Neurodevelopment of Infants with and without Craniofacial Microsomia

Abstract

Objectives

To determine whether infant cases with craniofacial microsomia (CFM) evidence poorer neurodevelopmental status than demographically-similar infants without craniofacial diagnoses (“controls”), and to examine cases’ neurodevelopmental outcomes by facial phenotype and hearing status.

Study design

Multi-center, observational study of 108 cases and 84 controls aged 12 to 24 months. Participants were assessed by the Bayley Scales of Infant and Toddler Development-Third Edition (Bayley-III) and the Preschool Language Scales, Fifth Edition (PLS-5). Facial features were classified with the Phenotypic Assessment Tool for Craniofacial Microsomia.

Results

After adjustment for demographic variables, there was little difference in Bayley-III or PLS-5 outcomes between cases and controls. Estimates of mean differences ranged from −0.23 to 1.79 corresponding to standardized effect sizes of −0.02 to 0.12 (P values from .30 to .88). Outcomes were better among females and those with higher socioeconomic status. Among cases, facial phenotype and hearing status showed little to no association with outcomes. Analysis of individual test scores indicated that 21% of cases and 16% of controls were developmentally delayed (OR=0.68, 95% CI 0.29 to 1.61).

Conclusions

Although learning problems have been observed in older children with CFM, we found no evidence of developmental or language delay among infants. Variation in outcomes across prior studies may reflect differences in ascertainment methods and CFM diagnostic criteria.

Craniofacial microsomia (CFM), also known as hemifacial microsomia, is a complex congenital condition typically involving underdevelopment of the mandible and ear (1, 2). CFM occurs in approximately 1 in 3,500 to 5,600 live births (3), with higher prevalence among Hispanic and Native American families (4, 5). CFM has been characterized as a spectrum of phenotypic anomalies ranging from isolated unilateral microtia to bilateral malformations of the ear, mandible, facial soft tissue, and orbit (2). Other cranial and extra-cranial malformations may co-occur (eg, lateral oral clefts, vertebral anomalies, cardiac defects).

Among the several functional problems associated with CFM (2), neurodevelopmental delays are perhaps the least understood and most difficult to recognize clinically. Although such deficits can strongly affect children’s quality of life (6, 7), their impact can be mitigated by early evaluation and intervention (8, 9). The few existing studies of neurodevelopment in the CFM population, mostly involving older children, have suggested that severe intellectual disability is rare, but mild to moderate neurodevelopmental delays and learning problems may occur at higher rates than those found in the general population (10–13). Given the heterogeneity of the CFM phenotype (2), and the theorized embryological mechanisms underlying the development of face and brain (14), it is possible that neurodevelopmental outcomes may vary in relation to the severity and pattern of anomalies observed in this condition (eg, microtia with or without microphthalmia or vertebral defects). Although previous investigations have examined neurodevelopmental outcomes by one or more selected features (e.g., presence or absence of extracranial anomalies) (10, 11), a consistent pattern of association between phenotype and developmental outcomes has yet to emerge.

Based on the findings from previous neurodevelopmental studies, we hypothesized that infants with CFM would show lower test scores on average and a higher frequency of developmental delay than controls. In secondary analyses, we examined neurodevelopmental outcomes by (a) facial phenotype using a modified version of the pictorial Orbit, Mandible, Ear, Nerve, and Soft tissue (OMENS) scoring system, which includes ratings for orbital asymmetry, mandibular hypoplasia, ear anomalies, facial nerve involvement, and soft-tissue deficiency (15, 16) and (b) the presence and severity of hearing impairment. We also estimated the effects of relevant covariates on neurodevelopmental status, including sex, socioeconomic status (SES), ethnicity, and bilingual home environment.

METHODS

Infants between the ages of 12 and 24 months were recruited to participate in an ongoing observational, longitudinal, multi-center project called Craniofacial Microsomia: Longitudinal Outcomes in Children Pre-Kindergarten (CLOCK), which tracks the neurodevelopmental, speech and hearing outcomes, and phenotypic features of infants with and without CFM (“cases” and “controls,” respectively). Participants were enrolled between 2012 and 2017 from one of 6 craniofacial centers: Children’s Hospital of Los Angeles (CHLA), Children’s Hospital of Philadelphia (CHOP), Seattle Children’s Hospital (SCH), University of Illinois-Chicago (UIC, including Shriners Hospital for Children, Chicago), and University of North Carolina (UNC) at Chapel Hill. This research was approved by the institutional review boards at all participating centers. All parents gave informed consent for their infant to participate in the study.

