Expert Interview With Judith K. Grether, PhD
The Best Evidence Study
Risk of Autism and Increasing Maternal and Paternal Age in a Large North American Population
Grether JK, Anderson MC, Croen LA, Smith D, Windham GC
Am J Epidemiol. 2009;170:1118-1126
Am J Epidemiol. 2009;170:1118-1126
This study was selected as the subject of this interview because of its high ranking in Medscape Best Evidence, which uses the McMaster Online Rating of Evidence System. Of a possible top score of 7, clinicians who used this system ranked this study as 5 for relevance and 7 for newsworthiness.
About the Interviewee
Judith Grether, PhD, is Senior Epidemiologist in the Environmental Health Investigations Branch, California Department of Health Services. Dr. Grether is a perinatal epidemiologist who has had the opportunity over many years, within the California Department of Public Health, to conduct research studies on the prevalence of and risk factors for developmental disabilities, primarily cerebral palsy and autism spectrum disorders. She has recently retired and is now working to complete some analyses and manuscripts.
Medscape: Your study, "Risk of Autism and Increasing Maternal and Paternal Age in a Large North American Population," was highly ranked by pediatric clinicians. Could you just give a brief description of it and its significance?
Judith Grether, PhD: Our study was the largest that has yet been conducted -- and, I would guess, is probably the largest that will be conducted -- to specifically address the question of whether the risk for autism in offspring increases with the advancing age of the mother and/or the father. With our very large study, we were also able to obtain statistically precise estimates of how much of an increase in risk there is as parents get older. A number of prior studies have looked at the question of a parental age effect; to my knowledge, none have reported adecreasing risk as parents get older, but some have found an increased risk only for advancing maternal age and others only for advancing paternal age. A couple of studies have pretty dramatic estimates of the amount of that increased risk -- particularly one from Israel that found only father's age to be important. From a scientific perspective, to get clues to what may underlie any increase in risk for autism associated with older parents, it is important to answer the question: is it both age of the father and the mother, or is it only mothers or only fathers? There are some tricky statistical issues here and it is very helpful to have a large database like we had, in which the subjects are not self-selected for the study.
Medscape: Where did you get your data from?
Dr. Grether: We analyzed data from the California Department of Developmental Services (DDS), which is a statewide service system in California for people with developmental disabilities. It is a well-established, well-known program that has been around since, I believe, the late 1970s. Children and adults with autism, mental retardation, cerebral palsy, epilepsy, and related conditions are eligible to receive services if they meet diagnostic and severity criteria. It is an entitlement program, so there are no income or citizenship criteria. The program has a database that, I believe, has used the same data collection instrument since 1987 to document eligibility for services.
With the appropriate approvals from DDS and our institutional review board (IRB), we were able to obtain a data file that contained eligibility diagnoses for children born in 1989-2002. We then linked the DDS data to the statewide birth certificate files, again with IRB approval and strict requirements to maintain the confidentiality of the data, to obtain demographic data, such as parental age, [that had been] recorded before there was any concern that a child might have a developmental disability. The DDS data are assembled for administrative purposes, so the database has definite limits for research studies, especially with regard to diagnostic details and potential underlying causal factors. However, DDS data are extremely valuable for looking at broad-stroke time trends and demographic patterns. And so, it is really very useful for some types of studies.
After we had excluded multiple births and subjects with missing data, we were able to analyze the demographic characteristics of more than 20,000 singleton children born to California residents who were receiving DDS services for autism and compare them to the remaining 7.5 million singleton children born to California residents during those same 14 years.
Medscape: Did these children in your dataset have classic autism or were they part of the autism spectrum?
Dr. Grether: Well, that is part of the complexity here. According to DDS eligibility guidelines, a child is supposed to have autistic disorder or have mental retardation with sufficient severity to meet the criteria to be getting services. In reality, partly because the diagnosis of autism is based on behavioral criteria, the distinction between autistic disorder and other spectrum conditions can be rather "soft." We know through experience that there are some children receiving DDS services for autism who may not, in some clinicians' eyes, meet the full DSM-IV [Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition] criteria for autistic disorder. However, these are clearly not typically functioning children; they need services and there may be nowhere else to send them. So, sometimes the eligibility criteria likely get interpreted a bit leniently. None of this, I think, is particularly relevant to our concerns in this study. It has restricted some other analyses, but is not so relevant here.
