The Early Course of Autism

Understanding the natural history of a disease is an important framework to have. It not only is critical for prognosis, but also informs us about diagnostic and screening strategies, is important to assessing interventions, and provides clues to causation.

There has been much debate about the early course of autism, specifically the earliest age at which autism may be detected. At present scientific evidence suggests that autism is dominantly genetic, and so researchers expect that there may be early signs of autism even in infancy. Traditionally, however, autism is not diagnosed until age 2-3, when parents bring their children to medical attention, or when signs are detected on routine well-child visits or in day-care.

Retrospective studies, largely involving review of home movies, have suggested that autism can be diagnosed as early as 6-12 months, suggesting that parental report is not an adequate screen because subtle signs are hard to detect without rigorous observation.

Now a group has published the first prospective study to address this question. They followed 25 children who were later diagnosed with autism spectrum disorder (ASD) (22 of which were high risk) and 25 low risk children who were later determined to have typical development (TD). They found:

These results suggest that behavioral signs of autism are not present at birth, as once suggested by Kanner, but emerge over time through a process of diminishment of key social communication behaviors. More children may present with a regressive course than previously thought, but parent report methods do not capture this phenomenon well. Implications for onset classification systems and clinical screening are also discussed.

More precisely, they carefully assessed the children, counting instances of eye contact and smiling, for example, and found that there were no statistically significant differences between the groups at 6 months, but that almost all measures were reduced in the ASD group by 12 months.

The study is rigorously designed, and its primary weakness is that it is a bit small. The authors also acknowledge that future studies should utilize longer periods of recorded observation. I would add that more frequent assessments between 6 and 12 months would help specify when the earliest divergence takes place. Replication of the study is therefore desirable.

But what these results indicate is that clear signs of autism emerge between 6 and 12 months of age. Further, social skills tend to be regressive in ASD between 6 and 18 months of age. It was previously thought that social regression was less common in ASD, but this study suggests it is the rule, not the exception. Meanwhile, language skills did not regress in this study, they continued to improve in the ASD group, just on a slower curve than the TD group.

Further the study documented that the parents generally did not recognize the social regression between 6 and 18 months. The authors refer to the well-known phenomenon of telescoping in which patients or family members will recall the onset of symptoms as being much more recent than they actually were.

Prior studies using home movies have shown that signs of autism can be detected between 8-12 months. A study looking at head circumference found statistical differences prior to 12 months. And one study looking at movements found differences between 4-6 months. So it seems the consensus of current evidence is that objective and detectable signs of autism emerge between 6-12 months. This study does not support detection prior to 6 months, but other studies do suggest this might be possible.

This study has implications for diagnostic categories of autism (specifically distinguishing regressive forms of autism from non-regressive forms, since most of the ASD children in this study showed social regression). It also is very informative regarding screening strategies. It suggests no utility to screening children prior to 6 months of age. Further, since ASD children continued to separate from TD children through 3 years of age, screening for ASD should not stop at 2 but continue to at least 3.

The authors, however, do not discuss one very significant implication of this study (although an implication already raised by prior studies demonstrating early signs of ASD) – the observation made by many parents that ASD symptom onset correlates with certain vaccinations. Many children are diagnosed between the age of 2 and 3, during the height of the childhood vaccine schedule. This lends itself to the assumption of correlation and causation on the part of some parents. The phenomenon of telescoping, whereby memories of time contract, will tend to reinforce this false correlation.

What this and other studies show is that not only is the assumption of causation fallacious, the observation of correlation is likely flawed as well. The true onset of autism in most ASD children likely began a year or two prior to the vaccines that are blamed as the cause.

This point was made most dramatically by the Cedillo case – one of the test cases brought before the Autism Omnibus court alleging vaccine injury causing autism. Cedillo’s parents alleged that their child developed autism as a result of a combination of the MMR vaccine and thimerosal from other vaccines. In courtroom testimony, however, experts were able to show with home movies that Cedillo showed clear signs of autism as an infant, prior to ever receiving the MMR vaccine.

