Investigation Into an Increase in Plagiocephaly in Texas From 1999 to 2007

Objective To examine factors that may explain a 9-fold increase in plagiocephaly in Texas from 1999 to 2007.

Design Descriptive epidemiologic study of time trends and a nested case-control study.

Setting Texas.

Participants Cases in the Texas Birth Defects Registry.

Outcome Measures Time trends in the birth prevalence of plagiocephaly overall and by region, demographic group, and clinical subgroup. Trends in percentage of cases using specific facilities or procedures.

Results From 1999 to 2007, the prevalence of plagiocephaly in Texas increased from 3.0 cases per 10 000 live births to 28.8, an average increase of 21.2% per year. This was highly statistically significant. The time trend was most pronounced in the Dallas/Fort Worth region and in certain health care facilities. It was observed in all demographic and clinical subgroups. Cases born in 2004 and 2005 were not more likely to be postnatally acquired when compared with cases born in 1999 and 2000. There was no commensurate decrease in other birth defects of the face or skull.

Conclusions A small part of the trend might have been due to delayed compliance with the infant supine sleeping recommendation and a slight increase in preterm births. It was not due to changes in birth defect coding practice or trends in multiple births, torticollis, or oligohydramnios. Because the plagiocephaly trend was observed mainly in patients visiting certain health care facilities, among mild cases, and among cases with minimally invasive procedures, we suspect it may be due mainly to changes in available therapies and insurance reimbursement practices.

Plagiocephaly is characterized by unilateral flattening of the head either in the frontal or occipital region. Previous studies have suggested several potential risk factors. Intrauterine constraint may result from plural births (eg, twins) or prolonged labor with the fetal head resting low in the pelvis^1-3 or oligohydramnios.⁴ Torticollis limits the infant's neck rotation thereby not allowing the infant to alternate head positions, possibly leading to plagiocephaly.^5-8 Preterm babies may be more likely to be put in a supine sleep position and therefore be more susceptible to secondary deformation.⁹ The prevalence of plagiocephaly has reportedly increased since 1992 when the American Academy of Pediatrics (AAP) recommended that infants be put to sleep on their back to reduce the risk of sudden infant death syndrome,^10-14 although the plagiocephaly is then an acquired and not congenital condition.

In Texas between 1999 and 2007, the prevalence of plagiocephaly increased more than 9-fold from 3.0 cases per 10 000 live births to 28.8. The aim of this study was to investigate factors that might help explain the increase.

Cases were identified using the Texas Birth Defects Registry (TBDR). The TBDR is a system with active methods of case ascertainment; trained staff routinely visit hospitals and other health care facilities where children with structural birth defects are diagnosed or treated. Continuous quality control helps ensure that the same surveillance procedures are followed by TBDR staff over time and between different regions of Texas. Diagnostic and therapeutic procedures are only abstracted if their results support or refute the existence of a particular birth defect; thus, a TBDR record does not contain all procedures done on a child. The diagnoses for each case are made by the person(s) completing the medical record in the health care facility. The information as provided in the medical record is often incomplete, sometimes comprising just a single-word diagnosis without a more detailed description.

Birth defects in the TBDR are coded using a 6-digit system (sometimes referred to as BPA codes) based on the British Pediatric Association and World Health Organization classification of disease, as modified by the US Centers for Disease Control and Prevention and the Texas Department of State Health Services. For this study, cases were selected who had a definite diagnosis of plagiocephaly (BPA code 754.050) and were born from January 1, 1999 (the first year the registry was statewide), through December 31, 2007 (the most recent year of cleaned registry data available at the time of this analysis). The TBDR collects as “plagiocephaly” infants or fetuses described as having “plagiocephaly” without other description; “unilateral posterior skull flattening” without other description; and more extensive descriptions consistent with the standard diagnosis of “skewed head.” If plagiocephaly is diagnosed concurrently with brachycephaly or scaphocephaly, both problems are coded. “Bilateral plagiocephaly” is collected as brachycephaly only (not as plagiocephaly). Children diagnosed with craniosynostosis of any suture are coded by TBDR staff with the appropriate craniosynostosis code; any cases of plagiocephaly with craniosynostosis were excluded from this study. The clinical reviewers of the TBDR have defined a set of findings related to plagiocephaly that are to be abstracted and coded if they are available. However, the registry is limited by the widely variable way in which plagiocephaly is diagnosed and recorded in medical records.

