Last time I talked about a recent study that looked at adults with severe intellectual disability (ID) in a city in Iceland and found that a decent number of them had undiagnosed autism. The numbers are fairly small (25 cases) but then again, so is the city that we are talking about (population 70,001).
But still, if this many cases of autism were missed in a population that presumably receives a lot of medical care, does that mean that there would be an equal or greater number of cases missed in the general population? Is there really a hidden horde of adults with autism? The data from the last study doesn't directly answer this question but that hasn't stopped some people from speculating that this study helps provide the answer.
So lets look at what numbers to see what we can possibly gleam from the study. As you read the following, please keep in mind that most of this is speculative in nature and could easily be completely wrong.
Lets start off with some basic assumptions.
The study talks about "adults" with severe ID and autism, but what does the label "adult" mean here? It is always important to know how old the "adults" are and when they were born. Any adult population that you are looking at should have an autism prevalence in line with the rest of their generation. That means if we are talking about "adults" born after the mid 1980s when the autism rates started shooting up we could reasonably expect then to have a higher autism rate than "adults" who were born in the 1960s.
The study says that the age range of the participants was from 18 to 67 years old, but it isn't exactly clear what these ages are in reference to. In one location the study seems to give the ages in reference to the year 2000 but then the results were not submitted for publication until this year. I can't believe that all of the data collection was done in 2000 and it has taken ten years to analyze the data. So based on the dates, the youngest participant was born sometime between 1982 and 1992. When you consider the fact that the majority of the cases found were younger (under 32), that means we are likely looking at people born between the the mid 1970s to the mid 1980s.
The next assumption is how many people with autism we expected to find. Fortunately, we have a study that looked at autism rates in Iceland during two different time period - 1974 to 1983 and 1983 to 1993. During the first time period the rate was approximately 3.8 per 10,000 while during the second it was 8.6 per 10,000.
One thing to note here is that the diagnosis during the earlier time period were largely based on an earlier version of the diagnostic criteria (ICD-9 vs ICD-10). The earlier criteria did not have as broad of a definition of autism and focused more heavily on people who would have had cognitive problems in addition to autism.
So, with that in mind, we get to the first question.
Did the initial number of people with autism in the severe ID population agree with the what the earlier estimates were for the time period? We know that the prevalence was thought to be 3.8 per 10,000 in the "adults" born between 1974 and 1983 which means that we would expect to find a total of 26 people with autism. The study initially had 13 people with autism - 11 in the population examined (119) and 2 in the part of the population that was not included (if you care about the details, ask in the comments and I can elaborate).
There would be two other groups of people with autism - those in the population with mild ID and those in the general population who didn't have ID. I don't have an exact figure for how many people have ID, but an accepted range seems to be between 1% and 3% of the population. Based on some discussion on the study, I am thinking that number in Iceland is little under 1% or approximately an additional 444 people with mild ID.
We know that autism is much more common in severe ID than in milder ID and it was present in about 5% of the severe cases. Lets assume that the rate is about half in the milder cases which would mean about an additional 11 cases. That brings the total to 24, but we aren't done yet.
The last part of the puzzle is how many people with autism are expected to have a form of ID. Again, there are no good numbers here but for now I am going to go with a rate of 80% as we are talking about the earlier criteria (again, if you are curious about this number, ask in the comments and I will try to explain).
Given this 80% number there would be an additional 6 cases which would bring the total up to 30 cases. That number is higher than the 26 that we would have expected, but considering the estimations used, the numbers aren't too far off.
Which brings us to the next question - what does the increase in the number of cases in the severe ID group mean for the entire population?
Since the second part of the study only examined half of the severe population, I am going to use the 21% rate that the researchers gave for the severe ID group and about 10% for the milder ID group. Putting these two figures together means that we would expect about 100 people with ID to have autism.
I used an 80% figure for the number of people with autism and ID in the earlier estimate, but for this estimate I am going to drop it. The reason is that we are talking about current diagnostic criteria which includes many milder cases of autism and we have some recent estimates from the CDC that we can use. The estimates from the CDC vary depending on the state you are looking at and how broad your definition of ID is. The highest estimate using the broadest definition of ID is about 80% while the lowest using the narrowest definition of ID is about 30%.
For the sake of simplicity, I am going to take the 41% figure that the CDC cites as an average. Using that figure, we arrive at a total of about 245 people with autism. This is a rate of about 35 per 10,000.
The final question is what would we have expected to find if autism is as common in adults as it is in children?
Looking back at some of the estimates from the few years, you would see numbers like 1 in 250, 1 in 150, and, most recently, 1 in 110. Those numbers translate into 40 per 10,000, 66 per 10,000 and 90 per 10,000.
So the answer is 35 per 10,000 compared to 90 per 10,000
Or in other words, the most recent estimate for autism is still 3 times higher than an aggressive extrapolation based on the data in the study.
The bottom line is that the hidden horde is still in hiding