Twenty-Year Outcome for Individuals With Autism and Average
or Near-Average Cognitive Abilities
Megan A. Farley, William M. McMahon, Eric Fombonne, William R. Jenson, Judith Miller,
Michael Gardner, Heidi Block, Carmen B. Pingree, Edward R. Ritvo, Riva Arielle Ritvo,
and Hilary Coon
Previous studies found substantial variability in adult outcome for people with autism whose cognitive functioning was
within the near-average and average ranges. This study examined adult outcome for 41 such individuals (38 men and 3
women) originally identified through an epidemiological survey of autism in Utah. Mean age at the time of their
previous cognitive assessment was 7.2 years (SD54.1, range53.1–25.9 years) and at follow-up was 32.5 years (SD55.7
years, range522.3–46.4 years). Outcome measures included standardized assessments of diagnostic status, cognitive
ability, and adaptive behavior. Additional information collected concerned demographic variables, indicators of
independence, social relationships, medical and psychiatric conditions, and social service use. Outcomes for this sample
were better than outcomes described in previous work on individuals with similar cognitive functioning. For example,
half of the participants were rated as ‘‘Very Good’’ or ‘‘Good’’ on a global outcome measure. As in previous studies, there
was considerable variability in measured cognitive ability over time. Over half of the sample had large gains or losses of
cognitive ability of greater than 1 standard deviation. Cognitive gain was associated with better outcome, as was better
adaptive functioning. While all participants had baseline IQs in the nonimpaired range, there was limited evidence to
support the use of other early childhood variables to predict adult outcome.
Keywords: autism; adult; outcome; longitudinal
Although much research has been conducted on autistic
disorder (AD) in the past 60 years, few studies address AD
in adults. Even fewer studies describe outcomes for adults
with AD and no cognitive impairment, a population
associated with greater variability in individual outcome
[Engstrom, Ekstrom, & Emilsson, 2003; Gillberg &
Steffenburg, 1987; Howlin, 2003; Howlin, Goode, Hut-
ton, & Rutter, 2004; Howlin, Mawhood, & Rutter, 2000;
Kanner, 1971; Lotter, 1974; Piven, Harper, Palmer, &
Arndt, 1996; Rumsey, Rapoport, & Sceery, 1985; Szatmari,
Bartolucci, Brenner, Bond, & Rich, 1989]. Communica-
tive phrase speech before age 6 and a relatively high
childhood IQ appear to be critical factors for a ‘‘good’’
outcome [Billstedt, Gillberg, & Gillberg, 2005; Kobayashi,
Murata, & Yoshinaga, 1992]. Cross-sectional studies and
case reports describe adults with AD who have achieved
significant independence in multiple domains, but long-
itudinal studies of those diagnosed in the 1970s and
earlier have shown that few adults with AD live outside
of their childhood residences or institutional settings.
Competitive and sheltered employment occupy less
than 40% of adults with AD studied longitudinally
[Howlin, 2003; Howlin et al., 2004]. Few adults with AD
in these studies marry or develop reciprocal friendships,
in spite of increased interest in developing social
relationships as individuals with AD age [Nordin &
Gillberg, 1998; Rumsey et al., 1985]. Some core char-
acteristics of AD abate moderately in many adults, and
functional abilities often improve [Piven et al., 1996;
Rumsey et al., 1985; Seltzer et al., 2003]. Low IQ,
adolescent onset of epilepsy, and comorbid psychiatric
disorders frequently complicate the course of AD [How-
lin, 2002]. These co-occuring disorders may be major
contributors to the higher mortality rate for individuals
with AD [Gillberg, 1991].
We now report our study of adult outcome in a sample
of individuals from a population-based group ascertained
20 years ago that comprised participants who, when first
Autism Research 2: 109–118, 2009109
Received November 2, 2008; revised February 13, 2009; accepted for publication February 18, 2009
Address for correspondence and reprints: Megan A. Farley, Utah Autism Research Program, Department of Psychiatry, University of Utah, 650 Komas
Dr., Ste. 206, Salt Lake City, UT 84108. E-mail: firstname.lastname@example.org
Grant sponsors: Utah Autism Foundation, University of Utah Research Foundation, Funding Incentives Seed Grant.
Published online 18 March 2009 in Wiley InterScience (www. interscience.wiley.com)
& 2009 International Society for Autism Research, Wiley Periodicals, Inc.
From the Department of Psychiatry, University of Utah, Salt Lake City, Utah (M.A.F., W.M.M., J.M., H.C.); Department of Educational Psychology,
University of Utah, Salt Lake City, Utah (M.A.F., W.R.J., M.G., H.B.); The Brain Institute, University of Utah, Salt Lake City, Utah (W.M.M.); Department
of Psychiatry, McGill University, Montreal, Quebec, Canada (E.F.); Carmen B. Pingree Center for Children with Autism, Salt Lake City, Utah (C.B.P.);
Neuropsychiatric Institute, Los Angeles Medical School, University of California, Los Angeles, California (E.R.R.); Yale Child Study Center, Yale
University School of Medicine, New Haven, Connecticut (R.A.R.)
assessed, were diagnosed with AD and had an IQZ70.
