Psychiatric and Medical Conditions in Transition-Aged Individuals With ASD

The study population was drawn from the membership of Kaiser Permanente Northern California (KPNC), a large integrated health care delivery system providing care to over 3.8 million residents in the greater San Francisco and Sacramento areas. KPNC members have similar socio-demographic characteristics to other community members locally and statewide, except at the extreme ends of the income distribution.⁶ The Kaiser Health Plan provides insurance coverage for inpatient, outpatient, and mental health care provided at KPNC hospitals, clinics, and other approved facilities. All patient encounters are recorded in electronic medical records (EMRs).

All individuals 14 to 25 years of age as of January 1, 2014, who were members of KPNC for at least 9 months in each calendar year from 2013 to 2015 were eligible for inclusion (N = 385 051). Individuals with an ASD diagnosis recorded in the EMR before 2014 were considered cases (n = 4123; see Supplemental Table 3 for International Classification of Diseases, Ninth Revision [ICD-9] codes). Approximately 34% were diagnosed by specialists at a KPNC ASD evaluation center after a comprehensive clinical assessment, including the Autism Diagnostic Observation Schedule,⁷ and 66% had a diagnosis recorded by nonspecialists on 2 separate occasions.

Three comparison populations were chosen from among all individuals without any ASD diagnoses recorded in the EMR by the end of 2015. A neurodevelopmental group was randomly sampled from individuals with attention-deficit/hyperactivity disorder (ADHD; n = 20 615; Supplemental Table 3). A medically defined group was composed of all individuals with diabetes mellitus (DM; n = 2156) in the KPNC diabetes registry.⁸ A typical control (TC) group (n = 20 516) was randomly sampled from individuals with no ADHD and DM diagnoses by the end of 2015. The ADHD and TC groups were sampled at a 5:1 ratio to cases and matched on sex and age group (14–17, 18–21, 22–25).

All diagnoses were identified from the KPNC EMR. Over 1000 ICD-9 codes were grouped into 175 conditions (eg, epilepsy) within 20 categories (eg, neurology). The definition of specific conditions was guided by phenotype-wide association study methodology⁹ and previously validated algorithms (Supplemental Table 3).

Intellectual disability (ID), developmental disability, and genetic disorders were defined by diagnoses documented in the EMR anytime through 2015. Psychiatric and medical conditions were defined by diagnoses documented in the EMR from 2014 to 2015. Overweight and obesity were defined by the average BMI recorded in the 2-year study period (Supplemental Table 3).

The prevalence of each condition was compared between ASD cases and each comparison group by using χ² tests. A separate multivariate logistic regression model was run for each condition to compare odds between cases and controls after controlling for sex, age, race, and total months of KPNC membership from 2013 to 2015. To address possible ascertainment bias, additional models were run also adjusting for the number of outpatient visits. The threshold for statistical significance was set at P < .05. All study procedures were approved by the KPNC Institutional Review Board.

The demographic characteristics of the study population are shown in Table 1. Among the 4123 individuals with ASD, the mean age was 18.4 years (SD 3.2 years), and the male to female ratio was 4:1. Individuals with ASD were more likely than TC and DM groups to be white, non-Hispanic, and more likely than all 3 comparison groups to have Medicaid.

Thirteen percent of transition-aged youth with ASD had a diagnosis of ID (15% mild, 9% moderate, 5% severe, 2% profound, 69% unspecified). The prevalence of ID increased with age (11% in 14–17-year-olds, 12% in 18–21-year-olds, 19% in 22–25-year-olds) and was higher in females (20%) than males (11%). In addition to their ASD diagnosis, 41% had received a developmental disorder (DD) diagnosis at some point in the past (20.7% global delay, 23% language disorder, 11% learning disorder, 4% motor delay). However, during the 2-year study period, only 4.3% had 1 or more of these diagnoses documented in their medical record (Table 2).

Over one-third (34%) of individuals with ASD had a co-occurring psychiatric condition; the most common were ADHD (15%), anxiety (14%), depression (10%), and bipolar disorder (6%) (Table 2). In adjusted analyses, most psychiatric conditions were significantly more common in the ASD group than in each comparison group, with 2 exceptions. Depression was significantly more common in the ASD group versus the TC group but significantly less common in the ASD group versus the ADHD group, and drug abuse and/or dependence was significantly less common in the ASD group than in all 3 comparison groups (Table 2). Less common disorders associated with high morbidity or mortality were also significantly more prevalent in individuals with ASD compared with those in the control group, including schizophrenic disorders (ASD: 1%; versus TC: 0.3%, adjusted odds ratio [aOR] = 3.6, 95% confidence interval (CI): 2.4–5.3; versus ADHD: 0.6%, aOR = 1.9, 95% CI: 1.3–2.6; versus DM: 0.7%, aOR = 2.0, 95% CI 1.1–3.8) and suicide and self-inflicted injury (ASD: 0.7%; versus TC: 0.3%, aOR = 3.6, 95% CI: 2.4–5.3; no significant difference versus ADHD [0.9%] or DM [0.9%]).

