Is the claim that a sample size of 1403 showed only 1 case of autism? Wouldn't you expect more? At a rate of 1:150, you'd expect around 9. And, if the expected rate of autism is indeed in this 1:1400 range, isn't the sample size too low to reach any conclusion at all? Wouldn't you want a sample size such that both the control and non-control group would be expected to contain such a case?