Our aims were to (1) define current function and service
needs; (2) explore the feasibility of studying a larger
statewide sample; and (3) examine relations among
historical and current variables not reported in other
recent outcome studies.
Participants were identified originally through a statewide
epidemiological survey of autism that occurred between
1984 and 1988 [Ritvo et al., 1989]. The aim of the survey
was to identify all cases of autism born between 1960 and
1984 living in Utah at the time of the survey. A team of
clinician–investigators used parent interviews, direct
observations, and records reviews to determine diagnostic
status for each case [cf. Ritvo et al., 1989 for details].
A total of 489 people were screened for autism. Of these,
241 were diagnosed with DSM-III autism, 138 were
determined not to have autism, and 110 were excluded
from the study because they were out of the study age
range, uncooperative, or lost to follow-up.
Inclusion criteria for this follow-up study were a
diagnosis of AD by the epidemiological survey team
and a nonverbal or full scale IQ score of 70 or more. The
exclusion criterion was the presence of a medical
condition that precluded completion of follow-up assess-
ments or which might have significantly affected in-
dependent living. Seventy-five participants were eligible
for this follow-up study.
Contact information for eligible individuals and their
parents was obtained using public records databases,
published telephone directories, and original contact
information from the 1980s epidemiological survey.
Letters of invitation to participate were mailed to
potential participants and their parents. Two weeks after
the letters were mailed, we attempted telephone contact
with those who had not yet scheduled an appointment.
One month after the first mailing, we mailed a second
invitation letter and again attempted telephone contact 2
We collected data for 41 participants (53%). Four
people (5%) declined to participate. We were unable to
locate 9 individuals (12%), and 14 others (19%) did not
respond to our contact attempts. Seven individuals (9%)
agreed to participate but did not complete the evalua-
tion. The follow-up sample included 38 men and 3
women (male to female ratio: 12.7:1.0). Mean age at the
time of the adult assessments was 32.5 years (SD55.7
years, range522.3–46.4 years). Time elapsed between
historical and current testing was 24.42 years (SD53.75
years, range515.25–34.17 years). Only one participant
was over age 14 when first assessed. There were two sets
of siblings among those participating in the follow-up.
Outcome measures between siblings were markedly
Approval for this study was obtained from the Institu-
tional Review Boards of the University of Utah and Utah’s
State Department of Human Services.
Most assessment procedures were conducted in partici-
pants’ homes, with the remainder performed in our
Interview—Revised [ADI; Lord, Rutter & LeCouteur,
1994] and Autism Diagnostic Observation Schedule–
Generic [ADOS; Lord et al., 2000] were used to obtain
lifetime and current information for autistic symptoms
and current diagnostic status. All assessment procedures
were performed by the first author and a research staff
person who had achieved research reliability standards in
ADOS and ADI administration. Two participants refused
to be assessed directly but allowed their parents to be
interviewed. Parents who were also legal guardians for 2
other participants refused to allow research staff to
interact with their children, citing previous negative
experiences with mental health workers. Thus, ADOS
scores were collected for 37 of 41 participants.
In all cases, a parent participated in completing the ADI
and other caregiver interviews. Some parents were aided
by their adult child’s social services case manager, spouse,
or adult sibling. Three parents reported substantial
difficulty recalling details of their child’s early develop-
ment, so valid lifetime ADI results could not be obtained
for these 3 participants.
Adaptive behavior and independent living.
behavior in adulthood was assessed using the Vineland
Adaptive Behavior Scales–Survey Edition [VABS; Sparrow,
Balla, & Cichetti, 1984]. Another interview concerning a
variety of outcome characteristics was developed for
significant adults. Questions addressed chronic medical
functioning; transportation, employment, and residential
and use of social services. We also asked questions about
the level of assistance participants required from caregivers
and from social services, using definitions for levels of
information concerning education histories was obtained
through the ADI.
this study, we asked parents a number of questions about
participants’ social relationships and involvement in
social organizations. Questions about friendships and
romantic relationships addressed historical and current
In the interview developed for
110Farley et al./Outcome for adults
relationships as well as the reporters’ impressions of the
participants’ level of interest in developing additional
scale of overall social and independent living functioning
was based on work status, residential situation, and
number and quality of friendships [Howlin et al., 2004].
Criteria for a ‘‘Very Good’’ rating were paid employment
without any extra supports to perform duties, existence
of important social relationships, and a high level of
independence in daily life. A ‘‘Good’’ outcome was
defined as a generally high level of independence in
work and home life, requiring some extra support.
A ‘‘Good’’ outcome also meant the person had a
friendship or some acquaintances. A ‘‘Fair’’ outcome
reflected the need for regular support in work or home
life but no need for a special residential facility. A person
with this rating was reported to have acquaintances
through regularly scheduled activities but no particular
friends. A ‘‘Poor’’ rating indicated the person needed a
very high level of support, such as that provided through
a special residential facility and day programming for
obtaining this rating would have no friendships outside
of the residential setting. A ‘‘Very Poor’’ rating meant the
person required a high level of care in a hospital setting
with no autonomy. A person with this rating would have
no friendships. Ratings were assigned independently by
the first author and a graduate research assistant. The
ratings was 0.97. Rating discrepancies were reviewed,
and consensus was reached through a case conference.
