Escolar Documentos
Profissional Documentos
Cultura Documentos
Departments of aNeurology/Neurosurgery, bPediatrics, and cEducational and Counseling Psychology and dSchool of Physical and Occupational Therapy, McGill
University, Montreal, Quebec, Canada
The authors have indicated they have no financial relationships relevant to this article to disclose.
ABSTRACT
OBJECTIVES. Our goal for this study was to prospectively test whether parent-com-
pleted questionnaires can be effectively used in the setting of a busy ambulatory
www.pediatrics.org/cgi/doi/10.1542/
pediatric clinic to accurately screen for developmental impairments. Specific ob- peds.2006-0466
jectives included (1) assessing the feasibility of using parent-report instruments in doi:10.1542/peds.2006-0466
the setting of a community pediatric clinic, (2) evaluating the accuracy of 2
Key Words
available screening tests (the Ages and Stages Questionnaire and Child Develop- developmental screening, parent-report,
ment Inventory), and (3) ascertaining if the pediatrician’s clinical judgment could questionnaires, infant
assigned to 1 of 2 groups and completed either the Ages and Stages Questionnaire Address correspondence to Michael I. Shevell,
MD, CM, Montreal Children’s Hospital, Room
or Child Development Inventory. The child’s pediatrician also completed a brief A-514, 2300 Tupper, Montreal, Quebec,
questionnaire regarding his or her opinion of the child’s development. Those Canada H3H 6P3. E-mail: michael.shevell@
muhc.mcgill.ca
children for whom concerns were identified by either questionnaire underwent PEDIATRICS (ISSN Numbers: Print, 0031-4005;
additional detailed testing by the Battelle Development Inventory, the “gold Online, 1098-4275). Copyright © 2006 by the
American Academy of Pediatrics
standard” for the purposes of this study. An equal number of children scoring
within the norms of the screening measures also underwent testing with the
Battelle Development Inventory.
RESULTS. Of the 356 parents contacted, 317 parents (90%) agreed to participate.
Most parents correctly completed the Ages and Stages Questionnaire (81%) and
the Child Development Inventory (75%). Predictive values were calculated for the
Ages and Stages Questionnaire and the Child Development Inventory (sensitivity:
0.67 and 0.50; specificity: 0.39 and 0.86; positive predictive value: 34% and 50%;
negative predictive value: 71% and 86%, respectively). Incorporating the physi-
e1178 RYDZ et al
Downloaded from pediatrics.aappublications.org at Nanyang Technological University on June 2, 2015
cian’s opinion regarding the developmental status of the a study from 1987 demonstrated that only 20% to 30%
child did not improve the accuracy of the screening of developmentally impaired children were being iden-
questionnaires. tified before school age by primary care practitioners
using developmental surveillance.5 For this reason, pro-
CONCLUSIONS. Three important conclusions were reached: fessional organizations advocate the use of standardized
(1) parent-completed questionnaires can be feasibly developmental screening tests.2 Developmental screen-
used in the setting of a pediatric clinic; (2) the pediatri- ing has been shown to improve the accuracy with which
cian’s opinion had little effect in ameliorating the accu- children’s developmental delays are identified when
racy of either questionnaire; and (3) single-point accu- compared with decisions based only on clinical judg-
racy of these screening instruments in a community ment.6–8 Indeed, the sensitivity and specificity of screen-
setting did not meet the requisite standard for develop- ing instruments are usually reported between 70% and
ment screening tests as set by current recommendations. 90%.9–12 Although the use of screening tests would im-
This study raises important questions about how devel- prove the rate and accuracy of identification, a recent
opmental screening can be performed, and we recom- survey demonstrated that only 23% of primary care
mend additional research to elucidate a successful clinicians used a standardized screening tool, leading the
screening procedure. authors to conclude: “Our findings do raise. . .the con-
cerns that systems of care that foster the proper use of
adequate detection methods in the primary care setting
e1180 RYDZ et al
Downloaded from pediatrics.aappublications.org at Nanyang Technological University on June 2, 2015
gold standard for the purpose of this study), which was were performed by using SPSS 12.0 (SPSS Inc, Chicago,
administered approximately 3 months after the initial IL), and all data entered were verified for errors.
