You are here:
-A A +A

IAN Research Report #6: Regression

Date Last Revised: 
September 5, 2008
Date Published: 
June 27, 2008

The Interactive Autism Network (IAN), a project collecting information online from families of children with autism spectrum disorders (ASDs) from throughout the United States, reports on developmental regression in autism. What does current research say about loss of language, social, motor, or living skills in children with ASDs? How many families participating in the IAN Project report that their child experienced such a loss?

Regression in Autism Spectrum Disorders

The term regression is often used in connection with autism. What does it mean? How often does it occur? And what are the implications for our understanding of the autism spectrum?

For scientific on. To skip to what IAN families reported about regression, click here.

Regression Defined

In a general sense, regression refers to any return to an earlier, less developed state. When speaking of developmental disorders and young children, regression refers more specifically to a loss of previously acquired skills, such as language, motor, or life skills.

Such a loss is associated with a variety of conditions. These include Landau-Kleffner syndrome -- also known as acquired epileptic aphasia -- and blindness from birth. 1,2,3  Rett's disorder and childhood disintegrative disorder -- which, like autism, are forms of pervasive developmental disorder, although much more rare -- also are characterized by a loss of previously acquired skills.4  It may be autism, however, that is most widely recognized as a cause of regression, especially in the realm of language. A study that selected children with language regression, and only then identified their diagnoses, found that 88% of them met criteria for autism or manifested autistic features.5  It is no wonder that a child's loss of previously acquired words has been described as a "useful red flag for ASD in a significant minority of cases." 6

Regression Research

As is implied by the phrase "a minority of cases," regression does not occur in all children with autism spectrum disorders. Some children never lose skills. Researchers have sought to better define how often regression occurs, what type of skills are most commonly lost, and if children with ASD who regress and those who do not regress differ in significant ways.

The first important issue was whether regression could be documented. Everything that was known about regression was based on parents' reports about what they remembered of their child's developmental history. Would impartial observers be able to validate these parent reports? Using home videotapes of children's first and second birthday parties, one research team set out to demonstrate whether regression occurred. They had individuals who had no idea which children were autistic (or not) or which children had regressed (or not) evaluate the videos, observing the children's communication, social interaction, emotional connection, repetitive behaviors, and play. They found that regression did indeed occur: "Analyses revealed that infants with ASD with regression showed similar use of joint attention, and more frequent use of words and babble, compared with typical infants at 12 months of age -- skills they had clearly lost by their second birthday." 7

A second group, also using home videotapes, confirmed these findings, and found that both parents and independent observers who watched and coded the videotapes were in close agreement about the onset and loss of expressive language, although there was minimal agreement regarding loss in nonlanguage areas like social or motor skills.8

Although these researchers sought to document the existence of regression, others worked to better describe it. For example, researcher Sally Rogers of the U.C. Davis M.I.N.D. Institute described three patterns of autism onset: 9

  1. No period of typical development -- autism symptoms present from birth
  2. Developmental plateau -- child begins to reach developmental milestones and then stops acquiring new skills
  3. Regression -- child loses skills

How many children develop normally prior to a loss of skills, and how many do not, has been the focus of much recent research.

A multisite study conducted by the Collaborative Programs for Excellence in Autism (CPEA) studied 351 children with ASD in the hope of learning more about skill loss. In the sample, 46% of children with ASD were reported to have regressed. The majority of those who experienced regression had not developed typically prior to losing skills, exhibiting what has been called "the delays-plus-regression phenotype." 10  Still, nearly 30% of those who regressed were reported to have been developing normally prior to the loss.11

Researchers at Yale, reporting on a sample including 237 children with autism, 199 with other ASDs, and 137 with other non-ASD developmental disabilities, and using a very strict definition of regression, found that only 11.8% of children with autism and 5.5% of children with other ASDs had experienced a "clear or possible loss" of skills, while an even smaller number had experienced a developmental plateau.12

Researchers also began to explore what types of skills children were losing. Did regression occur only in the domain of speech and communication? Or were there other types of losses? Studies revealed that loss of language was the most frequent type reported, but not the only one.