A composite rating on a 5-point
adulthood were measured using the Wechsler Adult
Intelligence Scale—Third Edition [WAIS-III; Wechsler,
1997]. Four participants declined to complete the
WAIS-III, and the parents/legal guardians of two other
participants refused to allow staff to assess their adult
children. Two participants attempted but were unable to
complete the tasks on the WAIS-III due to cognitive
impairments. Two participants obtained PIQ scores but
lacked sufficient language ability to complete verbal
subtests. One participant was unable to complete two
verbal subtests due to interference in his home testing
environment; therefore, only his PIQ is reported. In
summary, 30 participants obtained an FSIQ score, and
only the PIQ was obtained for 3 participants.
Historical cognitive abilities.
participants’ cognitive and language abilities were not
made as part of the epidemiological survey, yet all but
one participant had been evaluated by service providers
who obtained IQ scores prior to the epidemiological
survey. The previously untested man was given an IQ test
through that study. Data from a variety of assessment
instruments and age ranges were used to estimate
interpretations of results from the historical to current
Direct assessments of
inthis sample; thus,
assessments must be treated cautiously. The global
measure of cognitive ability nearest age 8 was the first
test selected for use; Wechsler tests were most preferred,
followed by the Stanford–Binet. For two children, the
Denver Developmental Screening Measure was the only
developmental age was the metric reported from this
measure, the formula DA/CA?100 was used to calculate
an estimate of cognitive ability. When a global measure
was unavailable, a measure of nonverbal ability was used
as the best estimate of cognitive ability. Again, the score
obtained closest to age 8 was used. For many participants,
multiple measures were available for global, nonverbal,
and verbal abilities. Data were analyzed using the best
measures available in each domain.
Data on the ages of single word use and phrase speech,
in months, were extracted from the historical records.
Parents were also asked for this information through the
ADI. If parental responses to the ADI differed from
historical information, details concerning the nature of
the participant’s language development were discussed in
order to isolate the best-estimated ages. Seven partici-
pants did not have these data in the records, nor were
their parents able to confidently recall the ages at which
their children acquired language milestones.
Data were entered twice to ensure accuracy and analyzed
using SPSS Version 16.0. For comparisons of interval-level
variables, two-tailed t-tests and ANOVA were used.
Spearman’s r tests were used in nonparametric analyses.
Representativeness of the Sample
To estimate potential sample bias, we compared cases in
the follow-up study to those not included. Comparisons
were made using the historical records and the Utah
Population Database (UPDB), a repository for genealogi-
cal, medical, and demographic information concerning
families in Utah. The groups were compared for sig-
nificant differences on 14 variables including historical
IQ scores, age at testing, gender, ages of single word
acquisition and phrase speech, availability of adult data
in the UPDB, marriage, driver’s license, birth of offspring,
and seizure disorder in childhood. A significant differ-
ence between groups emerged only for the age at which
they acquired single words. Participants were signifi-
cantly older (n532, M546.1 months, SD534.9) than
nonparticipants (n529, M528.0 months, SD526.0;
t(56.9)52.31, P50.024). This result could suggest that
the group participating in this study had greater
communicative impairments; however, there were no
significant differences between groups on measured
global cognitive or verbal ability measures.
Farley et al./Outcome for adults111
Thirty-six of the 38 participants for whom a valid ADI
was collected obtained a lifetime ADI diagnosis of autism.
Two participants fell short of the cutoff of 3 on the
Restricted Interests and Repetitive Behaviors Domain,
each obtaining a score of 2. Their scores exceeded the
cutoffs in the Social and Communication Domains.
Therefore, they amply met the DSM-IV [APA, 1994]
diagnostic criteria for AD. The ADOS was administered
to 37 participants. Due to impaired expressive language
abilities, two participants were administered Module 1,
and three participants completed Module 2. Although
both individuals who were administered a Module 1 were
described as using phrase speech regularly, their language
production appeared suppressed during the assessments.
The 5 individuals who were administered ADOS Modules
1 and 2 met ADOS criteria for autism. Twenty-two of the
32 participants who completed the ADOS Module 4
obtained an ADOS diagnosis of autism. Four scored in the
autism spectrum disorder range but did not meet ADOS
criteria for autism. Six participants did not score within
the autism or autism spectrum disorders ranges on the
ADOS. However, only one of these adults was described
by himself and others as no longer having any significant
autistic-type symptoms. Remaining participants who did
not meet full ADOS criteria for an autism spectrum
subtle social behaviors and participating with others in
a fully reciprocal manner.
Twenty-three participants (56%) attended public school
in special education programs until the maximum age of
21 years, 11 months. The remaining 18 (44%) completed
high school at grade 12 with a diploma, and 16 (39%)
went on to receive postsecondary education. Four of
these earned an associate’s degree or technical training
certificate, and 7 others earned one or more bachelor’s
degrees. Five participants were in full-time postsecondary
training or education at the time of the study, and 3 were
in school on a part-time basis. Of these 8 participants, 5
also worked full-time jobs, and 2 worked part-time.
