© The Author 2006. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved.
For Permissions, please email: email@example.com
Human Reproduction Page 1 of 8
Cognitive and motor development of 8-year-old
children born after ICSI compared to spontaneously
L.Leunens1,3, S.Celestin-Westreich1, M.Bonduelle2, I.Liebaers2 and I.Ponjaert-Kristoffersen1
1Developmental and Lifespan Psychology and 2Academisch Ziekenhuis, Centre for Medical Genetics, Vrije Universiteit Brussel,
3To whom correspondence should be addressed at: Pleinlaan 2, B-1050 Brussels, Belgium. E-mail: firstname.lastname@example.org
BACKGROUND: As a continuation of two large-scale, multicentre studies on the development of 5-year-old ICSI
children, we present results of the follow-up study undertaken on the cognitive and motor development of 8-year-old
ICSI children. METHODS: Developmental outcomes of 151 8-year-old singletons born through ICSI after 32 weeks
of gestation were compared with those of 153 singletons of the same age born after spontaneous conception (SC). Part
of this population was seen in a cohort at the age 5 years. Outcome measures include Wechsler Intelligence Scale for
Children-Revised (WISC-R) and Movement Assessment Battery for Children (ABC). RESULTS: Regarding intellec-
tual functioning, ICSI children tend to obtain significantly higher total (P < 0.01), verbal (P < 0.01) and performance
(P < 0.05) intelligence scores than SC children, nevertheless remaining in similar ranges. These effects are small
(Cohen’s d < 0.50). High maternal educational level stayed in the regression as a factor accounting for some of the
variance in total IQ between the groups. In terms of motor development, no significant differences were found
between ICSI and SC children regarding overall motor skills, manual, balance and ball skills. CONCLUSION: In this
follow-up study, ICSI and SC children show a comparable cognitive and motor development until the age of 8 years.
Key words: child follow-up/cognitive/ICSI/motor development
Since its introduction in the early 1990s, ICSI has become a
widespread assisted reproduction technique (ART) for couples
struggling with male infertility. However, the specificities of
this technique have raised concerns about its medical outcomes
for ICSI-born children. These concerns refer mainly to poten-
tial changes in genetic material, the possible transmission of
foreign genetic material, the use of immature or senescent
germ cells and associations between genetic disorders and
some forms of male infertility (Bowen et al., 1998; te Velde
et al., 1998; Tournaye, 2003). From a strictly medical perspec-
tive, an increased risk of adverse neonatal outcome has been
documented (Helmerhorst et al., 2004; Jackson et al., 2004;
Lie et al., 2005). Meta-analyses have shown a 30–40%
increase in congenital malformations in children born after
ICSI (Hansen et al., 2002; Rimm et al., 2004; Kallen et al.,
2005). Nevertheless, medical follow-up studies have so far
been reassuring overall, because they have failed to document
detrimental outcomes for children born through the ICSI tech-
nique (Kurinczuk, 2003; Devroey and Van Steirteghem, 2004;
Bonduelle et al., 2005). From a broader developmental per-
spective, it has been suggested that ICSI might also impede
children’s psychological development. Several studies have
addressed these questions and have so far yielded mixed to
reassuring findings regarding young ICSI children’s motor and
Overview of previous studies
One of the first studies in this field, in which 1-year-old ICSI-
born children were compared with spontaneously conceived
(SC) and IVF children, revealed an increased risk of mildly to
significantly delayed development (Bowen et al., 1998). How-
ever, a concurrent single-centre Belgian follow-up study found
no signs of delayed development in 2-year-old ICSI children
(Bonduelle et al., 1998).
Replication studies in a larger number of countries with ICSI
children aged up to 5 years show no delayed development
(Sutcliffe et al., 1999, 2001, 2003a,b; Neri et al., 2002; Bonduelle
et al., 2003; Leslie et al., 2003; Place and Englert, 2003;
Squires et al., 2003; Wennerholm et al., 2003; Papaligoura et al.,
2004), although generalization of these findings should be
treated with caution, given many methodological limitations, such
as relatively small sample sizes, lack of formal child assessment
and demographic matching of samples.
Large-scale, multicentre studies were undertaken to over-
come such limitations by investigating the long-term effects of
Hum. Reprod. Advance Access published August 2, 2006
by guest on May 29, 2013
L.Leunens et al.
ICSI across diverse medical systems. In this context, no signi-
ficant differences were found on WPPSI-R (Wechsler, 1990)
IQs between 5-year-old ICSI and SC children from Belgium,
Sweden and USA (Ponjaert-Kristoffersen et al., 2004). When
analysing subtest scores, however, ICSI children appeared
more likely to perform better on general factual/acquired
knowledge (information) and less well for some visual–spatial
abilities (mazes, block design and object assembly), which was
attributed to a higher incidence of low birth weight and prema-
turity in the ICSI study group.
A European collaborative follow-up study compared 5-year-old
singleton ICSI-born children with IVF and SC controls (Ponjaert-
Kristoffersen et al., 2005), recruited in Belgium, Sweden,
Denmark, Greece and UK and matched according to age, sex,
birth order, maternal education, parental socioeconomic status
and mother’s age at birth. Although no significant differences
for intelligence between groups were found, older maternal age
at birth was significantly linked to lower full-scale IQ (FSIQ)
and verbal IQ in ICSI/IVF children, as well as lower abilities to
anticipate relationships among parts (object assembly) in IVF
children. Although no significant differences for intelligence
between groups were found, older maternal age at birth was
significantly linked to lower full-scale IQ (FSIQ) and verbal IQ
in ICSI/IVF children, as well as lower abilities on a subtest of
the performance scale (object assembly) in IVF children. Higher
maternal educational level was also linked to better abilities on
the same performance subtest (object assembly) in ICSI/IVF
children, whereas a low educational level in this group was linked
to lower scores on spatial visualization an analysis (block design).