The current study adds nicely to the growing consensus that the true clinical onset of ASD is between 6 and 12 months of age. Whether or not there are biological markers of ASD prior to that remains to be seen, but is not unlikely. Early and fairly uniform onset is consistent with genetic causes of ASD, rather than environmental causes.


As several commenters have pointed out, I erred when I wrote that “Many children are diagnosed between the age of 2 and 3, during the height of the childhood vaccine schedule.” What I meant to convey was that this time period is still within the childhood vaccine schedule, and therefore it is likely that when parents first notice the signs of autism their child would have been recently vaccinated by chance alone.  The point remains, however, that many parents blame their children’s autism on vaccines they received after the true onset of symptoms.

Here I was thinking primarily of the MMR vaccine, the first to be blamed for autism (thanks to Andrew Wakefield). As you can see from the vaccine schedule, the first MMR vaccine in the series is scheduled for 12 months – after most children already have detectable signs of autism. This is why I used the Cedillo case to reflect this point.

Posted in: Science and Medicine

Leave a Comment (44) ↓

44 thoughts on “The Early Course of Autism

  1. Franklin says:

    Thanks for this very nice summary.

    Are there any studies as to whether some individuals would present at a later time? Is this type of disease usually uniform, or is there much variation across individuals? For example, could one child begin to show signs of autism at 9 months, where another wouldn’t begin until 24 months?

  2. Franklin,

    I’m not sure that autism is best understood as a disease.

    Autistics have strengths that NTs do not and some “very autistic” adults can function better than other “less autistic” adults.

    If you take autistic strengths into account, autism becomes a cognitive style with a spectrum rather than a disease. A cognitive style that is disabling at an extreme of the spectrum, yes. But not a disease.

  3. Watcher says:

    Wow, a very interesting study with both anti-vax ramifications as well as pinning down the developmental timeline. I can see many other clinical as well as animal model studies coming from this.

  4. KathyO says:

    I 100% understand that there’s no connection between vaccines and autism and have even written about that in my blog. However, I don’t see why this earlier onset of symptoms does anything one way or another about the vaccine correlation.

    My child is a teenager now, so I may not remember correctly, but it seems to me that she had an awful lot of vaccines during her first year, and definitely had some at two months. Maybe it wasn’t the MMR, but are the anti-vaxers that picky about which vaccines to blame?

  5. Enkidu says:

    KathyO: I was thinking the same thing. The “height” of the childhood vaccination schedule is actually 2 months-18 months, (they are vaccinated against 8 diseases alone at 2 months). At 2 years the only vaccine my daughter is getting is the flu shot.

    In the past, most anti-vaxers were all about blaming the MMR or mercury but now they blame everything starting with the HepB at birth.

  6. Nianya says:

    I agree that with regressive form, which my dd has, the symptoms continue to develop past 18 months; definitely through the 2s and 3s for us;

    I disagree about the speech issue though, since my dd had the severe speech issues associated with ASD and her speech never developed until she began to receive speech therapy at age 6.

  7. So why do anti-vax parents continue to refuse vaccines when their children are obviously health and well past the age of onset for autism?

  8. daedalus2u says:

    We know that a very reliable symptom correlated with autism is minicolumn morphology. Minicolums are the way that nerve cells are arranged, sort of like zillions of little microprocesssors arranged in parallel. People with autism tend to have larger numbers of minicolumns that are smaller, each with fewer and smaller neurons, but with larger total numbers of neurons in the whole brain. This is part of the association with larger brains. The number of minicolumns is fixed in utero during the first trimester. This is also the period at which some teratogens can “cause” autism, valproate, thalidomide and some others.