The most common medical conditions in transition-aged individuals with ASD included infections (42%), obesity (25%), neurologic conditions (18%), allergy and/or immunology conditions (16%), musculoskeletal conditions (15%), and GI conditions (11%). Other conditions previously reported in children with ASD had a relatively lower prevalence, including sleep disorders (7%), endocrine conditions (3%), and metabolic (3%) disorders (Table 2, Fig 1).

Compared with the TC group, individuals with ASD were at significantly increased risk for most medical conditions (Table 2, Fig 1). The highest risks were seen for epilepsy (aOR = 11.9), genetic disorders (aOR = 7.7), dyslipidemia (aOR = 4.9), constipation (aOR = 3.5), endocrine conditions (aOR = 3.2), and sleep disorders (aOR = 3.1). Antipsychotic medications, a known risk factor for obesity and dyslipidemia,¹⁰ were used by 19% of individuals with ASD. However, among individuals who did not take antipsychotic medications, obesity and dyslipidemia remained significantly more common in the ASD group versus the TC group (obesity: 23% vs 15%, P < .0001; dyslipidemia: 2.5% vs 0.7%, P < .0001). Functional GI disorders, genitourinary and renal conditions, infectious diseases as a group, and hereditary and degenerative disorders of the nervous system occurred at equal frequency in the ASD and TC groups; injury, musculoskeletal conditions, and migraines were significantly less common in the ASD group versus the TC group (Table 2).

Comparisons between the ASD and ADHD groups were largely similar to those between the ASD group and the TC group, with the following notable exceptions: allergy and/or immunology conditions, diabetes, diarrhea, gastroesophageal reflux, hematology and/or oncology, overweight, and ophthalmologic conditions did not differ in frequency between the ASD group and the ADHD group. GI disorders, genitourinary and/or renal conditions, infectious diseases as a group, and pulmonary conditions (asthma) were all significantly less prevalent in the comparisons between the ASD group versus the ADHD group (Table 2).

By contrast, compared with the DM group, individuals with ASD were at a similar or significantly lower risk of most medical conditions, with the exception of hearing impairment, genetic disorders, cerebral palsy, epilepsy, and dyssomnia, all of which were significantly more common in the ASD group compared with the DM group (Table 2). Smoking was significantly less common in individuals with ASD than in each comparison group (Table 2).

Additional adjustment for the number of outpatient visits resulted in slightly attenuated risk estimates, with only a few findings losing statistical significance: allergy and/or immunology (ASD versus DM), GI (ASD versus TD), and asthma (ASD versus TD) (data not shown).

Among individuals with ASD, most conditions increased with age (Fig 2). Infectious diseases and injury decreased with age, and musculoskeletal disorders fluctuated with age. For many conditions, age-related changes in the comparison groups were in the same direction as the ASD group. Most conditions were more common in females than males, with the largest sex differences observed for allergy and/or immunologic conditions, infections, musculoskeletal conditions, neurologic conditions, and psychiatric conditions (Fig 3). For each major comorbidity category, case-control differences were in the same direction for both males and females.

In this large population of insured, transition-aged individuals, psychiatric and medical conditions were more prevalent in individuals with ASD than in typical or ADHD controls, and most conditions increased with age and were more common in females.

Among individuals with ASD, the prevalence of ID (13.0%) was similar to the 16% prevalence reported for another population with a median age of 19 years,¹¹ although lower than previous reports for younger children² and older adults.¹² One reason for the lower prevalence may be that ID is typically identified and treated in school or community settings and thus not routinely noted in medical records. Prevalence of ID did increase with age, and our finding of 19% ID among 22- to 25-year-olds is consistent with our previous finding among all adults ≥18 years old, with a mean age of 29 years,⁴ perhaps because of the need for a medical diagnosis of ID to qualify for social security income.

