Number Needed to Treat
Here's an illustration of how the NNT concept works: Let's suppose somebody invents a new drug to prevent teenagers from becoming obsessed with gangsta rap. The manufacturer advertises that teenagers who are treated promptly with the drug when their parents first catch them listening to gangsta rap are 50 percent less likely to develop GROD (Gangsta Rap Obsessive Disorder). Sounds pretty good, doesn't it? Parents who value their hearing, their sanity, and household peace should all rush their teenagers to the doctor, right?
Well, no. Not exactly.
In this hypothetical scenario, what the manufacturer hasn't mentioned in the advertisement is that, out of 100 teenagers who listen to gangsta rap, only 4 of them become obsessed with it. Taking the new wonder-drug reduces that number to 2, making a gangsta rap obsession "50 percent less likely." As you can see, that statistic would seriously mislead parents because it was never very likely that any particular teen would become obsessed with gangsta rap. Because 100 teenagers had to be given the drug to prevent 2 instances of GROD, we would calculate the NNT for this particular drug as 50, meaning that the drug had to be administered to 50 people in order to benefit one.
You can find a more detailed explanation of the NNT concept in a recent article on the Slate website.
What does all of this have to do with autism? Well, we often see claims by biomed proponents that a particular treatment, supplement, or what-have-you is 50 percent effective at "rescuing" children from a lifetime of being trapped in their autism and unable to communicate. Leaving aside the awkward fact that not a single one of these claims has ever been supported by legitimate peer-reviewed studies, let's just wander back into the Land of Unlikely Hypotheticals for the moment and imagine that someone actually did perform a valid study confirming such a claim. What would it mean for parents who had just been told that their young child was on the autistic spectrum?
As I mentioned in a recent post, research has shown that approximately 90 percent of children on the autistic spectrum develop speech by age nine. (C. Lord et al, "Trajectory of language development in autistic spectrum disorders," in Developmental Language Disorders: From Phenotypes to Etiologies, 2004.)
Several older studies were discussed (thanks, Joseph and David) in the comments to that post, indicating that the percentage of Kanner autistics who developed speech by adulthood was between 30 percent and 50 percent. Not all autistics who lack functional speech as children are in the Kanner category, but many are. And we know that in past years, such children often were institutionalized at a young age, with no speech therapy and no reasonable opportunity to learn speech. According to the National Autistic Society, the ratio of Kanner autistics to the entire autistic spectrum is about 5:91 (this is based on studies conducted in the UK). So I think it's reasonable to estimate that at least 3 percent of today's autistic spectrum children will learn to speak after the age of nine.
There are also autistics who rely on alternative methods of communication. Facilitated speech has gotten a significant amount of media attention, but there are many methods. Unfortunately, some non-speaking autistics who could benefit from augmentative communication devices have been unable to get the devices as a result of being "written off" by ignorant people who assume that they are incapable of learning. As more advanced technology becomes widely available and society's attitudes toward autistic people improve, I expect that more autistics will be able to benefit from these devices. At present, however, I do not believe there are any accurate statistics on how many non-speaking autistics can learn to communicate by alternative means. Considering how often we see the topic mentioned on the Internet and elsewhere, though, we're probably talking about another 3 percent of the total autistic population (that's a conservative estimate).
Now let's add these percentages together. If 90 percent of all autistics learn to speak before age nine, and 3 percent begin speaking at a later age, and 3 percent do not speak but can communicate by other methods, what this means is that 96 percent of all autistic adults can communicate. Just as in my NNT illustration of teenagers and gangsta rap, only 4 percent of the children who would be receiving the hypothetical scientifically-proven biomed treatment would fall into the non-communicative category from which it was intended to "rescue" them, and only 2 percent actually would be "rescued." Once again, the NNT for this treatment would be 50, if such a treatment existed.
To put it another way, even if there really were a biomed treatment that had valid research studies supporting a claim that it could effectively "rescue" half of the autistic children who would otherwise spend their lives being unable to communicate, for every child who was "rescued" with this treatment, another 49 children would be dosed unnecessarily with drugs that could endanger their health. Those wouldn't be good odds even if there were legitimate scientific backing for the biomed peddlers' claims. Given the fact that no such proof exists—it's just lunacy to put children at risk like that.