Evaluation of the Relationship Between Environmental Factors, Nutrition, and Metabolic Changes in Children Diagnosed With Autism in North Cyprus: A Case-Control Study

Introduction Autism spectrum disorder (ASD) is a set of neurodevelopmental disorders characterized by deficits in social behaviors and nonverbal interactions. The disorder is believed to be multifactorial regarding etiopathology. This study aimed to investigate the possible risk factors associated with the development of autism in the prenatal and postnatal periods. Methods We conducted an unmatched case-control study composed of 56 autistic cases and 85 control children in North Cyprus. Cases recruited were previously diagnosed by a pediatric psychiatrist as being on the autistic spectrum. Parental questionnaires were distributed, and the collected data were analysed using Statistical Package for the Social Sciences (SPSS) version 24 (IBM Corp, Armonk, NY). Binary logistic regression was used to compute the adjusted odds ratios (aOR), adjusted for possible confounders. Results Our results showed increased odds of developing ASD in mothers with mental disorders such as depression and anxiety (aOR 6.99; 95% CI 1.94 - 25.24), mothers with medical conditions such as Hashimoto’s thyroiditis (aOR 2.68; 95% CI 1.06 - 6.78), mothers using aluminum-containing anti-acids (aOR 2.34; 95% CI 1.012 - 5.39), mothers exposed to loud noises during pregnancy (aOR 2.66; 95% CI 1.005 - 7.034), mothers with ≥ two previous miscarriages (aOR 4.19; 95% CI 1.17 -14.97), neonates with birth weight <2500 grams (aOR 4.19; 95% CI 1.16 - 14.84), male gender neonates (aOR 3.26; 95% CI 1.31 - 8.90), and neonates exposed to MRI or CT scan during the first year of life (aOR 6.94; 95% CI 1.15 - 42.07). Decreased odds of ASD development were observed in mothers using multivitamins during pregnancy (aOR 0.35; 95% CI 0.13 - 0.97), mothers consuming slight amounts of baking powder during pregnancy (aOR 0.235; 95% CI 0.09 - 0.60), mothers with threatened abortion (aOR 0.35; 95% CI 0.12 - 0.98), and neonates taking iron supplementation during the first six months of life (aOR 0.38; 95% 0.16 - 0.91). Conclusion There were various maternal and neonatal factors associated with ASD development in North Cyprus. Although there is some evidence to suggest that exposure to specific factors during prenatal or postnatal periods may increase the risk of ASD, there is insufficient evidence that implicates a specific factor for autism etiology. Future studies are recommended to be performed on larger scales to support further the factors associated with ASD development.


Introduction
The world health organization estimates that the prevalence of autism spectrum disorder (ASD) is one per 160 children worldwide [1], and the cost of autism reached 268 billion dollars in the United States of America in 2015 [2]. ASD encompasses various conditions manifested by social and language deficits, difficulties in nonverbal communication skills, and stereotyped repetitive behaviors [3]. Signs of ASD usually appear by age two or three; however, developmental delays may occur earlier, allowing the diagnosis to be made as early as 18 months of age [3]. Multiple factors have been implicated in the development of ASD, including environmental, biological, and genetic factors [4]. 1 1 1 1 1

Study sample
We conducted a case-control study in the Turkish Republic of Northern Cyprus in 2017 among children under the age of 18. According to the Statistical Yearbook of Northern Cyprus published in February 2016, the total number of children in North Cyprus under 18 years was 73,500, with an average of 4000 new births per year [9]. Approximately 300 children were enrolled in special education centers in North Cyprus, and around 100 of them were diagnosed as having an autism spectrum disorder. Based on the given numbers, we estimated the required sample size of cases as 56 cases using the "openepi" website for unmatched casecontrol studies [10]. The term "cases" refers to children who were diagnosed as being autistic by a pediatric psychiatrist in North Cyprus, based on the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) criteria [11]. Most autistic children were not getting any medical treatment, while only 5-10% were under psychotic or epileptic treatment. The term "controls" refers to children without a diagnosis of autism who were enrolled in public schools.

Procedure
We obtained ethical approval from the Ethics Committee of Eastern Mediterranean University. In addition, informed consent was obtained from the parents of autistic children by visiting eight special education centers in North Cyprus and calling the parents of autistic children registered in those centers. We then distributed hard-copy questionnaires to be filled by the mothers of autistic children and collected the questionnaires one month later. We also obtained informed consent from the mothers of controlled children by reaching out to two public schools in North Cyprus and distributing the questionnaires to the heads of those public schools. The questionnaire was distributed in the Turkish language and was developed based on a thorough literature review of similar studies. The questionnaire asked yes-no questions about the possible exposure to each variable during the pregnancy period.

