Article Text
Abstract
Background Deformational plagiocephaly (DP) is one of the most prevalent abnormal findings in infants and a frequent reason for parents to seek paediatric advice.
Objective To systematically review the literature and identify evidence and hypotheses on the aetiology and determinants of DP in otherwise healthy infants.
Design Systematic keyword search in all major biomedical databases to identify peer-reviewed publications reporting (a) empirical research or (b) hypotheses on the aetiology of DP in healthy, term infants. 3150 studies published between 1985 and 2016 and containing relevant keywords were screened. In a two-pronged approach, results were summarised separately for the body of empirical work (22 studies) and the body of hypotheses (110 articles).
Review findings Only a few empirical studies have examined risk factors in non-selected patient populations on a higher grade methodological level. The most commonly reported risk factors were: male gender, supine sleep position, limited neck rotation or preference in head position, first-born child, lower level of activity and lack of tummy time. Agreement between empirical studies was poor for most exposures, including supine sleep position, tummy time and use of car seats. The articles reporting hypotheses on the aetiology of DP cover a wide field of environmental and biological factors, but include little suggestions as to the potential influence of the everyday care environment of the baby.
Conclusions and relevance The evidence on the aetiology of DP is fragmentary and heterogeneous. In addition, factors possibly relevant to the development of DP have not been appreciated in the scientific discussion.
- deformational plagiocephaly
- positional plagiocephaly
- supine sleep position
- systematic review
- etiology
Statistics from Altmetric.com
What is already known on this topic?
Postnatally acquired posterior deformational plagiocephaly (DP) is one of the most prevalent abnormal paediatric findings and a frequent reason for seeking paediatric advice.
However, little is known about the factors that lead to DP in the normally developed, healthy child.
What this study adds?
Influences on head shape development are complex and not fully understood.
It is unclear if the supine sleep position is an independent risk factor for the development of DP.
Larger prospective cohort studies in primary care settings could clarify if there may be other risk factors for DP not detected as of yet.
Introduction
Flattening of the skull by external forces has been termed positional or deformational plagiocephaly (DP). Depending on the severity, DP presents as an abnormal head shape, facial asymmetry, frontal bossing, ear misalignment and asymmetrical orbits.1 The moulding can occur in utero, during birth, or it may develop postnatally. The latter form of DP is one of the most prevalent abnormal findings in otherwise healthy infants and a frequent reason for seeking paediatric advice.
While there is ample knowledge about risk factors for the development of DP in infants with diseases or developmental delay, much less is known about why DP should develop in a normal child. Risk factors proposed include supine sleep position, bottle propping, lack of tummy time, or frequent use of car seats, swings or bouncy seats.2–6 However, research has been ambiguous about their significance and relevance. It has been proposed that this ambiguity may reflect an incomplete understanding of how DP develops altogether.2–6
To gain a fuller understanding, we systematically reviewed the literature for empirical studies that have explicitly examined determinants of DP (key question 1) and for articles in which novel, as of yet untested, hypotheses on potentially contributing factors of DP have been proposed (key question 2).
Methods/literature search
In November/December 2013 and again in February 2016, we performed two identical systematic keyword searches in all major biomedical databases and study registries, including PubMed, Embase, Web of Science Core Collection, Cochrane Library, LILACS and CINAHL (for details on our search strategy, see online supplement). Peer-reviewed articles in English, Spanish and French published between 1985 and 2016 were eligible to be included.
supplement table
We defined case series as the minimum quality category for key question 1 and did not restrict study designs for key question 2. After elimination of duplicates, one person (JB) screened all articles by title with regard to the following inclusion criteria: (1) outcome–– DP, (2) sample––healthy children and (3) exposure––lifestyle or environmental risk factors for DP. Articles with samples of only newborns, children with specific conditions (like cerebral palsy), preterm children only or hospitalised children were excluded, as our interest was on the development of postnatally acquired DP in the normal, otherwise healthy child. Screening of abstracts and, if needed, full texts was performed in duplicate (HR-P and JB). Articles were categorised as containing either original, empirically tested data or hypotheses on the aetiology of DP.
Data extraction and quality rating in the set of n=22 empirical studies were performed in duplicate by two independent investigators (HR-P and JK). For aetiological factors with three or more comparable risk estimates from case–control or cohort studies, I2 was calculated to assess heterogeneity (see online supplementary table).
The quality rating of the identified 22 empirical studies included both study-level (eg, study design, recruitment) and outcome-level assessment (table 1, column 6) of potential bias. As many well-known quality ascertainment instruments for the assessment of study-level bias (eg, Cochrane,7 Critical Appraisal Skills Programme (CASP))8 could not be used for this sample of relatively low-grade evidence studies, we applied the Effective Public Health Practice Project Quality Assessment Tool (table 1, footnote and second row)9 and a score developed by the Oxford Centre for Evidence-based Medicine (table 1, footnote and last column). In the case of substantial inter-rater differences, discussion between the original duplicate raters (HR-P and JK) and a third experienced researcher (FDB) was carried out until agreement or consensus was reached. The review protocol according to PRISMA statement10 can be accessed via http://miph.umm.uni-heidelberg.de/miph/cms/upload/pdf/Plagiocepahly_Review_protocol_De_Bock_et_al..pdf.
