Association of Psychiatric and Neurologic Comorbidity With Mortality Among Persons With Autism Spectrum Disorder in a Danish Population

Importance  Increased mortality has been reported among persons with autism spectrum disorder (ASD), especially among those who also have the comorbid condition of epilepsy or intellectual disability. The effects of psychiatric and neurologic comorbidity on mortality among persons with ASD have not been rigorously examined in large, population-based studies.

Objective  To investigate the mortality patterns among persons with ASD overall and to assess the associations of comorbid mental, behavioral, and neurologic disorders with mortality among persons with ASD.

Design, Setting, and Participants  Longitudinal cohort study of children born in Denmark during the period from 1980 to 2010 who were alive at 1.5 years of age and followed up through 2013. This population-based sample of children (N = 1 912 904) was identified via linkage between the Danish Civil Registration Service and the Danish Medical Birth Register using a unique 10-digit identifier assigned to all live births and new residents in Denmark. Children were followed up for diagnoses of ASD (International Classification of Diseases, Eighth Revision [ICD-8] codes 299.00, 299.01, 299.02, and 299.03 and ICD-10 codes F84.0, F84.1, F84.5, F84.8, and F84.9) and other mental/behavioral disorders (ICD-8 codes 290-315 and ICD-10 codes F00-F99) in the Danish Psychiatric Central Research Register and for diagnoses of neurologic disorders (ICD-8 codes 320-359 and ICD-10 codes G00-G99) in the Danish National Patient Register. Data analysis was performed in December 2014.

Main Outcomes and Measures  Deaths and causes of death among cohort members were identified via the Danish Civil Registration Service and the Danish Cause of Death Register, respectively. Regressions analyses were performed using Cox regression.

Results  Of the 1 912 904 persons included in our study, 20 492 (1.1%) had ASD (15 901 [77.6%] were male). Of the 20 492 persons with ASD, 68 died (0.3%) (57 of 68 [83.8%] had comorbid mental/behavioral or neurologic disorders). The adjusted hazard ratio (aHR) for overall mortality was 2.0 (95% CI, 1.5-2.8) for ASD. The aHRs for ASD-associated mortality among cohort members who did not have neurologic (2.0 [95% CI, 1.4-3.0]) or other mental/behavioral disorders (1.7 [95% CI, 1.0-3.1]) were similar. The co-occurrence of ASD added no additional mortality risk for persons with neurologic (aHR, 0.7 [95% CI, 0.4-1.3]) or mental/behavioral disorders (aHR, 0.8 [95% CI, 0.5-1.2]) compared with persons with these disorders and no ASD.

Conclusions and Relevance  The mortality risk was 2-fold higher through young adulthood for persons with ASD than for persons without ASD, although mortality affected only 0.3% of persons with ASD. The mechanisms underlying ASD-associated mortality may be mediated through or shared with neurologic or mental/behavioral disorders, thereby providing insights into their potential neurobiological links. Health care professionals and family members should recognize the importance of these disorders with regard to the mortality risk for persons with ASD.

Autism spectrum disorder (ASD) is a neurodevelopmental disorder marked by deficits in social communication and interaction and by restricted, repetitive behaviors, interests, or activities.¹ The prevalence of reported ASD has dramatically increased in recent decades and is now in the range of 1% to 2% among children born since the 1990s.^2-4 There is little knowledge of the long-term outcome of persons with ASD. A few studies have reported a 2-fold or more increased mortality in persons with ASD compared with the general population, and mortality tended to be even higher among persons with the comorbid condition of epilepsy or intellectual disability.^5-12 Increased mortality among persons with ASD might be expected owing to the hazards arising from such factors as medications, poor lifestyle, behaviors such as elopement, or comorbidities. Prior studies of the effects of comorbidity on mortality among persons with ASD were limited owing to the use of cause of death as the only source of information on comorbidity, small sample sizes, or there being no population-based morbidity data on persons without ASD for comparison. Elucidating the effects of comorbid conditions on mortality risk for persons with ASD is crucial for informing medical care professionals, for patient and family counseling, and for gaining insights into the underlying pathogenesis that may lead to premature mortality. We followed a nation-wide Danish cohort to examine ASD-associated mortality patterns and estimate the potential effects of mental, behavioral, and neurologic comorbidity on mortality risk for persons with ASD into young adulthood.

Box Section Ref ID

We identified all children born in Denmark during the period from January 1, 1980, to December 31, 2010, who were alive at 1.5 years of age (N = 1 912 904) via linkage between the Danish Civil Registration Service and Medical Birth Register. The Medical Birth Register comprises information on all live births and stillbirths by women permanently residing in Denmark.¹³ Each new resident and live-born child in Denmark is assigned a 10-digit registration number that encodes a unique identifier. The Civil Registration Service also contains information on sex, date of birth, death, immigration status, maternal identity, paternal identity (if known), and sibling identity.¹⁴ The registration number remains unchanged throughout life and is used for accurate linkage of individual data between Danish registers.