Some unknown proportion of children enrolled with DDS are probably somewhere else on the spectrum, and there also children receiving services from DDS who have autism but whose eligibility diagnosis is mental retardation. We have no way of teasing this apart unless we actually go into the hard copy records, which we have done for some other studies, but not for this one, given the very large size of this study. The bottom line is that we can't make any claims about autistic disorder vs other spectrum disorders, so we have to be careful here, especially when looking at time trends.
Medscape: Would the fact that children are entering the system who actually don't meet the strict autism criteria be responsible for the increase observed over the past decades?
Dr. Grether: Clearly, there has been a huge increase in children with autism coming into the DDS system, and as epidemiologists, we have spent a lot of time trying to figure how much of that increase is likely attributable to children with other spectrum disorders coming into the system, and how much is likely due to other factors, such as more awareness or that there are now treatments for eligible children. We know all of those things go on, and also that diagnostic criteria have changed over time. The basic question is, from an etiologic standpoint, how much of the increase may be real, in the sense that something new is going on that was not there before and that contributes to some children having autism.
Medscape: So, you can't tell what percentage of the increase might be due to this broader diagnosis.
Dr. Grether: No, to really get a good handle on that, you would have to do a study in which you go into a community, you try to find every child who may be anywhere on the spectrum, and you give all of them the gold standard diagnostic evaluation. That is a very expensive and time-consuming kind of study.
Medscape: I suppose if you saw autism leveling off now, it would be some indication that a fairly significant percentage might be due to broader diagnostic or socioeconomic factors.
Dr. Grether: If the eligibility criteria for programs like the DDS included anybody anywhere on the spectrum, then you would expect there would be an influx of new kids coming in who meet the broader diagnostic criteria. The numbers would go up accordingly and then level off after that if nothing else was changing to lead to bigger numbers. Obviously, we cannot go backward in time and reconstruct such a scenario, so we have to live with uncertainty.
Medscape: Are there any other factors that you could hypothesize might be going into the increase in autism?
Dr. Grether: Many environmental factors have been raised as possibilities, because obviously, many things have changed in our environment. The difficulty is that most of the things that we worry about -- water and air contamination, plastic hardeners in the sports water bottles, etc. -- are very hard to study in a rigorous way. It is a huge methodological challenge and, fortunately for all of us, some talented scientists are trying to do these studies. Some years ago, after trying to tease out how-much-is-real-and-how-much-isn't kind of question, some of us got to the point of saying, you know, folks, we are never going to have the occurrence data to really figure this out in a satisfying way. Let's stop devoting some much of our precious time and energy to this question, and let's try to figure out what is going on in our environment that could be relevant here. Even if all these new environmental exposures and other changes in our environment are not contributing to autism, per se, we need to know more about them because autism is only one of many pediatric disorders that could be linked to environmental factors. Let's do our best to design studies to really try to narrow in on some of it. It is a huge challenge.
Medscape: What about genetic factors?
Dr. Grether: Genetic studies are going on, and talented researchers are finding more and more genetic factors that may contribute. But from everything we know about the human condition, environmental factors are likely to also play a role.
Medscape: To get back to your study, older parenting has been increasing, so might this also contribute to the increase in autism?
Dr. Grether: Yes, the typical age of parents has been going up somewhat. We have done an estimate, not yet published, in which we basically took the demographic profile of the statewide population in 1989 and applied it to the statewide population in 2002. This standardized the 2002 population for the demographic profile in 1989, permitting us to estimate how much of the increase in DDS-reported autism was either reduced or exacerbated by changing demographic patterns. It was somewhat reduced, but not by a lot. Overall, the changing demographics of the population, including increasing age at parenthood, seems to explain some small part of the increase we see. In a recent paper, colleagues at UC [University of California]-Davis used DDS data and came up with an estimate that 4.6% of the increase in DDS-reported autism from 1990-1999 may be attributable to changing parental age patterns. That's a very modest portion. Clearly, the fact that parents are typically somewhat older than they used to be is not the main explanation behind the increase in autism that we are seeing.
Medscape: What was the significance of your results on older parenting and autism?