screen by an experienced pediatric occupational thera-
pist who was blinded to both the screening instruments Measures
used and the results obtained. The 3-month interval
used in this study represents the 3-month referral wait BDI
typically experienced by families awaiting a more thor- The BDI,28 a psychometrically sound developmental as-
ough developmental assessment, and at 21 months of sessment for children from birth to 8 years, is widely
age, the child falls in the middle of the 18- to 23-month used in studies evaluating developmental delay and in
marking scheme of the BDI, giving the most accurate US-mandated early intervention programs. The BDI
estimation of his or her real abilities according to the gathers information, via a structured test format, to in-
parameters of this measure. clude interviews with the caregiver and direct observa-
To investigate false-negative estimates, an equal tion of the child. Items are assigned an age level at which
number of children scoring within the norms of the 75% of the norming population was able to perform. It
screening questionnaires were selected as controls and is composed of 341 items that evaluate 5 different do-
also underwent BDI testing. Controls were the next child mains: personal/social, adaptive, motor, communica-
of the same gender participating in the study who had tion, and cognitive, with each domain further subdi-
scored normally on the parent-completed screening in- vided into subdomains. Items are scored as typically
strument. A false-negative estimate for this approach signifying fully developed skills, sometimes signifying
was also obtained in a likewise manner. In this fashion, emerging abilities, or rarely signifying absent skills and
the control group was selected in an unbiased manner. are scored at values of 2, 1, or 0, respectively. The
Because of constraints posed by funding and time avail- authors of the test recommend the test for many func-
able from the occupational therapist, only 101 children tions including general screening, detailed assessment of
underwent BDI testing, after which ongoing study re- children who have previously been identified as having
cruitment was terminated. Although this is not as favor- developmental delay, or identifying strengths and weak-
able as testing all the participants with the BDI, 101 nesses of normal or developmentally impaired children.
completed BDIs represents a large enough sample to be The manual states that the assessment may take up to 2
a good indicator of the accuracy of the ASQ and CDI. hours to administer with children ⬍3 years of age.
The pediatrician who provided care to the child was
blinded to the actual questionnaire used and the results ASQ
obtained but was asked as part of the study to answer a The ASQs29 are 19 parent-report questionnaires that
question identical to the one given to the parent regard- span the age range of 4 to 60 months. Questionnaire
ing any possible concerns about the child’s development. points include 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 27,
This question was answered at the time of the subject’s 30, 33, 36, 42, 48, 54, and 60 months of age. Each
visit by simply checking the appropriate response questionnaire is composed of 3 sections: a brief set of
box(es). The results of this simple questionnaire were demographic items; 30 questions about the infant’s or
used in the analysis to establish what role the pediatri- child’s development assessing 5 different domains
cian could play in influencing the accuracy of the ASQ equally (communication, gross motor, fine motor, prob-
and CDI. lem-solving, and personal/social); and 7 open-ended
Parents who had agreed over the telephone to partic- questions eliciting parental concerns. The choice of re-
ipate but did not return the questionnaire were con- sponses for each item is “yes,” “sometimes,” or “not yet,”
tacted by telephone and reminded to return the ques- which are scored as 10, 5, or 0, respectively. The test is
tionnaire. If the questionnaire had not been returned 3 graded according to the domain tested and compared
months after it was due, the parents were sent a demo- with an empirically derived screening cutoff score and
graphic questionnaire to assess whether any parameters takes ⬃10 to 15 minutes to complete and ⬃5 minutes to
differentiated these parents from those who actually score.