Cases were recruited from each site’s hospital-based craniofacial centers, hospital-based clinics seeing infants or young children with CFM (e.g. hearing screening programs, ENT programs), and research study websites (eg, https://www.clinicaltrials.gov). To be eligible, cases had to have (1) at least one of the CFM inclusion criteria developed by the Facial Asymmetry Collaborative for Interdisciplinary Analysis and Learning (FACIAL) network (Figure; available at www.jpeds.com); (2) an age between 12–24 months (corrected for prematurity, when applicable, for children born between 34–36 weeks’ gestational age); and (3) a legal guardian who was able to provide informed written consent and be willing to participate for the duration of the study. Exclusion criteria for cases included: (1) diagnosis of a known syndrome (eg, Townes-Brocks, Treacher Collins, branchio-oto-renal, or Nager syndromes); (2) presence of an abnormal karyotype or major medical or neurological conditions (e.g., cancer, cerebral palsy); (3) premature birth (less than 34 weeks’ gestation); (4) any circumstance that would preclude the family’s ability to participate fully in the research; (5) a sibling already participating in the CLOCK study, or (6) a consenting parent who did not speak English or Spanish. Of the 219 potentially eligible cases that were approached, 108 (49%) were enrolled. Most non-participating families declined passively by not responding to study invitations.

Figure 1.

Eligibility Criteria for the CLOCK study

We identified eligible participants with demographic characteristics that met our frequency–matching criteria for the case cohort; these included infant age and sex, family SES, and language spoken in the home (English or Spanish). Exclusion criteria for controls included: (1) meeting one or more of the exclusionary criteria for cases; and (2) diagnosis or history of any disorder, condition, or injury that would affect facial features (e.g., craniofacial malformation or deformation; facial surgery or trauma). Of the 148 potentially eligible controls that were approached, 84 (57%) were enrolled.

Measures

We used the Bayley Scales of Infant and Toddler Development–Third Edition (Bayley-III) to assess cognitive and motor skills (17). The Bayley-III yields composite scores for both cognition and motor ability and subscale scores for fine and gross motor development. Raw scores are converted to norm-referenced standard scores (mean = 100, SD = 15) for composite scales and scaled scores (mean = 10, SD = 3) for the motor subscales. The Preschool Language Scale–Fifth Edition (PLS-5) was used to assess expressive and receptive language, using either the English or Spanish version of this norm-referenced, validated test (18). We selected the PLS-5 over the Bayley-III language scales as the latter are available only in English, and the Spanish version of the PLS-5 includes culturally relevant items with norms from a large sample of monolingual and bilingual Spanish-speaking children (19). The PLS-5 yields two scale scores, auditory comprehension (PLS-AC) and expressive communication (PLS-EC), as well as a total language score (PLS-TL), all based on a combination of child performance, examiner observations, and caregiver reports.

For both tests, gestational age was calculated using family report of due date and birth date. We corrected Bayley-III scores for prematurity for children born between 34 and 36 weeks’ gestation. Testing was completed in English (69%), Spanish (13%), or a combination of English and Spanish (18%) determined by families’ reports of their language use across settings, as well as examiners’ observations of participants’ language use during the study visit. Trained bilingual psychometrists used verbal Bayley-III prompts that were translated into Spanish for consistent use across sites. The PLS-5 has standardized versions in both English and Spanish. All assessments were videotaped and about 20% were double-scored for reliability by three of the authors (BC, KKS, and MS). Average level of agreement for item-by-item scoring was 97%, with agreement levels for single test administrations ranging from 70% to 100%.

Cases’ hearing status was based primarily on audiological information obtained as part of routine clinical care. Ninety-one (84%) of cases had such data available for review. Infants with audiology data were considered to have hearing loss when they demonstrated a greater than 20 decibel Pure Tone Average (PTA) over 4 frequencies (500, 1000, 20000, 4000 Hz; PTAs were calculated if at least 3 frequencies were present). For those cases without audiograms, hearing status was based on the absence of the external ear canal (i.e. aural atresia) or external ear canal stenosis, both of which result in conductive hearing loss. In a prior investigation, 98% of children with aural atresia or external ear canal stenosis had PTA of > 20 dB hearing loss (20). This allowed us to create the following categories of hearing status for all cases: (1) “No hearing loss” was defined as the absence of atresia and a negative audiometric finding; (2) “Unilateral hearing loss” referred to single-sided hearing loss based on audiometry or the presence of unilateral atresia; and (3) “Bilateral hearing loss” was assigned to cases meeting at least one of the following criteria: (a) presence of bilateral atresia, (b) audiological evidence of hearing loss in both ears, and/or (c) the presence of unilateral atresia plus audiological evidence of hearing loss in the contralateral ear.

