Plagiocephaly is defined as an asymmetrical distortion of the skull resulting in an oblique trapezoid or parallelogram head shape when viewed from the vertex position in the axial plane [1]. The severity of skull asymmetry can range from minimal focal flattening on one side of the cranial vault to severe deformation affecting the entire cranial vault, skull base, and facial skeleton. Plagiocephaly arises via 2 main mechanisms: premature fusion of ≥1 of the cranial sutures (craniosynostosis) or external mechanical forces acting on the cranial vault, which results in a distortion of the normally symmetric craniofacial skeleton (deformational plagiocephaly [DP]).

In craniosynostotic plagiocephaly involving any of the paired coronal or lambdoid sutures, the restriction of skull vault growth occurs perpendicular to the fused suture (Virchow’s law) [2,3]. Isolated craniosynostosis involving premature fusion of a single coronal suture results in an anterior plagiocephaly with brow retrusion on the affected side; similarly, craniosynostosis of a single lambdoid suture will restrict posterior cranial growth on the same side. The asymmetry is often accentuated as the remaining unfused sutures expand to enable accommodation of the rapidly growing infant brain. The majority of patients with craniosynostosis do not have an identifiable genetic cause, but this proportion is increased in patients with >1 suture involved [2,3].

Alternatively—and far more commonly—plagiocephaly is caused by deformational forces acting on one side of the back of the head, which distorts the normal symmetry of the skull in the absence of skull growth restriction due to craniosynostosis [4]. This deformity is characterized by a parallelogram-type deformity. This appears clinically as mild to severe occipital flattening, with or without ipsilateral anterior shift of the ear and orbital involvement [5]. The flattening of the posterior neurocranium, resulting from the external forces applied to the head, has led to the condition also being referred to as “flat head syndrome” [6]. The importance of external forces can be seen in the close relationship between DP and sleeping position [7], among other factors that may influence external head forces [8-12].

Since the 1980s, “back to sleep” campaigns have successfully publicized the benefits of supine sleeping for reducing the risk of sleep-related death, including sudden infant death syndrome [13]. Although these campaigns reduced the incidence of sudden infant death syndrome by 40%, an undesirable consequence has been an increase in the referrals of cases of DP, leading to more referrals to specialist centers [14,15]. The majority of cases of DP will resolve without intervention, and surgical treatment is not required [16]. However, in a subset of children, DP persists, even into teenage years [17]. Although physiotherapy and helmet therapy may play a role in improving head shape and limiting other long-term effects [18,19], understanding the factors that increase the risk of DP may help to prevent DP from developing. The aims of this systematic review and meta-analysis were to critically assess the evidence for risk factors for DP and to make evidence-based recommendations for reducing the prevalence of DP. This study was previously presented as a meeting abstract at the Royal College of Paediatrics and Child Health Conference on June 15, 2021.

A search of PubMed and Web of Science was performed covering the period from August 21, 2010, to August 21, 2022 (this included an initial search from August 21, 2010, to August 20, 2020, and an update search from August 21, 2010, to August 20, 2022, to ensure that the list of included studies was as up to date as possible). The combination of PubMed and Web of Science provides >97.5% coverage of published literature [23,24]. To balance comprehensive coverage with a pragmatic approach to ensure that the study was completed with limited available resources, additional databases were not searched, and hand searching and gray literature searches were also not performed. The search terms included “plagiocephaly” AND “risk factor.”

Study titles and abstracts were screened and assessed for relevance by a single reviewer (CRTH, NB, or AN). Specifically, original studies were included if they assessed risk factors for DP. Studies in a non-English language, those involving nonhuman subjects, and low-quality or nonoriginal studies (meeting abstracts, reviews, case series, case reports, and editorials) were excluded. Duplicates were identified by assessing study titles and removed manually by CRTH. Full-text review was performed after screening by a single reviewer (CRTH, NB, or AN). Studies not reporting nonobstetric risk factors for DP were excluded. Studies reporting preventive measures such as tummy time were included. However, studies involving treatments such as physical therapy and helmet therapy were not included. As screening was performed by a single reviewer, there were no discrepancies to be resolved.

Data extraction and reporting followed the PRISMA guidelines. However, due to resource limitations, data from eligible full-text articles were extracted by a single reviewer (CRTH, NB, or AN). The main outcomes of interest were odds ratios (ORs) and risk ratios (RRs) with 95% CIs, or significant associations, for risk factors for DP (or biomarkers for DP; eg, oblique diameter difference index). These outcomes were split into factors that were associated with DP and those that were not. Other items extracted from the eligible studies included authors, country of study, funding source, study design, study aims, total participants, percentage female, population assessed, age at baseline, and selection criteria.

