Perinatal Risk Factors in Children Diagnosed with Autism Spectrum Disorder in Dubai: A Case-Control Study

Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental disorder characterized by impaired social communication and restricted and repetitive patterns of behaviors [1]. The prevalence of this disorder is increasing globally, reaching a 1 in 54 children estimate in 2016 [2‒4]. ASD is recognized as a clinically heterogeneous disorder as its phenotypic presentation varies in typology and severity [5]. Similarly, the etiology of ASD is also recognized to be heterogeneous, with genetic factors playing a significant role with heritability estimates of up to 90% [6]. Further, twin studies consistently report higher concordance rates among monozygotic twins (60–90%) compared to dizygotic twins (up to 10%) [7]. Genome-wide association studies have identified a number of common genetic variants associated with ASD [8‒10], each, however, with a small effect size [11]. Candidate genes for ASD have also been identified through whole exome sequencing studies. Moreover, there is a strong association of de novo copy number variations in ASD compared to unaffected individuals [12]. ASD is therefore considered to involve multifactorial processes that each confer a small risk, but its exact pathogenesis is yet to be fully understood [13]. An interaction of environmental factors and various genetic predispositions is suggested [14].

The incomplete concordance rate in monozygotic twins and heritability studies point to the role of nongenetic factors in the etiology of ASD. Although pesticides and air pollution, for instance, have been identified as risk factors [15], this has not been replicated. Cumulative epidemiological evidence suggests that neonatal and perinatal factors such as abnormal fetal presentation, umbilical cord complications, fetal distress, birth injury or trauma, multiple births, maternal hemorrhage, summer birth, low birthweight, small for gestation, neonatal anemia, ABO or Rh incompatibility, neonatal hyperbilirubinemia, low APGAR scores, feeding difficulties, and meconium aspiration slightly increase the risk of developing ASD [7, 14]. Similarly, other studies have reported that maternal smoking, medication intake during pregnancy, neonatal dyspnea, and congenital anomalies, as well as maternal fever and infection during pregnancy and birth, all have a positive association with an increased risk of developing ASD [16, 17]. A meta-analysis by Wang et al. [18] found that age ≥35 years; being of Caucasian or Asian race; having a college degree; gestational diabetes and hypertension; antepartum and postpartum hemorrhage; caesarian delivery; gestational age ≤36 weeks; spontaneous, induced, or birth without labor; breech presentation; preeclampsia; fetal distress; low birthweight; male gender; and brain anomaly also led to increased risk of ASD, whereas maternal parity of ≥4 and female gender were associated with a decrease in the overall risk of developing autism.

The prevalence of ASD in the Arab Gulf countries and more specifically the United Arab Emirates (UAE) is significantly lower than worldwide, with the UAE reporting 29 per 10,000 children in comparison to worldwide (77 in 10,000) [19]. The lower numbers have been attributed to underdiagnosis and underreporting due to a lack of awareness and specialized healthcare services and cultural attitudes rather than the condition being less prevalent [19, 20]. Although there has been increasing interest in ASD research in these regions, the literature still remains limited, and local studies on perinatal risk factors are not prevalent [21]. According to Arafa et al. [22], risk factors for ASD vary across races, and the majority of the studies are based on Western populations. Studies by Salhia et al. [21], Arafa et al. [22], and Oommen et al. [23] provide some insights into ASD prevalence and risk factors in countries on the Arab peninsula, Egypt, and Saudi Arabia, respectively. Their findings suggest that the development of ASD in these countries may also be associated with consanguineous marriages, advanced maternal and paternal ages, medications taken by the mother, delivery via cesarean section, multiple pregnancy, and antenatal complications, among others [21‒24]. These studies have methodological issues, with some focusing on data collection through recall questionnaires, thus underscoring the need for more substantive research on both the prevalence and perinatal risk factors for ASD in the UAE. The purpose of this study was to characterize the sociodemographic and perinatal factors of a sample of children with ASD from the UAE, which are compared to those of age- and gender-controlled children.

The study sample consisted of children between the ages of 3–18 years who were assessed for ASD at Latifa Women and Children’s Hospital in Dubai, UAE. Patients were selected from September 2014 to September 2016 and diagnosed using the DSM-5 (Diagnostic Statistical Manual of Developmental Disorders 5th edition) criteria by a licensed physician with expertise in ASD. For the control group, age- and gender-matched children without a diagnosis of ASD or any subtype of developmental delay were randomly selected from a birth registry at the same hospital. An initial sample size was calculated at 61 subjects per group using an expected proportion of ASD cases to control subjects of 0.2 (20%), assumed odds ratio of 3, a confidence level of 0.95, and power of 0.8 (80%) [25].