    A syndrome that was previously defined as an ASD, Rett Syndrome has a known cause. It is caused by deletion of the MeCP2 gene, which is on the X chromosome. Females have two X chromosomes, and those X chromosomes are inactivated in a random pattern so that females are mosaic with respect to which X chromosome is activated in their cells. The MeCP2 protein is involved in regulating the readout of methylated DNA, that is DNA that has been epigenetically programmed to be read out differently over the lifespan of the individual cannot be in the absence of the MeCP2 protein.

    Rett Syndrome is present in utero and at birth, it only becomes apparent in early life, at about the same age as autism is observed to become apparent. There have been animal models of Rett Syndrome produced which recapitulate the symptoms of the disorder extremely well, including even stereotypical paw to mouth (hand to mouth) movements. When the MeCP2 gene is reactivated in these animals, the Rett Syndrome symptoms go away. There is essentially complete rescue from the MeCP2 deletion phenotype.

    My inference from this is that there are at least two factors in autism, the neuroanatomy (including epigenetic programming) that occurs in utero, and how physiology regulates that neuroanatomy later in life, very likely starting around the time autism symptoms become apparent.

    What the animal MeCP2 model shows is that the neuroanatomy is intact, even after quite severe symptoms are expressed. There are similar studies in humans showing temporary resolution of some ASD symptoms under certain conditions, fever for example, and in an in press PNAS paper discussed by Ed Yong on SB’s “Not exactly rocket science” nasal oxytocin.

    There are reports that some individuals do have a minicolumn morphology similar to that of people with autism, but who do not have autism (scientists for example). The problem is that minicolumn morphology can only be looked at postmortem. I suspect that such individuals have the in utero exposure that leads to the autism phenotype, but do not have the postnatal physiology that perpetuates the autism phenotype.

    I think this is tied up in nitric oxide physiology.

  9. dt says:

    Nice summary.
    One question though. You say: “Early and fairly uniform onset is consistent with genetic causes of ASD, rather than environmental causes.”

    Could there not be environmental causes to which the pregnant mother or the neonate are exposed to which are relevant or play some part?

  10. BKsea says:

    An interesting study, but it seems to me that comparing 25 high-risk infants to 25 low-risk infants opens the door to all sorts of confounding factors. I am curious why they did not compare 25 high-risk infants who went on to develop ASD to 25 HIGH-risk infants who went on to have TD.

  11. Scott says:

    An interesting study, but it seems to me that comparing 25 high-risk infants to 25 low-risk infants opens the door to all sorts of confounding factors.

    They didn’t, and Dr. Novella was mistaken to state that they did. They compared all 25 who developed autism to 25 low-risk infants. Of the former 25, 22 were high-risk and 3 low-risk. This is made clear in the abstract:

    A prospective longitudinal design was used to compare 25 infants later diagnosed with an autism spectrum disorder (ASD) with 25 gender-matched low-risk children later determined to have typical development.

    Note that there’s no mention of high-risk status in the former group. Table 1, of course, makes it crystal-clear by giving the actual risk breakdown.

    Of course, that doesn’t answer the question of why they didn’t attempt to match risk status as well – in particular, it seems plausible that learning from an older sibling (with ASD in the high-risk group, without in low-risk) would have some relevance.

  12. qetzal says:

    @BKsea & Dr. Novella,

    I looked at the paper itself (linked in Dr. Novella’s post), and it wasn’t actually 25 high-risk infants. Selected bits from the methods section:

    The sample reported in this paper was drawn from a larger longitudinal study of infant siblings of children with ASD (High-Risk group) or typical development (Low-Risk group), recruited at two sites (UC Davis, n = 33; UCLA, n = 17). The sole inclusion criterion for the High-Risk group was status as a younger sibling of a child with ASD….

    Of the 25 infants with ASD outcomes, 19 had older siblings with Autistic Disorder, 3 had older siblings with PDDNOS, and 3 were from the Low Risk group and did not have an older sibling with ASD….