A meta-analysis of ORs and RRs for factors associated with DP was performed. The inconsistency index (Ι²) and a Q statistic for chi-square significance for specific df were calculated to assess interstudy heterogeneity. Both fixed effects and random effects were reported. P values for 95% CIs were calculated. Excel (Microsoft Corp) and Prism (GraphPad Software) were used for statistical analysis.

Publication bias was assessed using funnel plots of ORs and RRs for DP risk factors against study precision (1/SE). The Egger test for asymmetry was conducted using linear regression, with P<.05 indicating publication bias [25].

The searches of PubMed and Web of Science yielded 159 articles; after removing 35 (22%) duplicates, 124 (78%) articles were screened based on abstract content. Of these 124 articles, 77 (62.1%) were not relevant. Full-text screening of the remaining 47 articles resulted in the exclusion of 20 (43%) irrelevant articles (these did not report nonobstetric risk factors for DP), 3 (6%) non-English articles, and 5 (11%) articles that were not accessible. Thus, of the initial 159 articles, 19 (11.9%) were eligible for inclusion in this study (Figure 1). The characteristics of the included studies are presented in Table 1.

A summary of 16 nonobstetric factors associated with DP is presented in Table 2. ORs are provided for 12 (75%) of these 16 nonobstetric factors.

This study assessed evidence of association with DP for 43 nonobstetric factors (demographic factors: n=3, 7%; developmental factors: n=6, 14%; dietary factors: n=4, 9%; maternal factors: n=8, 19%; medical and surgical factors: n=11, 26%; and positional and environmental factors: n=11, 26%). Of these 43 factors, 16 (37%) were associated with DP (demographic factors [male sex]: n=1, 6%; developmental factors [DDH, head circumference, and delay in motor milestones]: n=3, 19%; dietary factors [bottle-only feeding, formula feeding, and vitamin D intake]: n=3, 19%; maternal factors [maternal age and maternal education level]: n=2, 12%; medical and surgical factors [chronic lung disease grade II, mechanical ventilation, and obesity at diagnosis of DP]: n=3, 19%; and positional and environmental factors [head position preference, sleeping position, reduced tummy time, and time spent supine on the floor]: n=4, 25%). With the notable exceptions of maternal age, mechanical ventilation, and tummy time, these associations were either supported by nonconflicting evidence or a meta-analysis that resolved conflicting evidence into an association. Of the 16 factors, 12 (75%) had ORs that ranged from 1.10 (mechanical ventilation) to 7.15 (insufficient vitamin D intake). Of the 5 factors assessed by meta-analysis (male sex, reaching fewer motor milestones by age 6 months, maternal education level, head position preference, and sleeping position), only 1 (20%; male sex) was associated with interstudy heterogeneity (≥50%). No evidence of publication bias was detected.

Strategies to reduce the prevalence of DP have included guidance about the infant’s environment, positioning, and handling, with the goal of creating a nonrestrictive environment that promotes spontaneous and unhindered physical movement and symmetrical motor development [27,45].

On the basis of the evidence presented in this study, the following 11 recommendations, presented in order of importance, are offered with the aim of reducing the prevalence of DP.

First, to ensure bone health in infants, it is critical that vitamin D intake is adequate [46]. Vitamin D level should be assessed regularly during development, and dietary supplementation should be considered if vitamin D level is low.

Second, during the first months of life, babies develop a head position preference [47], and this preference is more often to the right [44,48]. The increased compressive forces on one side of the head for prolonged periods causes flattening on the side being compressed. It has been proposed that the position of the fetus in the later stages of pregnancy may, in part, be responsible for position preference [28,49]. However, there is also evidence that position preference can be modified by varying head position during sleep to encourage equal distribution of pressure [50]. Other strategies to mitigate head position preference should be used, such as gently moving the infant’s head to the unfavored side when asleep or physical therapy with or without kinesiological tape to reduce tightness in the neck muscles to facilitate easier neck turning [51,52].

Third, sufficient tummy time should be provided to strengthen the infant’s head, neck, and arms and reduce time spent supine when awake.

Fourth, as infants tend to turn away from windows and toward the center of a room, it is recommended to alternate the infant’s sleeping position by placing their head at different ends of the crib. This concept is supported by 2 studies suggesting no association between crib end change and DP [33,42]. However, it is important to state that the lack of association between crib end change and DP does not necessarily mean that changing the crib end will provide an effective treatment for established DP; it may merely help to reduce the risk of DP developing.