completed the questionnaire. Families with whom we Although the ASQ states that the questionnaire can
had no personal contact by telephone were not included be used in-clinic, the ASQ is meant for mail-out pur-
in the study. poses to be completed at home at the specific age inter-
Throughout this study, data obtained were stored and vals listed above. Because of this, the instructions ask the
managed on secure computers within the Montreal parents to attempt every activity with their child and, if
Children’s Hospital Division of Pediatric Neurology with the child is noncompliant, to try at a later time. Unfor-
separation of subject identifier variables. Ethical ap- tunately, this is not feasible in the context of a busy
proval for conduct of the study was obtained from the waiting room. Thus, these instructions were abridged,
Montreal Children’s Hospital–McGill University Health and parents were asked to base their responses on their
Centre Research Ethics Board. Data storage and analysis observations and previous experiences with their child.
e1182 RYDZ et al
Downloaded from pediatrics.aappublications.org at Nanyang Technological University on June 2, 2015
TABLE 2 Descriptive Data for 3 Populations ASQ and the CDI being negligible. Table 3 summarizes
True False Refusal these data.
Participants Participants
Child gender, n (%) Psychometric Values
Male 132 (53) 34 (53) 22 (56)
Female 116 (47) 30 (47) 17 (44) Psychometric Values of the Screening Questionnaires
Age Using the BDI as the gold standard, we assessed the ASQ
Mean ⫾ SD, mo 18.4 ⫾ .64 NA NA and the CDI for their psychometric properties: sensitiv-
Range, mo 17.03–20.47 NA NA ity, specificity, positive predictive value, and negative
Gestational age, mean ⫾ SD, wk 38.6 ⫾ 2.0 39.0 ⫾ 2.7 38.5 ⫾ 2.5
predictive value. Not one of these questionnaires proved
Preterm (⬍36 wk gestation), n (%) 20 (8) 3 (5) 2 (6)
Term (⬎36 wk gestation), n (%) 218 (92) 56 (95) 33 (94) to be an ideal screening instrument. The ASQ had mod-
NA indicates not applicable.
erate sensitivity (0.67) but poor specificity (0.39). Con-
versely, the CDI had poor sensitivity (0.50) but excellent
specificity (0.86). Table 4 summarizes the predictive val-
ues for the ASQ, CDI, and pediatrician’s questionnaire.
domains. Of the 53 children who failed the screening
tool, 41 continued study participation and underwent
Psychometric Values of Incorporating the Pediatrician’s
BDI assessment (ie, 12 dropped out).
Opinion With the Parent-Completed Questionnaires (ASQ
Conversely, of the 114 children who were assessed by
and CDI)
the CDI, 11 failed (failure rate: 9%). Expressive language
The physician’s opinion was incorporated with the par-
was again the most likely domain to be failed (5), fol-
ent-completed questionnaires to determine if this could
lowed by the gross motor, (5) fine motor, (1) self-help
be used as a potential modifier. As a result, those chil-
(1), and language comprehension (1 failure) domains.
dren (1) who had failed the parent-completed question-
Of the 11 children who failed the CDI, 8 continued study
naire and (2) for whom concern was listed by the pedi-
participation and undertook BDI assessment (ie, 3
atrician were grouped into a new category as those who
dropped out).
