Some readers may have seen the "Dateline" program "A Dose of Controversy" shown on Sunday, Aug. 30, 2009. This program examined the claims made by a British physician, Andrew Wakefield, about an association between vaccination of children with a combined measles, mumps, and rubella vaccine (MMR) and their later diagnosis as autistic. Wakefield's work is the basis of a movement to refuse vaccination for children which resulted in an abrupt drop in vaccination rates in the United Kingdom. The "Dateline" program also discussed studies on large numbers of children that failed to show any connection between vaccination and autism, strongly suggesting that Wakefield's conclusions were incorrect.
Of course, it is extremely frustrating to parents of autistic children-- and to other concerned community members--- to be told that no one really understands what autism is about, but that an idea that has been offered is certainly not true. But this is the nature of scientific investigation: it is possible to reject a hypothesis on the grounds that there is no evidence to support it, without at the same time being able to support another hypothesis. It is very rare to have a situation where the rejection of one hypothesis is combined with support for an alternative hypothesis. However frustrating people may find it, this is not one of those cases.
I want to lay out some ideas that I think are necessary for a thorough consideration of autism. The first is that there is no reason to assume that all children diagnosed as autistic have exactly the same basic problems. Failure to speak or communicate in other ways, lack of interest in other people, perseverative play or attentiveness to small sensory stimuli like flashing lights--- these can come about for more than one reason. Decades ago, all children with these characteristics were classed as autistic, but more recently variants of these developmental problems have been divided into a number of distinct Pervasive Developmental Disorders, of which one has been identified as involving a specific genetic problem. Although we retain the autism diagnosis, future work may find that there are distinct categories within the group of children who now receive that diagnosis. For example, work currently in progress is pursuing the idea that some children diagnosed as autistic show symptoms from the very early months of life, while others develop normally for a time and then regress, losing communicative abilities they had shown earlier. These two groups may have problems with very different causes and mechanisms.
A second point is that autism remains a relatively rare disorder. It takes a very large sample of children to give us a large enough group of autism diagnoses for any statistical conclusions to be drawn. The conclusions that can be drawn from small samples are not as meaningful as those from large groups. For example, suppose we were to look at a small-town hospital where only 20 babies were born in a year. What if we saw that 15 of the 20 in 2009 were boys? We would not be likely to conclude that the country was going to be in trouble, because so few girls were being born. We would realize that the small sample of births was not likely to be representative of the gender proportions for births throughout the country. Similarly, Andrew Wakefield's conclusions, based on the testing of 12 children, cannot be claimed to represent the whole population of children--- whereas a contradictory study involving almost 2 million children can be considered representative.
A third point is that ongoing research in child development has revealed phenomena that were unknown at the time when autism was initially described. For example, we now know that normally-developing infants of 7 or 8 months are likely to communicate through "joint attention"-- that is, they use the gaze ("eye contact") to signal to an adult the wish that he or she would look at something the baby finds interesting. The adult looks, the baby looks, they look back at each other and smile-- everybody enjoys this communication, which begins months before most babies have much capacity for using or understanding language. A failure to develop joint attention is a signal of developmental problems that occurs much earlier than the usual symptoms of autism. What other, earlier events will be discovered in the near future, giving us a better understanding of differences between typical and autistic development? We still have much to learn about both, and are not likely to come to any permanent conclusions on the basis of what we know today.
In fact, jumping to such conclusions is more likely to do harm than good to both autistic and typically-developing children. The evidence seems to be that Andrew Wakefield and his followers jumped to their conclusions rather than building them carefully on a foundation of reliable information.