The selection and collection of data on ASD cases were performed retrospectively for children diagnosed with ASD through the outpatient ASD diagnostic services at Latifa Women and Children’s Hospital in Dubai, UAE. The cases included a cohort of children (n = 87) diagnosed with ASD within a 2-year period (September 2014 to September 2016) and who were born between 1999 and 2014. Perinatal risk factors were ascertained using retrospective reviews of clinical charts in the Dubai hospital records section. Control subjects (n = 180) born between 1999 and 2014 were selected from the Dubai Health Authority (DHA) birth registry using a random method, obtaining roughly two controls for every ASD case after careful vetting for the presence of developmental delay or a diagnosis of ASD. After excluding subjects with insufficient perinatal data, a total of 134 controls were included in the study.

Electronic and paper medical records of DHA hospitals were reviewed to identify perinatal risk factors. This information was gathered mainly from neonatal ICU case records and other sources, including antenatal consults and delivery records. These data were entered on an anonymized and confidentially stored Excel database.

Ethical approval for the study was obtained from the Dubai Scientific Research Ethics Committee at the DHA.

The term “perinatal” pertains to the time around birth, specifically from 22 completed weeks of gestation up to seven completed days after birth. Furthermore, “ante-” and “postnatal” refer to the time before and after birth, respectively, whereas “natal” refers to the time during birth.

With the help of online journal databases, including PubMed, a comprehensive search was made to ascertain key perinatal risk factors implicated in ASD that are relevant to our study. Antenatal risk factors that were reviewed included the following: maternal age (≤35 years; >35 years), parity (primi-or multipara), degree of consanguinity, gestational length (preterm <33 weeks, preterm 33–37 weeks, term 37–42 weeks, or post-term >42 weeks), maternal fever, evidence of maternal infection (hepatitis B surface antigen, human immunodeficiency virus, syphilis screen [Venereal Disease Research Laboratory], Rubella immunoglobulin M, Group B Streptococcus status by high vaginal swab, chorioamnionitis), gestational diabetes mellitus (GDM), pregnancy-induced hypertension, premature rupture of membranes (<18 or >18 h), maternal blood group (ABO or Rh incompatibility), and antenatal ultrasound (normal or abnormal).

Natal risk factors reviewed included mode of delivery (normal vaginal delivery or lower segment cesarean section) with or without vacuum or forceps use, presentation (cephalic or abnormal), APGAR score (0–6; 7–10) at 5 min, birth weight (extremely low birth weight < 1,000 g, very low birth weight 1,000 g–1,500 g, low birth weight [LBW] 1,500 g–2,500 g, normal birth weight 2,500 g–4,000 g, or high birth weight [HBW] > 4,000 g), child’s blood group, ABO/Rh incompatibility, congenital anomalies, and neonatal ICU admission.

As descriptive statistics, mean (SD) or median (IQR) is presented for continuous variables as appropriate. The categorical variables were presented as number and percent according to case and control. The associations between the study variables and the cases and controls were studied using the χ² test and Fisher’s exact probability test. The variables that were significant at p ≤ 0.40 at the bivariate analyses were selected as potential variables for multivariable analyses. However, maternal fever is excluded due to the small numbers in the control arm. A multivariable analysis logistic regression was done. Odds ratio and 95% confidence intervals were reported. All tests were two tailed, and a p value <0.05 was considered as statistically significant. All statistical analysis was performed using IBM SPSS Statistics for Windows, version 21.0; IBM Corp., Armonk, NY, USA.

The total number of study participants included 87 ASD cases and 134 controls. Table 1 presents the distribution of baseline characteristics of the study subjects according to cases and controls. The average (SD) age of ASD cases was 7.2 (3.6) years, with a minimum age of 2 years and a maximum age of 18 years at the time of this study. The control subjects averaged an age of 7.0 (2.3) years with a minimum age of 2 years and a maximum age of 11 years at the time of this study. Gender distributions in both groups were similar, with an approximate ratio of four males to one female. The distribution of nationalities in the total study population included UAE nationals, who represented the vast majority (59.3%), followed by Indians (10%), Yemenis (3.2%), Egyptians (2.7%), Pakistanis (2.3%), Jordanians (2.3%), Filipinos (2.3%), and others (14.3%).

