The functional impact of cognition in adults with autism spectrum disorders

Abstract

Purpose and aim: The overall aim of this study was to examine the relationship between adaptive function and cognitive factors in young adults diagnosed with autism spectrum disorder (ASD) in adult age.

Methods: The study included 30 adults (age 18-30) diagnosed with ASD in adulthood. All participants were clinically referred to an adult psychiatric clinic for assessment. Adaptive functioning was measured with Adaptive Behavior Assessment System – 2nd edition (parent version). Wechsler scales of intelligence and Delis-Kaplan Executive Function System were used to measure intelligence and executive function.

Results: We found considerable adaptive functioning deficits regardless of Full Scale Intelligence Quotient (FSIQ) level. FSIQ, working memory and processing speed were positively associated with adaptive functioning. No associations were found between adaptive functioning and cognitive flexibility, inhibition, word generation or shifting. Regression analysis showed that working memory and processing speed predicted 23% of the variance in adaptive functioning in this group.

Conclusions: The results suggest that cognitive dysfunction could be an important area for intervention to improve adaptive functioning in ASD.

Keywords:

• Neurodevelopmental disorders
• autism spectrum disorder
• adaptive behaviour
• executive function
• cognitive ability

Introduction

Autism spectrum disorders (ASDs), which have an onset in early childhood, show heterogeneity of aetiology, clinical phenotype expression and severity in functional impairment, cognitive impairment, a strong genetic influence, probable environmental influences, and, a marked male preponderance [1]. Long-term follow-up studies of children with ASD show that ASD persists into adulthood [2–4] in a majority of cases. In addition, some individuals with ASD have received the diagnosis in adulthood even though the symptoms have been manifest from early childhood [5].

Studies have reported that children with ASD often show some degree of cognitive impairment, particularly executive dysfunction [6,7], and varying severity in functional impairment [1]. It has been suggested that adaptive skills are highly dependent on executive functions [8]. The executive function system is a higher-order control system that manages goal-directed behaviour and regulates lower-order functions such as motor skills, short-term memory, and language [9]. Executive functions include, for example, functions regulating attention, inhibition, planning and activity regulation [10]. Adaptive functioning, or adaptive skills, are key skills that every individual needs in order to manage the demands and expectations of everyday life. The American Association on Intellectual and Developmental Disabilities (AAIDD) has identified three clusters of adaptive skills. These are practical, social and conceptual skills needed to cope with everyday life [8]. Two of the psychometric assessment systems designed to measure the adaptive skills are the Adaptive Behaviour Assessment System (ABAS, [8]) and the Vineland Adaptive Behaviour Scales (VABS, [11]).

In the general population, intelligence (IQ) and adaptive behaviour are related. Correlation studies between ABAS-II and Wechsler Intelligence Scale for Children, fourth edition (WISC-IV, [12]) in the general population show a correlation between the global measure General Adaptive Composite and IQ of .41 (parent form) and .58 (teacher form) [8]. In recent research, clear discrepancies between IQ and adaptive functioning in children and adolescents with ASDs and without intellectual disabilities (IQ > 70) have been reported [13,14]. IQ has been reported to be superior to adaptive functioning in those with ASD and with no intellectual disability [15].

The knowledge regarding cognitive and adaptive functioning in adults with ASD is not as extensive as that of children and adolescents, and the knowledge regarding those that have not received their ASD diagnosis until adult age is even scarcer. All available research to this date points towards a substantial gap between IQ and adaptive function in persons with ASD across all age groups [16–19]. Recent research [17,19] has confirmed, as proposed by Pugliese et al. [18], that adaptive deficits in children and early adolescents worsen during late adolescence. Furthermore, Kraper et al. [17] showed substantial adaptive deficits in early adulthood, and Tillmann et al. [19] reported that adaptive deficits increase in the ASD group with age. Pugliese et al. [18] showed that executive function, and specifically measures of metacognition in the Behavior Rating Inventory of Executive Function [20], had a significant impact on adaptive skills in late adolescent age in individuals with ASD.