Variables
The demographic variables in our study include gender, child's age, mother's and father's ages at the birth of their child, child's order, birth weight, birth month, mother's country of birth, living in an old house, mother having medical conditions, and previous miscarriages. The other part of the survey focused on the possible risk factors or protective factors associated with the development of autism during the prenatal, perinatal, and postnatal periods. The odds ratio was adjusted for potential confounders, including mother's age, father's age, child's gender, birth weight, previous miscarriages, mother's medical conditions, and pregnancy-induced medical conditions. These confounders were chosen based on data from previous literature studies.

Statistical analysis
Data analysis was done through the Statistical Package for the Social Sciences (SPSS) version 24 software (IBM Corp., Armonk, NY) and statistical significance was set at a two-sided P-value of <0.05. Descriptive statistics were computed on all variables as the primary method of data evaluation. We compared the distribution of demographics and general characteristics between autistic and non-autistic children by performing Pearson's chi-square test, as well as an independent sample t-test for quantitative variables including child's age, mother's age, father's age, and order of the child. We also used the binary logistic regression analysis with a 95% Confidence Interval (CI) to calculate the unadjusted odds ratios (OR) and the adjusted odds ratios (aOR) for potential confounders. The statistical analysis on SPSS was done using the number of responses received for each variable, and the missing responses were not included in our statistical calculations.

Demographical characteristics
The total sample size included in our study comprised 141 subjects, 56 cases and 85 controls. The potentially eligible number of autistic cases in North Cyprus was estimated to be 100; out of these cases, we managed to collect responses from the mothers of 56 autistic children, which was the number of cases suggested by "openepi website". We also distributed 100 questionnaires for the control group, and we managed to collect responses from the mothers of 85 controlled children. Reasons for non-participation in this study included parental preferences, as some parents did not want to feel guilty for potentially causing the disease to their children. The means of ages in cases and controls were 9.6±5.3 and 6.3±2.8, respectively (P<0.0001). In the cases recruited, 78.6% were males, and 21.4% were females, while in the controls recruited, 56.5% were males, and 43.5% were females (P=0.007). These findings reveal the unmatching of ages and genders between the cases and controls in our study. The mean of mothers' ages at birth was also significantly different between cases (27.8±4.9) and controls (30.9±4.4) with P<0.0001. However, the mean of fathers' ages showed no significant difference between both groups (P=0.213). The mean order of a child was slightly higher in the cases (1.5±0.7) compared to controls (1.3±0.5) with P<0.049. The birth weight and birth month showed no differences between both groups (P >0.05). Table 1 shows the demographics and general characteristics of participants.  Continuous variables that were categorized in our study included birth weight, which was categorized into <2500g and >2500g; mother's and father's ages at delivery, which were categorized into <30 years and >30 years of age; and birth month, which was categorized into winter months (December, January, February) and non-winter months. Potential confounders were chosen based on data from previous literature studies.
Adjusted confounders in our study included mother's age, father's age, child's gender, birth weight, previous miscarriages, mother's medical conditions, and pregnancy-induced medical conditions. Several maternal factors have shown statistically significant results concerning the development of autism. Among these maternal factors, increased risk of autism was seen in mothers with mental disorders during pregnancy, such as anxiety and depression (aOR 6.99; 95% CI 1.94 -25.24), mothers with medical conditions during pregnancy, such as Hashimoto's thyroiditis and anemia (aOR 2.68; 95% CI 1.06 -6.78), mothers using aluminum-containing antacids during pregnancy (aOR 2.34; 95% CI 1.012 -5.39), and mothers exposed to prolonged periods of noise pollution, i.e., loud noises during pregnancy (aOR 2.66; 95% CI 1.005 -7.034). Decreased risk of autism was seen in mothers taking multivitamins during pregnancy (aOR 0.35; 95% CI 0.13 -0.97), mothers using baking powder twice a month while cooking (aOR 0.235; 95% CI 0.09 -0.60), and mothers with threatened abortion (aOR 0.35; 95% CI 0.12 -0.98). The following maternal factors showed no statistical significance in autism development: maternal smoking during pregnancy, maternal second-hand smoke exposure, pregnancy-induced conditions, such as gestational diabetes and hypertension, maternal vitamin D deficiency, hyperemesis, insecticide exposure, hair dying, consuming seafood and caffeine, heavy metal exposure, such as mercury, maternal vaccination, microwave usage, owning a pet, and medication use during pregnancy (P ≥ 0.05).

Neonatal factors
Several neonatal factors during and after delivery have shown statistically significant results related to the development of autism. Among these neonatal factors, an increased risk of autism was seen in neonates whose birth weight was less than 2500g (aOR 4.19; 95% CI 1. 16 -14.