The Childhood Autism Risks from Genetics and the Environment (CHARGE) Study reviewed the cases of 333 children with autism or ASD to learn more about skills lost. They found that 59% of the children had lost no skills, 26% had lost language or social skills, and 15% had lost language and social skills.13

In the Yale study mentioned previously, 66% of the children whose parents reported regression lost words or stopped talking; 50% lost eye contact or interest in social games or other people; and 34% began to engage in stereotyped behaviors (like hand flapping) or became irritable or anxious. A small number (3%) even lost the ability to walk or climb stairs, illustrating that motor skills can be impacted.14

Is there a Regressive 'Type' of Autism?

As researchers considered regression, they began to wonder: Were children on the autism spectrum who experienced regression different in any way from those who had no such history? Was there some special type of autism characterized by regression that was different from other types of autism? Identifying subtypes of autism was an important goal -- would regression be a factor that helped distinguish one type of autism from another?

So far, the answer has been mixed.

The CHARGE study investigated whether there were any differences between children who lost one type of skill and those who lost more than one, but researchers found none. They also reported there were few "developmental, demographic, or medical differences" between those who had regressed and those who had not.15  Similarly, groups studying regression at Yale, the University of Michigan, and the University of Washington did not find any difference in severity of autism between children who had regressed and those who hadn't. 16,17,18

Researchers at CPEA, meanwhile, focused on finding out whether there was evidence of some larger subtype of autism, perhaps consisting of a group of factors including regression. In part, they were responding to a theory of autism, later retracted by most of its authors, that proposed a link between the Measles-Mumps-Rubella (MMR) vaccine, gastrointestinal problems, and regressive autism.19,20  The CPEA group did find some differences between children who had regressed and those who had not. Although they found no evidence that regression in autism was associated with MMR vaccination, they did find that children in their study who had lost skills had poorer verbal IQs and social reciprocity, and more gastrointestinal problems, than children who had no reported history of regression. Few of the children who regressed had experienced normal development prior to the regression, and those who had were not the same children who had experienced GI symptoms. The notion of a regressive phenotype of autism characterized by near-normal development prior to loss and GI symptoms was therefore not supported by the study. 21 

A study led by Columbia University investigated similar matters, finding a higher rate of regression in children with autism and GI issues compared to that reported in the literature for children with autism in general. They were investigating possible links between measles virus in the GI tract (ostensibly from the MMR vaccine) in children with GI problems only and children with GI problems and autism. They found no difference between the two groups, but did find that 88% of the children with autism and GI issues in their study had experienced regression.  "Autism with GI disturbances is associated with elevated rates of regression in language or other skills," they reported, "and may represent an endophenotype distinct from other ASD."22 

Some consensus about regression in autism is beginning to emerge. In most studies, 40% to 50% of children appear to lose skills, but the great majority of them show signs of atypical development prior to the regression. Regression, if any, usually manifests itself in the realm of language, but losses in social, motor, and other skills have been noted. Most researchers have not found major differences between children who regress and those who do not, although at least two studies have noted more gastrointestinal problems among children with autism who regressed.

Interactive Autism Network: What Do IAN Families Report About Regression?

The Interactive Autism Network (IAN) is collecting information over the internet from thousands of families of children with ASD throughout the United States. What have these families been reporting about regression in their children with ASD? Does the information they are sharing confirm, refute, or expand upon what other researchers have found?

The following questions were asked:

  1. Did your child lose words, daily living skills, motor abilities, or social skills that he/she previously had?
  2. How significant was your child's loss of skills?
  3. Which type of skill was affected most (speech and language, motor abilities, social skills, or daily living skills)?
  4. What age was your child when you first noticed this loss of skills?
  5. Did your child's development plateau or halt such that he/she stopped gaining new skills but retained previously acquired skills?
  6. What age was your child when you first noticed this plateau or halt in his/her development?

Researchers investigating autism-related regression are careful to define exactly what constitutes a skill loss that is linked to autism and not to something else. Their aim is to confirm that autism-related regression occurred based on clinical observation by care providers and reports from parents.