    In this paper, we report on all infants at both sites who met criteria for an ASD outcome, as defined above, at 36 months of age (n = 25). They are compared to 25 infants who met criteria for Typical Development (TD) at 36 months, randomly selected from the Low-Risk group after matching the gender ratio to the ASD outcome group.

    Note that 3 of the ADS infants actually came from the low risk group (kids whose older siblings did not have ASD). However, I can’t find anything in the paper that addresses whether these 3 were notably different than the 22 ASD infants from the high rist group. (Not surprising, since n=3 would be basically useless anyway.)

    I agree it would be really interesting to have a third group: infants from the high risk group that showed no signes of ASD. I don’t know why they didn’t look at that. Maybe most of the kids in the high risk group are still too young to be sure they won’t develop ASD?

  13. qetzal says:

    Sorry for the botched quote tags above. The text that begins “Note that 3…” is mine, and should have been outside the quote box.

  14. BKsea says:

    Scott – thanks for the clarification. I think the concern about confounders still stands. In addition to your point about learning from older siblings, I also wondered if parents with older ASD kids have less time to spend with infants and whether genetic predispositions can yield developmental differences without necessarily leading to ASD.

  15. Scott says:

    I agree, the concern about confounders certainly does still stand. It just takes a different form.

  16. wales says:

    Others who have commented that the majority of vaccinations are given earlier than 2-3 years of age are correct. See the ACIP recommended schedule link below, the majority of vaccinations are administered in the 2 month to 15 month age range.

  17. Thanks for the clarifications. I did make that mistake. I will make corrections in the original.

    The issue of timing of ASD onset with vaccines – the point was that when parents notice ASD (usually 2-3 years of age) they sometime correlate with the vaccines given at that time, including MMR. This correlation is false.

    Also – there is no longer anything but trace thimerosal in the routine vaccine schedule, so that should not be an issue either.

    But the anti-vaxxers have retreated to vague mentions of vaccine toxins, so any vaccine is game.

  18. regarding the confounders, these are legitimate points, but I do not think they would explain the regressive social skills. Typical children never regress.

  19. pmoran says:

    “I don’t see why this earlier onset of symptoms does anything one way or another about the vaccine correlation.”

    It is important for public perceptions. We take it for granted that mothers are highly perceptive as to their child’s progress and will be stating a fact if they say their child was normal until they received certain vaccines. Such stories triggered the whole MMR debacle, I think.

  20. wales says:

    I reviewed the cited paper. From the discussion section: “Collectively, the present investigation and recent prospective studies suggest that signs of autism emerge over the first year or so of life in many children with ASD, rather than being present from close to birth, as once suggested by Kanner.”

    Dr. Novella states that this study weakens the correlation between vaccines and autism. I don’t understand the rationale for this line of thought given that peak vaccination occurs in the 2 month to 15 month range.

  21. daedalus2u says:

    The whole premise the study is based on, that siblings of children with autism are a high risk group is incompatible with a vaccine causation hypothesis.

  22. daedalus,

    I think the response to that is “biodiversity.” As in, kids who were vaccinated and then become autistic have a special genetic susceptibility to vaccine damage. Kids who were vaccinated and remain NT are the majority and drown out data from the biodiverse vaccine-susceptible kids so that vaccine testing seem to show that vaccines are safe when they are actually unsafe for a minority of special kids.

    And kids who were not vaccinated and become autistic anyway… well they aren’t very many of them, and a small number of kids were probably autistic before vaccines existed, so it’s probably the same phenomenon. Anyway we don’t think about them very hard.


  23. Dr.Jon says:

    To clarify about vaccines and timing – The “live” virus vaccines, varicella, MMR etc.. are not first given until 1 year of age, hence the perceived correlation with ASD.

  24. Plonit says:

    The article still isn’t 100% right, because some of the children later diagnosed with ASD were from the low-risk group (non-sibling group) – the logical corollary of some of the high-risk (sibling group) going on to be diagnosed with ASD.