Fifth, although bottle-only feeding should not be discouraged, it is recommended to alternate feeding positions, using both the dominant and nondominant sides when holding the infant. Bottle feeding, as opposed to exclusively breast feeding, may be positively associated with obesity (which has also been identified as a risk factor for DP) [53]. This coassociation may, in part, explain the association of bottle-only feeding with DP.

Sixth, although formula feeding should also not be discouraged, it is recommended to alternate feeding positions. It has been suggested by Weernink et al [43] that this particular risk factor is associated with lower maternal education level. Similar to bottle feeding, formula feeding, as opposed to exclusively breast feeding, may be positively associated with obesity (itself a risk factor for DP) [53].

Seventh, if motor milestones are delayed, infants should be referred for specialist assessment by a pediatrician. Similarly, obesity (BMI >97th percentile) should be identified and managed by a specialist pediatrician.

Eighth, to mitigate the impact of low maternal education level, educational resources and face-to-face education outlining factors associated with DP, as well as recommendations to reduce their impact, should be provided to all families at the 6-week postnatal infant check-up, with particular emphasis on those with low educational levels. A lower maternal education level results in worse health outcomes in infancy and later life [54]; although improving the education level of future mothers should be a national priority, there are other factors associated with low education level that may be immediately modifiable. These include access to information and access to resources [55].

Ninth, Information should be provided to families that male infants are at higher risk of DP. Losee et al [49] suggested that male infants tend to have larger, less flexible heads at birth, which are more likely to become deformed by compressive forces in utero and in the birth canal. Families should be particularly encouraged to engage with strategies to mitigate DP in male infants.

Tenth, information should be provided to families that infants with greater head circumference are at higher risk of DP. Families should be particularly encouraged to engage with strategies to mitigate DP in these infants. 

Eleventh, information should also be provided to families that infants who have had mechanical ventilation are at higher risk of DP. Families and care teams of infants requiring mechanical ventilation should be encouraged to engage with strategies to mitigate DP in these infants.

Alternative strategies for mitigating the impact of DP, such as helmet therapy, although costly for individuals and health care systems, may help avoid more invasive strategies [56]. It has been demonstrated that education provided by health care professionals, such as health visitors, midwives, and nurses, can successfully reduce the prevalence of DP [57]. Adopting these recommendations may lead to a reduction in the prevalence of DP.

The strengths of this study include a robust methodology. The protocol was registered with PROSPERO, and the reporting was in line with the PRISMA 2020 guidelines [22]. The databases searched provided >97.5% coverage [23]. Standardized data extraction was performed to minimize errors. Meta-analysis was performed where data allowed. Funnel plots suggested that publication bias did not impact the results; thus, a trim-and-fill analysis was not necessary to correct for asymmetry [58].

This study represents the most comprehensive analysis of nonobstetric factors associated with DP published to date. A systematic review by Bialocerkowski et al [59] only identified 5 factors associated with DP, of which 3 (60%) were nonobstetric factors (male sex, supine position, and neck problems) [59]. Another systematic review by De Bock et al [60] identified male sex, supine sleeping position, limited neck rotation, head position preference, lower level of activity, and reduced tummy time as the most important risk factors [60]. By contrast, our study provides evidence of association for 16 factors and confirms the association of male sex, supine sleeping position, head position preference, and reduced tummy time with DP.

A more recent systematic review by Inchingolo et al [61] provided only 2 recommendations to mitigate the impact of nonobstetric risk factors for DP: at least 30 minutes of tummy time and the use of a passive sleep curve mattress to improve harmonious skull growth. In addition, it did not include meta-analyses to support these recommendations. By contrast, our study makes 11 recommendations for nonobstetric factors to reduce the prevalence of DP. The systematic review by Inchingolo et al [61] also did not assess publication bias. Our review suggested that none of the recommendations were influenced by publication bias.