Phenotypic classifications were based on the integration of data collected from a health history interview with parent, medical record abstraction, and analysis of multiple craniofacial photographs taken of each participant. A modified version of the Phenotypic Assessment Tool for Craniofacial Microsomia (PAT-CFM (21–22)) was used by two craniofacial dysmorphologists (CH and DL) to classify the presence/absence and severity of facial features associated with CFM (for details see Luquetti et al (23)). Both phenotype coders were blinded to neurodevelopmental test scores. We categorized the presence of microtia according to the classification scheme developed by Marx and modified by Rogers (24,25). This is the classification used in the modified pictorial OMENS form and one of the most frequently used systems(16). Distinguishing the degree of mandibular and/or facial soft tissue deficiency on photographs is challenging and the two features are often associated (26, 27). Therefore, we used the term “mandibular hypoplasia” to include the presence of hypoplasia of either the facial soft tissue or mandible.

Three phenotypic subgroups were identified: 1) microtia only (in the absence of mandibular hypoplasia, epibulbar dermoids, lateral clefts, preauricular or facial tags, small and/or displaced orbit, nerve palsies); 2) microtia and mandibular hypoplasia; and 3) other combinations of CFM-associated malformations (two or more were required). We also developed phenotypic groups based on the laterality of features (e.g., right unilateral microtia plus left facial tag), the laterality of external ear involvement, the maximum grade of ear involvement, and the presence of extracranial anomalies (e.g., congenital heart defect).

Statistical Analyses

Group differences in neurodevelopment were estimated using linear regression with robust standard error estimates. Controls served as the referent for all analyses. Estimates were adjusted for age of the assessment (continuous), sex (male vs. female), SES (continuous), study site (categorical), race (white vs. non-white), and testing language (any Spanish vs. exclusively English). We also examined characteristics of cases and controls by the presence/absence of neurodevelopmental delay, defined as >1 standard deviation below the normative test mean for any one or more of the following: total PLS-5 scores and Bayley-III cognitive and motor scale composite scores. Logistic regression was used to estimate the odds of delay and adjusted for all demographic variables included in the primary analyses. We also conducted 2 post hoc sensitivity analyses: (1) excluding CHLA (which tested predominantly in Spanish) and (2) restricted to participants who had assessments in English only.

We conducted several secondary analyses to evaluate differences by cases’ phenotype, using controls as the referent category and adjusting for the same demographic factors as those listed above. We estimated differences across three categories of phenotype (microtia only, microtia plus mandibular hypoplasia, and other CFM-associated features), by overall laterality of features (unilateral or bilateral), by external ear involvement (unilateral or bilateral, excluding cases with no ear involvement), by grade of external ear involvement (0–2 and 3–4), and by the presence or absence of extracranial anomalies. We also evaluated whether cases’ neurodevelopmental outcomes differed by hearing status (no hearing loss, unilateral hearing loss, and bilateral hearing loss).

RESULTS

Demographic characteristics of cases and controls are provided in Table I. Cases were more likely than controls to be male, to be Hispanic or Latino, and to undergo their assessment in Spanish or a combination of Spanish and English. There was no evidence for differences between cases and controls in PLS-5 or Bayley-III outcomes after adjustment for demographic confounding variables. Estimates of mean differences from the linear regression models were small, ranging from −0.23 to 1.79 (p-values ranged from 0.30 to 0.88) (Table 2). Examination of covariates in the adjusted models revealed that outcome scores were higher in females than males and with increasing SES. PLS-5 and Bayley-III cognitive scores were also higher among infants from families who spoke some Spanish relative to those who spoke English exclusively. There was also evidence for differences by study site, with SCH participants consistently scoring higher and UIC participants consistently scoring lower compared with CHLA participants.

Table 1.

