Maternal and Child Health Journal

, Volume 11, Issue 1, pp 73–79

Prenatal and Birth Complications in Autism

Authors

    • Department of Preventive MedicineNew Jersey Medical School - UMDNJ
    • Department of NeurologyNew Jersey Medical School - UMDNJ
  • X. Ming
    • Department of NeurologyNew Jersey Medical School - UMDNJ
  • M. Lamendola
    • Department of Preventive MedicineNew Jersey Medical School - UMDNJ
Original Paper

DOI: 10.1007/s10995-006-0142-7

Cite this article as:
Brimacombe, M., Ming, X. & Lamendola, M. Matern Child Health J (2007) 11: 73. doi:10.1007/s10995-006-0142-7

Abstract

Objectives: Prenatal and birth history as potential sources of risk factors in relation to the onset of autism were examined. Methods: A cohort of 164 families of autistic children referred to The Autism Center at New Jersey Medical School—UMDNJ, Newark, New Jersey, over a two-year period was studied. Intake prenatal and birth history information was obtained from each family and reviewed by a clinician. Results: Prevalence rates in this cohort for vaginal bleeding, prolonged labor and prematurity were higher than comparable rates reported nationally and in New Jersey. Clustering of multiple prenatal risk factors was observed. This clustering was associated with the age of the mother, but uncorrelated with birth order. Conclusions: These findings support the general hypothesis that systemic problems at the prenatal stage may form a distinct dimension of risk associated with autism.

Keywords

Prenatal historyAutismRisk factorsEpidemiology

Introduction

Autism is a neurodevelopmental disorder defined by social and communication deficits and repetitive behaviors that are typically detectable in early childhood, continuing throughout life [1]. Autism is one of a group of disorders typically viewed as developmental disabilities comprising a spectrum, delineated by a set of observable dysfunctions [2, 3]. The prevalence of autism is estimated at 2-5/10,000 with a 4:1 male to female ratio [4]. Its etiology remains to be determined, but is thought to be an interaction of genetic and environmental factors [5, 6], with still unknown genetic structures associated with onset across a spectrum of severity and symptoms [7]. A biomedical basis in regard to language deficits has recently been detected [8].

Autistic disorder first surfaced in the DSM-III-Revised as a diagnosable disorder with 16 criteria in the areas of social development, communication and interests [1]. The current diagnostic criteria as operationalized in the DSM-IV, ICD-9 and ICD-10 require the presence of three features between the ages of 4–5 years of age (although symptoms must begin before age 3): failure to develop normal social interactions, lack of normal development of language, particularly the use of language for communication, and a restricted range of interests and behaviors [9].

One area of research interest is the association of autism with prenatal risk factors [1012]. Such risk factors may provide guidance in helping define the clinical sub-types of autism. In general, a growing body of autism research demonstrates neuroanatomic and biochemically based differences whose origins are prenatal [1316].

Autism and birth related risk factors

Prenatal risk factors have been the subject of several research studies examining potential associations between the number and severity of known prenatal and birth history related risk factors, the overall optimality of the birth, and the onset of autism. Comparisons of pre-, peri- and neonatal factor incidences in a cohort of autistic patients with incidences in the US population as reported in the Report of Final Natality Statistics in 1995 have been conducted [3]. Several factors, including uterine bleeding, showed statistically significant increases in the autistic group from national averages, supporting previous findings that suggested a consistent association of unfavorable events in pregnancy and delivery with pervasive developmental disorders.

A similar study was performed looking at genetic factors [17]. In a sample of children with autism spectrum disorder and their unaffected siblings, optimality, or an overall index rating reflecting freedom from complications of pregnancy and birth was used. Results showed that children with autism had lower optimality than unaffected siblings.

Prenatal factors in autistic children and their non-autistic siblings were examined [18], and use of medication during pregnancy was found to be relatively more prevalent in the autistic children. Several studies on prenatal and perinatal factors and the etiology of autism were reviewed [3], focusing on whether identified risk factors were consistent in the literature. Bleeding in pregnancy was reported as the most consistent pregnancy risk factor, surfacing in four separate studies. Short gestation was reported in two studies as a risk factor as well as abnormal presentation. The review concluded that studies using relatively recently evolved methodologies for design and analysis have often failed to confirm the findings of older studies on etiology.