Among individuals with language or learning disorders diagnosed at some point in the past, <2% had documentation of these disorders in the EMR during the transition years. Although language and learning skills may improve with age and treatment, significant deficits persist for the majority of individuals.^13,14 Thus, it is suggested in our data that language and learning disabilities are underrecognized in transition-aged individuals with ASD. These DDs may impact the ability of patients to understand verbal and written information from medical providers and may contribute to behavioral and emotional difficulties.¹⁰ Recognizing and accommodating the learning and communication needs of individuals with ASD is important for high-quality medical care.¹⁰ 

Approximately one-third of individuals with ASD had a psychiatric condition identified during the study period, a rate lower than the 54% previously reported for the KPNC adult population.⁴ Although the prevalence of most psychiatric conditions increased with age, the rate in 22- to 25-year-olds was still lower than that reported among all adults,⁴ which is consistent with expectations given that the typical age of onset for many conditions is in teen-aged and young adult years.¹ Of concern was the almost fourfold higher rate of suicide and self-inflicted injury among transition-aged individuals with ASD. Increased risks of suicide and self-harm have been reported previously in other populations of teenagers and adults with ASD .^4,11 On a positive note, individuals with ASD had lower drug abuse and smoking rates, consistent with previous reports.^4,15 Because these often begin as social behaviors, the difficulties with social cognition seen in those with ASD may be somewhat protective against these conditions.

Our finding that most medical conditions were more prevalent in transition-aged individuals with ASD than those in TC groups is similar to observations in children and adults.^4,16 Obesity is interesting to consider in relation to other comorbidities observed in the ASD group. Obesity in children with ASD has been linked to gross motor deficits limiting physical activity and with higher rates of poor sleep quality, constipation, selective eating, and other GI disturbances.¹⁷ In addition, obesity has known associations with several chronic medical conditions, including diabetes, hyperlipidemia, hypertension, and cardiovascular disease,¹⁰ all of which were elevated in the ASD group.

Although GI and sleep conditions were more common in transition-aged individuals with ASD than individuals in age-matched TC groups, prevalence was lower than previously reported for children and adults with ASD.^4,18,19 We also observed a lower rate of migraines in the ASD group versus comparison groups, in contrast to a previous report in which authors indicated that children with autism are more likely than those without to have had frequent severe headaches or migraines.¹⁶ Individuals with ASD and GI problems, sleep disorders, or headaches and/or migraines are more likely to have irritability, challenging behavior, social withdrawal, stereotypy, hyperactivity, inattention, and anxiety.^18,–20 These symptoms may lead to psychiatric diagnoses or treatment with psychotropic medications, which in turn may increase the risk of other health problems. Thus, although transition-aged individuals with ASD may experience relative health compared with children and adults, it is possible that certain medical conditions are underrecognized and undertreated in this age group, contributing to additional medical, behavioral, and mental health problems.

There are likely many contributors to the elevated rates of co-occurring conditions seen in individuals with ASD. Self-recognition of differences from others and social isolation may lead to anxiety and depression, and treatments used to address behavioral, psychiatric, and neurologic conditions may lead to or exacerbate other medical conditions, such as obesity, sleep, and GI problems. In addition, symptoms inherent to ASD, including sensory sensitivities and difficulties describing feelings and pain, interacting with others, transitioning, changing routines, and accommodating to unfamiliar people and places may make it more difficult for teenagers and young adults with ASD to access adequate medical care.¹⁰ Finally, clinical providers may be unfamiliar with how to accommodate the unique needs of transition-aged individuals with ASD, contributing to less successful office visits.^10,21 

Our findings must be considered in light of study limitations. To focus on conditions for which transition-aged youth are actively being treated, diagnoses were identified only over a 2-year time period, potentially resulting in underascertainment of chronic conditions that are well controlled and require infrequent visits. However, all included participants were members of KPNC for at least 9 months out of 1 year during the 2-year study period and the previous year, making it highly probable that active medical problems were captured. Although efforts were made to classify ICD-9 codes appropriately, phenotypic categorization may have resulted in occasional misclassification.

Psychiatric and medical comorbidities were common in this large, insured population of transition-aged individuals with ASD. Although more research is needed to identify the intrinsic and extrinsic factors that contribute to this excess burden, there is a pressing need for all clinicians, particularly general pediatricians and adult physicians, to approach ASD as a chronic health condition requiring regular follow-up and routine screening and treatment of medical and psychiatric issues, as is the standard of care for other chronic medical conditions.

ADHD

attention-deficit/hyperactivity disorder

aOR

adjusted odds ratio

ASD

autism spectrum disorder

CI

confidence interval

DD

developmental disorder

DM

diabetes mellitus

EMR

electronic medical record

GI

gastrointestinal

ICD-9

International Classification of Diseases, Ninth Revision

ID

intellectual disability

KPNC

Kaiser Permanente Northern California

TC

typical control