Overall, these results suggest that factors other than conception
mode, especially maternal age at birth and educational level,
may prevail in explaining the long-term cognitive development
of ICSI children.
Implications for the current study
Taken together, although findings on ICSI children’s motor and
cognitive development have so far been generally reassuring,
several needs for further study come forward from the existing
research literature. First, the ICSI population as a whole has
been studied only up to an early age. Further research is there-
fore needed to evaluate whether ICSI children continue to show
satisfactory development later in life. Additionally, some find-
ings remain inconsistent given certain methodological limita-
tions discussed above. Lastly, findings to date suggest that some
demographic variables may be more important in explaining
cognitive outcomes than children’s conception mode per se.
Further research should therefore clarify the relative impact of
conception mode compared with other demographics.
Responding to these needs, this study is the first long-term,
two-wave follow-up investigation at ages 8 and 10 years into
ICSI children’s psychological, family-relational and medical
development (Leunens et al., 2004a,b; 2005). This article dis-
cusses the first-wave study findings on the cognitive and motor
development of ICSI children at age 8 years. Given that this is
the first follow-up study on ICSI children’s development at this
age to our knowledge (Leunens et al., in press), the research
questions were mainly kept explorative: (i) Does ICSI have
any impact on children’s intellectual development at 8 years?
and (ii) Does ICSI have any impact on children’s motor devel-
opment at 8 years? Furthermore, we aimed to assess the relative
role of demographic variables such as maternal education level and
age at birth relative to conception mode. Overall, in line with
previous studies on 5-year-old ICSI children, we expected that no
effects resulting from the conception mode would be found on
ICSI children’s cognitive or motor development at 8 years.
Our study sample (n = 304) consisted of Dutch-speaking Belgian chil-
dren recruited and assessed when they were 8 years old (up to 8 years
11 months). Singletons born after ICSI (n = 151) and controls born
after SC (n = 153) were seen for individual assessment. The children’s
inclusion criteria were being born after 32 weeks of gestation, single-
ton and native language Dutch with at least one European parent.
These inclusion criteria were chosen to maximize follow-up rates, to
minimize confounding factors because of prematurity and exclude
potential influences because of multiple births, linguistic barriers and
Children were not matched for gestational age, but no significant
difference existed in mean gestational age between the study groups.
Gender was quite evenly distributed, with 77 boys (50.99%) and 74
girls (49.01%) in the ICSI group and 78 boys (50.98%) and 75 girls
(49.02%) in the SC group.
The ICSI children were recruited from initial birth cohorts established
at the Academic Hospital of the Vrije Universiteit Brussel. They had
already been assessed in their second and/or fifth year (Bonduelle
et al., 1998; Ponjaert-Kristoffersen et al., 2004, 2005). Approximately
two-thirds (n = 151/248, 61%) of the eligible Dutch-speaking ICSI
cohort aged 8 years between February 2001 and December 2003
(n = 248) was tested at 8 years. Of the remaining 39% of the cohort
(n = 97), 16.5% (n = 41/248) of the families could not be reached (lost
to follow-up) and 22.5% (n = 56/248) refused to participate in this study.
Of all the families in the eligible cohort who were actually reached by
telephone (n = 207/248), 73% responded positively (n = 151/207) and
27% refused participation (n = 56/207). A brief oral checklist was
administered to those parents preferring not to participate, to obtain
information on refusals and on major medical events concerning chil-
dren in this group. This checklist included items relative to hospitali-
zation, diseases, surgery, medication intake, additional therapies,
learning difficulties along with reasons for refusal. Taken together, at
least minimal information could thus be obtained on ~83.5% of the
eligible children from our centre.
SC children (controls) were selected to match children in the ICSI
group (age and gender) with further group level matching for maternal
education. The control children were recruited from surrounding
schools, with an average response rate of 37.5%. The reasons for
refusal in this group mainly pertained to a lack of interest or time to
participate (‘no time’, ‘don’t see why’ and ‘not interested’). No funda-
mental study-related objections were recorded.
The full study procedure consisted of a comprehensive psychological
assessment along with a pediatric neurological evaluation, as summa-
rized in Table I. All children were individually assessed by one
trained psychologist. Testing took place in a hospital consultation
room (ICSI children) or in an isolated classroom at school (SC chil-
dren), with ICSI children being assessed only on school-free periods,
because most parents would not participate if their child would have
by guest on May 29, 2013
ICSI and development at 8 years
to miss class. After briefing on assessment purpose and duration of 1.5 h,
the ICSI parents were handed over a stamped envelope with the
questionnaires (see Table I) and asked to fill out at least the medical
questionnaire on-site. Parents of the SC children received written
assessment briefing after the school head’s consent regarding partici-
pating in the study. Children’s assessment consisted of respectively
intelligence testing, the Family Relations Test and the Movement
Assessment Battery for Children (ABC) test. SC children were given
the stamped envelope with questionnaires to be filled out by their parents.
Although blind testing was not feasible, the use of standardized
measures with fixed study protocol is considered sufficient to mini-
mize potential observer bias. All participating families provided
written informed consent.