Results
Of 3150 studies retrieved by the above-mentioned search strategy (see flow chart, figure 1), 2951 were removed as they were duplicates or were identified as irrelevant by title, abstract and full-text screening, leaving 199 articles for detailed full-text screening. In the second step (review of the full texts), 71 more articles had to be excluded based on predefined exclusion criteria, such as being not peer-reviewed articles in the general press, not meeting language criteria or fulfilling content-related exclusion criteria (see figure 1 and Methods section). Another four articles (including one from 1981) were included post hoc by screening references of all relevant articles, leading to 132 relevant peer-reviewed articles containing information on the aetiology of this condition (figure 1). Of these, only 22 studies reported original, empirically tested data. Table 1 summarises these studies in terms of formal characteristics and potential bias (columns 8–14); table 2 reports the risk factors identified using a roster of exposure catergories. An online supplementary table informs about the heterogeneity of these studies.
The remaining 110 publications covered hypothetical discussions and debates on potential aetiological factors of DP. This body of non-empirical work was analysed and summarised in table 3, which reports all potential risk factors discussed over and above those found in the empirical studies.
Of the n=22 empirical studies on risk factors of DP (total study population of 27 782 children), 19 studies (with a total study population of 26 860 children) were conducted with highly selected populations. Only three studies were performed using a primary care patient base without significant selection bias. Study designs were all weak but for four studies (see table 1). Most studies (13 of 22; 59%) were retrospective chart reviews. Only six (27%) studies used a case–control or prospective cohort design. No controlled experiment was identified.
The methods of outcome measurement differed widely among the 22 studies: 11 (50%) defined the outcome ‘diagnosis of DP’ based upon the criteria that were either not defined or not reported in the majority of cases, 3 (14%) used anthropometric criteria and 8 (36%) applied imaging methods (eg, photography or neuroimaging) (table 1). The sample sizes of the studies range from n=23 children to n=19 685 children, with the largest studies showing the most serious selection bias, as data were taken from tertiary care registries. None of the 22 articles reported a statistical substantiation of sample sizes.
The concordance among empirical studies as to the 60 risk factors identified was poor (table 2). Not one single risk factor was reported in all studies. Most risk factors (63%) were only reported in one single study. The direction of effect was not homogeneous among the 22 studies for most exposures (eg, tummy time: significant protective effect in three studies, no significant effect in other three studies; see table 2 for more examples). Agreement for the direction of effect was only noted for two exposures (smoking and APGAR scores, both without significant effect). Supine sleep position was reported in only 14 out of the 22 studies, and was found to be significant in only 6 of them (table 2). Other exposures commonly reported were male gender, preferred head orientation or limited head rotation, developmental delay, level of activity, birth order and tummy time. All these factors were found to be significant in only about half of the higher quality empirical studies (prospective cohort and case–control studies; see table 1 and the online supplementary table).
In the 110 articles on putative or hypothetical aetiological factors of DP, the scope of risk factors mentioned was wide, ranging from metabolic influences (eg, folic acid supplementation during pregnancy) to influences of care routines (eg, placement of the crib in relation to the parents' bed). Eleven (64%) of the seventeen aetiological factors mentioned over and above the risk factors identified in the original, empirical studies were related to the concept of uniform positioning and stimulation by the caregivers (table 3). No publication was identified with reference to the social sleep arrangement (co-sleeping vs not co-sleeping) or modality of baby transport (use of pram or stroller vs babywearing).
Discussion
Our systematic review shows that empirically verified evidence on the aetiology of DP is rare and heterogeneous, with only a few studies examining risk factors in non-selected patient populations on a high-grade methodological level. The overwhelming majority of empirical data on determinants of DP has been generated through retrospective chart reviews and uncontrolled case series in tertiary referral centres. Most of the studies were also characterised by poor reporting of methods, population characteristics and outcome measurement. Thus, DP as a common concern both for parents and paediatricians is currently not well described in terms of its contributing factors.
The poor methodological quality of the empirical studies may explain why the evidence on the aetiology of DP is conflicting—another prominent finding in our review. Agreement between studies was poor for most exposures, including supine sleep position, tummy time and use of car seats (table 2). As a matter of fact, in our sample of 22 studies, not one single risk factor was uniformly identified as such if investigated by more than two research teams. For only two determinants, the concordance rate for significance of the exposure was more than 50% between studies (male gender and supine sleeping position).
Most determinants reported in the empirical studies had been identified as possible risk factors in an earlier review on the aetiology of DP by Bialocerkowski.2 The fact that obstetric factors seem less important in our review may be explained by our search strategy, which excluded studies of newborns only.
Most of the risk factors reported in our sample of studies fall into the clinical or biological category (eg, gender or birth order). Determinants related to lifestyle have been reported to a lesser extent (eg, feeding mode, different sleep or transport environments). Indeed, within the evidence base summarised in this review, it is currently not possible to identify a single environmental or lifestyle risk factor that could be advocated for preventive measures with any degree of scientific confidence.