Table 2 presents the maternal characteristics of the cases and controls. The average maternal age at the time of delivery for both cases and controls was 29 years. Subjects were grouped into two age ranges, those less than or equal to 35 years (85.7% of cases, 83.1% of controls) and those greater than 35 years (14.3% of cases, 16.9% of controls). Parity is statistically insignificant (p = 0.500), with 71.3% of mothers of ASD children being multipara, compared to 66.9% of mothers in the control group.

The distribution of consanguinity was about equal between cases and controls (p = 0.813), with 24.1% in the cases and 22.5% in the controls. Preterm delivery, both as <33 weeks and 33–37 weeks were found to be statistically insignificant, with p = 0.196 and p = 0.719, respectively. The proportion of gestational diabetes was nearly similar (about 19%) in both groups (p = 0.993). Pregnancy-induced hypertension was also statistically insignificant (p = 0.370), with 8.8% in the cases and 5.5% in the control group. Premature rupture of membranes (3% cases; 8% control) and multiple gestations (8% cases; 6.7% control) were also statistically insignificant, with p = 0.168 and p = 0.710, respectively. Gender distribution and mode of delivery and presentation were also statistically insignificant, with p = 0.357 and p = 0.383, respectively. Maternal fever was statistically significant in the bivariate analysis (p = 0.018), with 0% in the control group and 5% in the cases.

Low APGAR scores (0–6) at 5 min were 0% in the cases, and 1.6% in the controls, and were not associated with a risk of autism (p = 0.641). LBW (≤2,500 g) children were 11.4% in the cases, while this was 20.4% in the controls. The children who were born with LBW, that is, from 2,001 g to less than 2,500 g, were less likely to have ASD as compared to control children (p = 0.034). In cases, 25.3% of them had neonatal jaundice, which is compared to 0% in the control group. However, the proportion of other risk factors, including congenital anomalies and neonatal intensive care unit (NICU) transfer, was about the same in cases and controls.

Table 3 presents the results of the multivariable analysis. After adjusting for other obstetric and neonatal risk factors, there were no statistically significant associations. There was a trend for an association between high birthweight and ASD with an odds ratio of 6.76 (95% CI: 0.76–59.93, p = 0.086).

This retrospective case-control study aimed to identify perinatal risk factors for ASD in Dubai, UAE, especially given the lack of similar data from the UAE. Of the examined variables, maternal fever was significant in the bivariate analysis; however, it was not included in the multivariate analysis due to the small numbers in the case group. Initial studies showed inconsistent findings with regard to the significance of maternal fever [17, 26‒28]. However, more recent population-based studies with large sample sizes have found maternal fever to be significantly associated with ASD in the 2nd trimester (OR: 1.40; 95% CI: 1.09–1.79) (OR = 2.19, 95% CI: 1.14–4.23) [29, 30]. A more recent meta-analysis by Antoun et al. [31], however, found a positive association in the first trimester (OR: 1.15, 95% CI: 1.01–1.31), with suggested links between the disruption of normal cytokine levels and an imbalance of T-cell populations, and ASD, and acetaminophen use in the 3rd trimester and ASD onset in offspring [32].

Although LBW and preterm birth have been consistently reported as risk factors for neurodevelopmental disability [33], our results were not in line with this. Studies examining the associations between ASD and birth weight found relatively consistent results, indicating an increased risk associated with LBW [34]. Schendel and Bhasin also found that the risk for autism was consistently higher for LBW girls than that of LBW boys [34]. Although not statistically significant, our study revealed a trend for an association between ASD and HBW (95% CI: 0.76–59.93, p = 0.086). These findings are in line with existing evidence that suggests a correlation between HBW and ASD [35, 36]. Although an overwhelming body of evidence suggests that children who were born preterm are more likely to have ASD, our findings did not identify this. A meta-analysis done in 2018 that included over 3,300 preterm infants from 18 studies notes that the overall prevalence rate for ASD in children born preterm was around 7% compared to the general population (0.76%) [37]. Kuzniewicz et al. [38] further add that each week of shorter gestation was associated with an increased risk of ASD. This may be related to study-specific factors such as the sample size but require further evaluation in future studies from our region.

