Prenatal, Perinatal, and Postnatal Factors in a Cohort of Very Preterm and Very Low Birth Weight Toddlers with Suspected Autism Spectrum Disorder

A total of 133 neonates were included. From the screening conducted with the M-CHAT-R/F at 24 months of age, 80 (60.15%) were negative and 53 (39.85%) were positive. Of the 53 cases identified as positive, the evaluation with the ADOS-2 Toddler Module was completed in 50 (94.34%), with 35 administered to preverbal children or those with few words, and 15 to infants who regularly used some words. A total of 3 losses were recorded, all male children. A total of 24 subjects met the ADOS-2 most accurate cut-off point for ASD diagnosis, defined as a score of ≥ 12 for non-speaking children and ≥ 10 for children with some words, representing 18.46% of the entire sample.

In our study, the only prenatal factor significantly associated with suspected ASD was having a mother who was born abroad (Table 1).

As for perinatal factors, both low gestational age, categorized by days and weeks, and a weight of less than 750 g were associated with an increased risk of suspected ASD. Cesarean delivery and a complete course of corticosteroid maturation were associated with a lower probability of ASD. The presence of preeclampsia showed a trend towards significance (Table 2).

Finally, significant postnatal factors were bronchopulmonary dysplasia, hearing loss, longer NICU stay and low 10-minute Apgar scores. The presence of ROP, short length and cranial circumference at birth showed weak evidence (Table 3).

Additionally, some variables that were not found to be significantly associated with suspected ASD did show a significant relationship with a positive screening through the M-CHAT. These included paternal age of 35 years or older, maternal education up to secondary, multiple gestation and grade 2 or more of ROP.

This study was designed to investigate the association between specific factors and high probability of ASD diagnosis in a cohort of very preterm children. The results suggested, as hypothesized, that certain prenatal, perinatal, and postnatal factors might be associated with the likelihood of ASD. In accordance with previous findings, confirmed by this study, preterm children require thorough follow-up due to the high probability of neurodevelopmental deficits.

The only prenatal factor associated with suspected ASD identified in our study was maternal foreign birth, referring to mothers born outside of Spain. This is consistent with some prior findings (Becerra et al., 2014; Gardener et al., 2009), although other studies have reported contrasting results (Fairthorne et al., 2017). Regional characteristics may account for the discrepancies in outcomes. Susceptibility to relatively benign infections and the stress associated with residing in a new country have been proposed as potential explanations (Becerra et al., 2014; Gardener et al., 2009). Other factors commonly linked to ASD in the general population, such as assisted reproductive techniques and advanced paternal age, did not show any association in our study.

The results suggest, as hypothesized, that low gestational age was the most relevant perinatal variable associated with ASD in our study. The younger the gestational age, whether assessed in days or weeks, the higher the likelihood of suspected ASD. A recent meta-analysis found that premature infants are 3.3 times more likely to develop the disorder, although it did not find gestational age to be significantly related to autistic traits (Laverty et al., 2021). Nonetheless, several studies have consistently reported an association between lower gestational age and increased risk of ASD (Guo et al., 2022; Hwang et al., 2013; Johnson et al., 2010; Joseph et al., 2017a, b Kuzniewicz et al., 2014). The heterogeneity found may be due to differences in the categorization of the gestational age variable (Cogley et al., 2021). It has been suggested that the possible relationship is due to the vulnerability of the immature brain, the lack of neuroprotective factors or physiological instability (Joseph et al., 2017b). Another factor widely related to ASD is low birth weight (Gardener et al., 2011; Guo et al., 2022; Ma et al., 2022; Pyhälä et al., 2014; Wang et al., 2017). However, our results only indicated a weak relationship between these variables, except when the birth weight was extremely low (< 750 g). This suggests that in the premature population, gestational age may be a more determining factor than birth weight for the development of ASD.

Additionally, according to our results, cesarean delivery was associated with low probability of ASD diagnosis. This contrasts with findings from previous studies. Cesarean delivery has been associated with a higher likelihood of ASD in a comprehensive meta-analysis (Curran et al., 2015), although another study found no significant relationship in the premature population (Yip et al., 2016). Cesarean delivery is usually indicated when there are risks associated with prematurity. Our findings differ from other publications and likely require a larger study to evaluate the role of cesarean section, taking into account the indication for the cesarean section, the clinical status of the fetus and the mother, as well as the subsequent stabilization of the newborn. Another factor associated with a lower probability of ASD identified in our study was the complete administration of corticosteroids for fetal lung maturation. Corticosteroid administration is considered an effective means of reducing perinatal mortality and morbidity, as well as the risk of developmental delay (McGoldrick et al., 2020), although we are not aware of any previous studies linking it to ASD.

Preeclampsia scores trend towards significance as a protective factor for ASD, contrary to some studies suggesting it could be a risk factor for ASD (Jenabi et al., 2019; Maher et al., 2020; Wang et al., 2017). However, these studies involve a general population, without considering gestational age. One explanation could be that preeclampsia contributes to the development of ASD due to its association with prematurity, as the presence of the condition often leads to premature birth. In our study, the population consists exclusively of very premature infants, and if preeclampsia is the reason for prematurity, perhaps other variables more harmful to the fetus are not present.