Controversies, limitations of research and new hypotheses
While the risk profile of DP thus remains controversial, there seems to be agreement within the expert community on a basic biophysical framework of how DP develops.3 According to this pathogenetic framework, DP is the result of gravitational forces that act on the same spot of the skull for too long a period of time. Many of the risk factors summarised in our review, like the use of car seats, or supine sleep position, seem to fit well into this model. The pathogenetic influence of the supine sleeping position seems to be additionally supported by epidemiological data, according to which the incidence of DP has increased significantly since the ‘back to sleep campaign,’ launched in the 1990s to prevent sudden infant death syndrome.12–14
Yet, the evidence from this systematic review cautions this interpretation: the supine sleep position did not emerge as a consistent risk factor. The same holds true for the use of bouncers, rockers and car seats, which are not consistently associated with a higher likelihood of DP.
By separately presenting a review of hypotheses, we were able to identify potential risk factors currently not tested through empirical research which may fertilise future prospective studies on DP aetiology. We have reviewed this body of hypotheses for possible explanations on why such major everyday exposures such as infant sleep position should be so inconsistently related to the development of DP. However, in the 110 papers reviewed, we have not identified suggestions to advance the debate.
Notably, none of the hypotheses refers to the infant care practices where they differ most in contemporary Western societies, that is, in regard to the social sleep arrangement and the modality of baby transport. More than half of American mothers of infants from birth to 12 months of age report occasional bed-sharing,15 and the rate of American parents who routinely use co-bedding, co-sleeping or baby bays (‘side cars)’ as preferred sleep environments for infants has doubled to about 13.5% within 17 years.16 Also, a significant proportion uses slings, wraps or baby carriers to transport their infants instead of (or in addition to) prams, strollers or buggys. In our review, there is neither empirical evidence nor any hypothetical suggestion on which role these influences may play in the development of DP.
This may be a serious void as some of the exposures mentioned may have a significant bearing on head shape development and may therefore act as risk modifiers for DP. For example, there is good evidence that different sleep and transport environments are associated with vastly different moulding effects on the developing cranium. Experimental research in the sleep lab has shown that babies sleeping close to their nursing mother experience more body and head repositioning during sleep. In addition, they spend more time in active sleep phases during which the gravitational effects on the skull are minimised due to more arousals and a higher muscle tone.17 ,18 Thus, the social sleep arrangement seems to be able to modify or even nullify the effect of the supine sleep position exposure, making the latter a necessary rather than a sufficient exposure for DP to develop. This could explain why in this review supine sleep position appears as such an inconsistent determinant for DP.
Similar risk-modifying effects may come from alternative modes of baby transport: The deforming impact on the back of the skull is clearly different in quality and duration for a baby carried in a sling or wearing device than for one transported in a classic pram, stroller or car seat. Therefore, a baby routinely transported in a sling may show a different head shape development, even if subject to other risk factors, such as supine sleep position.
Strengths and limitations of the systematic review
To our knowledge, this study is the second2 systematic review on the aetiology of positional plagiocephaly and the first to use a two-pronged approach to summarise results, combining an in-detail summary of empirical studies on the determinants or risk factors of DP and a narrative summary of articles on novel hypotheses about potentially contributing factors to DP. The search strategy used was systematic and based on primary and secondary searches to identify published and unpublished evidence. Through a quality rating system (table 1), we were able to assess risk of selection, detection and assertion bias in all empirical studies. Raters of quality were not blinded to authors, which theoretically could lead to biased ratings. However, the independent raters' assessments were consistent in >95% of ratings. Although the PRISMA statement guided the methods in our systematic review, we could not consider publication bias as we only included completed studies.
Conclusion and implications
The lack of sound evidence for many of the measures advocated for the prevention of DP poses a challenge for the anticipatory guidance of parents. Yet, it has a benign face: many of the recommendations currently given to parents, such as providing adequate tummy time or positioning babies supine for sleep, are important, regardless of their effectiveness for the prevention of DP. On the other hand, parents do ask specific questions around their specific lifestyles: if I carry my baby in a sling instead of transporting her in a pram—will that prevent her from developing a flat head? To find more specific answers, we need more specific research. For the case of DP, this translates into larger prospective cohort studies in paediatric primary care settings with both rigorous assessment of potential confounders, outcome measures and risk modifiers—including the diverse lifestyle factors associated with today’s parenting practices.
Acknowledgments
The authors sincerely thank the two student assistants, Janina Beckmann (JB) and Juliane Klein (JK), for their important support in performing the systematic literature search for relevant articles.
References
Footnotes
Contributors FDB designed the study and conceptualised the methods, drafted the Methods section, reviewed in detail all other sections of the manuscript and approved the final manuscript as submitted. FDB had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. VB designed the search strategy, performed the searches in all databases, reviewed the manuscript and approved the final manuscript as submitted. HR-P conceptualised the study, extracted the data (supported by Juliane Klein and Janina Beckmann; see acknowledgements), drafted the main manuscript, reviewed the Methods section and approved the final manuscript as submitted.
Funding Internal (institutional) funding; no support from external sources.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.