Characteristics of infants with and without craniofacial microsomia

Characteristic	Cases		Controls
	N	%	N	%
Total	108	100.0	84	100.0
Sex
Male	67	62.0	37	44.0
Female	41	38.0	47	56.0
Age at visit
Mean, SD	14.5	2.8	14.4	2.7
<13 months	32	29.6	32	38.1
13–14 months	41	38.0	26	31.0
>14 months	35	32.4	26	31.0
Gestational age
Mean, SD	39.0	1.5	39.5	1.4
34–37 weeks	13	12.0	11	13.1
>37–40 weeks	73	67.6	48	57.1
>40 weeks	22	20.4	25	29.8
SES*
Mean, SD	35.2	14.2	37.5	15.0
I	16	14.8	15	17.9
II	20	18.5	17	20.2
III	28	25.9	23	27.4
IV	29	26.9	19	22.6
V	15	13.9	9	10.7
Hispanic or Latino*
No	49	45.4	51	60.7
Yes	59	54.6	31	36.9
Race*
White	80	74.1	48	57.1
Black/African American	3	2.8	5	6.0
Asian	7	6.5	0	0.0
American Indian/AK Native	3	2.8	3	3.6
Other race	3	2.8	4	4.8
Multiracial	10	9.3	23	27.4
Insurance
Private	50	46.3	44	52.4
Medicaid	58	53.7	40	47.6
Testing language (PDP)
100% English	69	63.9	63	75.0
100% Spanish	19	17.6	7	8.3
Combination English/Spanish	20	18.5	14	16.7
Study site
CHLA	40	37.0	8	9.5
CHOP	3	2.8	6	7.1
SCH	38	35.2	57	67.9
UNC	16	14.8	7	8.3
UIC	11	10.2	6	7.1
Phenotype
Microtia only	30	27.8	0	0.0
Microtia + Mandibular hypoplasia	59	54.6	0	0.0
Other	19	17.6	0	0.0
Atypical/No discernible anomaly	0	0.0	84	100.0
Hearing loss*
None	5	4.6	76	97.4
Unilateral	67	62.0	2	2.6
Bilateral	36	33.3	0	0.0
Hearing aid use
No	53	49.1	84	100.0
Yes (current or past use)	55	50.9	0	0.0
Hearing aid use (% based on those with hearing aid use)
<2 hrs/day	12	25.0	-	-
2–6 hrs/day	20	41.7	-	-
>6 hrs/day	16	33.3	-	-
Intervention services
Occupational therapy	7	6.5	0	0
Physical therapy	13	12.0	1	1.19
Speech, hearing, or language services	59	54.6	0	0.0
Other developmental services	8	7.4	0	0
Any services	70	64.8	1	1.2

^*Missingness: SES (1 control); ethnicity (2 controls); race (1 control, 2 cases); hearing loss (6 controls)

Table 2.

Comparison of mean neurodevelopmental scores for children with and without CFM

Test	Measure	Controls N	Cases N	Unadjusted				Adjusted*
				Mean Difference	95% CI		p-value	Mean Difference	95% CI		p-value
PLS-5	Total	82	108	−0.89	−3.60	1.81	0.52	0.80	−2.07	3.67	0.58
	Auditory Comprehension	82	108	−0.22	−2.80	2.36	0.87	1.20	−1.48	3.88	0.38
	Expressive Communication	82	108	−1.48	−4.49	1.54	0.34	0.34	−2.93	3.61	0.84
Bayley-III	Cognitive	82	108	−3.44	−6.71	−0.17	0.04	−0.23	−3.30	2.84	0.88
	Motor Composite	81	107	−1.37	−4.66	1.93	0.42	1.79	−1.61	5.19	0.30
	Fine Motor	82	108	−0.14	−0.87	0.60	0.71	0.30	−0.51	1.11	0.47
	Gross Motor	81	107	−0.32	−0.93	0.28	0.30	0.24	−0.36	0.84	0.44

^*Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish or combo)

The criterion for developmental delay was met in one or more areas by 13 (16%) of controls and 23 (21%) of cases (Table 3). Delayed cases were more likely to be male. In both cases and controls, delayed participants were of lower SES and more likely to have Medicaid insurance. After adjustment for demographic factors there was little evidence for case-control differences in the odds of developmental delay (OR=0.68, 95% CI 0.29 to 1.61).

Table 3.

Characteristics of infants with craniofacial microsomia by delay

Characteristic	Not delayed		Delayed
	N	%	N	%
Total	85	100.0	23	100.0
Sex
Male	50	58.8	17	73.9
Female	35	41.2	6	26.1
Age at visit
Mean, SD	14.3	2.4	15.3	4.0
<13 months	22	25.9	10	43.5
13–14 months	36	42.4	5	21.7
>14 months	27	31.8	8	34.8
SES
Mean, SD	37.0	14.1	28.9	12.7
I	15	17.6	1	4.3
II	17	20.0	3	13.0
III	23	27.1	5	21.7
IV	21	24.7	8	34.8
V	9	10.6	6	26.1
Hispanic or Latino
No	41	48.2	8	34.8
Yes	44	51.8	15	65.2
Race*
White	63	74.1	17	73.9
Black/African American	2	2.4	1	4.3
Asian	5	5.9	2	8.7
American Indian/AK Native	2	2.4	1	4.3
Other race	3	3.5	0	0.0
Multiracial	10	11.8	0	0.0
Insurance
Private	43	50.6	7	30.4
Medicaid	42	49.4	16	69.6
Testing language (PDP)
100% English	57	67.1	12	52.2
100% Spanish	14	16.5	5	21.7
Combination English/Spanish	14	16.5	6	26.1
Phenotype
Microtia only	20	23.5	10	43.5
Microtia + Mandibular hypoplasia	50	58.8	9	39.1
Other	15	17.6	4	17.4
Study site
CHLA	27	31.8	13	56.5
CHOP	2	2.4	1	4.3
SCH	36	42.4	2	8.7
UNC	12	14.1	4	17.4
UIC	8	9.4	3	13.0

