At this time, we lack any firm knowledge about the extent to which environmental chemicals or other environmental conditions might influence the etiology or manifestations of ASDs. That is, using the terminology of cancer risk assessment, do they act primarily as initiators, or primary causes, or are they more like promotors, simply enhancing or amplifying a process already underway or is it just a matter of confounding? For other neurodevelopmental endpoints, the past few decades have demonstrated convincingly that “normal” development is vulnerable to a variety of environmental chemicals. They include metals, pesticides, plasticizers, solvents (and alcohol), organohalogen compounds, and many other chemicals to which virtually all humans in developed countries are exposed. The length of the list of industrial chemicals and their metabolites found in the U.S. population, and published in the NHANES data base is staggering (CDC).
The current study found 72 children with parentally-reported ASD, corresponding to a prevalence of 1.5%. Such a figure is slightly higher than that reported in several other studies that showed ASD prevalence rates in the range of 0.3 to 1.2% (Baird et al., 2006
; Bertrand et al., 2001
; Fombonne, 2005
; Gillberg et al., 2006
; Rutter, 2005
). However, in a recent study on 9,430 children (7-9 y) in Bergen, Norway, parents completed an autism spectrum screening questionnaire (ASSQ), which consists of 27 items scored on a 3-point scale. The authors found an ASD prevalence of 1.8% based on this instrument. Tourette's syndrome was not included in the ASD definition used in that study (Posserud et al., 2006
). It is conceivable, too, that ASD may be underreported since non-responding families exhibit a significantly higher prevalence of autism than responding families (Posserud et al., 2006
What is novel about our data are the statistically significant links between the diagnosis of ASD and certain factors in the child's early environment. The most unexpected of these factors is flooring type. PVC flooring is an important source of airborne phthalates, already documented to be associated with asthma and allergy responses (Bornehag et al., 2004a
; Bornehag et al., 2004b; Jaakkola and Knight, 2008
). The greater the number of rooms covered by PVC flooring, the higher the levels of phthalates in dust.
Condensation on the inside of windows was also associated with ASD. Condensation, we believe, serves as an index of low air exchange rate in the home more than it serves as an independent risk factor. A low rate of exchange in the home will, in essence, increase indoor exposures to tobacco smoke and many other chemical (and biological) pollutants, including phthalates.
In interpreting these results, it is important to recognize that the diagnosis of neurodevelopmental disorders in Sweden is highly structured, unlike the system prevailing in the U.S. Virtually every pregnancy and every child are assessed with periodic examinations. The Swedish system embodies the following steps:
- >99.5% of all pregnant women are followed during pregnancy
- 95% start their controls in gestational week 10
- >99.5% of all children are followed at county level well baby clinics
- Nurses and GPs are given special pediatric training
- In every county, a pediatrician supervises the well baby clinics
- Physicians and nurses are trained to assess psychomotor development
- Assessment of autism was introduced in the 1980s
- Suspected cases are referred to a special neuropsychiatric team
- Practically all children with NP disorders are diagnosed before beginning school
Although we are confident about the clinical accuracy of the ASD diagnosis, our questionnaire data provide no information on when the disorder was detected or diagnosed. Abnormalities in social and cognitive function may have been present early in development or might have emerged much later. Within these limitations, the term, “incidence,” could be deceptive. For example, parental awareness of abnormal behavior may increase as the child ages, as indicated in , which shows a higher prevalence for older children. Until recently, however, a diagnosis of autism was relatively infrequent in children less than three years of age. Quite often, the diagnosis was not established until primary school or kindergarten, where the more demanding and less familiar environment would highlight the child's social and cognitive deficiencies (Landa, 2008
). Based on these considerations, we believe that our questionnaire data on environmental factors and lifestyles at baseline, collected for DBH-I, reflect the family's status prior to the formal diagnosis of ASD and provide a window on prenatal factors. Such a displacement in time should reduce the risk for recall bias. All the same, we recognize that the parents may have been aware of the child's problems at baseline, despite the lack of a formal diagnosis, and may have responded to the questionnaire in a systematically different way from parents whose children displayed no overt problems. In such a case, the risk for recall bias remains. Even so, we believe that associations between factors prevailing at baseline (2000) and parental reports of ASD five years later are more compelling and valid than those based on associations from cross-sectional data in the follow-up study. The exceptions consist of the effects that an ASD child may exert on factors such as economic stress and family unity. The follow-up data present a greater chance of confounding environmental and lifestyle factors with ASD because they were collected at the same time.