As adaptive skills are highly dependent on executive functions [8], deficits in executive functions would possibly help explain the gap between IQ and adaptive skills, at least to some extent. Studies focusing on cognitive correlates (IQ and executive functions) to adaptive function have so far relied on self-report scales and/or full-scale IQ or IQ estimates and no study has so far, to our knowledge, investigated directly assessed cognitive abilities’ association with adaptive function, apart from full-scale IQ or estimated IQ.

The aims of this study are to investigate the association between cognitive abilities, including measures of IQ and executive functions, and adaptive function in adults with ASD. This will add to the knowledge-base about adult ASD patients which are still understudied, cognitive deficits might also be a key area to target with supportive interventions in order to raise the patients’ functional level. Hopefully, we might be able to show clinically suspected substantial discrepancies between IQ and adaptive skills.

Methods

Procedure

This study used archival data from neuropsychological assessments made between 2010 and 2017 performed at the neurodevelopmental disorders (NDD) Assessment Team at the Adult Psychiatric Clinic in Helsingborg, Sweden. The team specialises in carrying out assessments of NDD in adult psychiatric patients. The NDD Assessment Team consists of one psychiatrist, two neuropsychologists, one occupational therapist and one social worker. Inclusion criteria in the study were aged between 18 and 30 at the time of the assessment, and completion of the assessment, including IQ test, executive function tests and ABAS-II. All participants included in the study were found to fill the criteria for an ASD diagnosis (see Table 1). The age group was chosen due to the ABAS-II norms. No exclusion criteria were used in this group. All were assessed for clinical purposes such as diagnostic evaluation. During the assessments, the patients met a neuropsychologist for neuropsychological assessment five times, a psychiatrist for psychiatric evaluation two or three times, and an occupational therapist for a functional assessment three times. A social worker met the patients’ relatives two or three times for interviews about the patients’ developmental history. Based on all the available information, diagnoses of ASD and, in eight cases, also diagnoses of co-morbid attention disorders were assigned according to the ICD-10 with members of the assessment team reaching consensus. All members of the NDD assessment team had a long experience (minimum 7, maximum 15 years) of diagnosing ASDs as well as attention disorders and other NDDs.

Table 1. Diagnostic data of the total sample.

	Total N = 30	Males n = 21	Females n = 9
Age mean (SD)	21.6 (2.9)	22.1 (3.0)	20.7 (2.6)
Diagnosis
F84.0 autistic disorder	10	8	2
F84.1 atypical autism	10	6	4
F84.5 AS	10	7	3
Comorbid ADHD/F90.0X	8	5	3

AS: Asperger syndrome; ADHD: attention-deficit/hyperactivity disorder.

Informed consent (oral or in writing) was obtained from all participants; all participants were given written and oral information about the study. The participants who gave oral consent (N = 7) did so because they themselves reported severe problems with activities in daily life and themselves believed that they would not remember signing and posting the consent form. Ethical approval was gained from the Regional Ethics Review Board in Lund, Sweden (Protocol no. 2015/696).

Participants

Thirty participants were included in this study; 21 males and 9 females, all with an ASD diagnosis according to ICD-10. They had all been diagnosed in adult age. For the presentation of age and diagnoses, see Table 1. No age differences between males or females were found. Eight participants received attention-deficit disorder diagnoses in addition to their ASD diagnosis (three with F90.0B ADHD, one with F90.0C ADD and four with F90.0X attention deficit disorder NOS). Nine of the participants had severe psychiatric comorbidity (recurrent depression, bipolar disorder, schizoaffective disorder, mixed anxiety/depression, personality disorder, and obsessive-compulsive disorder). The academic history in the study group was as follows: 26 (87%) had completed 9 years of compulsory school, and 10 (33%) had completed high school/upper secondary school. The current vocational status for the participants was as follows: 22 (73%) were unemployed with some form of welfare benefits, 6 (20%) were studying and 2 (7%) worked part-time. None worked full-time.