Dr. Grether: Our results were certainly consistent with what other people are reporting using California data and also with other studies conducted elsewhere in the world. Both mother's and father's age matter, there is both a maternal and a paternal age effect, and in our study, the maternal age effect is somewhat stronger. However, the increase in risk for autism in offspring born to older parents is pretty modest and probably not useful as a guide in personal decision-making. The real benefit of our study and similar ones is that we now have a scientific clue that may help point us toward underlying biological factors that contribute to the risk for autism in some children. We needed to pin down whether advancing parental age really matters or if it is just some kind of statistical artifact. We can now have confidence that there is a modest parental age effect for both mothers and fathers (at least in the US population). Our task now is to try to figure out why the risk of having a child with autism is greater for older parents.
Medscape: Can you give us some background on why parental age may affect the risk for autism?
Dr. Grether: There are 2 sets of factors, they are not mutually exclusive, and both may well be important. One set of factors involves biological changes that occur as men and women get older and that can affect the outcome of the pregnancy. Down syndrome would be an example of an outcome we know is sometimes related to maternal age. For women, as they get older, the hormonal balance of the womb changes in some ways, and older women are more likely to have infertility. There has been speculation that the hormonal changes that contribute to infertility, or the treatments for infertility, may increase the risk for autism, but we don't have very good data yet. As a woman ages, she will have an increased cumulative exposure to chemicals and toxins in her environment that may affect the neurodevelopment of the fetus. Again, there are not very much data there, but it makes good biological sense to conduct research on environmental factors to which the mother is exposed before or during pregnancy that may affect fetal neurodevelopment.
Medscape: How is male reproduction affected by aging that might influence the risk for autism?
Dr. Grether: Unlike eggs, which were formed during fetal development of the mother, the process of sperm creation and maturation is ongoing, and it is now recognized that as men get older, typically there are more new mutations that occur in the sperm, that are developing. Again, it could be related in some ways to cumulative toxic exposures as men get older. Several studies have demonstrated that, in families lacking a history of schizophrenia, there is an association between older age of fathers (but not mothers) and risk of having a child with schizophrenia, presumably due to more de novo mutations in sperm. These studies of young-adult schizophrenia provide a model that may be relevant to autism, but the studies needed to clarify this for autism have not yet been reported.
The data we and others report showing both a maternal and a paternal age effect for autism could indicate that genetic mutations to the sperm contribute to the risk for autism in the children, but clearly that is not the whole picture because we also see a maternal age effect. That is one reason why we are so interested in studying hormonal factors in the mother during or before pregnancy that may contribute to autism risk.
Medscape: Are there reasons, other than biology, why this risk exists in older parents?
Dr. Grether: There is another possibility here, another possible explanation that could also explain the parental age effect. Over and above whatever reproductive biological changes happen as people age, people who are predisposed to have an autistic child may be more likely to start having children later in life, for a variety of reasons. They may be going to college, pursuing more intensive careers, and putting off childbearing until later. We can speculate about these reasons, but we don't yet have studies on this and, unfortunately, we don't have relevant data from our study.
There is also speculation in the research literature that, to the extent there is a tendency for people to have children with a partner who is similar to themselves ("like marries like"), there may be a pattern for people with a genetic predisposition for autism to form a relationship with a partner who has a similar genetic predisposition. By waiting until later in life to have a family, perhaps there is more opportunity to find a partner with similar underlying genetics, thereby increasing the genetic risk in the child.
As I said, this is speculation at this point. The studies simply have not yet been done. The "takeaway" message here is that we now have some good hypotheses around which to conduct further studies that will help in disentangling the mix of genetic and environmental factors that contribute to autism.
Medscape: Did you find any greater association with autism and any ethnicity or race in older parents?
Dr. Grether: No, not with the increasing parental age. It was really very, very similar across racial ethnic groups, which is really interesting.
Medscape: What about socioeconomic groups?
Dr. Grether: We found an association with one SES [socioeconomic status] measure (private insurance payment for the delivery vs other sources of payment, primarily government programs for low income people) but not with another (educational level attained by the mother or father), so our findings must be considered to be inconclusive with regard to socioeconomic status.
Medscape: Haven't there been some studies indicating that higher economic groups are more prone to autism than lower?
Dr. Grether: Yes, but that is really a different question. Let me try to restate it, as this can be a real mind-twister. In our parental age study, we were trying to look at how much an increase in the age of parents is related to the risk for autism in a child. We were not looking at the baseline risk. So when I say that we did not see a difference in the increase in risk associated with age of parents for well-educated parents compared with less well-educated parents, what I am talking about is the increase associated with the parents being older. At any given age, well-educated parents are more likely to have children with autism than less well-educated parents (and we don't know why this is so), but the amount that the risk increases with increasing age of parents is the same for both groups. Their baseline risk may be different, but the rate of increase in risk represented by the upward slopes of the lines are parallel to each other.