Half of the participants were employed in either full-time
(N511, 27%) or part-time (N511, 27%) independent,
paid jobs. Three participants (7%) worked in supported
employment positions, 2 (5%) were part-time volunteers,
10 (24%) participated in day programs, and 4 (10%) were
Thirty-eight participants (93%) were members of the
Church of Jesus Christ of Latter Day Saints (LDS Church).
Many young men and some young women in the LDS
Church perform a 2-year mission in young adulthood.
Some members serve a service mission, working in local
church institutions, while others serve proselytizing
missions in the US or other countries. Proselytizing missions
are highly social endeavors in which the missionary is
with one or more assigned ‘‘companions’’ at all times,
except during tasks associated with personal hygiene.
They make frequent social overtures to strangers. Missions
typically last for 2 years. Three participants served interna-
tional missions for which they learned a new language.
Five served missions in US cities away from their homes,
and 2 served service missions while residing at their family
Friendships and Social Life
Three participants (7%) were married at the time of the
interviews, and all 3 had children, for a total of 7 offspring.
Grandparents of only one of these children expressed
concerns regarding their grandchild’s autistic-like charac-
teristics. Two other participants (5%) were divorced. One
previously married participant was newly engaged. Three
other participants (7%) were in long-term relationships.
Eighteen individuals (44%) had never dated. Thirteen
(32%) had dated in both group and single couples
interactions. Nine (22%) had dated only in group settings,
and one had dated only in a single couple arrangement.
When asked whether they felt their adult child wished to
have a romantic partner, parents of 15 (44%) participants
not already in a relationship said they believed their son or
daughter would like to be in a relationship. Parents of 17
participants (41%) said their son or daughter did not want a
romantic partner. Some of these reported their child’s
concern about passing autism on to children or about being
too difficult, themselves, for another person to manage.
Others were described as having no desire for a close
interpersonal relationship beyond pre-existing family rela-
tionships. Only 2 parents (5%) felt uncertain as to whether
their son or daughter wished to be involved in a romantic
Parents were also asked about their adult child’s
involvement in organized groups. Thirty-five participants
(85%) regularly attended church activities, and 7 (17%)
participated routinely in Special Olympics. Twelve parti-
cipants were also involved in other organized activities
(e.g., martial arts classes). Four participants (10%) were
uninvolved in any regularly occurring meetings with
Eleven participants had a driver’s license and drove
regularly (27%). Eleven others took public transit on a
regular basis. Four participants (10%) used a special
public transit service for people with disabilities or rode
regular public transit with support from others. Eight
112 Farley et al./Outcome for adults
participants (20%) relied on family or social services staff
to transport them.
Three participants (7%) lived in their own homes that
they had purchased themselves, and 2 (5%) lived
independently in their own apartments. One individual
lived in an apartment with a sibling, receiving limited
assistance from the sibling with daily living tasks.
Another participant lived in his own apartment but was
checked on several times a week by a parent. Parents of 2
participants established a separate apartment with an
independent entrance in their own home for their adult
children. Twenty-three participants (56%) lived with
their parents. In 6 cases, continued residence with
parents was due to a son or daughter being in school
and unable to afford their own independent home.
Parents of 3 of these participants expected their son or
daughter would move out in the near future. Three
participants lived alone in an apartment with support
from social services agencies, and 4 lived in group homes.
Two participants lived in a state facility for adults with
We used questions written by Engstrom et al.  to
ask parents how much help their son or daughter needed
from caregiving others. No participant had a rating of
‘‘None,’’ signifying ‘‘no contact with family.’’ Eleven
participants (27%) required a ‘‘Low’’ level of help,
meaning ‘‘normal or near-normal contact with parents,
siblings, and other relatives; support and practical
assistance from time to time.’’ Another 11 participants
required a ‘‘Moderate’’ level of help, meaning that they
required ‘‘regular practical assistance at home; daily
contact by phone or physically; help with local autho-
rities.’’ Nineteen individuals (46%) required a ‘‘High’’
level of assistance, meaning they needed ‘‘extensive help
with social contacts and employment sites; total control
of economic affairs; lives with relatives from time to
time’’ [Engstrom et al., 2003, p 103].
Overall Social Outcome
We applied Howlin et al.’s  method for estimating
overall social outcome to our sample. The number and
percent of participants by category was as follows: ‘‘Very
Good,’’ 10 (24%); ‘‘Good,’’ 10 (24%); ‘‘Fair,’’ 14 (34%);
‘‘Poor,’’ 7 (17%); and ‘‘Very Poor,’’ 0 (0%).
Comorbid Psychiatric Problems
Twenty-four participants (59%) had at least one diag-
nosed or suspected comorbid psychiatric condition, and
11 (27%) had two or more comorbid diagnoses. Fifteen
participants (37%) were described as taking prescription
medication aimed at regulating behavioral difficulties.
The most common psychiatric comorbidity was depres-
sion, which was diagnosed for 13 participants (32%) and
strongly suspected in 4 other cases (10%) who refused to
seek treatment. Nine participants (22%) were described as
having an anxiety disorder other than obsessive–compul-
sive disorder (OCD); three (7%) had a diagnosis of OCD.