To ensure the completion of the study, additional databases were not searched, and hand searching and gray literature searches were also not performed. This inevitably limited the comprehensiveness of this study, although no systematic review can claim to be truly comprehensive because this would necessitate continual inclusion of newly published studies. The search terms were designed to be specific yet pragmatic to ensure the completion of the work with limited resources; thus, some studies may have been overlooked. Due to resource limitations, it was not possible to record more detailed reasons for the exclusion of screened articles other than not reporting risk factors for DP. Resource constraints also prevented the screening process from being conducted by 2 independent reviewers. Instead, abstract and full-text screening was performed by a single reviewer. This impacts the reproducibility of the study due to the increased potential for errors in study selection when screening is undertaken by a single reviewer. That said, because the selection criteria were clearly set out, errors due to the application of these criteria during the screening process are likely to have had minimal impact on the study outcomes. In addition, it was not feasible, due to resource limitations, to contact study authors to attempt to collect raw data if a study did not report ORs for a risk factor. This may have limited the data that could have been used in meta-analysis. If contacting authors had been feasible, more factors could potentially have been analyzed quantitively to resolve discrepancies between the included studies. This represents a significant limitation of our study.

These factors may, in part, account for the apparent disparities reported in our results compared to other literature. These disparities include the apparent lack of association between developmental delay and DP [9], while demonstrating a significant association between reaching fewer motor milestones and DP [27,30]. A recent systematic review by Martinuik et al [62] included 19 studies that assessed the association between developmental delay and DP. Notwithstanding the fact that the authors included multiple studies that used the same study population more than once, a positive association between developmental delay and DP was reported in a majority of the studies [62]. The fact that our study did not demonstrate a similar association may, in part, be due to the inclusion of studies that assessed different populations, as well as methodological limitations in our search strategy that limited the identification of studies that may have met our eligibility criteria. Another study of a large primary care cohort of 77,108 patients has provided further evidence in support of an association between developmental delay and DP [63]. That said, the literature remains conflicting, with other studies unable to demonstrate an association between presence or degree of developmental delay and DP [64]. Commentators have highlighted the fact that most studies are retrospective and observational by design, and this limits conclusions about the correlative versus causative relationship between developmental delay and DP [65].

Another disparity reported in our study is that torticollis was not associated with DP [9,26,38]. A number of literature reviews have previously reported an association between congenital muscular torticollis and DP [8,59]. Although these studies were published a decade or so ago, it logically follows that if other positional factors, such as head position preference and sleeping position, have been found in more contemporaneous studies to be significantly associated with DP [28,33,34,39], then torticollis would also be expected to be found to be associated with DP. These conflicts reported in the literature may be due to the limitations of individual study design or differences between different study populations. Therefore, we recommend interpreting with caution the lack of association between torticollis and DP reported in our study, given the lack of a mechanistic explanation that reconciles this result with other positional factors that were found to be associated with DP.

In this review, we have included a study involving patients who developed plagiocephaly after ventriculoperitoneal shunt insertion [37]. Although the authors characterized the plagiocephaly as “positional,” implying that external forces had caused the skull deformity, an alternative hypothesis is that shunt-associated plagiocephaly is a different disease entity from DP [37]. This study has been included for comprehensiveness [37], but the results should be interpreted with caution alongside those of other studies.

Finally, another limitation includes the lack of a Grading of Recommendations Assessment, Development, and Evaluation (GRADE) assessment of evidence quality, which impacts the certainty of the conclusions and recommendations set out in this study.

In summary, this study provides the most comprehensive meta-analytic assessment of nonobstetric factors associated with DP published to date. It offers 11 evidence-based recommendations that can be adopted by health care systems globally to reduce the prevalence of DP. Future research should focus on investigating factors for which the literature is conflicting but quantitative data are lacking to enable meta-analyses to be performed; for example, maternal age was the only factor reported to be protective against DP, but conflicting studies reported that there was no association without providing quantitative data. Thus, maternal age as a protective factor for DP should be investigated further to provide quantitative data for meta-analytic approaches to determine its protective effect. Further research should also address the nature of specific relationships between risk factors and DP; for example, both bottle-only feeding and formula feeding have been associated with DP. Studies should focus on understanding the nature of this relationship, that is, whether this relationship is due to mechanical forces associated with bottle-feeding or whether a reduction in, or lack of, breast milk intake or a lack of complimentary foods alongside breast milk results in nutritional differences that impact skull development [66]. As our study also highlighted an association between obesity and DP, research on other early-life or environmental exposures should be conducted to elucidate their effects on growth and development, particularly skull growth and head shape [67,68]. Finally, randomized controlled trials, although considered the gold standard study design for obtaining reliable evidence of an intervention’s effectiveness, should not be conducted for the interventions herein that relate to risk factors for DP development that have already been assessed through meta-analysis. However, randomized controlled trials could provide further evidence for an intervention—if ethically appropriate—where the evidence for a particular risk factor is relatively weak, such as when only a single cohort study has provided the evidence for a risk factor’s influence on the development of DP.