IQ in Autism Spectrum Disorder: A Population-Based Birth Cohort Study

Olmsted County, Minnesota, is 90 miles from any urban center and is home to the Mayo Clinic. According to the US census data in 2000, the Olmsted County population is comparable to the US population with respect to median age (35.0 years versus 35.3 years) and male sex (49.1% vs 49.1%).¹⁹  The Olmsted County population is, however, more racially homogenous than the US population (White: 90.3% vs 75.1%), thus minimizing potential confounding due to race and ethnicity.²⁰ 

The birth cohort consists of all children born from January 1, 1976, through December 31, 2000, to mothers who were residents of Olmsted County at the time of the child’s birth (N = 43 215).²¹  Among these 43 215 children, medical record access for research purposes was available for 39 890, in accordance with Minnesota state privacy law, statute 144.335. The target population for this study consisted of the 31 220 children who still lived in Olmsted County at or after 3 years of age. These children were longitudinally followed from birth until adulthood, emigration, or death by using medical and school records. The cutoff age of 3 years was chosen because social, communication, and other behavioral characteristics of ASD are more consistently recognized at or beyond this age.²⁰ 

Medical and school records were used to identify cases with possible ASD. The Institutional Review Boards of the Mayo Clinic, Olmsted Medical Center, and the Independent School District 535 School Board approved a contractual research agreement to access school records of all birth cohort members registered at any of the 41 public, parochial, or private schools, including those who moved from the school district, were home-schooled, died, or graduated. The cumulative school record includes all school assessments, educational intervention reports, and notes related to any type of difficulty in learning or behavior, as observed by teachers, parents, school psychologists, physicians, social workers, school nurses, or counselors.

Medical records of all birth cohort members were also reviewed. The Mayo Clinic and Olmsted Medical Center provide virtually all primary and specialty care to residents of Olmsted County. Through the Rochester Epidemiology Project, all diagnoses recorded at Rochester medical facilities are indexed for automated retrieval.²¹  The medical record includes a detailed history of all medical encounters in the community, including social services, hospitalizations, emergency department and home visits as well as laboratory, psychiatry, and psychology reports and test results, from birth until the patient no longer resides in the community or is deceased.²² 

A comprehensive description of the ASD incident case ascertainment method has been previously published.²⁰  Three major steps were completed to identify all ASD incident cases based on two operational research definitions, and these were led by a study team that consisted of a senior child psychologist, two developmental-behavioral pediatricians, a speech language pathologist, and an MD epidemiologist. The process of identifying ASD cases began with identifying individuals with selected relevant neurodevelopmental and psychiatric disorders on the basis of medical and school records, then narrowing this group down to potential ASD subjects. The medical and school records of these potential ASD subjects were abstracted in detail through a multistage process, allowing accumulation of all details related to ASD symptoms. Finally, the third step consisted of determining ASD incident case status on the basis of the following two definitions: (1) an inclusive definition (research-defined autism spectrum disorder, inclusive criteria [ASD-R_I]) based on DSM-IV-TR criteria for autistic disorder, Asperger disorder, and pervasive developmental disorder not otherwise specified and (2) a more narrow, conservative definition (research-defined autism spectrum disorder, narrow criteria [ASD-R_N]) based on DSM-IV-TR criteria for autistic disorder. Clinical diagnoses (clinically diagnosed autism spectrum disorder [ASD-C]) were also abstracted from medical and school records on the basis of Hospital International Classification of Diseases Adapted (Mayo Clinic–specific) and International Classification of Diseases, Ninth Revision, diagnosis codes. This method yielded 1296 individuals who had at least two social criteria and at least one communication or restricted or repetitive criterion.²⁰  Patients with false-positive diagnoses were excluded by manual review to determine if their symptoms were related to a different condition, such as major depression. After exclusion of false-positive diagnoses, 1056 subjects met the criteria for ASD-R_I.

Intellectual ability was based on the last IQ score obtained between the ages of 3 and 21 years. A documented clinical diagnosis of ID was used for those subjects without available IQ scores. Full-scale IQ scores from the following IQ tests were used: Wechsler Intelligence Scale for Children, Wechsler Adult Intelligence Scale, Wechsler Abbreviated Scale of Intelligence, Wechsler Preschool and Primary Scale of Intelligence, Stanford-Binet Intelligence Scales, and Leiter International Performance Scale-Revised. Brief IQ score from the Leiter was included if there was no full-scale IQ score recorded for this test. In addition, the overall age summary score was used from the Woodcock-Johnson Tests of Cognitive Abilities. Intellectual ability was categorized as follows: moderate to profound ID (IQ: ≤50); mild ID (IQ: 51 to 70); borderline IQ (IQ: 71 to 85); average IQ (IQ: 86 to 115); above average IQ (IQ: 116 to 130); superior IQ (IQ: ≥131).

