lists selected characteristics of the sample. Of the 125 case-control pairs, 39 percent were female, and the mean age was 6.5 and 6.6 months in cases and controls, respectively. Children with and without craniosynostosis had similar distributions of race and family socioeconomic status. Among craniosynostosis cases, the majority had sagittal synostosis (n
= 62), followed by diagnoses of metopic (n
= 27), right or left coronal (n
= 28), and lambdoid (n
= 8). Sex of infant differed by diagnosis, with a larger proportion of male patients in the sagittal and metopic groups (79 and 67 percent, respectively); the left unilateral coronal group was composed only of female patients. These sex differences are consistent with population trends.4
In addition, sagittal cases tended to be younger than other diagnostic groups. In both groups, approximately 90 percent of children lived in households with both biological parents (not shown). Site of data collection (Seattle, Chicago, St. Louis, and Atlanta) was not associated with any dependent variable or covariate (p
Selected Characteristics of Cases and Controls at Presurgery Visit
Correlations among Measures
and show correlations among measures of developmental status (Bayley Scales of Infant Development, Second Edition and Preschool Language Scale, Third Edition), maternal IQ, family socioeconomic status, and infant age for cases and controls. In both groups, measures of cognitive, language, and motor functions were positively correlated with one another, with correlations ranging from 0.18 to 0.59. Although maternal IQ was positively correlated with socioeconomic status in both groups (0.40 and 0.44 for cases and controls, respectively), maternal IQ was not associated with any measure of infant developmental status. Only one of the four developmental measures was associated with age among cases: the Preschool Language Scale, Third Edition Auditory Comprehension scale (<0.24). Among controls, both the Mental Development Index and the Psychomotor Development Index were directly associated with age (0.21 and 0.20, respectively).
Correlations for Cases among Infant Test Scores, Mother’s IQ, Family Socioeconomic Status, and Infant Age*
Correlations for Controls among Infant Test Scores, Mother’s IQ, Family Socioeconomic Status, and Infant Age*
Comparison of Cases and Controls
lists group means and standard deviations (in parentheses) for cases and controls on the four developmental measures (Bayley Scales of Infant Development, Second Edition Mental Development Index, Bayley Scales of Infant Development, Second Edition Psychomotor Development Index, Preschool Language Scale, Third Edition Auditory Comprehension and Expressive Communication). also lists 95 percent confidence intervals (i.e., estimated range of values in the population) and effect sizes, which indicate the strength of association between group status (case versus control) and each dependent measure. Effect sizes were calculated by dividing group mean differences by pooled standard deviations. Hotelling’s t2 test was used to analyze the mean differences between groups. This procedure minimizes inflation of type I error caused by multiple significance tests with correlated dependent variables, essentially extending the standard t test to analyses of multiple variables. Alpha (criterion for statistical significance) was 0.05.
Univariate Comparisons of Mean Neurobehavioral Test Scores for Cases versus Matched Controls (Paired t Tests)
The overall effect of group was significant (F = 3.29; p = 0.016). Paired t tests were used to examine group differences on each measure. As shown in , cases had lower mean Mental Development Index scores (91.9) than controls (94.9; p = 0.005). Cases also had lower Psychomotor Development Index scores than controls (mean, 84.1 versus 88.8, respectively; p = 0.001). On average, cases scored 2.5 points lower on the Preschool Language Scale, Third Edition Auditory Comprehension scale than did controls, but this difference did not reach statistical significance (p = 0.069). No differences between cases and controls were found on the Preschool Language Scale, Third Edition Expressive Communication scale.
No statistically significant difference between groups was found for maternal IQ (p = 0.11). Both groups scored well within the average range for this instrument. Cases had a mean of 107.05 (SD = 12.08) and controls had a mean of 109.15 (SD = 12.29).
lists means and standard deviations for the four developmental measures by sex. A generalized estimating equation model with robust standard errors was used to test for the possibility that case-control differences on developmental measures differed by sex (i.e., testing the interaction between case status and sex). There was no evidence for interaction on any measure.
Comparisons of Mean (SD) Test Scores by Sex
Diagnostic Subgroup Differences
As shown in , mean Mental Development Index scores ranged from 88.3 in right unilateral coronal cases to 95.4 in lambdoid cases. Psychomotor Development Index scores were lowest in children with lambdoid synostosis (78.0) and highest in those with metopic synostosis (86.6). Performance on the Preschool Language Scale, Third Edition Auditory Comprehension and Expressive Communication domains was lowest in the right unilateral coronal group (89.1 and 90.0, respectively). Sagittal cases had the highest scores on the Preschool Language Scale, Third Edition Auditory Comprehension (93.7), whereas left unicoronal cases had the highest scores on the Preschool Language Scale, Third Edition Expressive Communication (102.7). Case-control differences in test performance across diagnostic subgroups were tested with linear regression models including a “dummy” variable for craniosynostosis diagnosis. Separate regressions were conducted for each of the four test scores. The sagittal group served as the reference group, as it contained the largest number of participants. These analyses found no significant group differences in case-control comparisons for any measure (p = 0.05).
Comparisons of Mean Test Scores across Diagnostic Groups*