Measures

Wechsler intelligence scales

Wechsler Adult Intelligence Scale – Fourth Edition (WAIS-IV, [21]) is one of the most commonly used intelligence tests for adults. An Index Score (mean = 100, s.d.=15) is generated for Full-Scale Intelligence Quotient (FSIQ) and for each of the four indexes; Verbal Comprehension Index (VCI), Perceptual Comprehension Index (PCI), Working Memory Index (WMI) and Processing Speed Index (PSI). Both WMI and PSI target different aspects of executive functions. WMI requires working memory and cognitive flexibility and PSI requires, besides visual processing ability, organized response strategy, cognitive flexibility, and concentration. Twenty-eight of the participants were assessed with the WAIS-IV. Two of the participants were assessed with the Wechsler Adult Intelligence Scale – Third edition [22]. WAIS-III has a similar construction as WAIS-IV and also generates FSIQ, VCI, PCI, WMI and PSI. WAIS-III has been found to correlate strongly with WAIS-IV. Reported corrected correlations between WAIS-IV and WAIS-III indexes are between r = 0.84 – 0.91 and for FSIQ r = 0.94 [21].

The adaptive behaviour assessment system (2nd edition) (ABAS-II), parent version

The ABAS-II [8] is a parent questionnaire (also available as a teacher questionnaire) assessing behavioural manifestation of adaptive functioning/activities of daily living in children, adolescents and young adults ages 5–21, but has also been used in older age groups [23,24]. ABAS-II was completed by a parent in the current study. Scores are presented as full-scale General Adaptive Composite (GAC) and also divided into the subscales Conceptual domain, Social domain and Practical domain. Composite scores have a norm-referenced mean of 100 and standard deviation of 15.

The Delis-Kaplan executive function system (D-KEFS)

D-KEFS [25] is a battery of neuropsychological tests, with a mean of 10 and a standard deviation of three, designed to measure different executive functions in children, adolescents and adults. Subtests from four of the nine D-KEFS tests were included in this study, all measuring functions that have been reported to be affected in NDDs [6]. The D-KEFS subtests used in this study were Trail Making Test (TMT), Verbal Fluency Test (VF), Colour-Word Interference Test (CWIT), and Proverb test.

D-KEFS – TMT. The TMT consists of five trials; visual scanning, number sequencing, letter sequencing, number-letter switching, and motor speed. The fourth condition was used in this study as a measure of mental flexibility demanding visual set-shifting and working memory. The primary measure is the time needed to complete the task; this time has then been derived to a scaled score (M = 10, SD = 3) that expresses the subject's performance compared to the normative group.

D-KEFS – VF test. There are three conditions in the VF test in which the examinee says as many words as possible by letter (condition 1), category (condition 2), and category switching prompts (condition 3). The primary measures in conditions 1 and 2 are the number of words produced in one minute; this number has been derived to a scaled score. In this study, we used the difference between conditions 1 and 2 as a measure of abstract word-generation and non-overlearned word-generation. In condition 3, we used the score measuring the number of words produced during one minute while switching categories. In condition 4 (measuring inhibition and switching), we used the score measuring the number of times the examinee switches verbal categories. This score was included as a measure of mental flexibility.

D-KEFS – CWIT. CWIT includes the following four conditions; colour naming (condition 1), word reading (condition 2), colour-word interference trials (condition 3), and switching between inhibitory and non-inhibitory responses (condition 4). The third and fourth conditions were used in the analysis. The third condition measures the ability to inhibit the salient, automatic task of reading words in order to name dissonant ink colours rapidly. The fourth condition measures the naming speed, reading speed, verbal inhibition and, most importantly, cognitive flexibility. On both these conditions, the primary measure is the time needed to complete the tasks, which has been derived from a scaled score. The third condition was included as inhibition, or the ability to not get distracted, which theoretically would be a helpful trait regarding ADL. The fourth condition was included as an additional measure of cognitive flexibility.

D-KEFS – Proverb test. The Proverb test requires the participant to interpret 8 sayings and includes two conditions; free inquiry (condition 1) and the multiple-choice (condition 2). The first condition was used in this study which measures fundamental language skills and verbal abstract thinking, semantic integration and generalisation. The primary measure is the number of correct answers, the grade of correctness and also grade of abstraction. This has been derived to a scaled score. This was included because abstract thinking, as well as generalisation theoretically, would be important abilities for succeeding with activities in daily life.

