Cathy Lord's Longitudinal Autism Study
Some time ago, I was lucky enough to be able to attend a talk by Cathy Lord on her recent longitudinal study of children with autism. Two hundred kids from several states entered the study at age 2 or 3 and were followed up at ages 5 or 6, age 9, and as adults. They are now about 21 or 22. Because the study began over 20 years ago when fewer high-IQ people were diagnosed with autism, many of the participants have intellectual disability.
Dr. Lord wanted to know:
Diagnosis: How stable is an autism diagnosis made at age 2 or 3? What are the best predictors of having an autism diagnosis at age 9?
Autism symptoms: How do DSM-IV autism symptoms change over time?
How does cognitive ability (esp. IQ) develop over time?
How does language develop over time? What predicts whether a person develops expressive language or remains nonverbal?
Outcome: How do people diagnosed at age 2 or 3 function in adulthood? How do they do in school? Do they work? Do they have friends?
Diagnostic stability
Of children diagnosed with autism at age 2, 76 retained this diagnosis at age 9, 13 were downgraded to PPD-NOS, and 1 was found not to be on the spectrum at all.
Of children diagnosed with PPD-NOS at age 2, 27 were diagnosed with autism at age 9, 11 retained this diagnosis, and 6 turned out not to be on the spectrum.
Of children found not to be on the spectrum at age 2, 2 were diagnosed with autism at age 9, 9 were diagnosed with PPD-NOS, and 34 continued to be NT.
In other words, diagnostic stability is pretty high both for the obviously autistic kids and the obviously NT kids. For kids in the "gray area" who receive a diagnosis of PPD-NOS, however, stability is quite low. Interestingly, it's not because "normal" kids are getting pathologized--most of the misdiagnosed kids turned out to be more severely autistic at age 9.
The best predictor of having an autism diagnosis at age 9 was clinical diagnosis at age 2.
Repetitive behaviors at age 2 also predicted diagnosis, as did parent reports of unusual communication.
Formal tests of verbal skills at age 2 predicted autism diagnosis at age 9, but interestingly, did not predict later verbal scores very well. (No, I'm not sure what this means).
Also interestingly, parent descriptions of social behaviors at age 2 did not predict an autism diagnosis at age 9.
Changes in autism symptoms
Between ages 2-14, calibrated ADOS scores changed in different ways in different kids.
40% of the group were called "persistent moderate." They consistently had moderate ADOS scores.
40% of the group were called "persistent severe." They consistently had severe ADOS scores.
10% of the group decreased in ADOS scores.
10% had increased ADOS scores, but actually this was driven by a mix of kids who went up and down a lot and kids who started out with mild scores and later increased.
Symptoms can change over development, but the diagnosis remains stable. For example, one child was always outgoing and socially directed, but as his language skills improved, his eye contact became less frequent and his facial expressions more muted.
There were changes in activity level and motor skills. Children who were coordinated at 2 were often less so at 9.
The relationship between verbal and nonverbal IQ can change over time.
Language Development-Predictors of Verbal Level at age 9
At age 2, nonverbal IQ was predictive of later language development, while verbal IQ was not.
Active parent participation in ANY sort of treatment before age 5 was related to improved language development.
Children with an initial PDD diagnosis ended up slightly better off than control NT kids with early language delay, while children with an initial autism diagnosis did way worse.
Participants broken into several groups with different developmental trajectories. While all started out at roughly the same (low) point, they differed both in rate of growth and where they ended up.
Of kids with autism diagnoses, a few had rapid growth in verbal ability and ended up with high scores, while most developed very slowly, with a few ending up with close to average scores and most below average.
For kids with extremely low verbal scores on the Mullen test, the gap between nonverbal and verbal IQ ranges from 10-40 points.
Nonverbal IQ tended to be higher than verbal IQ, but when verbal IQ increases, so does nonverbal IQ.
Verbal and nonverbal IQ both seem to predict language ability. For children who had poor expressive and receptive language, the average verbal IQ was 66 and the average nonverbal IQ was 76. For those who had poor expressive but intact receptive language, the average verbal IQ was 84 and the average nonverbal IQ was 100.
Out of the 200 original participants, only 84 had no words by age 2. 13 of these had low nonverbal as well as verbal IQ, and thus could be expected to have language delay. Of the remaining 61, 39 had some expressive language by age 5, while 22 still had no words. I have no notes on how many were still nonverbal by adulthood, unfortunately.
Life outcome
Dr. Lord reported that there were 2 highly successful groups of participants. In what she called the "optimum outcome group" were 8 people who have jobs and are going to university. While they originally had ADOS scores of 12, they now score close to 0 and do not look autistic to the untrained eye. The other successful group consisted of 10 participants who still look very autistic but have age-level adaptive skills, are doing well in school and at work, and had a steady but less dramatic decline in ADOS scores (from ~13 to 6).
Everyone in the "optimal outcome" group was white, from a two-parent family, continued living near where they first grew up, and had good academic skills.
These 2 successful groups were the only ones with dramatic verbal IQ increases, from the 50's to almost 100. They showed the most dramatic verbal IQ increase after age 5.
While most participants showed increasingly wide gaps in adaptive living skills compared to peers, the two successful groups went up.
There was no difference between the two successful groups in the interventions they participated in.
It looks pretty bad that only 10% (20/200) of Dr. Lord's participants ended up with "successful" outcomes. Cathy Lord thought this might be due to the large number of intellectually disabled people in the study, and said this was actually an improvement compared to previous longitudinal studies. In one earlier study, 50% were nonverbal as adults, whereas far fewer were in this study.
This was a very important study, but I had a few problems with it.
1. It stigmatized people simply for "looking autistic."
After Dr. Lord's talk, I asked her if there were any differences between her two successful groups other than whether they "looked autistic." While I didn't say this, if there were no functional differences between the two groups, calling one the "optimum outcome" group simply because they didn't look autistic is discriminatory. She replied that the two groups were very similar, but the "optimum" group did slightly better socially. While neither had many friends, members of the "optimum" group sometimes had a close friend or two. This is definitely a functional difference, but I'm not sure whether this is due to the "optimum" group having better social skills or the fact that NTs discriminate against them less because they look less autistic, and I think she should have been more reflective about the ethical issues here.
2. It didn't look at sensory symptoms, even though they're likely to relate to functioning and autism severity.
When I asked her about it, she said she did look at them, but indirectly, as these were in the "repetitive behavior" category. This despite the fact that as far as I know, it's been assumed rather than established that there's a close connection between sensory abnormalities and repetitive behavior. I think this was a major weakness of the study and I would have looked at sensory symptoms separately.
3. At least in this talk, it didn't address the role of support (other than autism-specific interventions) on outcome.
Yet the successful participants who learned to pass clearly had plenty of support--in particular, they never went through the stress of divorced parents or extensive moving. It seems their parents had plenty of social and economic capital to invest in them. I hope the data is published soon, because I'd like to see how much of functioning was due to participants' own early cognitive characteristics and how much was due to how they were raised and what sort of support they received.
4. What does it mean for the diagnosis to be stable if the underlying symptoms change? Is this actually meaningful at all?
TL;DR summary: It seems that people about my age who grew up with autism end up a lot better off than previous generations who've been studied, but there's still lots of room for improvement.