Six participants (15%) had a childhood diagnosis of an
attention-deficit disorder and were treated with medica-
tions, but 2 were no longer medicated as adults. Two had
Tourette’s Disorder. One had a diagnosis of schizophre-
nia, and another individual had been diagnosed with
bipolar disorder. One person had a history of pica and
had been diagnosed with antisocial personality disorder,
intermittent explosive disorder, and had a history of
severe self-inflicted violence.
participants had destroyed property in anger over the
past year, 6 reporters (15%) assented that this behavior
occurred, and 2 (7%) stated that it occurred very rarely.
When asked whether
Chronic Medical Conditions
Seven individuals (17%) were reported to have a seizure
disorder at some time in their lives. Five were treated for
seizures at the time of the study; one had febrile seizures
only in early childhood, and another man’s seizures
ended in his early 20s without recurrence.
Seven participants (17%) had confirmed sleep pro-
blems. Only 2 (5%) of the individuals affected with
sleeping difficulties were being treated. Reporters for 8
participants (20%) noted abnormalities in sleep behavior,
such as the apparent need for little sleep or unusual times
of sleeping and waking, but none reported that the
participant seemed excessively tired or adversely affected
by their sleep behavior.
The health history was good, with 26 participants
denying any chronic medical disorders. Medical condi-
tions reported by the others included food and/or
seasonal allergies in 7 participants (17%); 2 each (5%)
with chronic back pain or hemorrhoids; and one each
with a thyroid disorder, anemia, gout, flat feet, and high
Parents reported on their adult son’s or daughter’s social
services support, including Vocational Rehabilitation, the
Division of Services for People with Disabilities (DSPD),
Social Supplemental Income (SSI), and Medicaid. At the
time of the study, 16 individuals (39%) were DSPD
clients, and 3 (7%) were on a waiting list. Twenty-two
(54%) had never applied for DSPD funding. Of the 25
individuals not receiving funding for services through
DSPD, 13 (52%) had accessed Vocational Rehabilitation
services in the past. Twenty-three participants (56%)
received SSI and Medicaid benefits.
Farley et al./Outcome for adults113
We used Engstrom et al.’s ratings of the level of social
services needed. Twelve individuals (29%) required no
support from social service agencies in adulthood. Eight
participants (20%) required a ‘‘Low’’ level of support,
meaning that they required ‘‘advice and support from
habilitation; regular home help service.’’ ‘‘Moderate’’
support was required by another 8 adults (20%), suggest-
ing they needed ‘‘continuous home support; sheltered
job; job assistant; regular support from psychiatry and/or
habilitation.’’ Thirteen participants (32%) required a
‘‘High’’ level of social services support indicating a need
for ‘‘supported living; group home or institution; day
center; personal assistant’’ [Engstrom et al., 2003, p 103].
Parents reported their adult child’s history with law
enforcement, either due to actual infractions or misinter-
pretations of their child’s behavior. Fourteen participants
(34%) had previous interactions with law enforcement
officers, although in 2 cases these occurred only in early
childhood when their son or daughter escaped caregivers’
attention. The other 12 participants experienced some
problem in late adolescence or adulthood. Behaviors
resulting in intervention by law enforcement officers
included performing maintenance tasks in restricted areas
without any formal relationship to the business; observing
children in public from a short distance; engaging in
dangerous driving behavior under instructions from peers;
sexual behavior aimed at a peer with developmental
disabilities; stalking peers in pursuit of friendships; running
from a police officer in a reportedly suspicious manner;
and failing to pay parking tickets.
WAIS-III [Wechsler, 1997] full scale scores were available
for 30 participants and only PIQ scores were available for
3 others (Table I). A paired t-test of the VIQ and PIQ
scores for the 30 individuals who completed the entire
WAIS-III yielded no significant difference between these
scores (t(29)50.68, P50.505).
VABS scores were obtained for all participants. The
sample was 65.0 (SD532.8, range520–116). The mean
Communication domain score was 63.7 (SD533.9,
range520–109), and the mean Daily Living Skills
domain score was 76.2 (SD534.9, range520–119). The
group mean Socialization
(SD531.8, range520–125). Paired t-tests comparing
domain scores demonstrated that the Daily Living Skills
domain was significantly higher than either the Com-
munication (t(40)5?4.21, P50.000) or Socialization
between the Communication and Socialization domains
was not significant (t(40)50.56, P5n.s.).
Associations Among Adult Outcome Measures
Correlations between the social functioning composite and
other adult outcome measures were all large and signifi-
cant (ranging from 0.50 to 0.83). The strongest correlations
were between the social functioning composite and VABS
scores (r50.78–0.81; Fig. 1). WAIS-III PIQ scores shared the
weakest association with the social functioning composite,
but the relationships were still moderate to strong
(r50.55). High levels of support from caregivers and social
service agencies were negatively correlated with the social
functioning composite (r5?0.76 and ?0.73, respectively).