All outcome measures consisted of standardized tests with satisfactory
psychometric properties regarding reliability and validity (Wechsler,
1974; Vander Steene et al., 1986; Henderson and Sugden, 1998)
To address the present research questions, cognitive and motor
developments were respectively measured with the Wechsler Intelli-
gence Scale for Children-Revised (WISC-R) and Movement ABC.
The WISC-R (Wechsler, 1974; Vander Steene et al., 1986) is a widely
used, individually administered, standardized measure of intelligence
for children aged 6–18 years. It yields a full-scale, verbal and per-
formance intelligence quotient (mean 100, SD 15) along with six ver-
bal (information, similarities, arithmetic, vocabulary, comprehension
and digit span) and six performance subtest scores (picture comple-
tion, sequencing, block design, object assembly, substitution and
mazes) (mean 10, SD 3). Because intelligence scores cannot be
regarded as absolute figures, but are always embedded with a certain
probability within confidence intervals, the confidence intervals com-
puted by the test constructors of the Flemish/Dutch version of the
WISC-R (Vander Steene et al., 1986) were used in this article. This
method allows assessing clinical importance of differences in IQ
between the groups, which is impossible with other methodologies,
like z-scores for example. These confidence intervals from the Vander
Steene et al. (1986) test manual are computed based on estimated true
IQ scores, defined as IQt = (IQ0 – 100) rtt + 100 (Nunnally, 1978, in
Vander Steene et al., 1986), with rtt being the reliability factor for
Weighted Composite tests as defined by Mosier (1943, in Vander
Steene et al., 1986). The Flemish/Dutch WISC-R test constructors
derived the reliability factor for their sample of Flemish/Dutch chil-
dren, which multiplied with the SD of the observed IQ score (Stanley,
1971, in Vander Steene et al., 1986) leads to the confidence intervals
referred to in this article.
The Movement ABC (Henderson and Sugden, 1998) test measures
motor abilities for children aged 5–12 years on three scales: manual
skill (e.g. placing pins in a peg board), ball skill (e.g. throwing a bean-
bag over a 2-meter distance into a box) and balance (e.g. standing on
each foot for 20 s). The Movement ABC test yields percentile scores
with age norms. For Dutch-speaking children at age 8 years, the 15%
percentile threshold is situated at a total test score of 8.5 with scores
below 8.5 indicating above average motor skills with zero being the
best obtainable score.
Between-factor analyses were carried out on interval or ratio data
with two independent variables using ANOVA or t-tests in SPSS
13.0 for Windows. The independent measures used were the child’s
conception mode (ICSI or SC) and gender. The dependent measures
were the WISC-R and Movement ABC test scores. Nominal data
were analyzed using Pearson c2 tests. Significance levels of 0.05
were accepted throughout. ‘Cohen’s d’-values were computed to
account for effect sizes. This value is defined as d = (Experimental
mean – Control mean)/Pooled SD (Cohen, 1988; Rosnow and
Table I. Assessment measures in study protocol
Name of testMeasuring
Wechsler Intelligence Scale for Children-Revised
(Wechsler, 1974; Vander Steene et al., 1986)
Family Relations Test (Bene, 1985; Celestin-Westreich et al., 1999)
Movement Assessment Battery for Children (Henderson and Sugden, 1998)
Pediatric neurological examination
Genital examination and Tanner scores
Congenital malformations (ICD-9-BPA)
Physical examination and auxiological data
Child’s perception of relationship with parents and siblings
Status of prepubertal development
Heart auscultation, skin inspection, weight, height, head circumference,
Speech, tone, gait, balance, reflexes, strength against resistance, resistance
against passive movements and coordination
Monocular visual acuity
Neurological examination (Touwen’s criteria)
Pure tone audiometry
Test of Lang
Questionnaires administered to the parents
Child Behavior Checklist for Children (Achenbach and Edelbrock, 1983;
Verhulst et al., 1996)
Parenting Stress index (Abidin, 1990; de Brock et al., 1992)
Dyadic Adjustment Scale (Spanier, 1976)
Greenberger Scales (Greenberger and Goldberg, 1989)
Parental Acceptance/Rejection Questionnaire (Rohner, 1999)
General Health Questionnaire (Goldberg and Hillier, 1979;
Koeter and Ormel, 1991)
Quality of parental relationship
Commitment to parenting/to work
Nature of parent-child relationship
Parental emotional well-being
by guest on May 29, 2013
L.Leunens et al.
Stepwise linear regression analysis was conducted to control for
demographic differences between the conception groups. Because of
the pilot nature of this research on this population at this age, the anal-
yses did not set out to test a preconceived model of factors influencing
FSIQ, rather to explore the role of demographic variables on outcome
measures for the conception groups. Therefore, the stepwise regression
mode was preferred. (i) We started from the univariate variable
‘maternal educational level’. In linear regression, ‘standardized b’
values indicate how strongly each predictor variable influences the
criterion variable, in this case FSIQ. Beta is measured in units of SD.
In a first step, the following rationale was applied to ascertain which
factors should be included in a regression model of explaining vari-
ance in FSIQ between the conception mode groups. We started from
the univariate differences between the conception mode groups, so
conception mode was retained as a potential factor. (ii) Then, the
demographic variables showed a univariate conception group differ-
ence as summarized in Table II, namely ‘child’s age’, ‘maternal age at
birth’, ‘birth status firstborn’ and ‘admission to a neonatal intensive
care unit (NICU)’ were regressed separately as independent variables
on the outcome variable FSIQ to see which of these satisfied criteria
of association with both conception mode and outcome. (iii) Finally,
maternal educational levels were ‘admission to NICU’, ‘age of the
child’ and ‘firstborn’ and were not retained for further model-build-
ing. Only ‘maternal age at birth’ satisfied the association criteria.