Data were descriptively summarized by using frequency counts and percentages for categorical variables and mean and SD for continuous variables. For each ASD definition, we compared the proportion with average or higher intellectual ability between boys and girls using the χ² test and reported this ratio as a relative risk along with the corresponding 95% confidence interval. Results are presented overall and stratified by era (diagnosed with or met criteria for ASD before versus after January 1, 2000). Given that patients in this cohort were diagnosed with ASD between 1985 and 2015, the midpoint of January 1, 2000, was chosen as a cutoff to define the two eras. Comparisons between the two eras were evaluated by using the χ² test. Formal statistical comparisons between groups of subjects who met the ASD definitions were not performed because the groups were not mutually exclusive. Statistical analyses were performed by using the SAS version 9.4 software package (SAS Institute, Inc, Cary, NC).

A total of 1056 subjects met criteria for ASD-R_I, of which 533 met criteria for ASD-R_N. Of 242 subjects who met criteria for ASD-C, 210 met criteria for both ASD-R_I and ASD-R_N (24 for ASD-R_I only), and 8 only had a clinical diagnosis without documented ASD symptoms to fulfill either the ASD-R_I or ASD-R_N criteria. The 1056 subjects who met criteria for ASD-R_I included 774 (73.3%) boys and 282 (26.7%) girls, of which the documented race was White for 93.8% (967 of 1031).

Among 1056 subjects who met criteria for ASD-R_I, 890 (84.3%) had either documented IQ scores (N = 867) or only an ID diagnosis in the absence of an IQ score (N = 23). These 890 subjects were similar in terms of sex (% male: 73.3% vs 73.5%) and race (% White: 92.0% vs. 89.2%) to the 166 subjects without documented IQ scores or ID diagnoses. The following results are restricted to subjects with either documented IQ scores or only ID diagnosis. Characteristics of these subjects are summarized in Table 1.

Figure 1A summarizes the distribution of intellectual ability at last assessment obtained between ages of 3 and 21 years, according to ASD criteria. The proportion of subjects who scored in the average or higher IQ range increased as the ASD definition moved from only a clinical diagnosis (ASD-C: 42.8% [80 of 187]), to a narrow definition (ASD-R_N: 51.2% [232 of 453]) to the most inclusive definition (ASD-R_I: 59.1% [526 of 890]). This same pattern was observed when results were stratified by era in which the patients were diagnosed with or met criteria for ASD (Figs 1B and 1C). However, when the results were contrasted between the two eras, the proportion of subjects who scored in the average or higher IQ range was significantly higher in the later era (27.4% vs. 50.4%, P = .003) among those who met the ASD-C definition but not among those who met either ASD-R_I definition (58.0% vs 60.5%; P = .46) or ASD-R_N definition (49.1% vs 53.4%; P = .36).

Within ASD-R_I and ASD-R_N groups, boys were more likely than girls to have average or higher IQ with relative risks of 1.21 (P = .004) and 1.27 (P = .03), respectively (Fig 2A; Table 2). Although a relative risk of similar magnitude was observed in ASD-C group (relative risk: 1.39; P = .17), there was less statistical power, given that there were only 34 girls in the ASD-C group with IQ data available. Similar results were observed when stratified by era (Figs 2B and 2C; Table 2), although male to female relative risks in ASD-R_N group and ASD-C groups were a bit lower in the more recent era.

This large, longitudinal, population-based birth cohort study shows that, when using active ascertainment and uniform application of DSM-IV-TR–based research criteria for ASD, there is higher rate of average or higher IQ for ASD cases (51.2% to 59.1%) than in previous studies. The most inclusive definition of research-identified ASD cases had the highest percentage of individuals with average or above average IQ, and this proportion was similar when compared across two time periods. It is important to note that by all three ASD definitions in our study, the prevalence of ID (ASD-C: 34.8%; ASD-R_N: 26.1%; ASD-R_I: 18.6%) was greater than what would be expected in the general population (2.5%). We also confirmed that girls with ASD are less likely to have average or higher IQ than boys with ASD, no matter which research definition of ASD is considered. Finally, we ascertained that individuals who received a clinical diagnosis of ASD in our cohort had lower rates of average or higher IQ than individuals who met research criteria for ASD (42.8% vs 51.2% or 59.1%), but the proportion of subjects with a clinical diagnosis of ASD with average or higher IQ was significantly higher in the later era (27.4% vs 50.4%).

Using this epidemiologically rigorous study design, we found higher rates of average or higher IQ than found in previous reports.^4,8,10,11  According to the most recent data available from the CDC released in 2020, 42% of 8-year-old children with ASD and available IQ data were found to score in average or above average IQ range, which is comparable to our subjects with clinically diagnosed ASD but still a lower rate than our research-identified ASD cases.¹¹  Another study completed by the CDC showed that average annual prevalence rates of children with ASD and average to above average intellectual ability increased by 11.2% per year from 1996 to 2010.¹⁰  In contrast, prevalence rates of ASD with moderate to profound ID increased by only 3.1% per year in this study.¹⁰  These findings indicate that indeed more children with ASD and average or higher IQ are being identified. Another study, which relied on recruitment of ASD subjects via referrals to a specialty center, found only 29% of ASD cases scored in the average or higher IQ range.⁴  In contrast to some of these previous studies, our methods used a large, longitudinal, population-based birth cohort with detailed review of all medical and school records for research-defined ASD cases.

