Developmental Outcomes Among Young Children With Congenital Zika Syndrome in Brazil

Key Points Question What are the observed neurodevelopmental sequelae among young children with congenital Zika syndrome (CZS) identified during the Zika virus outbreak in Brazil from 2015 to 2016? Findings In this case series of 121 young children, nearly all children with CZS demonstrated profound developmental delays at age 2 to 3 years across all domains of functioning, with a relative strength in receptive communication. Severity of microcephaly at birth was the only significant factor associated with the severity of delays. Meaning In this study, most children diagnosed with CZS had profound developmental delays; however, there was variability within their developmental profile, providing direction for intervention.


Introduction
In 2015, Brazil experienced a widespread epidemic of Zika virus (ZIKV) infection, which ultimately spread across the Americas, with 87 countries and territories reporting autochthonous transmission. 1 Although estimates suggest a seroprevalence of more than 60% in some regions, 2 it is difficult to estimate exposure given that 60% to 80% of infections are asymptomatic 3 and adult symptoms were typically mild and nonspecific. However, fetal exposure has had severe consequences for a subset of infants. Early reports focused on findings of microcephaly. 4 However, it is now clear that congenital Zika syndrome (CZS) has variable presentation, severity, and prognosis. 5 Congenital Zika syndrome differs from other congenital infections in the 5 following ways: "severe microcephaly with partially collapsed skull; thin cerebral cortices with subcortical calcifications; macular scarring and focal pigmentary retinal mottling; congenital contractures; and marked early hypertonia and symptoms of extrapyramidal involvement." 5,6 According to the Pan American Health Organization, 31 countries reported microcephaly and/or central nervous system complications associated with ZIKV infection. 5 Approximately 3000 children were born with CZS during the 2015 outbreak in Brazil, mostly in the northeast. 7 At least 2 questions must be answered regarding the consequences of prenatal ZIKV exposure.
First, what are the effects on any exposed fetus? Second, what is the prognosis for those with obvious features of CZS at birth? Our study addresses the second question by assessing a sample of children who had prenatal ZIKV exposure and exhibited anomalies at birth not otherwise explained.
These children are at highest risk for severe, lifelong consequences on development and function.
Our broad goals were to characterize consequences and symptoms over time, identify potentially useful interventions for children and families, and document burden for families and costs to society.
Studies to date have provided important information about ZIKV exposure, including ocular manifestations, 8 neurological findings, 9 epileptogenic activity, arthrogryposis, and hearing loss. 10 Several studies have detailed functional impairments and neurodevelopmental skills in small samples, suggesting a high likelihood of severe to profound disabilities. Two early studies 10,11 reported profound delays as measured by a parent-report tool. A cross-sectional study of eight children (mean age, 21.1 months) with CZS using the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III), 12 reported that all children scored at or close to the floor on cognitive and motor domains. Another study 13 described developmental outcomes for 35 children with CZS, reporting below average BSID-III scores and abnormal brain findings confirmed by brain imaging. A larger study of 146 children aged 7 to 32 months whose mothers had confirmed gestational ZIKV infection 14 found that 40% had mild delays (<1 SD below BSID-III mean) and 12.3% had moderate to severe delays in at least 1 functional domain of the BSID-III. However, most children in this study who had exposure did not exhibit obvious anomalies at birth. To our knowledge, no published studies have explored profiles of development in large cohorts of children with obvious clinical features of CZS at birth.
In 2016, we launched a longitudinal study of children with CZS and their families. This study has 4 unique features. First, we have a well-characterized sample of children with laboratory-confirmed congenital ZIKV infection and clinical manifestations. Second, we are conducting a comprehensive battery of assessments over a 5-year period, documenting overall function, identifying secondary complications, and characterizing emerging profiles of strengths and weaknesses. Third, we are assessing the evolving consequences on families. 15 Finally, by selecting a sample of children who are receiving a range of services we will be able to identify and measure existing strategies that support child development and family adaptation.