^*Missingness: race (1 case with delay)

Phenotype differences

There was little evidence for differences by CFM phenotype. Relative to controls, cases with microtia only and those with microtia with mandibular hypoplasia scored slightly higher than controls for PLS-5 total language and auditory comprehension, but estimates included the null and were inconsistent in direction for other measures (Table 4). Mean differences for cases with other CFM features were inconsistent and imprecise across measures. Cases with bilateral CFM features scored lower than controls on all 7 measures and cases with unilateral involvement consistently scored higher than controls; however, effect sizes were modest and confidence intervals included the null for all estimates (Table 5; available at www.jpeds.com). A similar pattern was observed for laterality of external ear involvement (Table 6; available at www.jpeds.com). There was little evidence for differences by grade of ear involvement (Table 7; available at www.jpeds.com). Nineteen percent of cases had extracranial malformations and these cases scored lower than unaffected controls across all measures, although estimates were imprecise and included the null (Table 8; available at www.jpeds.com). Cases without extracranial malformations had modestly higher scores than controls, but all estimates included the null.

Table 4.

Comparison of mean neurodevelopmental scores for children with and without CFM, by phenotype

Test	Measure	Controls N=84	Cases
			Microtia only N=30				Microtia + Mandibular Hypoplasia N=59				Other CFM-associated features N=19
			Mean Difference*	95% CI		p-value	Mean Difference*	95% CI		p-value	Mean Difference*	95% CI		p-value
PLS-5	Total	Ref	1.20	−2.90	5.30	0.57	1.52	−1.81	4.84	0.37	−2.03	−7.04	2.98	0.43
	Auditory Comprehension	Ref	2.62	−0.74	5.98	0.13	1.52	−1.51	4.56	0.33	−1.53	−6.54	3.48	0.55
	Expressive Communication	Ref	−0.59	−5.74	4.56	0.82	1.39	−2.39	5.16	0.47	−2.05	−7.37	3.27	0.45
Bayley-III	Cognitive	Ref	−1.75	−6.48	2.99	0.47	0.00	−3.59	3.59	0.99	0.75	−3.89	5.38	0.75
	Motor Composite	Ref	−0.39	−5.43	4.64	0.88	2.86	−1.06	6.78	0.15	0.71	−6.86	8.29	0.85
	Fine Motor	Ref	0.43	−0.74	1.61	0.47	0.46	−0.42	1.35	0.31	−0.38	−1.93	1.16	0.63
	Gross Motor	Ref	−0.60	−1.53	0.33	0.21	0.48	−0.21	1.16	0.17	0.43	−0.91	1.78	0.53

^*Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish or combo)

Table 5.

Comparison of mean neurodevelopmental scores for children with and without CFM, by overall laterality of involved features

Test	Measure	Controls N=84	Cases*
			Unilateral N=66				Bilateral N=42
			Mean Difference**	95% CI		p-value	Mean Difference**	95% CI		p-value
PLS-5	Total	Ref	2.70	−0.37	5.78	0.08	−2.05	−5.85	1.75	0.29
	Auditory Comprehension	Ref	2.83	0.02	5.63	0.05	−1.25	−4.77	2.28	0.49
	Expressive Communication	Ref	2.24	−1.30	5.77	0.21	−2.50	−6.92	1.91	0.26
Bayley-III	Cognitive	Ref	0.54	−3.15	4.23	0.77	−1.38	−5.12	2.35	0.47
	Motor Composite	Ref	3.57	−0.13	7.28	0.06	−0.98	−5.96	4.00	0.70
	Fine Motor	Ref	0.69	−0.21	1.59	0.13	−0.28	−1.33	0.78	0.61
	Gross Motor	Ref	0.50	−0.16	1.15	0.14	−0.16	−1.06	0.73	0.72

^*Includes external ear, mandible, soft tissue, ear canal, coloboma, nerves, dermoids, tags, lateral cleft, cleft lip, and orbits (size + displacement)

^**Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish or combo)

Table 6.