Associations Between Historical Measures and
Correlations between historical and current cognitive
indices were moderate to high. The correlation between
historical global IQ and the WAIS-III FSIQ for the 30
individuals completing both measures was 0.51. Histor-
ical verbal IQ was correlated with current VIQ at r50.58,
Table I.IQ Test Results
Childhood assessmentAdulthood assessmenta
Age in years ScoreScore
nMean (SD) [range]Mean (SD) [range]nMean (SD) [range]
Full scale IQ
Best IQ estimateb
6.96 (2.82) [3.50–13.42]
7.28 (2.69) [3.75–13.42]
7.35 (4.12) [3.08–25.92]
7.17 (4.11) [3.08–25.92]
87.68 (13.17) [65–115]
78.72 (17.34) [51–114]
83.68 (17.56) [36–137]
86.66 (15.44) [69–137]
89.09 (23.36) [57–144]
89.10 (25.24) [51–139]
88.93 (25.73) [50–140]
aMean age at the time of the adult assessments was 32.48 years (SD55.67 years, range522.33–46.42 years).
bThe IQ score is the best available estimate of global cognitive ability as close to age 8 as possible. Seventy-five percent (31) were FSIQ, 17% (7) were
PIQ, and IQ test type for 7% (3) was unknown.
114Farley et al./Outcome for adults
and historical nonverbal ability was moderately corre-
lated to current PIQ (r50.41).
Historical full scale or global IQ was a moderately
useful metric for prediction of outcome level (df 3,37;
F53.94, P50.016), but historical verbal and nonverbal
scores were not. Early global IQ was available for 38 of
the 41 people assessed. When controlling for age
of assessment, this score accounted for approximately
23% of the variance in predicting outcome category,
close to current VIQ, which accounted for 27% of the
variance. Current PIQ accounted for only 14% of the
All participants were reported to have had commu-
nicative phrase speech or better language by age 6.
Neither the age of single word use nor the use of phrase
speech as a continuous variable was a useful factor in
discriminating between outcome groups overall. How-
ever, of the 7 participants in the ‘‘Very Good’’ outcome
group, 6 had acquired phrase speech by age 4. Of the 7
participants in the ‘‘Poor’’ outcome category with this
data, 5 were reported to have acquired phrase speech after
age 4, but before age 6.
Change in IQ Scores
Despite the moderate to high correlations between
historical and current IQ scores, there was substantial
change in individual scores. Global cognitive ability/FSIQ
changed by 1 standard deviation or more in 17 of the 30
people assessed on both occasions (10 increased and 7
declined). Almost half of those (n59 of 20) completing
measures of verbal ability on both occasions demon-
strated a change of at least 1 standard deviation
(7 increased and 2 declined), and 11 of the 24 partici-
pants who completed nonverbal measures on both
occasions obtained a difference in scores of at least 1
standard deviation (6 increased and 5 declined). As a
group, those with ‘‘Very Good/Good’’ outcomes showed
improved cognitive ability and those with ‘‘Fair/Poor’’
outcomes showed declines in measured cognitive ability
(t(29)54.47, P50.000); a large increase in IQ was
associated with better social outcome. Change in mea-
sured verbal (t(19)53.36, P50.004) and nonverbal
(t(23)52.25, P50.036) abilities also yielded significant
differences between ‘‘Very Good/Good’’ and ‘‘Fair/Poor’’
In spite of historical and ADI information strongly
supporting a lifetime diagnosis of AD, 6 participants
presented improved autistic symptoms to the point
where they obtained subclinical scores on the ADOS. In
cross-sectional samples, diagnostic procedures that assess
current symptoms may miss adults with childhood AD
who have made significant gains over time. Our results
suggest that the ADOS, when used alone, may lack
adequate sensitivity for correctly diagnosing some adults
with past diagnoses of AD and ongoing, related difficul-
ties. Furthermore, parent/caretaker recall of childhood
symptoms and temporal events appears diminished over
two decades. Our study highlights the importance of
treating AD as a lifetime diagnosis, corroborated by
Cognitive Functioning and Adaptive Behaviors
A change in global cognitive ability of one standard
deviation or more was evident in 17 out of 30
participants. Variability in individual cognitive ability
has also been noted in other samples [Howlin et al., 2004;
Mawhood, Howlin, & Rutter, 2000; Rumsey, 1992]. A
decrease in IQ may be a marker for important biological
changes in a subset of individuals with AD.
The finding that adaptive behavior measures were the
variables most closely correlated with outcome may have
implications for training/intervention programs to foster
Daily Living Skills domain was the most closely related
to better outcome. There were cases of individuals who
obtained high IQ scores but lacked the practical living
skills assessed on the VABS. It was clear that several of
these people felt frustrated by these practical limitations.