The clinically diagnosed ASD cases had the highest rate of ID, which suggests that ASD with associated ID is more easily recognizable by professionals. People with ASD and ID use the health care system at high rates.^23,24  Within our same research-defined ASD cohort, mean medical costs appeared to be much higher for ASD cases, with IQ <85.²⁵  Compared with children with ASD only, children with both ASD and ID tend to have more comorbidities, including seizures and self-injurious behaviors.^26,27  Children with ASD and average IQ may have a less obvious need for medical or special education services, placing them at higher risk of not being identified.

It remains crucial to identify children with average or higher IQ and ASD, as higher IQ does not necessarily correlate with improved future outcomes. Individuals with ASD and average or higher IQ have poor adaptive and social functioning and have difficulty with independent living.^28–31  People with ASD are, by definition, still impaired in their social communication abilities and have restricted and repetitive behaviors and interests. Despite average intelligence, these impairments may result in difficulties with developing meaningful relationships and maintaining employment. Thus, earlier identification and intervention for children with ASD and average IQ may improve social and occupational outcomes in this population.

Consistent with previous reports, our study also found that girls with ASD were less likely than boys with ASD to have average or higher IQ.^{6,7,12,13,15,32}  Our study also showed that, as the definition of ASD broadened, more girls scored in the average or higher IQ range. Some studies have found that girls are diagnosed with ASD later than boys,¹⁵  possibly indicating that ASD symptoms are not as apparent in girls at a young age. In addition, girls with ASD and average IQ may “camouflage” their difficulties with social communication and thus not meet full criteria for ASD.³²  Finally, there may be a subtler, female-specific autism phenotype that is not encompassed by the ASD criteria defined in the current Diagnostic and Statistical Manual of Mental Disorders.^34,35 

Our data also showed that individuals who received a clinical diagnosis of ASD documented in medical or school records (ASD-C) had lower rates of average or above average IQ, compared with that of research-identified ASD incident subjects. This finding suggests that children with ASD and average or higher IQ may be missed by clinicians. In another study, individuals with ASD and comorbid ID were more likely than those without ID to have existing community ASD diagnoses and were diagnosed with ASD earlier.²⁶  More accurate ascertainment of all ASD cases regardless of IQ is important for planning and delivery of medical, school, and community services.⁹ 

Our study had several potential limitations. First, according to the 2000 census data, the population of Olmsted County is more racially homogenous (White: 90.3% vs 75.1%), wealthier (median household income: $51 316 vs $41 994), and more highly educated (high school graduates: 91.1% vs 80.4%) than the general US population is, which may affect the generalizability of our findings.¹⁹  However, results from epidemiological studies in Olmsted County have generally been consistent with available national data.¹⁹  Second, our ASD incidence cases were identified via retrospective chart review, and we could not evaluate each case in person to confirm the diagnosis of ASD clinically. However, we employed rigorous training of research assistants and consistent review of symptoms noted in comprehensive medical and school records by a team of specialists in child development. Third, although ID is defined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition by impairments in both intellectual and adaptive functioning, our definition of ID relied only on standardized IQ measures. However, adaptive skills of children with ASD have been shown to be significantly lower than their IQ,⁸  whereas IQ has been shown to be a stable indicator of intelligence.³⁶  In addition, ID classification based on IQ alone is helpful for comparing epidemiological studies in which researchers investigate intellectual ability in the setting of ASD.³⁶  Finally, in our study, we used DSM-IV-TR criteria for our research-identified case definitions because ascertainment of ASD incidence cases was initiated before the release of Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition.

This large, longitudinal, population-based birth cohort reveals an increased rate of average to above average intelligence among individuals with ASD, compared with that of previous reports. Our data show that nearly half (42.8% to 59.1%) of individuals with ASD have average or higher IQ, and those identified with ASD clinically are less likely to have average or higher IQ than research-identified cases of ASD. By all three ASD definitions in our study, ASD is still associated with a much higher likelihood of ID compared with the prevalence in the general population. In addition, boys with ASD are more likely than girls with ASD to have average or higher IQ. Our results indicate that, despite public health efforts to promote early screening and identification of ASD, children (particularly girls) with ASD and average to above average IQ remain at risk for not being identified.

ASD

autism spectrum disorder

ASD-C

clinically diagnosed autism spectrum disorder

ASD-R_I

research-defined autism spectrum disorder, inclusive criteria

ASD-R_N

research-defined autism spectrum disorder, narrow criteria

CDC

Centers for Disease Control and Prevention

DSM-IV-TR

Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision

IQ

intelligence quotient