Finally, maternal educational levels were also retained because of the
documented association with child intelligence outcome in previous
In a second step, only the variables coming forward from step 1
[conception mode, maternal age at birth and maternal educational
level (high, medium and low)] were included in a stepwise regression
model, with a probability of F to enter £0.10. This criterion was cho-
sen to avoid the exclusion of meaningful variables with the conven-
tional significance level of 0.05, while keeping a reasonable balance
between type 1 error and power. After building the regression model,
it was tested for stability on randomly drawn samples of 66% of all
cases (see Results).
Despite ICSI and SC children’s very similar average age at
testing (Table II), statistical comparison of mean ages revealed
a significant difference. Also, there were fewer firstborns in the
SC group than in the ICSI group (Table II). Mean birth weight
and gestational age were similar in both groups. Significantly
more ICSI children had been admitted to a NICU, more specif-
ically for ‘short admissions’ (7 days or less). Thus, ICSI and
control children do not differ for ‘long admissions’ (over 7
days). The maternal ages at birth are significantly higher in the
ICSI than in the SC group. Maternal educational levels,
defined as high (postgraduate or a graduate degree), medium
(entered university or completely passed school matriculation)
and low (partially passed school matriculation or no qualifica-
tions at all), do not differ significantly between groups.
ICSI children’s FSIQs were significantly higher than those of
SC children [ICSI 112, CI 95% (105–118), SD 14.8; SC 107,
CI 95% (101–113), SD 13.6, P = 0.001, Cohen’s d = 0.35]. It
should also be noted that these IQ scores are still situated
within the same SD, thus resulting in similar IQ ranges as
shown in the confidence intervals indicated in the test manual,
relevant to clinical practice, with Cohen’s d showing that this
is a small effect (d < 0.50). Gender does not affect IQ scores in
either group. Additionally, there are no conception mode
Table II. Demographic description of the ICSI and spontaneous conception study populations
ICSI (n = 151)Spontaneous conception (n = 153) Statistical analysis
Mean age at testing (years)
Birth weight (g)
Gestatational age (weeks)
Maternal age at birth (years)
Maternal educational level
Paternal education level
No of days unknown
Admission status unknown
8.56 0.17 8.49
P = 0.013
P £ 0.001115/76.2%
P = 0.587
P = 0.5141.21.5
P £ 0.001
P = 0.726
P = 0.856
P = 0.082
P = 0.342
P = 0.248
P = 0.644
P = 0.012
by guest on May 29, 2013
ICSI and development at 8 years
effects for the small proportion of children showing marked
developmental delay [defined as scoring one SD below the
mean (<85)] on FSIQ (ICSI, four children; SC, three children;
P = 0.689), VIQ (ICSI, three children; SC, two children; P = 0.641)
and PIQ (ICSI, 12 children; SC, 12 children; P = 0.973).
Similarly, ICSI children obtain significantly higher verbal
IQs than SC children. Again, however, effect size is small
(Cohen’s d) and mean scores remain in similar ranges (see con-
fidence intervals). Boys obtain higher verbal IQ scores than
girls. ICSI children also obtain slightly higher performance scale
scores; yet again this effect is small and mean scores fall in simi-
lar confidence intervals. No gender effects are detected in either
group. Table III summarizes verbal and performance IQ scores
between the groups, as well as detailed subscale scores.
Several additional analyses were carried out to refine our
insights in the variables that may explain the observed yet
small differences in favour of ICSI children’s IQs.
First, FSIQ scores were compared for ICSI and SC children
born between 32 and 36 weeks of gestation, to check for pos-
sible prematurity effects. Premature ICSI children’s FSIQ was
still significantly higher compared with premature controls
(ICSI 125.7, SD 9.8; SC 100.5, SD 15.3; P = 0.008). Mean
FSIQ scores overall remain ‘average’ to ‘high’.
Second, because of (i) the substantial number of missing
data on maternal educational level in the SC group, (ii) the
trend of higher educational levels in the ICSI group and (iii)
documented effects of maternal educational level in previous
ICSI research, we reran the analysis excluding all children
whose maternal educational level was unknown. This analysis
resulted in IQ outcomes comparable to the initial results
(ICSI 112.1, SD 14.5; SC 107.9, SD 12.6; P = 0.022), leading
to the assumption there is no manifest detrimental effect of
missing maternal educational level data on the initial results.
However, when comparing FSIQs of children with highly edu-
cated mothers, the significant difference in FSIQ between the
groups disappeared (ICSI: 114.3; SC: 111.6; P = 0.207).
When, in an additional analysis, assuming all mothers whose
educational level was unknown were highly educated and com-
paring only FSIQs of children with these ‘highly educated
mothers’ (real highly educated mothers plus mothers with missing
info on educational level), the difference in FSIQ reappeared
(ICSI: 114.2; SC: 108.7; P = 0.004). These additional analyses
indicate (i) an effect of educational level on the FSIQ scores of
children in both groups, with high maternal educational level
thus being responsible for the slightly higher FSIQ scores in
the ICSI group and (ii) that the mothers, whose educational
level was unknown, were most probably not highly educated.