In contrast, the IAN data is based on parent report only. It should also be noted that we did not ask: "Did your child experience autism-related regression?" We asked about skill loss and plateaus. This made our questions more family-friendly, but also means that types of skill loss that do not represent autism-related regression may have been reported. For example, there are parents who indicated that their child lost skills after a major emotional trauma; this can occur with typically developing children, too, and would probably not be considered autism-related. Still, most IAN parents reporting skill loss and developmental plateaus described them occurring in children under age three, and in the communication domain. This is a good indication that the questions above did capture autism-related regression in most cases.


We began by looking at what age children had been reported to lose skills. It soon became clear that, no matter the specific ASD diagnosis, the majority of them had lost skills between 13 and 18 months of age. (See Figure 1.) Even some children with Asperger's syndrome -- a diagnosis not commonly associated with regression -- were reported to have lost skills by IAN parents.

Figure 1.
Line graph showing age at skill loss by ASD diagnosis.
Note: ASD (Autism Spectrum Disorder) and PDD (Pervasive Developmental Disorder) are "catch-all" diagnostic categories sometimes used as a preliminary diagnosis. These are often later replaced by more specific diagnoses, such as Autism or Asperger's syndrome.

As we examined our data, we realized that some parents were reporting skill loss taking place well after their child reached 36 months of age. According to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, (DSM-IV), children with an autism diagnosis are supposed to have shown symptoms before age 36 months. (American Psychiatric Association, 2000) In the literature, furthermore, regression has been most frequently reported to take place during the toddler years. To understand what type of skill loss parents were reporting in older children, we turned to the comments that parents are able to leave for researchers when they fill in the IAN questionnaires.

From parents' comments, it appears that this late skill loss is something very different from what is typically thought of as "regression." It does not generally involve language loss, but involves losses in social functioning. Some parents reported the increasing distress that can impact higher-functioning children on the spectrum when their peers' social development begins to outstrip their own. Unfortunately, this was perhaps not so much about a loss of skills as about falling ever more behind in social functioning compared to typical peers, a growing realization of difference, a loss of self-esteem, and the struggles that accompany this:

  • "He went from being cheerful and full of smiles to being sarcastic and sullen to others." -- of a child with PDD-NOS who had difficulty at age 5
  • "Lost social skills, or his anxiety over social failure increased to the point he no longer had the confidence he once had had." -- of a child with Asperger's who had difficulty at age 6
  • "My son has lost social skills and the ability to cope with stress. He used to be very social and always had friends. He now has very low self-esteem, is very negative, and does not make friends easily." -- of a child with Asperger's who had difficulty at age 9

These losses and struggles of older children are an important topic in their own right. They do not truly represent regression as defined by most researchers, however. In order to compare our figures for regression with those of other researchers, we decided to focus only on skill loss reported at 36 months of age or earlier. That way, we would avoid mixing regression with other varieties of loss experienced by older children. (See Figure 2.)

Figure 2.
Line graph showing skill loss by ASD diagnosis by age of 3.

IAN data confirms that nearly half of children with a diagnosis of autism experience regression before age 3. Children with other autism spectrum diagnoses also experience regression, although to a lesser degree. (See Table 1.)


Table 1. Regression by ASD Diagnosis
Autism Spectrum Diagnosis

Number of
Children with Skill
Loss 0-36 months

Percentage of Children
with Skill Loss
0-36 months

Autism (n=2622)



PDD-NOS (n=1220)



ASD (n=524)



Asperger's syndrome (n=973)



PDD (n=174)



All (n=5513)



Parents of children with autism were the most likely to describe the regression as severe, and the majority of parents of children with other diagnoses rated the regression as moderate. (See Figure 3.)

Figure 3.
Bar graph showing regression severity by ASD diagnosis.

As would be expected from previous research, language was the skill most frequently identified as "affected most" for all children on the spectrum, followed by social skills. Even some children with Asperger's syndrome, who supposedly do not experience speech delays, were reported to have lost speech or language skills. (See Table 2.)


Table 2. Of Those Who Were Reported to Regress,
Lost Skill that Was Affected Most: Autism, PDD-NOS, Asperger's



Asperger's Syndrome





Social Skills
(like eye contact and playing with other children)




Motor Abilities
(like walking, jumping, and playing with small toys)




Daily Living Skills
(like feeding oneself and toileting)




Table 2 shows only what skill was affected most, and does not address the question of whether multiple skills were affected. A number of parents wrote in the comments section that all these skills were affected at the same time, so it was difficult to choose among them.