In contrast, 3 participants with relatively low IQ
scores were able to care for themselves with little
assistance. These individuals obtained ‘‘Good’’ or ‘‘Very
Good’’ outcome ratings in spite of borderline cognitive
Outcome Adjustment Status
Overall, this sample obtained comparable or better
outcomes in adulthood than other samples of similar
early cognitive ability. Several factors could account for
the relative success of this sample. We recruited only half
of the eligible individuals, so we may have undersampled
Trends in adult outcome variables by social function-
Farley et al./Outcome for adults115
lower functioning cases. However, our analyses of the
representativeness of the follow-up sample do not
support this notion. Some participants in this study
were in early adolescence at the time of the initial
assessment,and their cognitive
other abilities may have stabilized by the time of their
assessment. Participants in some previous studies were
identified through clinical referrals for developmental
identified through research aimed at identifying all
cases of autism in Utah using previously identified
cases, reviews of clinical records, referrals from education
and other service agencies, and self-referrals in response
to media campaigns. Thus, some participants from
purely clinical samples may have had more severe
developmental concerns than some participants in this
Another possible factor is the support of the LDS
communities in which many participants grew up and
continue to live. The church building that members
attend is determined by their address; thus, children go
to school, attend church activities, and reside in a
neighborhood as a cohort. Inclusion of individuals with
disabilities is a strong LDS cultural value. Parents attend
services and activities with their neighbors, sharing joys
and concerns about their families. Families are often
large, and in many cases a sibling’s social circle extended
to include the participant. Some participants had assis-
tance finding work through these networks.
in thisstudy were
Comorbid Psychiatric Conditions
Rates of comorbid psychiatric difficulties were lower in
this sample than in other samples, possibly due to
limitations in our methodology [Eaves & Ho, 2008;
Ghaziuddin, Weidmer-Mikhail, & Ghaziuddin, 1998;
Howlin, 2003; Rumsey et al., 1985; Szatmari et al.,
1989]. In some cases, comorbid conditions were so
pronounced that they seemed to be the participants’
primary difficulty. In spite of high IQ scores and adequate
practical skills, a handful of participants were unable to
seek employment and independence due to difficulties
Limitations of the Study
This study is limited by a number of factors. Our
recruitment rate was just over half of those eligible.
While the epidemiological survey records are a rich data
source, standardized data collected from that time is
limited. Much information was derived from records of
service providers who used a wide range of assessment
instruments. There was substantial variation in the ages
at which participants were evaluated by these service
providers and by the epidemiological survey team. There
was also variation among participants in the time elapsed
between initial assessments and follow-up. Given that
this sample comes from the extreme end of the range of
cognitive functioning, it is likely that measurement
error and regression to the mean are at play, and the
true score for a participant with a historical IQ score of 70
could actually be much lower. Therefore, conclusions
drawn from the early cognitive data must be treated
Another limitation related to measurement was that
the follow-up assessment protocol did not include
alternative measures of cognitive ability for the 5
individuals who were unable to complete the WAIS-III.
While results on alternative measures for those who
appeared to lack the cognitive ability to complete the
WAIS-III would likely have been low, this information
may have increased the predictive power of other
Other limitations are related to the nature of the
sample. The calculated prevalence rate in the epidemio-
logical survey sample was lower than in other studies
during the same period [Fombonne, 2005], and it is likely
that cases were missed. Participants in this study were
diagnosed using DSM-III criteria [APA, 1980] and may
have had more severe symptoms in childhood than
would children diagnosed using the broader DSM-IV
criteria [APA, 1994].
The sample was drawn from a unique geographic and
cultural group with a significant religious influence,
offering advantages and disadvantages to this project.
The advantages were many, including a general will-
ingness to help others and a strong focus on family
relationships that may benefit individuals with AD.
A disadvantage for this project could be that this sample
may be less representative of the experience of other
adults with autism and average-range cognitive abilities
than would be the case with a sample from a more
This study demonstrates both the feasibility and value of
longitudinal autism research. While progress is being
made by cross-sectional research in understanding aut-
ism spectrum disorders, longitudinal work is required to
uncover childhood variables associated with level of
functioning in adults with AD. While some individuals
in our DSM-III defined sample were quite successful as
adults, others were not. Identifying environmental sup-
ports associated with better outcome may guide decisions
made by parents, policy makers, and clinicians.
This study encourages further research with the larger
sample that was part of the epidemiological survey 20
years ago. Results from the current sample of individuals
with DSM-III diagnosed AD and IQZ70 should be
complemented by the full sample of 241 DSM-III positive
116 Farley et al./Outcome for adults
and 138 children who did not meet DSM-III criteria.
This unique source of longitudinal data is a valuable
resource thatis highly likely
new information on the natural history of autism.
Challenges lay in recruitment, retention, and the time
remaining for meaningful participation by the subject’s
aging parents. Nevertheless, the potential gains from
such future studies are visible on the horizon and call for
We are grateful to the adults with autism and their family
members who participated in this project. Partial support
for all datasets within the Utah Population Database
(UPDB) was provided by the University of Utah Hunts-
man Cancer Institute. Our thanks also go to Jubel
Morgan, R.N., Annie Ashcraft, M.S., Michele Villalobos,
M.S., Deborah Bilder, M.D., and Dale Cannon, Ph.D. for
their work on this project.
American Psychiatric Association. (1980). Diagnostic and
statistical manual of mental disorders, 3e. Washington, DC:
American Psychiatric Association.
American Psychiatric Association. (1994). Diagnostic and
statistical manual of mental disorders, 4e. Washington, DC:
American Psychiatric Association.
Billstedt, E., Gillberg, C., & Gillberg, C. (2005). Autism after
adolescence: population-based 13- to 22-year follow-up study
of 120 individuals with autism diagnosed in childhood.
Journalof Autismand Developmental
Eaves, L.C., & Ho, H. (2008). Young adult outcome of autism
spectrum disorders. Journal of Autism and Developmental
Disorders, 38, 739–747.