Third, to explore the influence of demographic differences
between conception groups’ FSIQ’s, a stepwise regression ana-
lysis was carried out. Among those demographic descriptors
showing a univariate group difference between the ICSI and
SC groups, the following separately entered factors also
showed an association with FSIQ: ‘conception mode’ (stand-
ardized b = –0.181, t = –3.206, P = 0.001) and ‘maternal age at
birth’ (standardized b = 0.127, t = 2.202, P = 0.028). Addition-
ally, ‘high’ (standardized b = 0.295, t = 4.729, P = 0.000),
‘medium’ (standardized b = –0.250, t = –4.006, P = 0.000) and
‘low’ (standardized b = –0.163, t = –2.567, P = 0.011) mater-
nal educational level also showed to be associated with FSIQ.
The factors ‘admission to NICU’, ‘age of the child’ and ‘first-
born’ showed no significant association with FSIQ and were
therefore excluded from the regression model. Given these
results, the regression model (adjusted R2 = 0.099; P = 0.000)
retained the factors ‘high maternal educational level’ (stand-
ardized b = 0.292, t = 4.782, P = 0.000) and ‘conception mode’
(standardized b = –0.140, t = –2.290, P = 0.023), whereby
higher FSIQ is linked to highly educated mothers and to ICSI
as conception mode. Furthermore, when running this model on
randomly drawn samples of 66% of the total population, the
same two factors remain in many trials, confirming the stabil-
ity of this regression model. Likewise, when composing a sam-
ple of all SC cases with full availability of demographic
Table III. Mean (M) of the verbal and performance IQs and subscales as measured by the Wechsler Intelligence Scale for Children-Revised
aSignificant conception mode effect.
bSignificant gender effect.
cVander Steene et al. (1986).
ICSI Spontaneous conceptionStatistical analysis
P = 0.001a, P = 0.028b
Cohen’s d = 0.39
P = 0.001
P = 0.117
P = 0.159
P = 0.002
P = 0.006
P = 0.054
P = 0.024a;
Cohen’s d = 0.26
P = 0.000
P = 0.009
P = 0.035
P = 0.024a
P = 0.014
P = 0.849
by guest on May 29, 2013
L.Leunens et al.
information, and a random sample of an equivalent number of
ICSI cases (also with full demographics), the same two factors
remain in the regression.
In sum, these results show that ICSI children obtain signifi-
cantly, yet only slightly, higher IQ’s compared with SC con-
trols, probably because of higher maternal educational levels in
the ICSI group. However, the difference is not clinically
important, high maternal educational level is positively related
to IQ overall, and the model explains only 10% of the variability
in FSIQ scores. Future analyses should investigate additional
family-related characteristics to shed light on the possible role
of other factors.
As summarized in Table IV, no significant differences were
found between ICSI and SC children’s motor skills as meas-
ured by the Movement ABC. However, girls did perform sig-
nificantly better on manual skills in both groups (P £ 0.001).
This study aimed to explore the cognitive and motor develop-
ment of 8-year-old ICSI children compared with spontaneously
conceived age-matched controls. Overall, the results regarding
ICSI children’s cognitive and motor development appear to be
reassuring for parents and clinicians. Because this is the first
follow-up study of ICSI children at the age of 8 years, these
findings cannot be compared with outcomes of similar studies.
The present results are however in line with previous large-
scale studies indicating ICSI children’s reassuring cognitive
and motor development at younger ages (e.g. age 5 years,
Ponjaert-Kristoffersen et al., 2005).
Thus, 8-year-old ICSI children do not display any signific-
ant developmental delays as compared with SC children, be it
regarding cognitive or motor functioning. The statistical power
of the analysis to detect a difference of 5 IQ points at a 0.05
significance level, given the sample sizes, was found to be at
0.826. ICSI children’s intelligence scores are slightly higher in
this study, although these effects are small and the difference is
not clinically important. Besides WISC-R test characteristics
(mean 100, SD 15), which clearly indicate that ICSI and SC
children’s mean FSIQ scores fall within the same SD of the
test, it is accepted in clinical practice that intelligence scores
should not be regarded as absolute figures but rather embedded
between confidence intervals. Therefore, we compared confid-
ence intervals for ICSI and SC children’s mean FSIQ scores
provided by the Dutch/Flemish test constructors, which indic-
ate that ICSI and control children’s IQs still remain in similar
ranges when situated within the intelligence levels applied in
practice. It is worth remembering in this context that the statis-
tically significant univariate group differences in favour of
ICSI children’s IQs cannot be directly transposed to the inter-
pretation of individual cases. Hereafter, we further discuss the
implications of these quite straightforward findings in the light
of the study limitations and the possible role of influencing
One study limitation that limits generalizing these findings
refers to the response rate. Indeed, although 61% of the eligible
ICSI cohort was formally assessed, approximately one-third of
the initial birth cohort could not be seen in this study, which
requires questioning potential participation bias. This difficulty
is evidently inherent to the nature of long-term follow-up stud-
ies and underlines the complexity of longitudinal research.
