Professional Documents
Culture Documents
Kualiti Hidup
Kualiti Hidup
DOI 10.1007/s11136-004-6013-z
Heleen M.E. van Agt1, Marie-Louise Essink-Bot1, Heleen A. van der Stege2, J.G. de Ridder-Sluiter2 &
Harry J. de Koning1
1
Department of Public Health, Erasmus MC, Rotterdam (E-mail: h.vanagt@erasmusmc.nl); 2Dutch
Foundation for the Deaf and Hard of Hearing Child, Amsterdam, The Netherlands
Abstract
We investigated health-related quality of life (HRQOL) of children with language problems and controls.
Data on language development (Language Screening Instrument 3-years-olds, Van Wiechen items) and
HRQOL by means of the TNO-AZL Pre-school children Quality of Life-questionnaire (TAPQOL) were
collected at age 3 in a population-based cohort by parental questionnaire (n=8877, response 78%; mean
age 39.1 months (SD 2.0), 4347 were girls). Cronbach’s a (internal consistency) ranged between 0.63 and
0.85. Dependent on the definition of language problem, 131 to 316 children appeared to be language
impaired. Receiver Operating Characteristic analyses (ROC-curves) to assess the discriminative ability of
six TAPQOL scales revealed that the Communication scale and Social Functioning scale discriminated best
between children with language problems and children without these problems. Language-impaired chil-
dren had significantly lower scores on the Communication scale and Social Functioning scale as compared
to children without language problems (p<0.01). The findings indicate that language problems at age three
can have an impact on children’s social life. These results provide additional evidence for the importance of
monitoring the language development and its consequences during childhood.
Key words: Health Related Quality of Life, Language development, Pre-school children
structed. The discriminative abilities of the TAP- in general, the percentages of missing values in all
QOL-scales were compared by calculating the area groups were low, ranging from 1.9 to 6.7 %, which
under the ROC curves [25, 26]: a value of 0.50 indicates that the TAPQOL is feasible in this
means that discrimination is equal to chance; if population. Five from the six scales appeared to
values are between 0.50 and 1.0, this means that meet the standard of Cronbach’s a for group
discrimination is better than chance. comparison.
Construct validity
Results
Table 4 shows the PCC correlation coefficients
Feasibility and reliability between the TAPQOL-scales and the language
development measures (comprehension and pro-
Table 3 presents an overview of the missing values duction) based on the data from all children in the
and Cronbach’s a of the TAPQOL-scales in the response group. It appeared that the Communi-
total population and in subgroups of children cation scale correlated relatively high with both
identified with language problems (according to language development measures, whereas for all
the expert-panel or clinicians and according to other scales including Social Functioning the cor-
parents). As expected, the completion of the 2-step relation between these measures was very low.
items of the Communication scale (Appendix II) Table 4 also presents the correlations between the
was more complex than the other items, yielding objective and subjective part of the Communica-
the highest percentage of missing values. However, tion scale and the language development measures.
Table 3. Cronbach’s as of 6 TAPQOL-scales in all children and in subgroups of children identified as having diagnosed language
problems according to two grouping variables
Table 4. Construct validity: Pearson correlation coefficients between TAPQOL scales and the measures of language comprehension
(LSI-CT) and language production (LSI-PF + VW) in all children
0.83
0.52
0.52
0.51
0.53
0.60
The first factor seemed to represent a language
development dimension and the second appeared
p-Value (MWU)
to reflect a HRQOL dimension. All TAPQOL
Table 5. Discriminative ability of TAPQOL and differences in TAPQOL-scores between children with and without language problems (N=8877)
scales except for Communication had high factor
loadings (P0.30) on the second dimension and low
factor loadings on the first one. The language
0.25
0.40
0.44
0.01
<0.01
<0.01
measures and the Communication scale had high
factor loadings on the first factor and low factor
loadings on the second factor.
In the second factor analysis, the separate scores
No (N=8561)/Language
of the ‘objective’ and ‘subjective’ part of the
(according to panel)
Communication scale were included. Both Com-
problem (N=316)
(19)
(19)
(22)
(11)
(16)
(23)
munication scores had high factor loadings on the
(13)/68.4
(16)/59.3
(19)/69.5
(10)/96.6
(12)/93.1
(15)/83.4
‘language development’ factor (0.57 and 0.43).