    There’s something about this study that gives me pause, and I think I was alerted by your formulation that they “followed 25 children who were later diagnosed with autism spectrum disorder (ASD)” alongside the description of the study as prospective.

    So, immediately I think “huh? How does one know to follow the 25 children who will be diagnosed with ASD?” Of course, they didn’t: they followed 50 children in a prospective study, 25 of whom happened to later to be diagnosed with ASD. But the weirdly back-to-front description (prospective study….of those later diagnosed) comes from the study authors and not Novella.

    The rate of diagnosis (1:2) is very different from the real world, because researchers have drawn from a population where they expected very high rates of ASD diagnosis. But I would like to see some real world data on the likelihood of a younger sibling of someone diagnosed with ASD also being diagnosed with ASD. In real world situations, is it as high as 22:25 (88%)?* If so, that’s reassuring (in terms of the study validity), but if it’s lower than I wonder about the issue of blinding in a study like this. The only mention of blinding is in relation to ASD diagnosis, and only in one validation procedure. There is no mention of blinding researchers assessing children to the risk-group status of the participant children, so I wonder if that is an issue – especially since coding is subjectively perceived (the observer is the instrument). It’s just a thought.

    *My back-of-the-envelope stuff also reminds me that confidence intervals are always nice to have.

  25. Plonit says:

    I apologise, on closer reading, assessors were unaware of the group membership.

  26. Plonit says:

    Actually, on closer reading – is this truly a prospective study?

    I’m really unclear on how they’ve selected participants for analysis from the larger study into which this one was nested.

  27. Scott says:

    They measured the participants prospectively. Then, at the predetermined stopping point, checked how many of them developed autism, selected an equal number of sex-matched low-risk-controls, and did the analysis.

    So it’s prospective, but with group categorization taking place at the end of the observation period instead of the beginning.

  28. passionlessDrone says:

    Hello friends –

    Interesting study, but the acrobatics necessary to contend that the ‘majority of the vaccine schedule’ is between two to three years shows either someone didn’t know the first thing about our vaccine schedule, or intentional wordsmithing for purposes unknown.

    The bulk of our increase in the vaccination schedule has occurred in at the two, four, and six month appointments over the past two decades. In fact, of the recommended vaccines, only two, MMR and varicella are recommended to start after the six month timeframe; the rest, Hep-B, HIB, DTaP, Rotavirus, Polio, Pneumococcal and Flu have all been given at least once, and up to three times by one year of age. These are simple to validate facts.

    Why not just have a discussion of the study without the need to defend the (understudied) vaccine schedule?

    – pD

  29. PD – I wrote an addendum to clarify the point I was trying to make. It was an unfortunate choice of words.

  30. edgar says:

    With my epi hat on,
    This just occurred to me, so bear with me, but maybe there is some link between vaccines and autism? Maybe an autistic child behavioral reaction to mild vaccine ickiness is significantly different from non-autistic, or at least significantly different from their normal behavior, thus leading parents to think it is a ’cause’?

  31. David Gorski says:

    It figures. J.B. Handley has characteristically gone on the attack:

    Can anyone count the number of fallacious arguments in his post?

  32. Matt says:

    There is a lot of speculation going on about vaccines. My quick read of the paper didn’t have any mention of vaccine status of these children.

    Most of the ASD kids were baby siblings of other ASD kids. The families are working with MIND and UCLA.

    It is not a stretch to assume that some fraction of these children were alternately or un-vaccinated.

  33. dt says:

    I see Handley admits that autism is not usually diagnosed after vaccine shots (yet still makes the erroneous causal claim):

    It is exceptionally rare that I hear the story, “my son was 100% fine, and at 2 years old after one vaccine appointment he lost everything.” I have heard that story, but very rarely.
    More commonly, I hear from parents about a chronic slide into autism with a progression of health issues accompanying the slide.