Given that the main consideration in this context would be that
because of dropout rates we might miss out on information
regarding ICSI children who do present difficulties (in contrast
to the participating ICSI children), several indicators however
can be considered as reassuring to this regard. Thus, it can be
assumed that the ICSI families who were ‘lost to follow-up’
would be randomly distributed in the birth cohort, therefore not
necessarily presenting more problems. Apart from these, 73%
of the ICSI families who had actually been reached could also
be assessed. Moreover, although the possibility of a participa-
tion bias cannot entirely be ruled out, we did gather informa-
tion on around 83% of the eligible cohort (including the
refusing families) by means of a telephone checklist concern-
ing major medical events, learning difficulties and additional
therapies. This allowed us to ascertain that no major problems
were identified in the dropout ICSI group (Belva et al., in
press), suggesting that our findings are not biased by partici-
pant versus dropout ICSI populations. Likewise, a participation
bias cannot be ruled out for the control group given the low
response rate of 37%, which tends to be typical when participa-
tion is voluntary [e.g. recruitment not based on established
(medical) structures such as a national birth registry existing in
other countries, no financial rewards]. As a result, the recruited
control group may have been biased towards children with
either few or many problems. However, our findings on ICSI
and SC children’s cognitive and motor development are based
on the use of statistically normed instruments, thus resulting
from comparison with normative data for Dutch-speaking
Belgian children (cf Methods and Outcome measures). Com-
parison of the mean and SD IQ’s in our sample with the
Belgian norms indicate that these are situated in the (above)
The generalization of our findings may also be limited given
that only children born after 32 weeks of gestation were
included. It should be noted however that this inclusion crite-
rion still contains children who were born as premature infants
between 32 and 37 weeks of gestation, whose IQs were aver-
age to high in both groups regardless of possible major birth
defects and low birth weights, which were not per se excluded
from this study.
Table IV. Mean (M) total motor score, manual skill score, ball skill score and
balance score as measured by the Movement ABC
aSignificant gender effect.
M SDM SD
Total motor score
Manual skill score
Ball skill score
P = 0.377
P = 0.584, P = 0.001a
P = 0.743
P = 0.232
by guest on May 29, 2013
ICSI and development at 8 years
Regarding the impact of demographic factors in our analyses,
high maternal educational levels contributed to the small
amount of the variance in the FSIQ. These findings are consist-
ent with the knowledge that children’s intelligence scores,
especially verbal ones, tend to be sensitive to stimulation,
including a positive influence of higher maternal educational
level in particular (Neiss and Rowe, 2000). Our regression ana-
lysis confirms this tendency, showing that both ICSI and SC
children with highly educated mothers obtain higher FSIQ’s,
persistently so when analysing only the subsample of children
with full demographic information (no missing data). How-
ever, one could suspect higher levels of education in the ICSI
group given the amount of missing data on this variable in the
control group. This assumption was confirmed by the addi-
tional analyses performed on subgroups of the sample; children
with highly educated mothers in both groups have comparable
IQs. When comparing children with full demographic data, the
significant FSIQ difference reappeared. When finally assuming
all children with missing data on maternal educational level
had highly educated mothers, and comparing this ‘new’ sample
of children with highly educated mothers in both groups, the
significant FSIQ difference also reappeared. These analyses
imply that indeed maternal educational level is responsible for
the slightly higher intelligence scores in the ICSI group,
through the missing data on this variable in the control group
and the fact these mothers whose educational level was
unknown probably were not highly educated. Additionally,
because in our sample ICSI children’s mothers were signifi-
cantly older when they gave birth, it could be hypothesized that
ICSI children’s mothers being older when they give birth and
overall highly educated tend to stimulate their children more.
From a more practical perspective, the small difference in IQ
scores in favour of the ICSI children may also have been facil-
itated by the fact that these were accompanied by their parents
at the hospital. Parental presence may have heightened imme-
diate testing motivation for ICSI children, compared with chil-
dren in the SC group who were tested in their schools without
such potentially encouraging parental presence. Although the
difference in testing situation was a practical necessity and the
assessment measures were chosen for their relative robustness
against immediate environmental influences, these cannot be
entirely excluded for the ICSI group (Wechsler, 1974; Vander
Steene et al., 1986; Henderson and Sugden, 1998).
Finally, a broader contextual influence pertinent to the gen-
eration of ICSI children being investigated at present may con-
tribute to the observed slight intellectual ‘advantage’ at age of
8 years. The children in this study are indeed part of the first
generation of families in the world to use the ICSI technique.
Given that ICSI was initially unknown to the public at large,
we can suppose that the families constituting this initial cohort
are ‘privileged’ to some extent, for instance as regards access
to specialized (medical) information sources. This ‘select first
generation effect’ may positively influence the outcomes of
ICSI children currently studied. To cite Barnes et al. (2004),
‘Families who conceived from fertility treatment are undoubt-
edly the successful ones, (...) a selected group whose intrinsic
qualities have contributed to the success of their treatment’.
Given the documented effects of contextual and family factors
on children’s development, continuing follow-up research on
subsequent ICSI cohorts from more socioeconomically and
cultural-ethnically diverse populations are needed (Neiss and
Rowe, 2000; Barnes et al., 2004; Celestin-Westreich and
Celestin, in press).
In conclusion, the results of the present study are reassuring
regarding singleton ICSI children’s cognitive and motor abili-
ties up to middle elementary school age. Our findings so far
suggest that these ICSI children do not suffer any significant
developmental delay as compared with naturally conceived
children, displaying at least average cognitive abilities. Further
research should evaluate emotional, behavioural and family-
relational components to investigate to what extent these con-
clusions apply to other areas of ICSI children’s development.
Also, the possible role of other personal, family and contextual
variables in the development of children born after ICSI should
be explored. As mentioned earlier, the follow-up of later
cohorts will benefit from including socioculturally more diverse
populations. Furthermore, given the pilot nature of research on
ICSI youth, continuing follow-up is needed to make sure that
they continue to do well, as is the case in currently ongoing re-
assessment of these children at ages 10 years and above.
This study was supported by a research grant from the Fonds voor
Wetenschappelijk Onderzoek Vlaanderen (FWO) and the Onderzoek-
sraad Vrije Universiteit Brussel. We thank all the families and the
schools who kindly agreed to take part in the study.