had failed both screening instruments. Likewise, those
In total, 101 BDI assessments were performed. Twen-
(1) who had passed the parent-completed questionnaire
ty-nine children received a failing score. Similarly, the
and (2) for whom no concern was listed by the pedia-
domain that was failed most often was communication
(17), followed by the gross motor (11), cognition (6),
personal-social (3), and adaptive (1) domains. Of the
101 BDIs performed, 49 were children who had previ- TABLE 3 Assessment of the Questionnaires’ Feasibility
ously failed one of the questionnaires, whereas 52 in- ASQ
fants were control subjects. Of the 101 BDI tests per- N 183
No. of questionnaires distributed 165
formed, 41 failed ASQs and 8 failed CDIs were assessed
No. (%) completed 134 (81)
with the BDI, and the remaining 52 BDI assessments No. returned late 37
were performed on control subjects. No. (%) of questionnaires not returned 28 (17)
No. (%) of questionnaires incomplete 3 (2)
Ease of Questionnaire Completion Parent’s opinion of questionnaire (N ⫽ 1; mean: 1.5 ⫾ .6), n (%)
Very easy 75 (57)
Closer examination was given to the rates of completion
Easy 50 (38)
and the parents’ opinion of the questionnaires to assess Neutral 6 (5)
feasibility and ease of use. Overall, the CDI was more Difficult 1 (1)
likely to be either returned late (27% for CDI vs 22% for Very difficult 0
ASQ) or not returned at all (23% for CDI vs 17% for CDI
N 171
ASQ). This difference was to be expected, because the
No. of questionnaires distributed 152
CDI contains more items to answer and, thus, is more No. (%) completed 114 (75)
labor intensive. Overall, 81% of the ASQs were returned No. returned late 41
completed, and 75% of the CDIs were returned com- No. (%) of questionnaires not returned 36 (23)
pleted. No. (%) of questionnaires incomplete 2 (2)
Parent’s opinion of questionnaire (N ⫽ 112; mean: 1.6 ⫾ .7), n (%)
The Likert-type questionnaire was used to assess the
Very easy 54 (48)
parent’s opinion of the questionnaires. The options (very Easy 49 (44)
easy, easy, neutral, difficult, and very difficult) were Neutral 8 (7)
given the numerical values 1 through 5, respectively. Difficult 1 (1)
The ASQ received a mean value of 1.5 ⫾ 0.6, whereas Very difficult 0
the CDI received a mean value of 1.6 ⫾ 0.7. Therefore, Completion rate and the parental ranking of the ASQ and the CDI. (To collect data on the
parent’s opinion of the questionnaire, the question “Did you find this questionnaire easy to
the majority of the parents ranked the questionnaires as complete? Place a check mark in the box that corresponds to your answer” was asked. The
either very easy or easy, with the difference between the number and types of responses are listed.)
e1184 RYDZ et al
Downloaded from pediatrics.aappublications.org at Nanyang Technological University on June 2, 2015
area. This is a problem that seems to have persisted over can incorporate this particular strength within a screen-
decades. In 1979 Shonkoff et al lamented that “four ing procedure. Moreover, we found that a “one-shot”
fifths of physicians . . . viewed their formal training in screening protocol might not be sufficient to identify
this area as inadequate. . . . [A]lmost 2/3 did not feel developmental delay accurately. Indeed, the AAP31 and
that practical experience was an adequate substitute for other researchers have come to the same conclusion. For
formal training in developmental assessment skills.”3 example, Darrah et al32 concluded that screening should
This reality still persists ⬎20 years later. A 2000 study involve multiple time points and multiple domains be-
performed by the AAP found that 64% of primary care fore referrals are made. A similar statement was issued
physicians reported inadequate training in developmen- by the AAP Committee on Children With Disabilities.2
tal assessment.31 As a result, an improvement in screen- This study has several limitations. First, the ASQ was
ing accuracy does not just necessitate more accurate used as an in-clinic questionnaire; although the manual
developmental screening tests but also a joint effort from states that it can be used in this fashion, this screening
different facets of the professional, public, and research instrument was originally constructed as a take-home
community advocating for more-thorough training in questionnaire so that the parent could attempt to ad-
childhood development and impairments. dress the items over several days with the child. By
limiting the time to ⬃15 minutes, with the parent rely-
Accuracy of the ASQ and CDI ing on past experiences to answer the questions, the
The psychometric properties of the ASQ and CDI were accuracy of the instrument could have been somewhat
surprising, because several studies for both the ASQ and compromised. Second, only 36 subjects underwent both
the CDI have advocated their accuracy in screening in- the CDI and BDI. This limits the ability to evaluate the
fants for developmental delay. A factor that may have accuracy of the CDI, in particular with respect to sensi-
reduced the accuracy of the ASQ and CDI in our study
tivity, because there were few subjects who actually
was the time difference between screening the child and
failed the CDI. Thus, a larger sample size may be needed
performing the full assessment. Unlike other projects in
before drawing clearer conclusions regarding the CDI’s
which the gold standard was applied immediately after
performance as a screening tool. Finally, the community
the screening instrument, our study imposed a 3-month
in which we were conducting the study was predomi-
waiting period representing the “real-life” delay typically
nantly middle class with a high school education and
experienced between screening and referral. Previous
some postsecondary attendance. Our ability to general-
studies have demonstrated that 3 months represents a
ize the results to the general population, particularly to
time interval within which an infant’s developmental
disadvantaged populations, is limited.