Medscape: Are there any other factors that might affect the risk of having a child with autism among older vs younger parents?
Dr. Grether: One finding that we have, which most other studies have not examined, is that both the maternal and paternal age effect is stronger among first-born children than among later-born children. For first-born children, both for mothers and fathers, that slope with increasing parental age is steeper than for later-born children. So, if it is the first time a parent is having a child, the risk of that child having autism is greater than for a child of a parent of comparable age who already has children.
The explanation for this pattern may be that parents starting their families later in life have a pre-existing risk and so we would see this pattern more strongly among first-born children. It certainly should not matter if it is the first-born child for that older father or the fifth-born, since the sperm at that older age would still be going through the same genetic changes.
Medscape: Does the higher risk in first-born children hold up with women as well as men?
Dr. Grether: Yes, the maternal age effect is stronger in first-born than in later-born children. For a woman has had prior children, prior pregnancies may change the in utero environment in ways that affect her fetus differently than is typical for a women of the same age having her first child. Or perhaps the effect of toxic exposures from the environment is different, depending on the reproductive history of a woman giving birth for the first time compared with one who has given birth before.
My guess is that when we have the answers to all the questions raised by our study and others that have addressed the question of a parental age effect, we will see that both delayed childbearing and age-related biological changes contribute to the higher risk for older parents. But until the research is conducted, we won't know. A recent letter to the editor in the American Journal of Public Health by some colleagues basically said, and I fully agree, that studies, including ours, have consistently established a parental age effect. Now it is time to move on and figure out why.
Medscape: Do you see prenatal screening anytime in the future?
Dr. Grether: No. I don't think we are close to having prenatal screening. In the meantime, as a society, we need to work through the deep ethical issues involved. Our current best understanding is that autism spectrum disorders are at one end of a human continuum, raising complex questions about the kind of society we want to fashion as science changes and the possibilities for medical interventions increase.
Medscape: Just as an aside, our news group covered the Lancet article that retracted the study associating autism with vaccinations. Do you have any thoughts on this and the problem of convincing fearful parents to immunize their children?
Dr. Grether: It has always surprised me how many people confuse the various concerns that have been raised regarding childhood immunizations and risk for autism. The Lancet retraction focused on the MMR [measles, mumps, and rubella] vaccination and the allegations by a team of researchers in Great Britain about MMR playing a role in autism in some children. The MMR-autism connection has now been demonstrated to have been based on very shoddy, perhaps even falsified, evidence. There is no scientific basis that I am aware of for being concerned about MMR and autism. Another issue about vaccines is the thimerosal preservative that was included in multidose vials of some other childhood vaccines (thimerosal was not in the MMR vaccine). Here there have been many studies and no credible study has found a link between thimerosal exposure through vaccines and autism. In the United States, thimerosal exposure from vaccines is no longer an issue, except perhaps through some flu shots. The other concern that I have heard voiced is that the recommended childhood vaccine schedule now includes so many vaccines and that children are thereby being exposed to more antigens than previously. Although there are now more vaccines recommended for children, the total antigen exposure from today's vaccines is considerably less than in earlier years because of advances in vaccine composition. And compared to the sheer volume and diversity of antigens in a typical environment, the antigen dose received in vaccines is minor. While it is understandable that parents, and perhaps some clinicians, have raised the alarm about vaccines, I am not aware of any credible evidence to support these concerns.
Medscape: To conclude, and I think you really answered this already, do your results have any implications for personal decision-making or clinical practice?
Dr. Grether: You know, I wish they did, but I really don't think that they do at this point. The increase in risk that comes with older parents is really quite modest and we don't know what is behind it. I think the main benefit of our study and other similar ones is to alert us to an important etiologic clue, and it is now time to find out what is going on.
- Shelton JF, Tancredi DJ, Hertz-Picciotto I. Independent and dependent contributions of advanced maternal and paternal ages to autism risk. Autism Res. 2010;3:30-39.
- Reichenberg A, Gross R, Sandin S, Susser ES. Advancing paternal and maternal age are both important for autism risk [letter]. Am J Public Health. 2010;100:772-773