Abidin RR (1990) Parenting Stress Index Test Manual, 3rd edn. Paediatric
Psychology Press, Charlottesville, VA.
Achenbach TM and Edelbrock C (1983) Manual for the Child Behavior
Checklist and Revised Child Behavior Profile. University of Vermont,
Department of Psychiatry, Burlington, VT.
Barnes J, Sutcliffe AG, Kristoffersen I, Loft A, Wennerholm U, Tarlatzis BC,
Kantaris X, Nekkebroeck J, Hagberg BS, Madsen SV et al. (2004) The
influence of assisted reproduction on family functioning and children’s
socio-emotional development: results from a European study. Hum Reprod
Belva F, Henriet S, Liebaers I, Van Steirteghem A, Célestin-Westreich S and
Bonduelle M. Medical outcome of eight year old. ICSI Children, in press.
Bene E (1985) Manual for the Family Relations Test, 2nd edn. NFER, Slough.
Bonduelle M, Joris H, Hofmans K, Liebaers I and Van Steirteghem A (1998)
Mental development of 201 ICSI children at 2 years of age. Lancet
Bonduelle M, Ponjaert I, Van Steirteghem A, Derde MP, Devroey P and
Liebaers I (2003) Developmental outcome at 2 years of age for children
born after ICSI compared with children born after IVF. Hum Reprod
Bonduelle M, Wennerholm UB, Loft A, Tarlatzis BC, Peters C, Henriet S,
Mau C, Victorin-Cederquist A, Van Steirteghem A, Balaska A et al. (2005)
A multi-centre cohort study of the physical health of 5-year-old children
conceived after intracytoplasmic sperm injection, in vitro fertilization and
natural conception. Hum Reprod 20,413–419.
Bowen JR, Gibson FL, Leslie GI and Saunders DM (1998) Medical and devel-
opmental outcome at 1 year for children conceived by intracytoplasmic
sperm injection. Lancet 351,1529–1534.
de Brock AJLL, Vermulst AA, Gerris JRM and Abidin RR (1992) Nijmeegse
Ouderlijke Stress Index. Swets & Zeitlinger B.V., Lisse, The Netherlands.
Celestin-Westreich S and Celestin LP (2005) [Families’ cognitive-emotional
adjustment processes when facing Attention-Deficit/Hyperactivity Disorder]
Annales Médico-Psychologiques, [Epub ahead of print 23 November 2005].
Celestin-Westreich S, Baarda B and Ponjaert-Kristoffersen I (1999) Familie
Relatie Test. Swets & Zeitlinger B.V., Lisse, The Netherlands.
by guest on May 29, 2013
L.Leunens et al.
Cohen J (1988) Statistical Power Analysis for the Behavioral Sciences, 2nd
edn. Lawrence Erlbaum Associates, Hillsdale, NJ.
Devroey P and Van Steirteghem A (2004) A review of ten years experience of
ICSI. Hum Reprod Update 1,19–28.
Goldberg DP and Hillier VF (1979) A scaled version of the general health
questionnaire. Psychol Med 9,139–145.
Greenberger E and Goldberg WA (1989) Work, parenting, and the socializa-
tion of children. Dev Psychol 25,22–35.
Hansen M, Kurinczuk JJ, Bower C and Webb S (2002) The risk of major birth
defects after intracytoplasmic sperm injection and in vitro fertilization. N
Engl J Med 346,725–730.
Helmerhorst FM, Perquin DA, Donker D and Keirse MJ (2004) Perinatal out-
come of singletons and twins after assisted conception: a systematic review
of controlled studies. BMJ 328,261.
Henderson SE and Sugden DA (1998) Movement Assessment Battery for
Children. Handleiding. Swets & Zeitlinger B.V., Lisse, The Netherlands.
Jackson RA, Gibson KA, Wu YW and Croughan MS (2004) Perinatal out-
comes in singletons following in vitro fertilization: a meta-analysis. Obstet
Kallen B, Finnstrom O, Nygren KG and Olausson PO (2005) In vitro fertiliza-
tion (IVF) in Sweden: risk for congenital malformations after different IVF
methods. Birth Defects Res a Clin Mol Teratol 73,162–169.
Koeter MWJ and Ormel J (1991) GHQ-28. Translated and adapted with per-
mission of the publisher from the General Health Questionnaire, 1978,
David Goldberg and the Institute of Psychiatry, Published by. NFER-
NELSON Publishing Co. Limited, United Kingdom.
Kurinczuk JJ (2003) Safety issues in assisted reproduction technology – From
theory to reality – just what are the data telling us about ICSI offspring health
and future fertility and should we be concerned? Hum Reprod 18,925–931.
Leslie GI, Gibson FL, McMahon C, Cohen J, Saunders DM and Tennant C
(2003) Children conceived using ICSI do not have an increased risk of
delayed mental development at 5 years of age. Hum Reprod 18,2067–2072.
Leunens L, Célestin-Westreich S, Belva F, Bonduelle M and Ponjaert-Kristof-
fersen I (2004a) Developmental outcome of 8-year-old ICSI children. Hum
Leunens L, Célestin-Westreich S and Ponjaert-Kristoffersen I (2004b) Cogni-
tive and motor development of 8-year-old ICSI children. In Findling R,
Davidson PB and Figge K (eds) Abstracts of the 51st Annual Meeting of the
American Academy of Child and Adolescent Psychiatry, Washington D.C.,
USA, 19–24 October 2004, p. 141.