During the neonatal period, our findings indicated an association between autism and a series of complications that commonly occur in premature infants. Specifically, bronchopulmonary dysplasia (BPD) and hearing loss were identified. These relationships remained significant even when controlling for gestational age. Previous studies have found a link between BPD and ASD (Hack et al., 2009) or associated characteristics (Brumbaugh et al., 2018; Sriram et al., 2018). Additionally, the relationship between hearing impairment and ASD has been extensively explored. Several studies have found a higher prevalence of hearing impairment in the ASD population (Demopoulos & Lewine, 2016; Ocak et al., 2018; Rydzewska et al., 2019; Ting et al., 2023), although a previous literature review did not find conclusive data (Beers et al., 2014). Structural and functional abnormalities have been found in the auditory brainstem in individuals with ASD, which could explain this relationship (Smith et al., 2019). Another associated factor identified in our study is the duration of stay in the NICU. This variable has previously been associated with ASD (Subedi et al., 2017; Winkler-Schwartz et al., 2014). Exposure to stressors associated with the NICU can alter the brain both functionally and structurally (Smith et al., 2011), which could explain the emergence of neurodevelopmental disorders such as ASD. Lastly, an Apgar score at 10 min lower than 7 has also been identified as an associated factor for ASD. This relationship remains significant when controlling for gestational age. Most previous studies agree that a low Apgar score at 5 min is associated with the presence of ASD (Gardener et al., 2011; Getahun et al., 2017; Modabbernia et al., 2019; Wang et al., 2017). It has been postulated that neonatal vitality and ASD share a common genetic background, or that the role of environmental factors such as birth complications in the development of ASD could be possible explanations for this relationship (Modabbernia et al., 2019). The variables that showed a trend towards significance were low birth length, head circumference, and ROP grade 2 or higher. Regarding ROP, it has also been associated with ASD, suggesting a common pathway between retinal issues and neurodevelopmental disorders.

Our results suggest that some factors associated with ASD in the preterm population could differ from those found in the general population. While gestational age and having a mother born abroad or hearing loss remain associated, cesarean delivery or complete maturation with corticosteroids could reduce the probability of ASD diagnosis. Certain aspects related to prematurity, such as ROP, bronchopulmonary dysplasia, low Apgar score or length of stay in the NICU, would be associated factors even when controlling for gestational age as a confounding variable.

Some variables in our study were associated with an increased likelihood of obtaining a positive screening on the M-CHAT, but not with a higher probability of subsequent ASD; specifically, paternal age equal to or greater than 35 years, mother’s education up to secondary, multiple gestation, preeclampsia, short birth length and grade 2 ROP. Several previous studies indicated that those registering false positives on the M-CHAT are at risk of cognitive, sensory, and behavioral problems (Luyster et al., 2011; Moore et al., 2012), so these factors are likely to pose a higher risk of other disorders or difficulties.

The rate of positive M-CHAT-R/F screens in our study is higher than that reported by other recent cohorts of premature infants (Shuster et al., 2023). This difference may be partially explained by variations in exclusion criteria or by cultural differences influencing parental reporting and interpretation of developmental behaviors within our population. Additionally, the higher rates of suspected ASD cases observed in our cohort compared to larger population-based studies (Crump et al., 2021; Joseph et al., 2017b) likely result from our early assessment at 24 months using the ADOS-2 Toddler Module, whereas those studies relied on formal clinical diagnoses at older ages.

Some limitations of the study need to be considered when interpreting the results. Firstly, the sample size, as although the sample size calculation indicated that the number of subjects was adequate, larger samples would enhance statistical power and representativeness. Additionally, a limited number of cases hinders discussion of causal relationships and, consequently, risk factors, but it does allow for establishing positive and negative associations between diverse variables and laying a foundation for future research. Moreover, standardized cognitive, language, or motor scores were not included in the analysis. Given the neurodevelopmental vulnerability of preterm children, these scores could provide valuable insights into their developmental trajectories and potential confounding factors. Finally, it should be noted that the detection of ASD is based on a cut-off point and, as indicated by the Spanish protocol (Pallás Alonso et al., 2018), follow-up should continue until adolescence to confirm the diagnosis.

Although previous prospective studies (Joseph et al., 2017b; Shuster et al., 2023), have explored risk factors for ASD in extremely preterm infants, research on this topic remains limited, particularly regarding early identification and prospective follow-up in clinical settings. This design allows for earlier detection of warning signs, which is essential for prevention and targeted intervention programs. Lastly, the administration of the M-CHAT-R/F follow-up interview and the ADOS-2 was carried out by the same rater in all cases. While having a single examiner may limit inter-rater reliability and introduce potential biases related to examiner expectations or interpretation, it also minimizes discrepancies in scoring due to subjective differences between multiple raters. Additionally, using a single evaluator enhances family adherence to the study and facilitates coordination with other professionals involved in the child’s care.

The high probability of neurodevelopmental disorders such as ASD in the very preterm population underscores the need for future prospective longitudinal studies with larger samples aimed at shedding light on these findings and elucidating modifiable associated factors in early childhood. This follow-up promotes early detection and intervention, thus improving the quality of life for individuals with ASD.