  34. passionlessDrone says:

    Hi Edgar –

    You might be interested in knowing that children with autism have been shown time and time again to respond with a much more robust innate immune resonse when compared to their non autistic peers. By way of example, you might consider Enstrom, Jyonouchi, or Ashwood.

    The luxury we have always had in the fact that we’ve only really studied the MMR or thimerosal regarding autism is that we really couldn’t think of a mechanism by which other vaccines, given much earlier and in combination, might be affecting our infants in ways that might affect future neurodevelopment outside of immediately obvious, difficult to misinterpret, adverse reactions. It seems, however, that the immune system is to some extent, maleable in during some times of development; for a less controversial example, check out the relationship between getting RSV as an infant and having a asthma or wheezing later in childhood.

    It turns out, we have many studies on the effect of immune challenges in animal models that show sublte but real effects on behavior, seizure succeptibility, and immune function. These changes have been shown to have variety of contributory factors, including time dependence, ties to specific inflammatory cytokines, and TLR selectivity. For a good review paper, you might check out Early-life programming of later-life brain and behavior: a critical role for the immune system

    – pD

  35. Matt says:

    So Handley shifts goalposts again. This is a surprise?

    Does Wakefield’s Lancet study, claiming regression within days of MMR is incorrect?

    One of the key questions Handley claims as a benchmark for his “fourteen studies” website is:

    Did the study actually contemplate the real world example of a parent vaccinating their child with 5 or more vaccines and then seeing a regression into autism afterward?

    Handley’s day is over. It was before this, and it is even more true now.

  36. superdave says:

    I mean this with no sense or irony, sarcasm, cynicism. Handley’s rreply to this post saddens me. That a professional adult reached the level he reached in that reply is truly demoralizing. I don’t think I even want to read AOA any more, it’s like watching a turtle struggle to get off it’s back.

  37. cedge20 says:

    Oh, for goodness’ sake. The original study described makes perfect sense to me as the parent of a teenager with HFA/Asperger’s. We were doting parents, but we did notice signs at 1 yr that he wasn’t relating to us like other children might. He did not spontaneously smile us waving through the window and smiling; it was more like he studied us.

    Elizabeth Moon, in Speed of Dark, has a great line: All babies are autistic; some grow out of it. I think this has some application to the study; you see this divergence of behavior at a certain point. It may correlate to other environmental factors, but they aren’t the cause of it.

  38. BillyJoe says:

    “In fact, of the recommended vaccines, only two, MMR and varicella are recommended to start after the six month timeframe;”

    Actually, that is 4 vaccines (measles, mumps, rubella, and varicella) ;)

  39. BillyJoe says:

    Steven Novella said:
    “Here I was thinking primarily of the MMR vaccine, the first to be blamed for autism (thanks to Andrew Wakefield). As you can see from the vaccine schedule, the first MMR vaccine in the series is scheduled for 12 months – after most children already have detectable signs of autism. ”

    There is already a perfect riposte to this on the web:

    “…we have believed that mercury is seriously implicated in the worldwide epidemic of autism. MMR has been no more than a bit player in this saga. No doubt to the satisfaction of the small coterie of companies who comprise the huge international vaccine cartel which has known all along that the real culprit DTP, given at two, three and four months of age does the real damage.

    Unfortunately for parents and babies MMR, given at about twelve to fifteen months, can coincide with the visible onset of autism thus cloaking the true cause of harm. MMR could probably tip the balance in the case of a child seriously harmed by DTP but whom so far has escaped the worst of vaccine damage.”

    So, you see, it’s DTP, not MMR, and your whole post actually supports the vaccine-autism link!
    How ignorant are you then?

  40. annejaa says:

    Well people with autism have issues with non-verbal communication,a wide range of social interactions and activities that include an element of play.If the symptoms are not severe,the person can be taught that eye contact is important for most people and he/she will remember to look people in the eye.You have done a good job and i will share this information with my friends too.
    hp laptop coupon

Comments are closed.