Discussion
Our case-control study was able to establish various neonatal and maternal risk factors for the development of ASD. When considering maternal factors, compared to their control counterparts, significant risk factors for autism were observed in mothers with mental disorders, mothers with medical conditions, mothers using aluminum-containing antacids, and mothers exposed to loud noises while pregnant. On the other hand, protective factors were multivitamins usage during pregnancy, small amounts of baking powder usage during pregnancy, and mothers with threatened abortion. Neonatal factors implicated in the development of ASD were male gender, birth weight below 2500g, and exposure to MRI or CT in the first year of life.
Several meta-analyses and review articles were found in the literature combining multiple case-control studies related to ASD development. Studies that showed similar findings to our results include a metaanalysis done by Chen et al. (2016), which revealed that maternal autoimmune disease is a significant risk factor for ASD development with a pooled OR of 1.34 (95% CI 1.23 -1.46) [12]. Maternal depression was also found to be a substantial risk factor for ASD in a meta-analysis performed by Kobayashi et al. (2016), especially in mothers using selective serotonin reuptake inhibitors (SSRIs) throughout their pregnancy with a pooled OR of 1.45 (95% CI 1.15 -1.82) [13]. Moreover, Gardener et al. (2011) showed an increased risk of autism in neonates who have low birth weights (<2500g) with a pooled OR at 1.63 (95% CI 1.19 -2.33), showing similarity to our findings [6]. Another meta-analysis was done by Wang et al. (2017) showed that male gender is associated with increased risk of autism with a relative risk (RR) of 1.47 (95% CI: 1.39 -1.55), a clear correlation that was also demonstrated in our study [7]. Wang et al. (2017) revealed that cigarette smoking was a non-significant factor [7]. However, Zhang et al. (2010) argued that several chemicals involved in second-hand smoking, such as polycyclic aromatic hydrocarbons and metals, have adverse health effects in fetal hypoxia and brain development [14]. These effects might hypothetically affect the social development of autistic children; nonetheless, our study did not show a relation to smoke exposure. Another systematic review was done by Cheng et al. (2019) showed that low birth weights in neonates together with autoimmune diseases and maternal depression increase the risk of developing ASD, further supporting our results [8]. Additionally, in the review of Elsabbagh (2019), there was some evidence of protective effects of multivitamins supplementation, including the use of folic acid and vitamin D, against the development of ASD [15]. Other studies also showed that iron, vitamin D, and fatty acids played a protective role against ASD development [16]. A meta-analysis by Sulaiman et al.
(2020) demonstrated significant relation between all the metals and autism. Aluminum levels in blood were negatively associated with ASD, while aluminum levels in hair and urine were positively associated [17], focusing the lights on the effects of aluminum-containing anti-acids and their relation to autism found in our study. The majority of the studies mentioned above showed similarities with our study findings.
The meta-analysis done by Wu et al. (2016) focused solely on finding a connection between ASD and parental age. The findings differed from our study, as they showed a significant risk of developing ASD with increased parental ages [18]. A 10-year increase in paternal and maternal ages was associated with 21%, and 18% increased risk of autism in the offspring, with OR 1.21 (95% CI 1.18-1.24) and OR 1.18 (95% CI 1.10 -1.26), respectively [18]. In the meta-analysis of Curran et al. (2015), cesarean section was associated with increased risk of ASD with pooled aOR at 1.23 (95% CI 1.07 -1.40) [19], compared to the insignificant results in our study. According to Xu et al. (2014), mothers with diabetes mellitus (DM) were shown to have an increased risk of having an autistic child compared to mothers without DM with a pooled RR of 1.48 (95% CI 1.25 to 1.75) [20]. Wang et al. (2017) also showed that gestational DM and gestational hypertension, in addition to cesarean delivery, were associated with increased ASD risk in the offspring, showing different findings from the results obtained in our study [7]. Additionally, Wang et al. demonstrated that threatened abortion actually increased the risk of autism with RR 1.54 (95% CI 1.28 -1.87) [7], in contrast to our finding that showed it to be a protective factor. This equivocal finding in our study might be because women with threatened abortions in North Cyprus tend to be more cautious throughout their pregnancy and their delivery method. Some perinatal factors were also proven to be significantly related to ASD development, including hypoxic-ischemic damage, fetal distress during delivery, low Apgar (appearance, pulse, grimace, activity, and respiration) score, preterm birth, hyperbilirubinemia, small for gestational age, and congenital disabilities [6,15,21]. These findings were not demonstrated to be significant in our study.

Limitations
The differences in the results found in our study compared to literature might be attributed to the smaller sample size used, as opposed to the wider sample sizes used in the meta-analyses. Additionally, most of the meta-analyses included multiple pieces of research from different geographical areas, allowing the inclusion of various factors related to the studied population. A major limitation of our study is its retrospective design, predisposing it to recall bias. Prenatal, obstetric, and postnatal characteristics were evaluated based on surveys answered by mothers depending on their ability to remember past events. Furthermore, matching was not accomplished in our study due to the small sample size of the Northern Cyprus population, making it difficult to find many autistic children of similar ages. The ages of autistic children included in our study were increased to include children up to 18 years of age, aiming to reach larger numbers of autistic children.
One difficulty that may also arise when exploring this topic is that many of these risk factors are connected, causing a possible confounder effect; mothers with medical conditions usually give birth to underweight neonates, pinpointing which risk factor is the most significant remains a challenge.