Another reason for our confidence in these results is the association we found with smoking. Smoking, especially during pregnancy, is a well-documented risk factor for child behaviour problems, including ASD (Fergusson et al., 1998
; Indredavik et al., 2007
; Linnet et al., 2003
; Wakschlag et al., 2002
; Williams et al., 1998
). A case-control study showed daily maternal smoking during early pregnancy to be linked to autism in the child (Hultman et al., 2002
). Our study confirms that prenatal smoking and smoking during the child's first year of life can be coupled to ASD. Further, since data on smoking during this early life period were collected at baseline, we strongly suspect that tobacco smoke exposure occurred before the diagnosis of ASD, limiting the probability of recall bias. Similarly, the data collected in 2000, when the children were 1-3 years of age, almost certainly reflect conditions prevailing during pregnancy.
Children with ASD were reported to have around twice the odds of airway symptoms (i.e., wheezing and doctor diagnosed asthma) compared to children without ASD. Such an association could be seen in both the baseline and the follow up data. The mechanism for this association is not obvious. It is interesting to point out that boys have more doctor-diagnosed asthma than girls in the age range 6-8 years (11.8 vs. 5.6%, p<0.001 in the current study), and a similar pattern is evident for ASD (Fombonne, 1999
; Wong and Hui, 2008
; Yeargin-Allsopp et al., 2003
); i.e., boys suffer more from both asthma/allergy and ASD. However, in the current data, the association between asthma and ASD was mainly found for girls. Of the 12 girls with ASD, 27.3% (3 girls) were reported to have doctor-diagnosed asthma at baseline compared to 3.6% asthma among girls without ASD. Among the 60 boys with ASD, 5 had asthma at baseline, corresponding to a prevalence of 8.3% compared to 6.8% among boys without ASD.
The reason why allergic symptoms in the mothers at baseline were weakly associated with ASD in the child while such problems among the fathers were not associated at all is not obvious. However, such associations have been reported in a study from the U.S., where the risk for ASD doubled in the offspring of mothers with allergic diseases during pregnancy (Croen et al., 2005
), indicating that such diseases could be associated with ASD. A number of potential mechanisms for such an association can be extracted from the neurodevelopmental literature, for example, those studies showing associations between schizophrenia risk and maternal infections (Brown and Susser, 2002
; Meyer et al., 2005
). Infections, of course, activate the immune system, whose potential links to ASDs have been the subject of many recent publications (Becker, 2007
; Fatemi et al., 2008
; Gregg et al., 2008
; Pessah et al., 2008
Our study showed that families with an ASD child report more economic problems and proxies for lower economic status, such as living in multi-family house and a smaller size home, compared with families without an ASD child. Unfortunately, data on economic problems (i.e., problems in paying bills) were only collected in the follow-up study, meaning that we cannot establish temporal relationships. Information about the type of building and size of the home were collected both at baseline and at follow up. Here there are indications of a stronger association between ASD and these factors in data from 2005 compared to data from 2000, indicating that economic problems are more likely a result of ASD rather than a risk factor. Divorce in the family (i.e., most often a single motherhood), which in Sweden is associated with lower socioeconomic status (Fritzell et al., 2007
), was also a stronger risk factor in the 2005 data, compared to the 2000 data, indicating again that divorce and associated financial problems are more a result of ASD disorder in the family than the opposite. It is well known that parents with children that are disabled or have chronic diseases are more apt to divorce. However, one complicating issue is that smoking is more common in families with lower socioeconomic status. For example, in the families experiencing divorce, about 23% of the mothers smoked while only 10% smoked in families where the child lived with both parents. In families that often have problems in paying bills, 37% of the mothers smoked while 25% of the mothers smoked in families that sometimes have problem with paying bills and in families without such problems only about 11% smoked. Therefore, it is difficult to separate the effects of socioeconomic status from those of smoking when considering risk for ASD.
Sick Building Syndrome is applied to a situation in which occupants of a building complain of nonspecific symptoms such as headache, dizziness, nausea, difficulty concentrating, eye or throat irritation, cough, and dry or itchy skin. It is assumed to result from factors such as inadequate ventilation, chemical contaminants, and biological contaminants. We used the list in as an index of comfort level, so to speak, in the home environment which might then be traced more specifically to the indoor environmental variables.