Comparison of mean neurodevelopmental scores for children with and without CFM, by external ear involvement

Test	Measure	Controls N=84	Cases*
			Unilateral N=72				Bilateral N=31
			Mean Difference**	95% CI		p-value	Mean Difference**	95% CI		p-value
PLS-5	Total	Ref	2.26	−0.83	5.35	0.15	−1.26	−5.43	2.91	0.55
	Auditory Comprehension	Ref	2.30	−0.63	5.23	0.12	−0.83	−4.32	2.66	0.64
	Expressive Communication	Ref	1.99	−1.50	5.48	0.26	−1.67	−6.77	3.44	0.52
Bayley-III	Cognitive	Ref	0.52	−2.96	4.01	0.77	−1.66	−5.92	2.59	0.44
	Motor Composite	Ref	3.65	0.06	7.24	0.05	−3.58	−8.71	1.54	0.17
	Fine Motor	Ref	0.76	−0.11	1.63	0.09	−0.66	−1.73	0.40	0.22
	Gross Motor	Ref	0.38	−0.26	1.03	0.25	−0.55	−1.50	0.41	0.26

^*Excludes 5 controls without ear involvement

^**Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish)

Table 7.

Comparison of mean neurodevelopmental scores for children with and without CFM, by maximum external ear grade

Test	Measure	Controls N=84	Cases
			Grades 0–2 N=38				Grades 3–4 N=70
			Mean Difference*	95% CI		p-value	Mean Difference*	95% CI		p-value
PLS-5	Total	Ref	−1.22	−4.76	2.32	0.50	2.16	−0.98	5.30	0.18
	Auditory Comprehension	Ref	0.78	−2.62	4.18	0.65	1.48	−1.43	4.38	0.32
	Expressive Communication	Ref	−2.83	−6.81	1.15	0.16	2.48	−1.12	6.08	0.18
Bayley-III	Cognitive	Ref	−0.27	−3.75	3.21	0.88	−0.21	−3.93	3.52	0.91
	Motor Composite	Ref	0.55	−3.72	4.83	0.80	2.65	−1.30	6.61	0.19
	Fine Motor	Ref	0.21	−0.78	1.20	0.68	0.37	−0.58	1.31	0.44
	Gross Motor	Ref	−0.03	−0.79	0.74	0.95	0.42	−0.28	1.13	0.23

^*Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish)

Table 8.

Comparison of mean neurodevelopmental scores for children with and without CFM, by presence of extracranial malformations

Test	Measure	Controls* N=83	Cases
			Absent N=87				Present N=21
			Mean Difference**	95% CI		p-value	Mean Difference**	95% CI		p-value
PLS-5	Total	Ref	1.97	−1.02	4.96	0.20	−3.15	−7.85	1.55	0.19
	Auditory Comprehension	Ref	1.85	−1.01	4.71	0.20	−1.18	−4.88	2.52	0.53
	Expressive Communication	Ref	1.89	−1.46	5.24	0.27	−4.79	−10.69	1.11	0.11
Bayley-III	Cognitive	Ref	0.34	−3.03	3.70	0.84	−2.29	−6.78	2.19	0.31
	Motor Composite	Ref	3.21	−0.36	6.77	0.08	−4.75	−9.60	0.09	0.05
	Fine Motor	Ref	0.59	−0.25	1.43	0.16	−1.04	−2.26	0.18	0.09
	Gross Motor	Ref	0.40	−0.25	1.05	0.23	−0.55	−1.51	0.41	0.26

^*Excludes 1 control with an extracranial malformation

^**Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish)

Hearing

Among cases, 33% had bilateral hearing loss, 62% had unilateral hearing loss, and 5% had no evidence for hearing loss. Cases with hearing loss and without use of a hearing aid had higher scores than controls on most measures, in particular PLS-5 auditory comprehension and Bayley-III fine and composite motor scores, although all estimates included the null (mean differences ranged from −0.47 to 3.13, p-values ranged from 0.12 to 0.98; Table 9; available at www.jpeds.com). There was little evidence for differences in cases with hearing loss and hearing aid use relative to controls. Cases with no hearing loss and no hearing aid use had lower scores across most outcomes, although only one, Bayley-III gross motor, included the null (estimates ranged from −4.80 to 0.97, p-values ranged from 0.002 to 0.88).

Table 9.