The issue of whether genetic components and environmental risks linked to early fetal development were present in autistic patients was examined using a case-control design in a Swedish setting [19]. Caesarian delivery, vaginal bleeding and fetal distress were found to be significantly associated with autism. A recent Danish study [20] found that low Apgar score, breech presentation and gestational age at birth and parental psychiatric history were all significantly associated with onset of autism supporting the idea that prenatal environmental and genetic causes are important sources of risk.

In this study prenatal and birth related risk factors were examined in a cohort of autistic patients. The incidence and overall clustering of these prenatal risk factors was examined with the aim of identifying prenatal risk factors that are relatively more prevalent among autistic children than in the general population at the state and national level. To examine the frequency of multiple co-incident factors, the number of typical prenatal risk factors present for each birth was determined. Potential correlation with age and birth order was investigated.

Methods

This study was conducted at The Autism Center at New Jersey Medical School – UMDNJ. Children presenting to the pediatric neurologist at The Autism Center NJMS – UMDNJ between October 2001 and August 2003 were assessed by a pediatric neurology specialist and formally diagnosed and evaluated. As part of the assessment, parents or caregivers were asked to complete an intake survey that was reviewed and validated by the attending specialist. One component of the intake form captures pregnancy, labor and delivery and newborn history of the children and their mothers. Children with diagnosed autism, autism spectrum disorder (ASD) or pervasive developmental disorders (PDD) were included in the cohort.

The NJMS Autism Center is the major referring Center for Northern New Jersey. The subjects were referred largely by parents (>95%) after the reporting of the establishment of the Center in a state newspaper. Reasons for referral included confirmation of diagnosis, developmental assessment, medical evaluations, treatments for medical or behavioral disorders, seeking innovative treatment, seeking multidisciplinary comprehensive evaluation and care, or possibility of enrollment in research programs. The diagnosis of ASD (autism, PDD-NOS or Asperger's syndrome) was made or confirmed based on DSM-IV criteria. Autism Diagnostic Interview-Revised, Autism Diagnostic Observation Schedule-Generic, and/or Childhood Autism Rating Scale were used for confirmation of diagnosis in some of the subjects, performed in our center or elsewhere. Children with Rett's syndrome or disintegrative disorders were excluded.

Rates for the various factors in the cohort were calculated by dividing the total number of cases by the number of reporting subjects. The denominator did not vary greatly across risk factors. The issue of how to interpret these rates is a challenge. A longer-term prevalence approach is not really appropriate for a disorder such as autism, recently redefined in terms of its diagnostic criteria and may itself have become more prevalent over the past decade. After examining weighted averages of 1998, 2000 and 2002 state and national incidence rates respectively to estimate a localized prevalence value and finding them stable, we decided that year 2000 state and national incidence rates served as representative benchmark null values against which the autism cohort prevalence rates in this time frame could reasonably be compared. The national incidence rates for prenatal risk factors were taken from the National Vital Statistics Reports database for year 2000 births maintained by the Center for Disease Control and Prevention. New Jersey state incidence rates for the year 2000 were obtained from the Center for Health Statistics – New Jersey Department of Health and Senior Services.

https://static-content.springer.com/image/art%3A10.1007%2Fs10995-006-0142-7/MediaObjects/10995_2006_142_Fig1_HTML.jpg
Fig. 1

Geographic distribution of NJ autism cases

As racial sub-groups typically have an effect on birth history related risk factors, the racial composition of the cohort was examined. While sample sizes were limited for detailed sub-group breakdowns for the various risk factors, the composition was typical of New Jersey and comparable in overall terms to the state population. In terms of autism prevalence, the racially representative nature of the cohort is comparable to other findings [4].

To assess potential multi-factor clustering among prenatal risk factors in the sample, a basic measure of overall birth complications was defined, similar to that employed in [17]. Eleven common prenatal risk factors were identified (birth type, length of pregnancy, C-section risk, fetal distress, induced birth, toxemia, gestational diabetes, age >35, vaginal bleeding, frequent infection, hypoxia). The total number of these risk factors present for each mother was determined and its distribution examined, both overall and across age categories and gender of child with mean and standard deviations reported. The effect of birth order was also examined as first-born births often have more complications. This has also been found to be a risk factor in some cohorts of autistic children [11, 12].

A clinical database was developed consisting of clinical and demographic prenatal and birth history information for the cohort was accomplished by abstracting prenatal and birth history related information from clinician validated intake forms. All patients were included except for those whose birth histories were unavailable such as the adopted children. The database was created in Microsoft Access version 2000. The data was imported into SPSS (version 8.0). Frequency tallies were performed on all of the categorical variables and prevalence rates were determined. Risk factor rates were then compared with national and state rates. Chi-square tests for differences in proportions were used in the analysis.