status may change. Darrah et al32 found that, when
This study does lay the foundation for future studies
assessing fine motor skills, gross motor skills, and com-
on screening. First, and most important, there is the
munication, the majority of children tested serially had
40 to 60 percentile fluctuations in performance. Changes possibility of assessing the benefits of ⱖ2 serial screening
of this magnitude took place at least once within 5 tests at different points in time. Because the develop-
testing points (ie, 9, 11, 13, 16, and 21 months of age). mental trajectory, especially at younger ages, is subject
Consequently, a wait of 3 months could result in a to large variability, a second screening effort (either
different developmental estimate. Because of the insta- completed in-clinic or sent by mail) has the potential to
bility and discontinuity of a child’s development, what improve the accuracy of the initial identification process.
might have been identified as a transient weakness at 18 To decrease the amount of time used to score the ques-
months could have resolved itself by 21 months. The tionnaire, application of computer software could be
developmental domain of communication also seems to implemented (ie, questionnaire completed and scored
be too variable at the 18-month age to provide a reliable on a hand-held device). This would reinforce the prac-
indicator between normality and abnormality. At this ticality of these types of screening instruments by de-
age, the “normal” differences in the actual times that creasing the effort required from office staff and person-
communication milestones are reached by children are nel. Finally, a study conducted on a more representative
too variable,2 which demonstrates the importance of the population incorporating a larger sample size would
time of actual assessment and the advantage of regular, clarify issues of applicability, feasibility, and yield. With
ongoing multiple screening efforts. additional research, the knowledge gap between avail-
Overall, neither test provided acceptable screening able screening tests and developing a practical protocol
properties at 18 months of age; therefore, a screening for their application can be bridged. Such efforts would
protocol suitable to the setting of a clinic cannot consist have considerable clinical implications with respect to
solely of either instrument applied at one point in time. systematically enhancing the capability for earlier iden-
However, our results demonstrate that the CDI identi- tification of children with developmental delays, which
fied children with normal development with high accu- should expedite intervention and, theoretically, opti-
racy (ie, high negative predictive value). Future studies mize eventual outcome.
e1186 RYDZ et al
Downloaded from pediatrics.aappublications.org at Nanyang Technological University on June 2, 2015
Screening for Developmental Delay in the Setting of a Community Pediatric
Clinic: A Prospective Assessment of Parent-Report Questionnaires
David Rydz, Myriam Srour, Maryam Oskoui, Nancy Marget, Mitchell Shiller, Rena
Birnbaum, Annette Majnemer and Michael I. Shevell
Pediatrics 2006;118;e1178
DOI: 10.1542/peds.2006-0466
Updated Information & including high resolution figures, can be found at:
Services http://pediatrics.aappublications.org/content/118/4/e1178.full.
html
References This article cites 27 articles, 17 of which can be accessed free
at:
http://pediatrics.aappublications.org/content/118/4/e1178.full.
html#ref-list-1
Citations This article has been cited by 12 HighWire-hosted articles:
http://pediatrics.aappublications.org/content/118/4/e1178.full.
html#related-urls
Permissions & Licensing Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
http://pediatrics.aappublications.org/site/misc/Permissions.xht
ml
Reprints Information about ordering reprints can be found online:
http://pediatrics.aappublications.org/site/misc/reprints.xhtml
The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://pediatrics.aappublications.org/content/118/4/e1178.full.html