Mean (SD)
97.4
89.0
62.8
71.8
95.7
92.2
Discriminative ability
Area under ROC
0.54
0.62
according to parents). It shows that the Commu-
nication scale and the Social Functioning scale are
p-Value (MWU)
0.01
<0.01
<0.01
(17)
(24)
(12)/92.1
(15)/82.3
95.7
92.2
language problems
Social functioning
Positive mood
the TAPQOL, as it was a unique opportunity to However, we could not use this instrument, as it
collect data on HRQOL in a new population, is only suitable for children up to 30 months of
moreover taking account of the large scale of the age (we sent the parent questionnaire just around
study. Secondly, we selected the scales in a the child’s third birthday). However, besides the
careful way with the help and consent of the items in the language measures there were more
designers of the TAPQOL. As a result, the scales items in the questionnaire measuring explicitly
referring for the most part to physical health, linguistic knowledge, e.g. one item inquiring
namely gastric problems, appetite, skin problems, about the type of language problem of the child
motor functioning, pulmonary problems and (‘My child’s vocabulary contained too few words
sleep were not selected. Thirdly, we re-examined for his/her age’ was one of the answer categories)
reliability and validity of the included scales and another item asked ‘does your child say more
which proved to be satisfactory as was shown in than 50 words?’. Other items more implicitly refer
this paper. Given these careful considerations, to linguistic knowledge. So, we think the parent
we think the selection of scales is justified in this questionnaire contained a sufficient number of
particular instance. However, we certainly should items referring explicitly or implicitly to linguistic
not recommend this in general. knowledge.
In defining language problems the gold stan- Differences in HRQOL between children with
dard is less clearly stated as for well-defined language problems and children without these
diseases such as cancer or diabetes. With devel- problems may have been (partly) due to differences
opmental disorders such as language delays the in socio-demographic characteristics between these
distinction between delays that fall within the groups. Children with language problems were
natural biological variation and delays or disor- more often boys and had less often highly educated
ders that require intervention is much more dif- mothers than children without language problems.
ficult. There is a norm for language production However, adjusting for these factors in the analyses
(>50 words at 2 years of age) [3] which was did not affect the observed differences in TAPQOL-
used to identify serious language delays at the scores between children with and without language
SHC in our study, but we had to make use of problems. This means that differences in HRQOL
proxies to assess whether there was or had been can be attributed to language problems.
a language problem or not in the total cohort. If the severity of language problems differed
To identify children with language problems we between responders and non-responders, non-re-
made use of parent reports as well as specialists sponse could have caused bias. Therefore we
and an expert-panel. Although two of our cri- compared the scores on the language production
teria to identify children with language problems score and the language comprehension score
consisted of external sources, namely informa- between responders and non-responders in chil-
tion from specialists who had actually seen the dren who were assessed for language delay at
child (at SHC or otherwise) and an expert-panel, Speech and Hearing Centres. We found that the
who blindly assessed written case descriptions, average scores on both language measures in the
we ourselves did not actually see these children non-response group were significantly lower than
which could be a limitation of the study. Inter- the average scores in the response group. So, the
estingly, despite these different sources it turned impact of language problems on HRQOL might
out that the results seem to apply irrespective of in fact be higher in the total group than esti-
the criterion used. mated in this study.
Many items of the language measures in the We think the results are generalisable, because
parent questionnaire pertain to the social use of our population consisted of children from the
language (as ‘speech fluency’, ‘speech understood general population. All children resident in the
by strangers’ or understanding questions), working area of the health centres in the selected
whereas the number of items concerning linguistic regions were included. The selection of regions
knowledge seemed to be limited. The MacArthur was pragmatic and not based on knowledge
CDI should be a useful instrument to measure about the prevalence of language problems in
productive vocabulary and grammar skills [30]. these areas.