Methods
Participants were 121 toddlers with CZS receiving clinical services at a rehabilitation center in Recife, Brazil. Primary caregivers provided written informed consent while present at the rehabilitation center after being read aloud the consent forms. This study was approved by the institutional review board at the Altino Ventura Foundation (FAV) in Pernambuco, Brazil. Study data were collected and managed using REDCap electronic data capture tools. 16 Reporting conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
The sample was limited to children with serologic confirmation of ZIKV, obtained by the detection of immunoglobulin M antibody capture enzyme-linked immunosorbent assay or real-time reverse transcription-polymerase chain reaction testing from cerebrospinal fluid (collected at age 3 months or younger), who met clinical criteria accompanied by parental report of suspected exposure to ZIKV during pregnancy. 17 Symptoms reported by mothers included rash, fever, joint pain, conjunctivitis, headaches, and ocular pain during pregnancy. Before the 2016 outbreak of ZIKV, assays to detect ZIKV were only available for research or from the US Centers for Disease Control and Prevention. [18][19][20] Inclusion in this study was limited to children who presented with clinical abnormalities associated with CZS (ie, microcephaly, central nervous system damage, subcortical calcifications, ocular damage, congenital contractures, hypertonia extrapyramidal involvement) 4 ; had serology confirmation ruling out toxoplasmosis, rubella, cytomegalovirus, herpes, syphilis, or HIV 4 ; and had neuroimaging findings consistent with CZS brain abnormalities. Microcephaly and severe microcephaly were defined based on International Fetal and Newborn Growth Consortium for the 21st Century fetal growth standards. 21 Children were seen at the clinic for a developmental assessment using the Brazilian Portuguese version of the BSID-III, 22 a widely used norm-referenced measure of development that spans 3 domains (ie, cognitive, language, and motor), with 5 subdomains (ie, cognitive, receptive language, expressive language, fine motor, and gross motor). Three scores are provided for each domain, as follows: raw scores, age-equivalent scores, and scaled scores (mean [SD], 10 [3]; range, [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. Standard scores, with a mean of 100 and an SD of 15, can be computed for the cognitive scale, and composite scores for language (combining the receptive and expressive scales) and motor (combining the fine and gross motor scales) scales.
We used age-equivalent scores to explore profiles of strengths and weaknesses across the 5 subdomains. Age-equivalent scores allow comparison of performance on domains (in contrast to raw scores, for which ranges vary). However, because of the lack of variability in norm-referenced scores, we used domain raw scores for correlation and regression analyses. Raw scores reflect the number of items a child can complete in contrast to scaled scores, which provide a comparison to the normed population. While raw scores are generally not used in analyses because of varying expectations based on age, the children in our sample were within a relatively small age span (mean [SD] age, 31.2 [1.9] months) and thus raw scores provided a more robust method for exploring variability within the sample.
Children were assessed by trained research assistants with experience working with children with CZS and similar disabilities. They received extensive training on BSID-III administration and acceptable accommodations for children with severe disabilities. All administrations were videotaped and coded for fidelity by an independent psychologist (K.O.) with expertise in the BSID-III. Errors in scoring or administration were corrected with feedback to the research assistants for future administrations. No errors were reported that compromised the validity of the data. The BSID-III was administered using standard procedures, with accommodations for the child's visual or motor impairments, as suggested by the manual. 22 For example, for children with light vision only, overhead lights were turned off and a flashlight was used to provide contrast. Parents or guardians were present during the BSID-III administration. Demographic information and medical histories were obtained through parent or caregiver interview and review of medical records.

Statistical Analysis
Descriptive analyses included frequencies, means, SDs, and ranges, calculated for general characteristics including age, sex, gestational age, type of delivery, and weight and head circumference at birth. Descriptive analyses of the total raw scores and age-equivalent scores were also included across the 5 developmental domains of the BSID-III.
Comparison of means tests (t tests) were conducted to assess differences in sex among the 5 domains. Correlation analyses were performed to examine the association of independent variables with raw scores. Pearson correlation coefficients between developmental scores and demographic variables (ie, age, maternal age, household income) and clinical variables (ie, birth weight, gestational age, head circumference at birth) were included in the analysis. A stepwise multivariate linear regression was calculated to identify potential factors associated with development, including sex, age, head circumference at birth, gestational age, birth weight, and household income. All analyses were performed using the SAS version 9 (SAS Institute). Statistical significance was set at P < .05, and all tests were 2-tailed.

Results
The 121 children had a mean age at assessment of 31.2 (1.9) months, with 58 (49.6%) boys and 61 (50.4%) girls. A total of 99 children (81.8%) had microcephaly at birth, 74 (61.2%) of whom were classified as severe (>3 SDs below mean head circumference) ( Table 1). Means, SDs, and ranges for standard scores, raw scores, and age-equivalent scores are provided in Table 2. Nearly all (106 [87.6%] to 118 [97.5%]) participants scored at the floor of at least 1 scale. However, there was variability in raw scores and age-equivalent scores. For example, a standard score of 55 for the cognitive domain was assigned to children with raw scores ranging from 1 to 42. Table 3 and Table 4 provide results for correlation and regression analyses testing factors' associations with outcomes.
Comparison tests revealed no differences by sex of child on any BSID-III scale or control variable.     suspension, and while being carried. Head circumference at birth was positively correlated with gross motor skills (r = 0.236; P < .001). The multiple linear regression indicated that head circumference at birth was a significant factor in gross motor scores (β = 1.12; SE, 0.64; P = .08); however, the overall model did not fit.