Admitting diagnoses for our cohort of children ranged in complexity from term newborns with jaundice to extreme preterms with multiple problems. Consequently, NICU admission was not found to be statistically significant (OR = 0.95, 95% CI: 0.47–1.93, p = 0.879) between the two groups in this study, which is contrary to available evidence in the literature. In a matched case-control study by Karmel et al. [39], a substantial subset of their cohort of NICU graduates received a diagnosis of ASD. They also found that NICU graduates were around 2 weeks younger at birth. In addition, poor recovery of function 1 month post-discharge from NICU, accompanied by asymmetric visual tracking and abnormal upper extremity tone, suggests atypical behaviors that may be very early markers of ASD. For future research, it may be worthwhile to explore prospective clinical outcomes of NICU graduates at 1 month (as suggested by Karmel et al. [39]), or even at 1-year and 18-month intervals to assess neurobehavioural effects of NICU admissions, including ASD.

Multiparity has been positively correlated with ASD in the region [22]. Our study, however, showed no significant association even though there was a high incidence of multiparous women in the study. Although we did not stratify multiparity to analyze each multiparous state, studies have shown that successive births had a lowering of risk associated with ASD [40‒42]. On the other hand, ASD association with increasing parity has also been reported [7, 43]. Gardner et al. [7] in their meta-analysis have noted numerous studies reporting increased risk of ASD among first, fourth, and later born children.

Consanguinity is highly prevalent in the Middle East region and is considered a relevant risk factor for ASD [19, 21, 23, 44, 45]. In our sample, 24.1% of the cohort of children with ASD had a history of consanguinity, out of which 67% were born of consanguineous marriages between Emirati parents. Research from the region compared DNA of family members in cousin marriages with high incidences of autism and found missing chunks of DNA, affecting at least 6 genes that play a role in autism [19, 23]. Similarly, through whole genome sequencing and homozygous mapping, Tuncay et al. [45] identified potential etiological ASD variants in a consanguineous cohort. In this study, multivariate analysis on consanguinity and its risk contribution to LBW and maternal age were not performed; a focus on this, together with data on Emirati children born of consanguineous unions, could be an area for future research.

GDM is on the rise in the UAE, with historical data recording an increase in prevalence among Emirati nationals of 16.9% in the last 2 decades (8–24.9%) [36, 46]. Although our study did not show any significant association between maternal GDM and ASD, a large multiethnic clinical cohort by Xiang et al. [47] showed that maternal GDM diagnosed by 26 weeks of gestation was associated with an increased risk of ASD in offspring. Additionally, a meta-analysis of 18 ASD studies found an increased risk of ASD (OR: 1.42, 95% CI: 1.22, 1.65) [48], making this a potential topic for further research.

The retrospective design and a limited sample size pose a major limitation to our study. The source of our data to identify perinatal risk factors was restricted to reviewing medical files. To further validate our results, a larger sample size may be considered in addition to adding a prospective arm to the study. Furthermore, although the DSM-5 diagnostic criteria were used to diagnose ASD by autism specialists, other research diagnostic tools such as the Autism Diagnostic Interview-Revised (ADI-R) and Autism Diagnostic Observation Schedule-2 (ADOS-2) were not administered. Although awareness of ASD in the UAE and wider Middle East is on the rise, there exists a perceived stigma toward mental health, as do other barriers such as high cost of mental health among individuals from diverse nationalities residing in the region. These limitations may adversely affect the generalizability of studies such as ours.

This study identified maternal fever and HBW as two potential perinatal risk factors that are associated with ASD in a population sample from the UAE. The results are in agreement with the premise that ASD is a result of multifactorial processes and influences. There is a need for further prospective research on larger samples by exploring the role of the environment and, in particular, genetic associations of ASD in the UAE, given the unique cultural dynamics of the country and region.

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. Ethical approval for the study was obtained from the Dubai Scientific Research Ethics Committee at the Dubai Health Authority (DHA) under the file number DSREC/RRP/2016/06 on March 6, 2016. Patients visited DHA facilities and signed a general consent form, permitting use of their deidentified data for education and research purposes through the electronic medical record (EMR) system. Data collection was performed between the years 2016 and 2017, within the validity of the ethical approval.

The authors have no conflicts of interest to declare.

There are no funding sources to declare.

H.A.B. took the lead in drafting the content, with an extensive literature review, and took the lead in writing the manuscript. S.H. assisted in the production and critical revision of the report. A.A. and J.L. oversaw the creation of the report and provided critical feedback.

Data in this study were obtained from Dubai Health Authority where patient confidentiality restricts public sharing of this information. The collected data can be requested from the corresponding author, upon reasonable request.