child interaction)
How often playing together, what is favorite
play, playing alone
3. Language comprehension
Body parts: eyes, mouth, belly, feet, hair, hand
TABLE 4 Primary Outcome Measures at Age 8: Screened With Specific Instrument (Completed Screens) and Never Screened With Specific
Instrument
Screened Not Screened Total RR 95% CI P Intracluster
Correlation
2-Sided
Testing
1-Sided
Testing
In special school 1980 (41 关2.1兴) 3142 (114 关3.6兴) 5122 (115 关3.0兴) 0.75
a
0.62–0.92 .005 .003 0.0044
0.75
b
0.62–0.91 .003 .002 0.0024
Repeating a grade 1961 (265 关13.5兴) 3092 (447 关14.5兴) 5053 (712 关14.1兴) 0.95
a
0.86–1.04 .256 .129 0.0000
0.95
b
0.86–1.04 .282 .141 0.0000
Repeating a grade because of language 1585 (92 关5.8兴) 2469 (142 关5.8兴) 4054 (234 关5.8兴) 0.98
a
0.84–1.15 .821 .411 0.0051
problems (in regular primary school) 0.98
b
0.84–1.15 .830 .415 0.0056
Below 10th percentile of oral language 817 (55 关6.7兴) 1271 (137 关10.8兴) 2088 (192 关9.2兴) 0.74
a
0.62–0.90 .002 .001 0.0083
tests 0.74
b
0.62–0.90 .002 .001 0.0070
Below 10th percentile of reading tests 1188 (55 关4.6兴) 1829 (88 关4.8兴) 3017 (143 关4.7兴) 0.98
a
0.82–1.16 .819 .410 0.0000
in grade 2 0.98
b
0.82–1.16 .791 .396 0.0000
Below 10th percentile of spelling tests 1127 (30 关2.7兴) 1685 (65 关3.9兴) 2812 (95 关3.4兴) 0.87
a
0.68–1.12 .287 .144 0.0099
in grade 2 0.84
b
0.66–1.08 .175 .088 0.0032
Do you think that in the future the child 1124 (221 关12.3兴) 1793 (134 关11.9兴) 2917 (355 关12.2兴) 0.98
a
0.88–1.10 .745 .373 0.0000
would develop in a normal way
(answer teacher “no”)
0.98
b
0.88–1.10 .770 .385 0.0000
a
Calculated according to logistic regression with distinguishable binomial random effect, which takes cluster randomization into account.
b
Calculated according to logistic regression with distinguishable binomial random effect, which takes cluster randomization into account, adjusted for region.
1324 VAN AGT et al
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
REFERENCES
1. First LR, Palfrey JS. The infant or young child with develop-
mental delay. N Engl J Med. 1994;330:478 – 483
2. Felsenfeld S, Broen PA, McGue M. A 28-year follow-up of
adults with a history of moderate phonological disorder: edu-
cational and occupational results. J Speech Hear Res. 1994;37:
1341–1353
3. Stothard SE, Snowling MJ, Bishop DV, Chipchase BB, Kaplan
CA. Language-impaired preschoolers: a follow-up into adoles-
cence. J Speech Lang Hear Res. 1998;41:407– 418
4. Johnson CJ, Beitchman JH, Young A, et al. Fourteen-year
follow-up of children with and without speech/language
impairments: speech/language stability and outcomes. J Speech
Lang Hear Res. 1999;42:744 –760
5. Law J, Boyle J, Harris F, Harkness A, Nye C. Screening for
speech and language delay: a systematic review of the litera-
ture. Health Technol Assess. 1998;2:1–184
6. de Koning HJ, de Ridder-Sluiter JG, van Agt HM, et al. A
cluster-randomised trial of screening for language disorders in
toddlers. J Med Screen. 2004;11:109 –116
7. Miniscalco C, Nygren G, Hagberg B, Kadesjo B, Gillberg C.
Neuropsychiatric and neurodevelopmental outcome of chil-
dren at age 6 and 7 years who screened positive for language
problems at 30 months. Dev Med Child Neurol. 2006;48:
361–366
8. Law J, Garrett Z, Nye C. The efficacy of treatment for children
with developmental speech and language delay/disorder: a
meta-analysis. J Speech Lang Hear Res. 2004;47:924 –943
9. Nelson HD, Nygren P, Walker M, Panoscha R. Screening for
speech and language delay in preschool children: systematic
evidence review for the US Preventive Services Task Force.
Pediatrics. 2006;117(2). Available at: www.pediatrics.org/cgi/
content/full/117/2/e298
10. Klee T, Carson DK, Gavin WJ, Hall L, Kent A, Reece S. Con-
current and predictive validity of an early language screening
program. J Speech Lang Hear Res. 1998;41:627– 641
11. Laing GJ, Law J, Levin A, Logan S. Evaluation of a structured
test and a parent led method for screening for speech and
language problems: prospective population based study. BMJ.