Data analysis

Analyses were conducted using IBM SPSS Statistics, version 22. All variables were controlled for normal distribution in order to make sure that parametric tests could be used. Pearson correlation analyses were made to test the correlations between the ABAS-II domains, GAC (the global, general adaptive function measure of the ABAS-II) and the results from all cognitive and executive function tests. To explore the contribution of the WAIS-IV indexes and executive function measures to GAC, linear regressions were run with GAC as the dependent variable and, respectively, each correlating WAIS subscale and executive function measure. A post-hoc analysis (multiple linear regression) was run with GAC as the dependent variable and WMI and PSI as independent variables. The statistical significance criterion was set a priori at (alfa) p = 0.05. As this is an explorative study, a large number of analyses were made, especially the correlation analyses. Corrections were not made in the correlation tests because the resulting alfa would be extremely low and cause very large amounts of type 2 errors.

Results

Intellectual, executive and adaptive function

According to WAIS-IV, the intellectual function was in the average range (Table 2) with one outlier, with an FSIQ range between 64 and 123. The outlier had substantial irregularities in the WAIS profile and was not assessed as filling criteria for mild intellectual disability. Scores on the EF tests were also in the average range, except for the Proverb test which was lower. All domains regarding adaptive function in ABAS-II were significantly below the average range according to the norms, except for the practical composite.

Table 2. Intellectual, executive and adaptive function compared to the normative sample.

Measure	Mean (SD)
WAIS index (M = 100, SD = 15)
Verbal comprehension index n = 30	91.8 (13.2)
Perceptual reasoning index n = 30	101.3 (16.9)
Working memory index n = 30	91.3 (12.0)
Processing speed index n = 30	92.4 (14.4)
Full scale IQ n = 30	93.3 (13.0)
ABAS scores (M = 100, SD = 15)
Conceptual domain n = 30	61.5 (18.1)
Social domain n = 30	55.3 (9.4)
Practical domain n = 30	72.5 (18.6)
Global adaptive composite n = 30	62.1 (15.7)
D-KEFS (M = 10, SD =3)
Trail Making Test 4, n = 28	7.4 (3.3)
Verbal fluency 1, n = 27	7.9 (2.8)
Verbal fluency 2, n = 27	9.7 (4.0)
Verbal fluency 3, n = 27	7.9 (3.9)
Verbal fluency 2-1, n = 27	8.2 (2.7)
Colour word interference 3, n = 27	8.5 (4.0)
Colour word interference 4, n = 27	7.9 (3.9)
Proverb test, n = 26	5.7 (4.0)

Correlations between ABAS-II, WAIS, and D-KEFS

Moderate correlations were found between GAC (which summarises performance across all ABAS-II skill areas) and WMI, and between GAC and PSI (Table 3). The analysis also showed a correlation between GAC and FSIQ of .38, indicating a weak correlation. No correlations were found between GAC and any of the executive function tests in D-KEFS.

Table 3. Statistically significant Pearson’s correlations between ABAS-II, WAIS and D-KEFS.

	n	p	Pearson's r
Correlation to general adaptive composite (ABAS-II)
WAIS working memory index	30	.02	0.44
WAIS processing speed index	30	.01	0.45
WAIS full scale IQ	30	.04	0.38
Correlation to conceptual domain (ABAS-II)
WAIS working memory index	30	.01	0.46
WAIS processing speed index	30	<.01	0.49
WAIS full scale IQ	30	<.01	0.49
D-KEFS TMT condition 4	28	<.01	0.48
D-KEFS CWI condition 3	27	.03	0.43
D-KEFS CWI condition 4	27	.05	0.38
Correlation to social domain (ABAS-II)
D-KEFS verbal fluency condition 4	27	.04	–0.39
Correlations to practical domain (ABAS-II)
WAIS verbal comprehension index	30	.05	0.36
WAIS working memory index	30	<.01	0.47
WAIS processing speed index	30	.03	0.40

Regarding ABAS-II skill domains we found moderate correlations between the Conceptual domain and the WAIS indexes WMI, PSI and FSIQ as well as between Conceptual domain and D-KEFS subtest TMT 4 and CWI 3. The correlation between the Conceptual domain and CWI 4 was significant albeit weaker. A weak, negative correlation was found between the Social domain and D-KEFS Verbal Fluency condition 4. Finally, the Practical domain was moderately correlated to the WAIS indexes WMI and PSI, whereas a weak correlation was found to VCI.