Leunens L, Célestin-Westreich S, Bonduelle M and Ponjaert-Kristoffersen I
(2005) Cognitive and motor development of ICSI children at age 8. Hum
Leunens L, Célestin-Westreich S, Bonduelle M and Ponjaert-Kristoffersen.
Development of children born after ICSI – a review of recent literature, in
Lie RT, Lyngstadaas A, Ørstavik KH, Bakketeig L, Jacobsen G and Tanbo T
(2005) Birth defects in children conceived by ICSI compared with children
conceived by other IVF-methods; a meta-analysis. Int J Epidemiol 34,696–701.
Neiss M and Rowe DC (2000) Parental education and Child’s verbal IQ in
adoptive and biological families in the national longitudinal study of adoles-
cent health. Behav Genet 30,487–495.
Neri QV, Squires J, Wang A, Rosenwaks Z and Palermo GD (2002) Screening
for cognitive and motor development of 3-year-old ICSI and IVF children.
Fertil Steril 78,S250–S251.
Papaligoura Z, Panopoulou-Maratou O, Solman M, Arvaniti K and Sarafidou J
(2004) Cognitive development of 12 month old Greek infants conceived
after ICSI and the effects of the method on their parents. Hum Reprod
Place I and Englert Y (2003) A prospective longitudinal study of the physical,
psychomotor, and intellectual development of singleton children up to 5
years who were conceived by intracytoplasmic sperm injection compared
with children conceived spontaneously and by in vitro fertilization. Fertil
Ponjaert-Kristoffersen I, Tjus T, Nekkebroeck J, Squires J, Verté D, Heimann
M, Bonduelle M and Wennerholm UB (2004) Psychological follow-up
study of 5-year-old ICSI children. Hum Reprod 19,2791–2797.
Ponjaert-Kristoffersen I, Bonduelle M, Barnes J, Nekkebroeck J, Loft A,
Wennerholm UB, Tarlatzis BC, Peters C, Hagberg BS, Madsen SV et al.
(2005) An international collaborative study of ICSI conceived 5-year-
olds, Child Outcomes: Cognitive and Motor assessment. Pediatrics
Rimm AA, Katayama AC, Diaz M and Katayama KP (2004) A meta-ana-
lysis of controlled studies comparing major malformation rates in IVF
and ICSI infants with naturally conceived children. J Assist Reprod Genet
Rohner RP (1999) Handbook for the Study of Parental Acceptance and Rejec-
tion. University of Connecticut, Storrs, CT.
Rosnow RL and Rosenthal R (1996) Computing contrasts, effect sizes, and
counternulls on other people’s published data: general procedures for
research consumers. Psychol Methods 1,331–340.
Spanier GB (1976) Measuring dyadic adjustment: new scales for assessing the
quality of marriage and similar dyads. J Marriage Fam 38,15–28.
Squires J, Carter A and Kaplan P (2003) Developmental monitoring of chil-
dren conceived by intracytoplasmic sperm injection and in vitro fertilization.
Fertil Steril 79,453–454.
Sutcliffe AG, Taylor B, Li J, Thornton S, Grudzinskas JG and Lieberman BA
(1999) Children born after intracytoplasmic sperm injection: population
control study. BMJ 318,704–705.
Sutcliffe AG, Taylor B, Saunders K, Thornton S, Lieberman BA and Grudzinskas
JG (2001) Outcome in the second year of life after in-vitro fertilisation by intra-
cytoplasmic sperm injection: a UK case-control study. Lancet 357,2080–2084.
Sutcliffe AG, Saunders K, McLachlan R, Taylor B, Edwards P, Grudzinskas
G, Lieberman BA and Thornton S (2003a) A retrospective case-control
study of developmental and other outcomes in a cohort of Australian chil-
dren conceived by intracytoplasmic sperm injection compared with a similar
group in the United Kingdom. Fertil Steril 79,512–516.
Sutcliffe AG, Peters CC, Chrysostomou X and Edwards P (2003b) National
study of five-year-old ICSI conceived children compared with naturally and
IVF conceived controls. Arch Dis Child 88(Suppl I),A52.
Tournaye H (2003) ICSI: a technique too far? Int J Androl 26,63–69.
Vander Steene G, van Haasen PP, De Bruyn EEJ, Coetsier P, Pijl YL, Poort-
inga YH, lutje Spelberg HC, Spoelders-Claes R and Stinissen J (1986)
Wechsler Intelligence Scale for Children-Revised. Nederlandstalige Uit-
gave. Swets & Zeitlinger B.V., Lisse, The Netherlands.
te Velde ER, van Baar AL and van Kooij RJ (1998) Concerns about assisted
reproduction. Lancet 351,1524–1525.
Verhulst C, van der Ende J and Koot HM (1996) Handleiding voor de CBCL/4-18.
(Dutch version). Universiteit Rotterdam.
Wechsler D (1974) Wechsler Intelligence Scale for Children-Revised: Manual.
The Psychological Corporation, New York.
Wechsler D (1990) Manual for the Wechsler Preschool and Primary Scales of
Intelligence-Revised. The Psychological Corporation, New York.
Wennerholm UB, Bergh C, Niklasson A, Borres M, Amark K, Heiman M,
Tjus T, Hök Wikstrand M, Niklasson A, Tjus T et al. (2003) Health and
development in children born after ICSI. Hum Reprod, abstracts of the 19th
annual meeting of the ESHRE, Madrid, Spain.
Submitted on November 10, 2005; resubmitted on January 31, 2006, April 3,
2006, May 24, 2006; accepted on June 5, 2006
by guest on May 29, 2013