The prevalence of parents' reports of sick building syndrome symptoms (SBS) is in accordance with other studies (Eriksson and Stenberg, 2006
). There was a tendency for parents of ASD children to report more SBS symptoms than other parents. It is important to point out that the SBS reports come from the baseline investigation (2000), meaning that reports of SBS symptoms most probably are not a result of a subsequent ASD diagnosis in the child.
Perhaps the most intriguing of our findings is that PVC flooring material in the home, when the child was 1-3 years of age, was associated with ASD five years later. In accordance with the discussion above, even if a parent knew that their child had ASD at baseline, there is no reason to believe that parent's reports on type of flooring would be biased in the absence of any discussion of the relationship between type of flooring or building material and ASD. PVC flooring, however, might be seen as a proxy for lower socioeconomic status in Swedish society. As said earlier, socioeconomic factors may be a result of ASD in the family while PVC flooring was associated with ASD more strongly at baseline than in the follow-up study five years later. Another finding indicated that PVC flooring in the parent's bedroom showed a stronger association with ASD compared to PVC flooring in the child's bedroom. Many children in Sweden sleep in their parent's bedroom during the first years of life, so that one explanation for the association could be that children with problems in their cognitive functioning more often sleep with their parents and for a longer period of time compared with children without ASD. We also found that ASD children slightly more often slept in the parent's bedroom compared with other children at baseline (47.9 vs. 43.2%).
As reported by Bornehag (Bornehag et al., 2005c
) it is uncommon to find wall-to-wall carpeting in Swedish homes. They estimated a frequency of less than half a percent. In contrast, floors in U.S. homes, especially newer models, most often are carpeted, except for kitchens and bathrooms, because wood flooring is considerably more expensive. Hwang (Hwang et al., 2008
) found indoor dust samples highly contaminated by phthalates such as di-(2-ethylhexyl) phthalate (DEHP). Mendell (Mendell, 2007
) summarized 21 epidemiologic studies on associations between indoor residential chemical emissions, or emission-related materials or activities, and respiratory health or allergy in infants or children. Strong associations were reported between many risk factors and respiratory or allergic effects. Risk factors identified most frequently included formaldehyde or particleboard, phthalates or plastic materials, and recent painting. Other studies have also found tributyl tin, pyrethroids, and Polybrominated Diphenyl Ethers (PBDEs) in carpets. The U.S. situation, for this reason, may magnify the associations with ASD reported here, and, we believe, deserves at least exploratory investigations.
Phase II of the DBH study showed that PVC flooring is one important source of phthalates in indoor dust in Swedish homes (Bornehag et al., 2005a
). Moreover, exposure to selected phthalate esters in indoor dust was associated with asthma in children in a dose-response relationship (Bornehag et al., 2004b
), a result confirmed in a DBH-study in Bulgaria (Kolarik et al., 2008
). These findings evoke the question of potential mechanisms. Phthalates are viewed as potential endocrine-disrupting chemicals, acting as anti-androgens to disrupt male development (Wilson et al., 2008
). At the same time, it has been suggested that phthalates can induce asthma and allergic responses (Jaakkola and Knight, 2008
). The challenge to investigators is to define how these toxic manifestations might be connected to the etiology or expression of ASDs.
Up to now, no environmental factors have been documented in the etiology of ASDs. Although gestational exposures to thalidomide (Stromland and Miller, 1993
) and valproic acid (Christianson et al., 1994
) are risk factors for ASDs, they can hardly be considered environmental sources. Their main value, similar to the role that MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) played in Parkinson's disease (Langston, 1985
, Langston et al., 1999
) is to illustrate the possibility that exogenous chemicals may be risk factors for the disease.
A limitation of this study is that we have only parental reports of ASD from a questionnaire with limited confirmatory information. However, our validation study of ten families showed that the parental reports of ASD in the questionnaire reflected a clinical diagnosis of ASD by medical professionals. Furthermore, our experience tells us that there is no reason to believe that the parents would report autism, Asperger or Tourette's syndrome without a clinical diagnosis or other relevant information as a basis for their reports. We also believe that our data on ASD are valid because our estimated prevalence is in accordance with other studies, and associated factors, such as male sex and smoking among mothers, are in agreement with findings from other studies. The association of PVC flooring with ASD, which we presume to reflect exposure to airborne phthalates, suggests that studies of other chemical contaminants with endocrine disruptor properties might yield useful insights into the genesis of ASDs.