Predictive factors

WMI was a significant positive predictor of GAC – scores, accounting for 16.6% of the variance (F_{1, 28}=6.78, p < .05), FSIQ accounted for 11.5% of the variance (F_{1, 28}=4.75, p < .05) and PSI accounted for 17.0% of the variance (F_{1, 28}=6.93, p < .05) (Table 4).

Table 4. Linear regression models with the dependent factor general adaptive composite (ABAS-II) and independent variables.

Independent variable	$\beta$	SE β	t	p	Constant
WAIS working memory index	0.58	0.22	2.60	.02	9.46
WAIS processing speed index	0.49	0.19	2.63	.01	17.22
WAIS full scale IQ	0.46	0.21	2.18	.04	18.97

Due to the limited sample size, only two independent factors (WMI and PSI) were entered into a regression model with GAC as the dependent factor. The model was significant (F_{2, 27}=5.54, p = 0.01) and accounted for 23.8% of the variance in GAC. The differences within the model were not significant.

Difference between groups

Mann–Whitney U-test was performed to investigate differences between the group with co-morbid attention diagnoses (ADHD/ADD/ADD NOS, n = 8) and the group without co-morbid attention diagnoses (n = 22). The only factor that differed significantly between the two groups was the word context test, where the group with co-morbid ADHD diagnoses had significantly better results.

Discussion

This study aims at bringing further understanding of the adaptive functioning of adults with ASDs; if and how the adaptive functioning relates to neuropsychological factors. On the one hand, we found IQ levels in the normal range, slightly below the population average and on the other hand, we noticed large deficits in adaptive function, two standard deviations below the population average or lower, globally and in every domain (except for the Practical domain, which was only slightly above that definition). We also found results on executive function tests within the normal range, though below the population average. The exception was the Proverb test, where the results were more than two standard deviations below the population average.

The global measure of adaptive function, the GAC, was weakly correlated to IQ, a correlation comparable with what is reported in the general population [8]. There were moderate correlations between GAC and the WAIS indexes WMI and PSI. WMI and PSI, which measure different aspects of EF, combined proved to explain one fifth to one-fourth of the variance in GAC. Several correlations were also found between D-KEFS tests of executive function and the Conceptual domain of ABAS-II, but no associations could be found between the executive function tests and GAC or the sub-domains, except for a peculiar weak negative correlation between social domain and the Verbal Fluency Test.

The results indicate that, in this group, an adult person with ASD, and regardless of average or above-average IQ, has substantial adaptive deficits, which is also supported in other studies [7,14]. The relationship between FSIQ and GAC is evident throughout the IQ scale. This might be a bit surprising since a higher IQ theoretically would mean better prerequisites for compensating other deficiencies. However, the low adaptive function even in individuals with superior IQ and ASD mirrors clinical experience.

Few studies have so far focused on how cognitive abilities impact on adaptive aspects of adults with ASD. Wallace et al. [7] found that executive function, measured by parent rating, was robustly associated with both adaptive functioning and co-morbid symptoms of depression and anxiety in ASD, which lead the authors to suggest that executive function problems should be a focus of evaluation and intervention amongst adults with ASD. The current study contributes by adding a bit of nuance; it seems like working memory and processing speed, which are broader and more general measures of executive function, are more central in the functional deficits than the more narrow neuropsychological functions such as cognitive flexibility or inhibition. It seems logical that disturbances in the distributed networks of several executive functions rather than specific, isolated executive components have an influence on adaptive functioning.