Conclusions
Autism is one of the leading mental causes of disability in children worldwide. Looking at potential risk factors might give us insight into the possible preventative measures. Our study among children in Northern Cyprus showed evidence suggesting several maternal and neonatal factors associated with autism. The results showed increased odds of developing ASD in mothers with mental disorders such as depression and anxiety, mothers with medical conditions such as Hashimoto's thyroiditis, mothers using aluminumcontaining anti-acids, mothers exposed to loud noises during pregnancy, mothers with ≥ two previous miscarriages, neonates with birth weight <2500 grams, male gender neonates, and neonates exposed to MRI or CT during their first year of life. The factors that demonstrated decreased odds of ASD development were observed in mothers using multivitamins during pregnancy, mothers consuming slight amounts of baking powder during pregnancy, mothers with threatened abortion, and neonates taking iron supplementation during the first six months of life. Although there is some evidence to suggest that exposure to specific factors during the prenatal or postnatal periods may increase or decrease ASD risk, there is insufficient evidence that implicates a specific aetiological factor for autism. It is recommended that future studies be performed on a larger scale to support further the factors associated with ASD development. reported "Despite being a common condition with a lifelong course, ASD showed no apparent connection with a single specific risk factor [2]. Instead, a combination of both genetic and environmental factors is suggested [5]." "Identifying the possible risk factors linked to ASD is essential in preventing the condition. In addition to the substantial economic burden of ASD, the quality of life for affected individuals and their families is decreased substantially. Therefore, targeting disease prevention is a crucial element in improving the quality of life in recent years." Describe any efforts to address potential sources of bias "The odds ratio was adjusted for potential confounders, including mother's age, father's age, child's gender, birth weight, previous miscarriages, mother's medical conditions, and pregnancy-induced medical conditions. These confounders were chosen based on data from previous literature studies."

Study size 10
Explain how the study size was version 24 software, and statistical significance was set at a two-sided P-value of <0.05.
Descriptive statistics were computed on all variables as the primary method of data evaluation.
We compared the distribution of demographics and general characteristics between autistic and non-autistic children by performing Pearson's chi-square test, as well as independent sample ttest for quantitative variables including child's age, mother's age, father's age, and order of child. We also used the binary logistic regression analysis with a 95% Confidence Interval (CI) to calculate the unadjusted odds ratio (OR) and the adjusted odds ratios (aOR) for potential confounders." The potentially eligible number of autistic cases in North Cyprus was estimated to be 100 cases; out of these cases, we managed to collect responses from the mothers of 56 autistic children, which was the number of cases suggested by "openepi website". We also distributed 100 questionnaires for the control group, and we managed to collect responses from the mothers of      [7], in contrast to our finding that showed it to be a protective factor. This equivocal finding in our study might be due to the fact that women with threatened abortion in North Cyprus tend to be more cautious throughout their pregnancy and their method of delivery. Some perinatal factors were also proven to be significantly related to ASD development, including hypoxic-ischemic damage, fetal distress during delivery, low Apgar score, preterm birth, hyperbilirubinemia, small for gestational age, and birth defects [6,15,21]. These findings were not demonstrated to be significant in our study.
Discuss the generalisability "Several meta-analyses and review articles were found in the literature combining multiple casecontrol studies related to ASD development. Studies that showed similar findings to our results include a meta-analysis done by Chen et al. (2016), which revealed that maternal autoimmune disease is a significant risk factor for ASD development with a pooled OR of 1.34 (95% CI 1.23 -1.46) [12]. Maternal depression was also found to be a substantial risk factor for ASD in a metaanalysis performed by Kobayashi  of the study results our findings [6]. Another meta-analysis was done by Wang et al. (2017) showed that male gender is associated with increased risk of autism with a relative risk (RR) of 1.47 (95% CI: 1.39 -1.55), a clear correlation that was also demonstrated in our study [7]. Wang et al. (2017) revealed that cigarette smoking was a non-significant factor [7]." "Another systematic review was done by Cheng et al.
(2019) showed that low birth weights in neonates together with autoimmune diseases and maternal depression increase the risk of developing ASD, further supporting our results [8]. Additionally, in the review of Elsabbagh (2019), there was some evidence of a protective effect of multivitamins supplementation, including the use of folic acid and vitamin D, against the development of ASD [15].
Other studies also showed that iron, vitamin D, and fatty acids played a protective role against ASD development [16]." Other information  In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.