This study was approved by UMDNJ Institutional Review Board.

Results

In total there were 164 subjects in this autistic cohort, 88% of whom were from the Northern New Jersey area. See Fig. 1. Complete case data was available for 128 of these. Of the entire cohort, 137 (82%) were male, giving a male to female ratio of approximately 4:1. The mean age of was 6.8 years with a standard deviation of 3.5 and an overall range of 2 to 18 years. In terms of ethnic breakdown, 103 (62.8%) were white, 31 (18.9%) black, 23 (14.0%) hispanic and 7 (4.3%) asian. In year 2000 the population distribution of these categories was 78.8%, 14.8%, 6.1% and 13.3% respectively [21]. The sample cohort is reflective of the overall ethnic composition of New Jersey, with a slightly higher percentage of hispanic children in the sample.

Frequencies of birth history related risk factors in the cohort are summarized in Table 1. Risk factors with cohort prevalence rates over 20% included the following; infection during pregnancy (24.3%), induced delivery (25.7%) and caesarian delivery (29.9%). Prevalence rates of over 15% in the cohort include prematurity (16.7%), vaginal bleeding (16.7%) and newborn jaundice (18.1%). Some typical prenatal risk factors not having high prevalence were low birthweight, toxemia (3.9%) and hypoxia (1.7%).
Table 1

comparison of cohort prevalence rates of birth history risk factors with national rates (year 2000)

Risk factor

Autism cohort rate (Overall sample size, count, %)

National (2000) Population rate* (%)

P-Value**

Multiple birth

143, 14, 9.7

3.1

.001

Prematurity (under 37 wks)

143, 24, 16.7

11.6

.044

Vaginal bleeding

143, 24, 16.7

6.6

.001

Infection during

   

pregnancy

144, 35, 24.3

Diabetes

142, 9, 6.3

2.9

.025

Drug/alcohol use

142, 6, 4.2

Prolonged labor

144, 11, 7.6

7.8

NS

Induced delivery

144, 37, 25.7

19.9

.054

Caesarian delivery

144, 43, 29.9

22.9

.032

Fetal distress

143, 9, 6.3

3.9

NS

Low birth weight

144, 12, 8.3

7.6

NS

Lack of oxygen

143, 3, 2.1

Newborn jaundice

144, 26, 18.1

*CDC National vital statistics reports (2000).**P-values calculated using Chi-square test.

Comparison with national rates

The cohort rates were then compared to known year 2000 national rates for the general population for each risk factor. P-values for chi-square tests of difference in rates are given in Table 1. Five risk factors with significantly higher rates for the autism cohort were identified. They were multiple delivery, prematurity, vaginal bleeding, diabetes, caesarian delivery.

In addition, 25.6% of the cohort (mothers) had taken medications during pregnancy (terbulatin (4 cases), antibiotics (5 cases) and combinations (37 cases)), 20.6% had induced labor and 32.5% reported frequent infections, defined here as two or more infections, during pregnancy. Fetal distress was observed in 5.3% of cases. The degree of stress was not available for analysis. Accurate comparative national rates were not available for all factors.

Multiple deliveries in the cohort consisted of twins (6 fraternal, 3 identical, 7.3% total) and triplets (5, 4.0%). Most of the multiple births resulted in low birth weights and reduced length of pregnancy.

Comparison with New Jersey rates

As a large majority of cases were from New Jersey or mothers who had lived in New Jersey while pregnant (88%), the prevalence rates for the birth history risk factors in autism were compared to New Jersey state population prevalence rates. See Table 2.
Table 2

Comparison of cohort prevalence rates of birth history risk factors with new jersey rates (year 2000)

Risk factor

Autism cohort rate

NJ (2000) population rate* (%)

P-Value**

Multiple delivery

9.7

3.9

.002

Prematurity (under 37 wks)

16.7

9.4

.004

Vaginal bleeding

16.7

6.6

.001

Diabetes

6.3

4.0

NS

Low birth weight

8.3

7.5

NS

Prolonged labor

7.6

1.8

.001

Caesarian delivery

29.9

26.5

NS

*Center for health statistics – New Jersey Department of Health and Senior Services.**Chi-square test.

Differences were identified for several risk factors; multiple delivery, prematurity, vaginal bleeding and prolonged labor. There were no significant differences for diabetes and caesarian delivery, even though these risk factors differed significantly different from national rates.