In a study by Kenworthy et al. [26], no significant correlation between IQ and adaptive function was found in an ASD group between 12 and 21 years of age. This might be explained by the fact that during adolescence daily life is more structured and adolescents receive more support during this time (from parents, school, etc.) which might explain why IQ in this age group is not correlating with adaptive function. The gap between IQ and adaptive functioning in our study might be explained by the fact that our study group has passed through childhood and adolescence with an undiagnosed ASD and with a lack of support and understanding for their ASD. A substantial number of adult patients with undiagnosed ASD in the psychiatric system present with severe psychiatric symptoms and disorders. The core ASD is often clouded by these, in effect, co-morbid psychiatric disorders. Psychiatric treatment is in many cases seriously hampered and effective treatment is delayed as the team treating the patient is unaware of the underlying ASD. As it is neither difficult nor time-consuming to measure adaptive behaviour long-term and quantitatively, it may prove to be a useful tool – among others – for cutting through the co-morbidities and spotting the core deficits. Making greater use of adaptive function measures might help identify which patients in both out- and inpatient adult psychiatric care need to be referred for a more thorough neuropsychiatric assessment.

A number of factors seem to be influencing adaptive functioning in adults with ASDs. Kraper et al. [17] found significant correlations between levels of psychiatric comorbidity and adaptive function, and Tillmann et al. [19] found that the core social communications deficits contribute to deficits in adaptive function. We have shown that impairments in executive function (WMI and PSI – the broader measures) also make substantial contributions to the adaptive function deficits in adults with ASDs. At the same time, the more narrow measures of executive function did not affect adaptive function.

Psychiatric comorbid disorders, such as depression, and anxiety, as well as other NDDs such as ADHD, which frequently occurs in adults with ASD [3,5], causes major cognitive and functional deficits [27–29]. It is not fully clear how much of the functional impairment is related to co-morbid disorders. Gillberg and Fernell [30] has suggested that co-morbidities have a greater impact on prognosis and function in ASD than the core ASD symptoms themselves. It is reasonable to assume that the presence of a co-morbid disorder would also have an impact on cognitive function, but the current knowledge-base is somewhat incongruent. There is e.g. currently conflicting evidence regarding whether psychiatric symptoms co-morbid to the core ASD (for example ADHD, depression and anxiety disorders) are associated with adaptive deficits [17] or not [19]. So far, the clinical features most often associated with adaptive deficits in the ASD group seem to be age [15,19], the severity of autistic symptoms [19], and executive function – which was shown by Pugliese et al. [18] and also in this study.

Limitations in the study are the small sample size, with an even smaller proportion of women. There were high rates of psychiatric comorbidity in the study group, which could affect the test results negatively; though we still argue that this is a representative group as the risk of developing psychiatric symptoms in the ASD group as a whole is greatly increased compared to the non-ASD population [31,32]. This study relied on archival data; therefore, it was not possible to include a control group. Instead, we used the test norms as a means of control; the norms of the different tests and measures used are drawn from a general population and not from a psychiatric population. This is to our knowledge the first study of cognitive and adaptive function, based on testing, in adults who are identified as having ASD in adult age. Most studies regarding cognition and adaptive function in adults with ASDs are follow-up studies, based on adults who received diagnosis during childhood or early adolescence (see for example Billstedt et al. [2]).

Ethical approval

All procedures performed in this study were in accordance with the ethical codes of the Regional Ethics Review Board in Lund, Sweden.

Informed consent

Informed consent was obtained from all participants included in the study.

Acknowledgments

The authors thank the participants who willingly participated in this study and the staff at the NDD Assessment team in Helsingborg, Sweden for contributing to the assessments of the study participants.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by Region Skåne (the public healthcare district of Skåne, Sweden).

Notes on contributors

Johan Nyrenius

Johan Nyrenius is a clinical psychologist and a specialist in neuropsychology. He is a PhD student at the University of Gothenburg. His PhD project is aimed at exploring adult psychiatry patients with previously undiagnosed ASD in terms of psychiatric co-morbidity, adaptive functioning, psychosocial and cognitive factors, suicidality and alcohol/substance use.

Eva Billstedt

Eva Billstedt, PhD is a professor and director of the Gillberg Neuropsychiatry Centre at the Institute of Neuroscience and Physiology at the University of Gothenburg. She is the author of over 65 original peer-reviewed publications on neurodevelopmental disorders. Her research interests include all aspects of neurodevelopmental disorders, especially autism and outcomes of neurodevelopmental disorders. She is supervising Johan Nyrenius’ PhD – project.