Isolated plagiocephaly was defined for the purposes of this study as having only plagiocephaly or plagiocephaly with co-occurring birth defects that are all considered “minor” by the National Birth Defects Prevention Study.¹⁵ A case was considered to be “severe” if there was a documented invasive diagnostic or therapeutic procedure, the child died within the first year of life, or the pregnancy outcome was not a live birth; all other cases were considered to be “mild.” Maternal race/ethnicity was self-assessed in the vital record (eg, birth certificate) or if that was missing, taken from the medical record. We included that in our analysis because occurrence of some birth defects can vary widely by race/ethnic group.¹⁶

Poisson regression with year as a continuous variable was used to assess the time trends in plagiocephaly cases overall, different demographic groups (eg, maternal race/ethnicity), clinical subgroups (eg, with or without co-occurring torticollis), and other skull or face defects that might have been miscoded or recoded to plagiocephaly (all BPA codes beginning with 754 or 756). By including live births as an offset, we were effectively analyzing time trends in birth prevalence rates. The slope of the fitted trend line was expressed as average annual percentage increase or decrease. Time trends were considered to be statistically significant if their 95% confidence interval excluded zero. Using an interaction term allowed us to test whether the time trend was significantly different between subgroups (eg, cases with vs without torticollis).

Health care facilities and procedures documented for at least 5% of the total cases were also assessed for trends. However, it was not feasible to determine relevant denominators, and a case could visit multiple facilities or have multiple procedures. Hence, instead of Poisson regression, we chose to assess the percentage of cases each year visiting a facility or having a documented procedure using the Cochran-Armitage trend test. Cases diagnosed and managed in private provider offices or other clinics not directly affiliated with one of the facilities covered by the TBDR were not ascertained, ie, the case had to generate some documentation in a covered facility medical record to be ascertained by the TBDR, even if that documentation was limited to a radiologic procedure report.

Recognizing that there may be a referral and/or treatment bias in those facilities with some medical and/or surgical specialties, those with some level of craniofacial specialty care were identified. The TBDR staff contacted each of the facilities of interest and queried whether there was an affiliated “craniofacial clinic.” The composition of the individual clinic, physician specialty, type(s) of surgery performed, and level of activity were not queried.

As part of an intern project, we had manually reviewed TBDR records for 75 randomly selected plagiocephaly cases delivered in 1999 and 2000 (operationally defined as “early” cases) and 75 delivered in 2004 and 2005 (defined as “later” cases) to determine if there was a change over time in plagiocephaly description, location, laterality, or age at first diagnosis. Assignment of age at diagnosis and other factors was done blinded to early vs later case status. The distribution of description, location, and laterality among early and later cases was compared using a χ² test; mean age at diagnosis was compared using a t test.

This analysis was not reviewed by the Texas Department of State Health Services institutional review board because it was deemed to be standard public health practice.

From 1999 to 2007, the prevalence of plagiocephaly in Texas increased more than 9-fold, from 3.0 cases per 10 000 live births to 28.8 (Figure and Table 1), resulting in 6295 total cases with a definite diagnosis. This was equivalent to an average annual increase of 21.2% per year (95% confidence interval, 12.7-30.7), which is highly statistically significant. Time trends were widely variable across the state. The bulk of the increase came from the Dallas/Fort Worth region, where the birth prevalence of plagiocephaly increased 23.2 times from 2.6 cases per 10 000 live births in 1999 to 60.5 cases per 10 000 live births in 2007.

At least 5% of the cases visited 1 or more of 7 health care facilities (Table 2), of a total of 245 facilities in Texas accessed by the TBDR. Five of the 7 were in the Dallas/Fort Worth region. Facility C in Houston was visited by the majority of cases in 1999, but this quickly dwindled. Beginning in 2001 through 2007, roughly half the cases were seen in facilities A or B, both in Dallas/Fort Worth. Six of the 7 facilities had a craniofacial clinic.

Similar significant and increasing trends in plagiocephaly prevalence were evident in all demographic subgroups whether based on maternal age or race/ethnicity, infant sex, plurality, or gestational age (Table 3). While all maternal education groups showed significant increasing time trends, the trend was greatest among mothers with less education (P for interaction of year with education = .01). The proportion of plagiocephaly cases from multiple births remained stable over time, but the proportion of cases born preterm increased very slightly from 27.2% in 1999 to 29.0% in 2007 (data not shown).