Developmental Plateaus

Some families did not report out-and-out regression, but a developmental plateau -- one of the types of autism onset described by Rogers. How common was this among IAN families?


Developmental stagnation was reported far less often than loss of skills, but it was definitely reported. More than 900 children were reported to have reached a developmental plateau, with nearly 800 experiencing such a plateau at age 36 months or younger.

Would the age at which children reached a developmental plateau be the same or different from the age at which children were reported to regress? Based on the IAN data, it appears that the typical age for regression and developmental plateau is similar. Just as for regression, the most typical age for a reported developmental plateau was between 13 and 18 months of age.

Looking at loss and stagnation of skills together we found that of the 5,472 children whose parents answered the "lost skills" or "plateau" questions, and who were age 3 or younger when a loss or stagnation of skills occurred, 2,502 (46%) experienced neither a developmental regression nor plateau; 2,114 (39%) experienced regression; and 856 (15%) experienced a plateau. (See Table 3.)


Table 3. Regression vs. Plateau
Child's Status

Number of Children
(through 36 mos. of age)

Percentage of Children
(through 36 mos. of age)

Neither Regression nor Plateau



Regression Only



Plateau Only



Total with Regression or Plateau






All together, 54% of the children with ASD participating in IAN have been affected by some kind of loss or stagnation of skills by age 3. As we have seen, this varies to some degree by ASD diagnosis. (See Figure 4.)

Figure 4.
Bar graph showing regression or plateau by ASD diagnosis.

In Conclusion

According to the reports of IAN families, 39% of children with a variety of ASDs, and 49% of children with a diagnosis of autism, experience a loss of skills by age 3. These figures are in the same range as those reported by the CPEA study (46%) and the CHARGE study (41%). In contrast, the Yale study, using a very strict definition, reported a lower rate: 22.4% for children with autism by parent report, but 11.8% if the strictest definition possible was used.


IAN's data also demonstrate the importance of including numbers on children who experience a developmental plateau such that they stop gaining skills without actually losing any. Fifteen percent of children participating in IAN were reported to have hit such a developmental plateau, which brings the total number of children experiencing a developmental loss or halt to 54%.

IAN's findings also confirm the type of skills that are most frequently lost. No matter the ASD diagnosis, speech and language was the skill area reported as "most affected," followed by social skills. As in other studies, a small number of families reported loss of daily living skills or motor skills.

The IAN Project plans to gather more data in order to make possible additional exploration of issues surrounding regression. For example, a questionnaire on gastrointestinal issues is under development. Once deployed, we will be able to see whether we, like the researchers at CHARGE and Columbia University, find a link between regression and GI issues.

As we examine all the information we have on skill loss or stagnation, it becomes clear how important it is for those studying these issues to precisely define what constitutes a regression or plateau. Do losses in only certain skill categories count? Should children with Asperger's be included in studying the phenomenon, or does their participation "wash down" the numbers? (In our study, children with Asperger's were far less likely to regress than children with other ASD diagnoses.) Should only certain age categories be considered? And if there are losses of some type occurring at later ages, how should these be characterized and studied? It is our hope that the IAN data on developmental regression and plateaus -- currently based on the experience of more than 5,000 children with ASD -- will help researchers answer these and other questions.

For More Research Findings and Features...

Read these articles recently published here, at IAN Community:

Please Note: These Findings Are Preliminary


The analyses presented here by the Interactive Autism Network are preliminary. They are based on information submitted via the Internet by parents of children with autism spectrum disorders (ASDs) from the United States who choose to participate. They may not generalize to the larger population of parents of children with ASDs. The data have not been peer-reviewed -- that is, undergone evaluation by researchers expert in a particular field -- or been submitted for publication. IAN views participating families as research partners, and shares such preliminary information to thank them and demonstrate the importance of their ongoing involvement.

We encourage autism researchers investigating these topics to apply for access to the IAN database. Contact

View References for this Article

Please rate the helpfulness of this article: 
Average: 4 (1 vote)