1353
The results seem to indicate that children with and may lead at a later age to behavioral
and without language problems did not differ in problems and learning problems at school. These
problem behaviour like short-tempered or results provide additional evidence for the
hostility, nor did they differ in being cheerful, importance of monitoring the language develop-
content and happy, nor in being afraid, tense and ment during childhood [31].
anxious. However, it is not possible to determine
whether this was actually true or because these
scales were less sensitive, as all children had high
scores on these scales. This seems to be an issue Acknowledgements
for further investigation. However, at age three,
children with and without language problems not The Health Care Insurance Board (‘College voor
only appeared to differ in communication abilities zorgverzekeringen’) financially supported this
like understanding what other people say and study. The authors wish to thank all personnel of
expressing one’s will, they also seemed to differ the child health care centres, the staff of the six
on social functioning capabilities like being at Home Care institutions (Het Groene Kruis
ease with other children and feeling sure with Oostelijk Zuid-Limburg, Het Groene Kruis
other children. In some subgroups, children with Heuvelland, Stichting Thuiszorg Midden Lim-
language problems seemed to be less full of en- burg, Stichting Thuiszorg Westelijke Mijnstreek,
ergy, less active and less brisk than children Stichting Thuiszorg Midden Brabant and Stich-
without these problems. ting Thuiszorg Den Haag), the staff and per-
We demonstrated that language problems at sonnel of the Speech and Hearing Centres
the age of three can have an impact on chil- (Stichting Audiologisch Centrum Hoensbroeck,
dren’s social life. Parents say they experience Audiologisch Centrum Eindhoven, Stichting Au-
more problems in understanding other people, in diologisch Centrum voor Tilburg en omstreken,
speaking clearly and coherently and in expressing Haags Audiologisch Centrum ‘Effatha’, Centrum
one’s will. Moreover, they seem to be less voor Auditief en Communicatief Beperkten ‘St.
capable in playing with other children pleasantly, Marie’ Eindhoven and Stichting Thuiszorg Rot-
they are less at ease with other children and they terdam), as well as the General Practitioners,
feel less sure of themselves with other children. If Speech and Language therapists, ear-, nose- and
children’s social life is affected this may have throat- specialists, paediatricians and other pri-
serious consequences for their cognitive and mary-care workers in the research area for their
social-emotional development and well being, collaboration.
22. Gerritsen FME. VTO Language-screening 3- to 6-years 27. Veen S, et al. Quality of life in preschool children born
Olds: The Development of Language Screening Instru- preterm. Dev Med Child Neurol 2001; 43(7): 460–465.
ments to Use in Youth Health Care (in Dutch). Leiden: 28. Stoneham G. Friendship skills in children with specific
Rijksuniversiteit Leiden, 1988. language impairment. Int J Lang Commun Disord 2001;
23. Brouwer-de Jong E, Burgmeijer R, Laurent de Angelo M. 36(Suppl): 276–281.
Monitoring Development at the Child Health Care Cen- 29. Jerome AC, et al. Self-esteem in children with specific lan-
tre. Manual for the revised Van Wiechen examination (in guage impairment. J Speech Lang Hear Res 2002; 45(4):
Dutch). Assen: van Gorkum, 1996. 700–714.
24. Streiner DL, Norman GR. Health Measurement Scales: A 30. Fenson, L., et al., Variability in early communicative de-
Practical Guide to their Development and Use. Oxford/ velopment. Monogr Soc Res Child Dev 1994; 59(5): 1–173;
New York/Tokyo: Oxford University Press, 1989. discussion 174-185.
25. Hanley JA, McNeil BJ. The meaning and use of the area 31. Law J, Conti-Ramsden. Treating children with speech and
under a receiver operating characteristic (ROC) curve. language impairments. Br Med J 2000; 321(7266): 908–909.
Radiology 1982; 143(1): 29–36.
26. Essink-Bot ML, et al. An empirical comparison of four
generic health status measures. The Nottingham Health Address for correspondence: Heleen M.E. van Agt, Department
Profile, the Medical Outcomes Study 36-item Short- of Public Health, Erasmus MC, P.O. Box 1738, 3000 DR
Form Health Survey, the COOP/WONCA charts, and Rotterdam, The Netherlands
the EuroQol instrument. Med Care 1997; 35(5): 522– Phone: +31-0-10-4087714; Fax: +31-0-10-4089449
537. E-mail: h.vanagt@erasmusmc.nl