Clinical subgroups also showed similar significant increasing time trends; thus, the trend did not differ depending on whether cases also had torticollis, for example (Table 4). However, the great majority of total plagiocephaly cases were definite diagnosis, mild, and without torticollis or oligohydramnios. There was no change over time in the proportion of plagiocephaly cases with documented oligohydramnios, while the proportion of cases with torticollis decreased from 40.8% in 1999 to 27.9% in 2007 (data not shown).

We examined diagnostic or therapeutic procedures documented in the TBDR for at least 5% of the total cases (Table 5). All of these procedures with the exception of physical examination were documented in increasing percentages of plagiocephaly cases over time. Overall, however, most cases had records of “physical examination,” “consultation,” and “radiography.”

In our substudy looking at early vs later cases, the mean age at first diagnosis was 156.9 days for cases born in 1999 and 2000 and 148.8 days for those born in 2004 and 2005; this difference was not statistically significant. Laterality of plagiocephaly did not differ significantly between early and late cases. We could not adequately compare the distribution of descriptions and skull locations because too many records were missing that information.

Among other birth defects of the skull or face, only 1 exhibited a statistically significant downward time trend in 1999 to 2007, “asymmetric head” (BPA code 754.055), but it decreased at only 4.3% per year (Table 6). Several defects significantly increased: “depressions in skull” (BPA code 754.040), “other skull deformity” (BPA codes 754.080 and 754.090), “craniosynostosis” (BPA codes 756.000-756.030), “Goldenhar syndrome/hemifacial microsomia” (BPA codes 756.060 and 756.065), “hypertelorism, telecanthus, wide-set eyes” (BPA code 756.085), and “other skull or face bone anomalies” (BPA codes 756.080 and 756.090). No other defects of the skull or face exhibited significant time trends in 1999 to 2007. BPA code 754.040 (“depressions in skull”) is used for abnormalities in fontanelle size or shape and these defects compose the majority of cases in this code.

We observed greater than a 9-fold increase in the prevalence of plagiocephaly from 1999 to 2007. This was consistent with previous reports of increased prevalence of plagiocephaly since 1992, when supine sleep position was first recommended by the AAP.^1,10-14 However, that recommendation was not likely to explain our observed dramatic increase, because our study period began 7 years after the recommendation was released. Also, because the mean age at first diagnosis was similar for cases in born 1999 and 2000 and 2004 and 2005, there is no evidence that the temporal increase was caused by an increase in the proportion of acquired (vs congenital) plagiocephaly, as might be expected from adopting the AAP recommendation.¹⁰ On the other hand, barriers to following the supine sleep recommendation among lower-income mothers have been identified¹⁷ and we observed the greatest increases over time among lower-education groups, so this remains a possibility.

The time trend was observed in most regions of Texas but was most pronounced in the Dallas/Fort Worth region. This seems to have been driven largely by certain facilities; 5 of the 7 facilities visited by at least 5% of the cases were in the Dallas/Fort Worth area. Six of the 7 facilities had self-identified craniofacial clinics, which may have encouraged increased referrals to them but which also might have been in response to increased referrals and need for care. Other than tending to have self-identified craniofacial clinics, we were unable to distinguish characteristics of facilities showing dramatic increases from those that did not, either in Texas or specifically within the Dallas/Fort Worth region.

The plagiocephaly increase was not explained by changes in coding practice in the TBDR or health care facilities. Only the skull defect “asymmetric head” decreased significantly over time, but at 4.3% per year among 745 total cases, this is unlikely to have supplied the 21.2% per year increase among 6295 cases of plagiocephaly.

Clinical subgroups all showed similar increasing time trends, but the majority of plagiocephaly cases were definite cases, mild, and without torticollis or oligohydramnios. There was no change over time in the proportion of cases that had documented oligohydramnios, and the proportion with torticollis actually decreased, so those could not explain the temporal pattern. Fetal crowding in multiple births may deform the skull.^1-3 However, increasing occurrence of twins and triplets, for example, could not explain the temporal increase in plagiocephaly because the proportion of plagiocephaly cases from multiple births remained stable over time. Also, skull deformation from in utero restriction or delivery typically corrects spontaneously in the newborn period and would not lead to diagnosis of deformational plagiocephaly at 150 days of age. There was a slight upward trend in the proportion of cases that were born preterm, so that might have been partly responsible. However, it was a very slight increase (from 27.2% of cases to 29.0% of cases), so it would not explain the 9-fold increase in plagiocephaly.

