The design of this study differed from that of previous investigations in ≥3 ways. First, craniofacial dysmorphologists who were blinded to case status rated the head shape of every participant, including control subjects, rather than DP being diagnosed on the basis of a single practitioner’s clinical judgment and it being assumed that nonreferred control subjects did not have DP. Second, in contrast to previous investigations that relied on normative test values for comparison,2,3
we compared case subjects’ test scores directly with those of a controlgroup. The use of normative test values alone is problematic, because normative data may not adequately represent children referred to a particular clinical program, which introduces the possibility of demographic bias. Normative test values also are prone to cohort effects, which are of particular concern in research on DP because of the effects of the “Back to Sleep” campaign on infant sleep positioning and the possibility of associated increases in the prevalence of motor delays in the general population.11–14
Normative values for instruments developed before this campaign (eg, the first and second editions of the Bayley Scales of Infant Development used in previous studies of DP) may lead to erroneous conclusions about the developmental status of an index group in relation to those normative values. Third, in addition to including control subjects, we used the BSID-III, for which normative values were established in 2004, well after initiation of the Back to Sleep campaign.
With adjustment for potential confounders, we observed differences between case subjects and control subjects on all BSID-III composite scales and subscales. The cognitive and language scale differences were clinically modest (~0.3 SD), but the motor scale composite difference was of a clinically important magnitude (nearly 0.7-SD difference favoring control group participants). This difference was more evident for the gross-motor subscale (eg, sitting up, rolling from back to side, and crawling) than the fine-motor subscale (eg, transferring objects from hand to hand). Equivalent scores were observed for case subjects with and without torticollis, which suggests that restricted neck motion did not account for the differences between case subjects and control subjects.
It is important to note that motor, language, and cognitive skills are highly correlated in early infancy, which makes it difficult to differentiate among them. For example, cognitive tasks often rely at least in part on a motor response (eg, reaching for a target stimulus). Therefore, it may be that we are detecting a fundamental motor deficit among infants with DP, which is manifested in other areas of the BSID-III because of this overlap. Follow-up assessments of this sample at later ages (currently in progress at ages 18 and 36 months) should help us to distinguish specific differences.
Despite the differences noted between case subjects and control subjects, average composite scores (range: 92–107) were near the BSID-III normative average of 100. This likely reflects the relatively low demographic risk of our sample, because both case and control families tended to have relatively high SES backgrounds (ie, ~70% of control subjects and case subjects were in the 2 highest SES categories). Similarly, although case subjects were twice as likely as control subjects to score in the delayed range of motor development (~20% and 9%, respectively), nearly 16% of the population would be expected to perform in this range on a test with standardized scores.
Our findings do not necessarily imply that DP causes neurodevelopmental delays. We have proposed several hypotheses that might account for this association,15
although these hypotheses are tentative, untested, and an issue of some debate.16
Possible explanations for the observed association include a direct effect of DP, in which skull asymmetry affects brain development directly, and the reverse of this situation, in which DP is a consequence rather than a cause of early neurodevelopmental delays (eg, motor impairments limit infant mobility, which promotes skull asymmetry). Unfortunately, the cross-sectional data generated here cannot distinguish these possibilities.
Blinded ratings of the 3-dimensional images confirmed DP for the vast majority of case subjects, classifying >99% of referred infants as having some degree of cranial deformation. However, the raters identified a relatively large proportion of control group participants with mildly asymmetric or flat occipital head shapes. In secondary analyses, this 30% subgroup tended to have lower BSID-III language and motor scores, compared with control subjects with no evidence of DP. We do not know how this proportion of presumably typical infants with undiagnosed DP compares with the population at large, because documented prevalence estimates have varied widely, and data from population-based studies are limited. With a birth registry sample similar to ours, Hutchison et al17
calculated clinical cutoff scores on the basis of cephalic indices and oblique cranial length ratios and identified DP or brachycephaly in ~20% of 4-month-old infants, 9% of 8-month-old infants, and 7% of 12-month-old infants. Bias in the ascertainment of our control group might have led to an elevated rate of undiagnosed DP. During telephone screening, we explicitly queried potential control group families about head shape and excluded those whose child either had been formally diagnosed as having DP or was suspected of having DP. However, some parents might have been motivated to participate by a head shape concern that they did not mention. It also is possible that we found a high rate of undiagnosed DP because of the precision of the rating method used, which might have promoted a lower-than-usual threshold for diagnosis.
In addition to the possibility of bias in ascertainment of the control group, there might have been other sources of bias in the recruitment of both case subjects and control subjects. Among case subjects, the consent rate was relatively low (52%), which might be partly attributable to the short time frame in which families were required to participate and the clinical impression conveyed to most parents that DP is a benign craniofacial disorder, in relation to other disorders with greater morbidity (eg, craniosynostosis). We were able to assess several potential sources of bias in case ascertainment, including infant age, gender, and ethnicity, severity of DP, and family health insurance status. These analyses provided no evidence of biased case participation. However, there might have been unmeasured sources of bias that affected our results. For example, parents with greater concern about their child’s development might have been more likely to participate. To the extent that such concerns were justified, they might have contributed to overrepresentation of children with delays, compared with the general population.
Control subjects were recruited from a registry of families that agreed to be contacted for research. Although this group was similar to the case sample in terms of measured demographic characteristics, such as SES, families that participate in such a registry may differ in unquantified ways that affect infant development. To be included in the registry, parents needed to have filled out and returned a postcard within a few weeks after their infant’s birth. Such parents might have had more general intellectual curiosity and greater commitment to research, and these qualities might somehow be associated with increased BSID-III scores among their children. A population cohort-based approach could address these various potential sources of bias but would require a much larger sample to ensure adequate representation of infants with significant DP.