Comparison of mean neurodevelopmental scores for children with and without CFM, by hearing loss and hearing aid use

Test	Measure	Controls * N=76	Cases*
			No hearing loss, no hearing aid use N=4				Hearing loss + hearing aid use N=36				Hearing loss w/out hearing aid use N=61
			Mean Difference**	95% CI		p-value	Mean Difference**	95% CI		p-value	Mean Difference**	95% CI		p-value
PLS-5	Total	Ref	−4.06	−10.87	2.75	0.24	1.48	−2.78	5.74	0.50	0.83	−2.34	4.00	0.61
	Auditory Comprehension	Ref	−2.68	−9.60	4.24	0.45	0.59	−3.11	4.28	0.76	2.05	−1.04	5.13	0.20
	Expressive Communication	Ref	−4.80	−12.14	2.54	0.20	2.23	−2.82	7.27	0.39	−0.47	−3.99	3.06	0.80
Bayley-III	Cognitive	Ref	0.37	−4.41	5.15	0.88	1.25	−3.60	6.11	0.61	−0.04	−3.60	3.52	0.98
	Motor Composite	Ref	−2.56	−10.04	4.91	0.50	−0.32	−5.73	5.09	0.91	3.13	−0.81	7.07	0.12
	Fine Motor	Ref	0.97	−0.42	2.37	0.17	0.11	−1.06	1.28	0.85	0.37	−0.56	1.31	0.43
	Gross Motor	Ref	−1.91	−3.13	−0.70	0.002	−0.26	−1.24	0.71	0.60	0.60	−0.08	1.29	0.09

^*Cases who used a hearing aid for <2 hrs/day were considered non-users; cases without documented hearing loss but who reported using a hearing aid were considered to have hearing loss. Excludes 7 cases with missing information on time spent using a hearing aid, 2 controls with unilateral hearing loss, and 6 controls missing information on hearing loss.

^**Adjusted for age (months), SES (continuous), sex, study site (categorical), race (white/non-white), language (English/Spanish or combo)

DISCUSSION

In this case-control study we found no evidence for neurodevelopmental differences between infants with and without CFM in the first to second year of life. Group differences in mean test scores were negligible before and after statistical adjustment for socio-demographic variables that may affect neurodevelopment, including age, sex, SES, ethnicity, and the language in which testing was conducted (English or Spanish). Null findings were consistently observed across tested development domains, including cognition, language, and fine and gross motor skills. These results were only modestly influenced by facial phenotype, including the graded severity of ear anomalies. The presence of extracranial anomalies was associated with lower test scores across all measured domains, although these effects were also modest. There was no indication that hearing loss or use of hearing aids was related to developmental outcomes among infants with CFM, a finding that is likely due to their limited exposure to hearing intervention at this age. Positive intervention effects are more likely to be observed when these children are assessed at age 3.

Our findings regarding group differences are largely discrepant from those observed in previous neurodevelopmental investigations of children with CFM. The most notable differences are observed in comparisons with the Cohen et al (10) report of 24 infants and preschoolers with CFM (average age 26 months). Standardized developmental tests were given as part of a clinical work-up, including an earlier version of the Bayley exam (28) than the one we used. Nearly 60% of the Cohen sample scored more than two standard deviations below the mean in one or more developmental domains and developmental status varied with phenotypic presentation as indicated by substantially lower test performance among patients with bilateral (vs. unilateral) anomalies, low muscle tone, and cervical spine abnormalities.

Two other studies published previously followed participants in a longitudinal cohort recruited by Werler et al (7). Neurobehavioral assessments were administered in childhood (average age 7) and adolescence (average age 13). Testing at age 7 indicated that children with CFM were 2 to 3 times more likely than children without craniofacial anomalies to score in the delayed range on tests of receptive vocabulary and visual-motor skills (11). In the adolescent follow-up (12), average scores on tests of IQ and academic achievement were lower in cases than controls, with group differences most prominent in language functions. However, the magnitude of most differences was small (standardized effect sizes, −0.04 to −0.2). Analyses by facial phenotype showed that case and control group differences were slightly larger for cases with both microtia and mandibular hypoplasia than either condition alone, but with small effect sizes. As in the present study, cases’ hearing status unexpectedly had little effect on these outcomes, which the investigators attributed to the use of non-standardized hearing assessments.

Such discrepancies among studies may be due to variation in ascertainment methods (e.g., recruitment from audiology vs. genetics clinics), case sample definitions in relation to the CFM spectrum, and exclusions of related or other medical conditions. For example, nearly 20% of the patients examined by Cohen et al (10) had anomalies identified by brain imaging, a subgroup that generated an average IQ of 65. In contrast, Werler et al (7) observed < 5% of their sample to have central nervous system (CNS) anomalies (as indicated by medical records) and these participants were excluded from their case to control comparisons. In the current study, we also found < 5% of cases to have documented indications of CNS involvement. In another example, Collett et al (2011)(11) excluded children who had unilateral microtia without other features of CFM, but in the current sample we included these cases because isolated unilateral microtia represents an important clinical and theoretical endpoint for the CFM spectrum. This is suggested by the overlapping epidemiology and associated extracranial anomalies (e.g. renal and cervical spine anomalies) between this subgroup and the rest of the CFM spectrum and the resulting potential for misclassification of individuals with CFM-features as having “isolated” microtia (29–33). Wide discrepancies in neurodevelopmental (and other) outcomes will likely persist until investigators adopt the same diagnostic and exclusionary criteria in their definitions of CFM case samples.