The current treatment modalities for plagiocephaly include head repositioning and/or physical therapy, orthotic helmet (molding) therapy, and surgery.¹⁴ Although it appears that helmet therapy may be more effective than repositioning, definitive conclusions are not available and cost may also influence which approach is adopted.^18,19 In our study, information on repositioning or helmet therapy was unavailable. Of the diagnostic or therapeutic procedures we could examine, prevalence among cases for all except “physical examination” increased over time, although most cases in total had noninvasive procedures like “physical examination,” “consultation,” and “radiography.” The Healthcare Common Procedure Coding System Alpha Numeric Panel issued a new billing code in October 2002: S1040, cranial remolding orthosis, rigid, with soft interface material, custom fabricated (includes fitting and adjustment).²⁰

Because the largest increases were in tertiary care and referral centers, it is possible that the increase in cases is related to increased referral, increased treatment, or both. Treatment patterns can vary even among medical specialists, eg, practitioners of one specialty might be more likely to treat with helmets than members of another specialty would be, and these subtleties in treatment patterns could also affect referrals. For example, parents may be eager to obtain a helmet rather than follow a “passive” repositioning regimen, and this might lead more pediatricians to refer patients to a particular specialty practitioner. However, it was not possible to determine the driving force behind increased referral. Possibilities include parental request, primary care physician recommendations, increased marketing of either the evaluation services or treatment services, and reimbursement incentives.

This study had several strengths. It documented and described a dramatic increase of plagiocephaly prevalence over time. It examined a variety of hypotheses using data from a large population-based birth defects registry. The descriptive epidemiology part of the study allowed us to examine data on a large number of plagiocephaly cases, while the nested case-control design looking at early (1999-2000) vs later (2004-2005) cases permitted examination of detailed information such as age at first diagnosis. Potential observational bias was decreased because cases were abstracted using hospital records, unless the location of diagnosis had shifted from facilities the TBDR does not access (private physicians' offices) to facilities we do access (hospitals).

The main limitation was that, using the data available to the TBDR, we were unable to pin down an explanation for the temporal increase. To evaluate this, an analysis of referral patterns and use of active therapies and interventions would be necessary. It is possible that development and marketing of reimbursable procedures (such as orthotic helmets) prompted increased referral. It is also possible that parental discomfort with the cosmetically nonstandard head shape played a role. In Texas, prescriptions for the orthotic helmets may be written by any physician, but some insurance providers require(d) that the prescriptions be generated by a specialist (craniofacial surgeon, neurosurgeon, or geneticist) for payment. Thus, increased use of the helmets, for whatever reason, would lead to increased specialist referral and thus increased ascertainment by the TBDR. The TBDR does not collect sufficiently detailed information to test that.

We observed a statistically significant increase in the prevalence of plagiocephaly in Texas from 1999 to 2007. A small part of this might have been due to delayed compliance with the AAP recommendation for supine infant sleeping and a slight increase in preterm births. It was definitely not due to changes in birth defect coding practice, trends in multiple births, or trends in clinical conditions like torticollis or oligohydramnios. Because the plagiocephaly trend was observed mainly in patients visiting certain health care facilities, among mild cases, and among cases with minimally invasive procedures, we suspect it may be due mainly to changes in available therapies and insurance reimbursement practices.

Correspondence: Peter H. Langlois, PhD, Birth Defects Epidemiology and Surveillance Branch, Texas Department of State Health Services, MC 1964, PO Box 149347, Austin, TX 78714-9347 (peter.langlois@dshs.state.tx.us).

Accepted for Publication: January 26, 2011.

Published Online: April 4, 2011. doi:10.1001/archpediatrics.2011.42

Author Contributions: Dr Langlois 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. Study concept and design: Ethen, Scheuerle, and Langlois. Acquisition of data: Ethen and Langlois. Analysis and interpretation of data: Sheu, Ethen, Scheuerle, and Langlois. Drafting of the manuscript: Sheu and Langlois. Critical revision of the manuscript for important intellectual content: Ethen, Scheuerle, and Langlois. Statistical analysis: Ethen and Langlois. Administrative, technical, and material support: Sheu, Ethen, and Langlois. Study supervision: Langlois.

Funding/Support: This project was supported in part through cooperative agreement U01DD000494 from the Centers for Disease Control and Prevention, as well as through Title V Maternal and Child Health Block Grant funds from the Health Resources and Services Administration, with the Texas Department of State Health Services.

Additional Contributions: We are grateful to Hyrum Kanady, Dallas/Fort Worth regional supervisor for the Birth Defects Epidemiology and Surveillance Branch, for first alerting us to an increase in cases.