2002;325:1152
12. Westerlund M, Sundelin C. Can severe language disability be
identified in three-year-olds? Evaluation of a routine screening
procedure. Acta Paediatr. 2000;89:94 –100
13. Mattsson CM, Marild S, Pehrsson NG. Evaluation of a lan-
guage-screening programme for 2.5-year-olds at child health
centres in Sweden. Acta Paediatr. 2001;90:339 –344
14. McGinty C. An investigation into aspects of the Mayo early
language screening test. Child Care Health Dev. 2000;26:
111–128
15. Brouwers-de Jong EA, Burgmeijer RJ, Laurent de Angulo MS,
eds. Monitoring development at the Child Health Center: Manual for
the Revised Van Wiechen Examination [in Dutch]. Assen,
Germany: Van Gorcum; 1996
16. de Ridder-Sluiter JG. Early detection of communicative disor-
ders. XXV International Congress of Psychology; 19 –24 July 1992;
Brussels, Belgium: Lawrence Erlbaum Associates
17. de Ridder-Sluiter JG. Early Detection of Developmental Delays in
Communication Abilities [in Dutch]. Leiden, Netherlands: Rijks-
universiteit Leiden; 1990
18. Bishop DVM. The biological basis of specific language impair-
ment. In: Fletcher P, Hall D, eds. Specific Speech and Language
Disorders in Children: Correlates, Characteristics and Outcomes. Lon-
don, United Kingdom: Whurr Publishers; 1991:2–17
19. Stanton-Chapman TL, Chapman DA, Bainbridge NL, Scott KG.
Identification of early risk factors for language impairment. Res
Dev Disabil. 2002;23:390–405
20. Verhoeven LTW. Vocabulary Test, Manual [in Dutch]. Arnhem,
Netherlands: Cito; 1996
21. Verhoeven LTW. Test of Progress in Spelling Skills, Manual [in
Dutch]. Arnhem, Netherlands: Cito; 1997
22. Verhoeven LTW. Reading Comprehension, Manual [in Dutch].
Arnhem, Netherlands: Cito, 1997
23. Verhoeven LTW, Vermeer A. Language Test All Children, Assess-
ment of Oral Language Skills Dutch for Children in Group 1– 4 [in
Dutch]. Arnhem, Netherlands: Citogroep, 2001
24. Verhoeven LTW. Three-Minutes Test (Technical Reading) [in
Dutch]. Arnhem, Netherlands: Cito; 1997
25. Donner A. Some aspects of the design and analysis of cluster
randomization trials. Appl Stat. 1998;47:95–113
26. Diez Roux AV. A glossary for multilevel analysis. J Epidemiol
Community Health. 2005;56:588–594
27. Bland JM, Altman DG. One and two sided tests of significance.
BMJ. 1994;309:248
28. Droop M, Verhoeven L. Language proficiency and reading
ability in first- and second-language learners. Read Res Q. 2003;
38:78 –103
29. Klee T, Pearce K, Carson DK. Improving the positive predictive
value of screening for developmental language disorder. J
Speech Lang Hear Res. 2000;43:821– 833
30. Stott CM, Merricks MJ, Bolton PF, Goodyer IM. Screening for
speech and language disorders: the reliability, validity and
accuracy of the General Language Screen. Int J Lang Commun
Disord. 2002;37:133–151
31. Westerlund M. Relationship between a global rating of speech
ability at the age of 3 yrs and a phonological screening 1 yr
later: a prospective field study. Scand J Caring Sci. 2001;15(3):
222–227
APPENDIX 2 Question 1 of the Screening Instrument at 24 Months
of Age
17
Let us start with the little sounds and words of (name of the child)
Thinking about the past period, can you tell me:
a. How does (name of the child) call people in his/her proximity?
b. What does (name of the child) say if (name of the child) wants
something to eat or drink?
c. What does (name of the child) say if (name of the child) wants
to play with toys?
Possible answer categories
Sentence
Word or name
Calls by sound or indicates; for instance, br ⫽ car
Daddy, mama
Pointing out with sound
a
Pointing out without sound
a
Not indicating anything
b
a
If the parent’s answer is “pointing out,” then ask, “Does (name of the child) make any sounds
while pointing?”
b
Does (name of the child) make little sounds? If yes, then what are these sounds like?
PEDIATRICS Volume 120, Number 6, December 2007 1325
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
DOI: 10.1542/peds.2006-3145
2007;120;1317Pediatrics
W. Verhoeven and Harry J. de Koning
Heleen M.E. van Agt, Heleen A. van der Stege, Hanneke de Ridder-Sluiter, Ludo T.
Effects on School Performance and Language Development at Age 8
A Cluster-Randomized Trial of Screening for Language Delay in Toddlers:
Services
Updated Information &
tml
http://pediatrics.aappublications.org/content/120/6/1317.full.h
including high resolution figures, can be found at:
References
tml#ref-list-1
http://pediatrics.aappublications.org/content/120/6/1317.full.h
at:
This article cites 21 articles, 7 of which can be accessed free
Citations
tml#related-urls
http://pediatrics.aappublications.org/content/120/6/1317.full.h
This article has been cited by 9 HighWire-hosted articles:
Subspecialty Collections
ctice
http://pediatrics.aappublications.org/cgi/collection/office_pra
Office Practice
the following collection(s):
This article, along with others on similar topics, appears in
Permissions & Licensing
ml
http://pediatrics.aappublications.org/site/misc/Permissions.xht
tables) or in its entirety can be found online at:
Information about reproducing this article in parts (figures,
Reprints
http://pediatrics.aappublications.org/site/misc/reprints.xhtml
Information about ordering reprints can be found online:
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Grove Village, Illinois, 60007. Copyright © 2007 by the American Academy of Pediatrics. All
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
by guest on June 5, 2013pediatrics.aappublications.orgDownloaded from