Engstrom, I., Ekstrom, L., & Emilsson, B. (2003). Psychosocial
functioning in a group of Swedish adults with Asperger
syndrome or high-functioning autism. Autism, 7, 99–110.
Fombonne, E. (2005). The changing epidemiology of autism.
Journal of Applied Research in Intellectual Disabilities, 2005,
Ghaziuddin, M., Weidmer-Mikhail, E., & Ghaziuddin, N.
(1998). Comorbidity of Asperger syndrome: a preliminary
report. Journalof Intellectual
Gillberg, C. (1991). Outcome in autism and autistic-like
conditions. Journal of the American Academy of Child and
Adolescent Psychiatry, 30, 375–382.
Gillberg, C., & Steffenburg, S. (1987). Outcome and prognostic
factors in infantile autism and similar conditions: a popula-
tion-based study of 46 cases followed through puberty.
Journalof Autismand Developmental
Howlin, P. (2002). Autistic disorders. In: Howlin P, editor.
Outcomes in neurodevelopmental and genetic disorders.
NewYork: Cambridge University Press.
Howlin, P. (2003). Outcome in high-functioning adults with
autism with and without early language delays: implications
for the differentiation between autism and Asperger syn-
drome. Journal of Autism and Developmental Disabilities, 33,
Howlin, P., Goode, S., Hutton, J., & Rutter, M. (2004). Adult
outcome for children with autism. Journal of Child Psychol-
ogy and Psychiatry, 45, 212–229.
Howlin, P., Mawhood, L., & Rutter, M. (2000). Autism and
developmental receptive language disorder—a follow-up
comparison in early adult life. II: social, behavioural, and
psychiatric outcomes. Journal of Child Psychology and
Psychiatry, 41, 561–578.
Kanner, L. (1971). Follow-up study of eleven autistic children
originally reported in 1943. Journal of Autism and Childhood
Schizophrenia, 1, 119–145.
follow-up study of 201 children with autism in Kyushu and
Yamaguchi areas, Japan. Journal of Autism and Develop-
mental Disorders, 22, 395–411.
Lord, C., Risi, S., Lambrecht, L., Cook, E.H., Leventhal, B.L., et al.
(2000). The Autism Diagnostic Observation Schedule—Gen-
eric: a standard measure of social and communication deficits
associated with the spectrum of autism. Journal of Autism
and Developmental Disorders, 30, 205–223.
Lord, C., Rutter, M., & LeCouteur, A. (1994). Autism Diagnostic
Interview—Revised: a revised version of a diagnostic inter-
view for caregivers of individuals with possible pervasive
developmental disorders. Journal of Autism and Develop-
mental Disorders, 24, 659–685.
Lotter, V. (1974). Social adjustment and placement of autistic
children in Middlesex: a follow-up study. Journal of Autism
and Childhood Schizophrenia, 4, 11–32.
Mawhood, L., Howlin, P., & Rutter, M. (2000). Autism and
developmental receptive language disorder—a comparative
follow-up in early adult life. I: Cognitive and language
outcomes. Journal of Child Psychology & Psychiatry, 41,
Nordin, V., & Gillberg, C. (1998). The long-term course of
autistic disorders: update on follow-up studies. Acta Psychia-
trica Scandinavica, 97, 99–108.
Piven, J., Harper, J., Palmer, P., & Arndt, S. (1996). Course of
behavioral change in autism: a retrospective study of high-IQ
adolescents and adults. Journal of the American Academy of
Child and Adolescent Psychiatry, 35(4), 523–529.
Ritvo, E.R., Freeman, B.J., Pingree, C., Mason-Brothers, A., Jorde, L.,
et al. (1989). The UCLA—University of Utah Epidemiological
Survey of Autism: prevalence. American Journal of Psychiatry,
Rumsey, J.M. (1992). Neuropsychological studies of high-level
autism. In: Schopler E, Mesibov GB, editors. High-function-
ing individuals with autism. New York: Plenum, pp 41–64.
Rumsey, J.M., Rapoport, J.L., & Sceery, W.R. (1985). Autistic
children as adults: psychiatric, social, and behavioral out-
comes. Journal of the American Academy of Child and
Adolescent Psychiatry, 24, 465–473.
& Yoshinaga,K. (1992).A
Farley et al./Outcome for adults117
Seltzer, M.M., Krauss, M.W., Shattuck, P.T., Orsmond, G., Swe, A., Download full-text
& Lord, C. (2003). The symptoms of autism spectrum
disorders in adolescence and adulthood. Journal of Autism
and Developmental Disorders, 33, 565–581.
Sparrow, S.S., Balla, D.A., & Cicchetti, D.V. (1984). Survey form
manual for the interview edition of the Vineland Adaptive
Behavior Scales. Circle Pines, MN: American Guidance Service.
Szatmari, P., Bartolucci, G., Brenner, R., Bond, S., & Rich, S.
(1989). A follow-up study of high-functioning autistic
children. Journal of Autism and Developmental Disorders,
Wechsler, D. (1997). Manual for the Wechsler Adult Intelligence
118 Farley et al./Outcome for adults