Are Lovaas's 'Recovered' Kids Really Normal?

Most of you have probably heard of Ivar Lovaas's 1987 study in which he reported that 47% of autistic kids who got early ABA had 'normal functioning' and were therefore cured of autism.I just found a follow-up study published in 1993 that examined his experimental and control groups, including reporting test results for each of the 9 'recovered' kids (10-16 years old). The results are very interesting, and not exactly consistent with his claims.
WISC-R verbal, performance and full-scale IQ and subtest scores are reported. According to this source, a difference of 11 or more points between verbal IQ and performance IQ is clinically significant. Out of the 9 'recovered' kids, 7 showed a clinically significant difference between VIQ and PIQ by that definition (3 higher verbal IQ and 4 higher performance IQ).
The same source describes the use of a discrepancy calculation of subtests that are significant strengths and weaknesses, defined by being 3 or more points from the average. Normal kids typically have no significant strengths/weaknesses on verbal IQ, one on performance IQ and two on full-scale IQ. Assuming that more than that number of discrepant subtest scores indicates abnormal scatter, 5 children had abnormal scatter (all of them showing VIQ/PIQ discrepancies).
On the Vineland Adaptive Behavior scales, all (except one kid who wasn't tested) scored within the normal range on all areas, but 5 had at least one score in the borderline/low normal range (70-85). I don't know of any norms for discrepancy between Vineland scores and IQ, but 6/8 of them had Vineland scores at least 11 points below IQ (Vineland scores have the same normal range as IQ). This suggests they were functioning more poorly in practical areas than their IQs would predict. (And makes me very concerned for these kids, as an unrecognized mild self-care impairment can cause serious problems once you enter adulthood.) In terms of the Vineland Maladaptive Behavior Scale, 3 kids had clinically significant behavior problems on that scale.
On the Personality Inventory for Children, 6 scored in the clinical range on at least one subscale (2 not tested). Lovaas noted high scores on the Intellectual-Screening subscale could be explained by their history of classic autism, because many of the questions on that scale are retrospective, such as when the child started talking. Excluding that subscale (on which 3 had scores in the clinical range), 4 had scores in the clinical range on at least one subscale (not counting the Lie, Frequency and Defensiveness scales that try to test how honest the person was, on which 2 kids scored highly - one of whom had no other elevated scores). Another 2 had at least one score in the borderline range, though none in the clinically significant range.
Overall, combining all three measures, all the children had definite abnormalities (VIQ/PIQ discrepancy and/or clinically significant behavior problems). Not enough information is available to decide if they were still autistic (though many of the abnormalities they showed are common among autistics) but they certainly weren't neurotypical.

References:
Lovaas (1987): Behavioral treatment and normal educational and intellectual functioning in young autistic children. Journal of Consulting and Clinical Psychology, 55, 3-9

McEachin, Smith & Lovaas (1993): Long-Term Outcome for Children With Autism Who Received Early Intensive Behavioral Treatment. American Journal on Mental Retardation, 97(4), 359-372

[Correction: one child, RS, should actually be classified as possibly abnormal rather than abnormal. I misread one of his PIC scores.]