Language Composite
There was more variability in overall language scores than in cognitive or motor scores. The mean (SD) standard score was 47.7 (2.89), with a range of 47 to 71. Overall, 106 (87.6%) scored at the floor of 47 on the language domain. Scores between 0 and 18 resulted in a standard score at the floor.

Receptive Language Scale
The mean (SD) receptive language age-equivalent score was 3.

Discussion
To our knowledge, this study is the largest report to date of children with obvious birth defects as a result of ZIKV infection. Children with the most severe cases of CZS were expected to have substantial delays and impairments, 11,23 but the magnitude of delay and relative strengths and weaknesses has not been previously documented.
We found profound delays in all developmental domains. At age 2.5 years, most children were functioning more like a child aged 2 to 3 months. Most scored at the floor of the standardized scales, although we found a wide range of raw scores associated with the floor. 24,25 Because we are unable to meaningfully interpret analyses based on standard scores owing to the lack of variability, we used age-equivalent scores and raw scores to examine within-and between-participant variance and describe developmental profiles.
We observed a relative strength in receptive language. Given the significant motor and visual impairments, these children's primary means for interacting with their environment appears to be through auditory input. Most could take in and respond to the environment, react to sounds and voices, differentiate between familiar and unfamiliar voices, and respond to their names. They also interacted vocally with their environment, making sounds to convey moods and engage in social hypothesis, it is possible that older mothers talk more often and provide more verbal stimulation to their children, resulting in higher receptive language scores. This is an important area for future research.
In contrast to language outcomes, we found more substantial motor impairments. Previous studies have suggested that many children with CZS meet criteria for cerebral palsy, 26 and arthrogryposis, or congenital joint contractures, is a well-documented comorbidity in CZS. 27 Intentional use of hands was a significant challenge for most children. 26,27 This finding is consistent with earlier studies of children with ZIKV-associated microcephaly. More refined assessments are needed to understand the full extent of motor impairments. For example, previous studies have documented abnormal persistence of primitive reflexes 12 and motor functioning in the severe level on the Gross Motor Function Classification System. 28 We will assess these motor variables longitudinally to measure progression or regression over time.
Because most children did not have functional use of hands or the ability to control movement in their limbs, they were unable to demonstrate higher-order cognitive skills, such as object permanence or basic problem-solving. Novel techniques, such as heart-defined attention 29 or habituation procedures, 30 may be necessary to assess the extent to which children with severe CZS are capable of learning from their environment. 29 The extent of microcephaly, as defined by head circumference at birth, was the only significant factor associated with cognitive and motor outcomes. More than 60% of children had head circumferences more than 3 SDs below the mean at birth. Nearly all also had significant brain calcifications. Even with no microcephaly or other birth defects, higher-than-expected rates of brain and ocular anomalies have been reported among those with laboratory evidence of ZIKV infections. 31,32 These findings suggest that neurodevelopmental differences are likely to manifest as children get older, even in those without obvious challenges in infancy. Indeed, Neilsen-Saines et al 14 reported that approximately one-third of children showed significant delays on the BSID-III, which included only 6 participants with unresolved microcephaly. Interestingly, in this cohort with less impairment, language function was more delayed than cognitive or motor function, the opposite of our findings. This may reflect a compensation of auditory pathways in children whose vision and motor skills are significantly challenged.

Limitations and Strengths
This study highlights the weaknesses and strengths of using the BSID-III. Standard scoring procedures are not likely to provide useful information beyond demonstrating the extent of developmental impairment in children with CZS. Furthermore, although items span the full developmental spectrum from birth through toddlerhood, the reliance on visual and motor output may penalize children with CZS in demonstrating what they can do. Conversely, the BSID-III raw scores and age-equivalent scores provide a sensitive measure of potential change over time, allowing for monitoring of gain or loss of skills in response to time, treatment, or seizures. Additionally, adapting the BSID-III to accommodate visual and motor difficulties made it possible to more accurately assess the developmental function of children with CZS. However, there are few appropriate, validated assessment tools that are appropriate for children with such a range of functional skills.
Two other limitations should be noted. First, standards for confirming ZIKV have improved over the last 2 years, and so we do not have criterion-standard methods for ZIKV testing on our participants. Furthermore, although this is a large cohort of children with CZS, they were all ascertained from 1 center in Brazil; therefore, we are limited in our ability to generalize findings to other cohorts of affected children who may have different parent demographic characteristics or access to care.