Clustering in the number of risk factors present was examined for all mothers for whom there was complete data (n=128) on eleven standard prenatal risk factors. These factors were mother's age greater than 35, multiple delivery, length of pregnancy, induced birth, lack of oxygen, caesarian delivery, toxemia, diabetes, frequent infection, fetal distress, vaginal bleeding. A frequency histogram is shown in Fig. 2. The range of clustered risk factors ran from 0 to 7 with 90% of mothers having at least one risk factor present and 51.5% having 3 or more risk factors. The overall mean number of risk factors per mother was 2.03 with a standard error of 0.15. Mothers over the age of 35 had on average 3.05 risk factors with a standard error of 0.31, while those under 35 had an average of 1.74 risk factors with a standard error of 0.15 (t-test, p-value = 0.001). Comparison between first-born and non first-born cases showed no significant difference in the number of clustered risk factors (mean and standard deviation respectively of 2.23 (1.58), 1.95 (1.36), t-test p-value = 0.41).
https://static-content.springer.com/image/art%3A10.1007%2Fs10995-006-0142-7/MediaObjects/10995_2006_142_Fig2_HTML.gif
Fig. 2

Frequency distribution: clustering of multiple risk factors

Discussion

These results support findings in previous studies [3, 16, 19] that have found birth history related risk factors, including vaginal bleeding related to the onset of autism. This may also be supportive of recent findings regarding abnormally high brain volume in some autistic children [22]. The observed rate for vaginal bleeding in this autism cohort is significantly higher than both the national and New Jersey levels for the general population.

Multiple deliveries are significantly higher in the autism cohort than in the national and New Jersey levels for the general population. While this is consistent with findings in [5], and may support a genetic role in the etiology of autism, it should be noted that various other environmental factors may be relevant here including nutrients, uterus condition, associated vaginal bleeding and other factors. Genetic background may only be a contributing factor, possibly interacting with environmental exposure.

The Caesarian delivery rate in the cohort was higher than the national rate. Similar results supporting caesarian delivery as a potential risk factor for autism were found in [19]. It is important to note that the cause of caesarian delivery may affect the relative importance of this risk factor. While repeat caesarian should have little impact on pregnancy, emergency caesarian may produce great stress on both mother and child.

Increased concordance rates for autism spectrum disorders in dizygotic twins whose mothers were exposed to Terbutaline as treatment for pre-term labor has been found [23]. The same authors found a significantly increased prevalence of polymorphisms of the β2-adrenergic receptors themselves (substitutions with glycine and glutamic acid at codons 16 and 27 respectively) in autistic patients compared to the general population (Connors, personal communication).

Even though there were multiple significant differences between the autism cohort and the national and New Jersey rates, no overwhelming evidence points towards one particular prenatal risk factor. This is in agreement with recent work examining an Australian cohort of autistic patients [24]. In our study, the autism cohort had a high level of multiple birth complications, agreeing with work in [17] and specifically vaginal bleeding was found to be more common, agreeing with [2].

The risk factors examined here for clustering are standard determinants of maternal health. While age had the expected effect, birth order was not relevant. While there is no typical number of risk factors to be expected in healthy mothers, reasonable interpretation might set this rather low. Here 90% of mothers had at least one risk factor and over half the mothers in the cohort had 3 or more risk factors. This type of systemic pattern in birth related complications supports recent findings in a large Danish study [20] where low Apgar score was found to be a risk factor in regard to autism.

Limitations

Our study reflects a cohort of limited sample size. The cohort reflects a primarily Northern New Jersey population. Further, the data here is primarily self-report, although verified by physician with history taking and with access to medical records brought by patients to the Autism Clinic. Birth history risk factors and related outcomes may be greatly affected by differences in the quality of service provided by hospitals, obstetricians and mother's life style, as can the occurrence of multiple risk factors. We also did not collect data on mother's lifestyle in the period immediately preceding pregnancy.

Summary

The results here support broader investigation of maternal prenatal health as a dimension of risk in relation to autism and ASD. Such clusters of birth history related problems may represent systemic illnesses or insults which may affect fetal development and be relevant to the onset of developmental disorders. The risk factors of multiple deliveries, prematurity, vaginal bleeding, and prolonged labor should all be considered for further investigation in larger U.S. autistic cohorts, augmented with detailed environmental exposure and genetic information, and the clustering patterns of such risk factors examined in detail.

Copyright information

© Springer Science+Business Media, LLC 2006