The lack of differences observed here between cases and controls and among case subgroups may be due in large part to the age range within which our baseline assessment was conducted (12 to 24 months). In the second year of life, CFM risk factors such as compromised speech and expressive language have yet to influence substantially cognitive development (e.g., assessment of problem-solving skills at this age is largely based on nonverbal functions), but may affect outcomes at later ages. Moreover, given the challenge of obtaining specific vocal or verbal responses from infants in a testing situation, it is difficult to measure speech and language reliably at this age, necessitating reliance upon parents’ recall of language functions. We will be assessing this cohort again at age 3, when we will have a better opportunity to observe directly the impact of speech and expressive language on overall development

Our finding of equivalent developmental status among infants and toddlers with and without CFM may indicate that the academic problems observed among older children with CFM might be influenced by post-infancy risk factors, such as the impact of hearing loss on language acquisition and later, the effects of speech dysfluency or anomalous facial appearance on socialization and achievement motivation. However, the design of this study cannot rule out the possibility that differences might not be accurately assessed or present at this age, or that prenatal neuropathology also contributes to the school-age outcomes of children with CFM (14).

Test scores in the current study were examined categorically using a common criterion for developmental delay (34, 35). Only slightly more cases than controls met this criterion in one or more areas of functioning (21% vs. 16% respectively). Nevertheless, it is important clinically to understand which variables within the case group may distinguish delayed from non-delayed infants. Although we had too few delayed cases to evaluate such differences statistically, the descriptive data in Table 4 suggest that lower SES may be a distinguishing variable. Over 60% of delayed infants with CFM were in families from the lowest two categories of SES (vs. 35% of non-delayed cases). Correspondingly, 70% of delayed cases received Medicaid funding for their medical services (vs. 49% of non-delayed cases).

Consistent with prior epidemiological studies of CFM(5), over half the case families in this cohort (55%) were of Hispanic/Latino ancestry, with 63% of parents in this subgroup speaking Spanish exclusively or a combination of Spanish and English. These demographic characteristics required a bilingual approach to infant assessment, including the use of English and Spanish versions of the PLS-5 as well as examiners fluent in both English and Spanish for Bayley-III administration. Incorporation of bilingual assessment is a complex process that introduces several methodological complications and potential sources of error. For example, different language versions of the same test differ slightly in item content, different normative groups are involved, bilingual test administrations can vary and inter-tester reliability may differ by language. We sought to attenuate the impact of these issues through careful test selection, written protocols to standardize bilingual test administrations, recruitment of frequency-matched controls, and inclusion of related covariates in statistical analyses (eg, test language). In our view, the methodological complications of bilingual assessment are far outweighed by the benefits of having a representative, diverse sample of children with CFM.

Our study does have limitations. First, like prior studies, our sample size was limited, particularly for subgroup analyses, and that could have led to an inability to detect differences. Second, only about half of identified eligible case families consented to participate, which could limit the study’s generalizability of findings to the broader population of children with CFM. However, comparisons of participating and nonparticipating eligible cases indicated that these two groups were reasonably similar with respect to both demographics and phenotype, although some exceptions were observed (eg,, participants were slightly more likely to be male and Hispanic; see Luquetti et al(21) for details). Finally, we were unable to obtain complete audiological data for ~ 15% of our cohort. Consequently, we classified the hearing status of these individuals in relation to the documented presence of aural atresia or external ear canal stenosis. Although not ideal, we have confidence in this classification method, as previous research has shown that nearly all children with either of these conditions have definite hearing loss, as determined by audiological evaluations (20).

CFM is a highly variable diagnosis that has multiple risk factors potentially influencing children’s development. The learning problems observed at school age and in adolescence(11, 12) cannot be understood fully without the longitudinal tracking of development beginning in infancy. This study represents an important first step in elucidating the trajectory of functional outcomes among young children with CFM and opportunities for intervention. We plan to follow this CFM cohort longitudinally. In this baseline assessment of we found no evidence for neurodevelopmental differences among infants with CFM between the ages of one and two. This suggests that the academic problems observed in some older children with CFM might not be accurately assessed or present at this age, or might be influenced by the increasing toll of CFM-related risk factors on post-infancy development (e.g., hearing and speech impairments). As many older children with CFM show average or above average achievement(11, 12), it is important to understand which variables in early childhood best distinguish delayed from non-delayed cases. These issues will provide the investigative framework for our next evaluation of this sample at age 3.