Children and adolescents suspected of ASD or other mental or behavioral disorders are referred by general practitioners or school psychologists to a child and adolescent psychiatric department for a multidisciplinary evaluation, and their conditions are diagnosed by a child and adolescent psychiatrist; Danish health care is universal and free of charge. All diagnoses are reported to the Danish Psychiatric Central Research Register once a diagnosis is established and without regard to the need for treatment or educational provisions. The Psychiatric Register holds information on all inpatient admissions to psychiatric hospitals and wards since 1969, and since 1995, it also includes outpatient admissions. The International Classification of Diseases, Eighth Revision (ICD-8) was used as the diagnostic classification system for reporting to the Psychiatric Register from 1969 through 1993. In 1994, the ICD-8 was replaced by the ICD-10. Registry reporting is done only by psychiatrists following mandatory training in the use of the ICD. Although diagnostic misclassification is possible, validation of select diagnoses (eg, schizophrenia, single-episode depression, dementia, and childhood autism) has been performed with good results.¹⁵ The following ASD diagnosis codes were used: ICD-8 codes 299.00, 299.01, 299.02, and 299.03 and ICD-10 codes F84.0, F84.1, F84.5, F84.8, and F84.9. For childhood autism, ICD-8 code 299.00 and ICD-10 code F84.0 were used. Age at ASD diagnosis was defined as the age at first inpatient or outpatient admission leading to an ASD diagnosis.

Information on other mental or behavioral disorder diagnoses was obtained from the Psychiatric Register. Information on neurologic disorder diagnoses was obtained from the Danish National Patient Register, which holds all data on inpatient diagnoses given at discharge from somatic wards in all hospitals since 1977; since 1995, it also holds all data on discharge diagnoses from outpatient and emergency department contacts.¹⁶ The ICD-8 was used for diagnostic reporting to the National Patient Register through 1993; thereafter, the ICD-10 has been used. See eTable 1 in the Supplement for diagnosis codes.

All deaths after 1.5 years of age were identified in the Civil Registration Service. Cause of death information was obtained from the Danish Cause of Death Register, which has computerized records since 1970.¹⁷ Cause of death was reported to the register using the ICD-8 through 1993; thereafter, the ICD-10 has been used.

Data on children’s sex, date of birth, birth weight, and gestational age and on parental ages at birth were obtained from the Medical Birth Register and the Civil Registration Service. Parental history of any psychiatric diagnosis prior to the birth of the child was obtained from the Psychiatric Register (ICD-8 codes 290-315 and ICD-10 codes F00-F99).

All members of the study cohort were followed up from 1.5 years of age until they died (study outcome) or emigrated, or until the end of follow-up on December 31, 2013, whichever occurred first. The primary exposure was a diagnosis of ASD. Therefore, cohort members who received a diagnosis of ASD during follow-up were included thereafter in the ASD group at risk of death from the time of diagnosis (time-dependent exposure), otherwise they remained in the non-ASD group at risk of death. The same approach was used for other morbidity diagnoses received during follow-up. Mortality curves by age for cohort members with or without ASD were estimated based on a flexible, parametric Royston-Parmar model on the hazard scale with 4 degrees of freedom and standard placement of knots.¹⁸ Semiparametric Cox regression was used to estimate the mortality hazard ratio (HR) and 95% CI associated with ASD, with age as the underlying time scale. Hazard ratios were calculated only in exposure groups with 5 or more deaths associated with ASD. Adjusted HRs (aHRs) were calculated using 2 models. In model 1, we used separate baseline ASD diagnostic rates in strata of 3-year birth-year groups to adjust for the increasing prevalence of ASD over time. Model 2 is the same as model 1 but includes sex (except for the sex-specific aHRs), birth weight (≤2500, 2501-3000, 3001-4000, and ≥4001 g), gestational age (<37, 37-40, and ≥41 weeks), and parental ages at birth (<35 and ≥35 years). Model 2 was run both with and without terms for parental psychiatric history prior to birth (yes or no), with the same results. Only the model 2 results without parental psychiatric history included are reported. Because mortality among youth and young adults is rare (<10% incidence), the HR for mortality can be interpreted as a relative risk.¹⁹ The proportional hazards assumption was evaluated for all variables by comparing estimated log − log survivor curves over the different categories of variables investigated.

The overall characteristics of mortality in the ASD and non-ASD groups (without considering comorbidity) included estimation of the median age at death and, for the ASD group only, the median time between diagnosis and death. Hazard ratios were calculated for ASD overall and in strata of sex, birth year (1980-1995 and 1996-2010), age at diagnosis (≤9 years and ≥10 years), and cause of death. Adjusted HRs were calculated for ASD (overall and by sex) and childhood autism separately. Analyses of the risk of death considering the presence or absence of ASD and other mental/behavioral or neurologic morbid conditions were conducted in 4 comparisons based on model 2. Three comparisons were designed to assess the risk of death from other morbid conditions in the absence of ASD, from ASD in the absence of other morbid conditions, and from ASD and other morbid conditions combined; the reference groups for all 3 comparisons consisted of persons with no ASD and no other morbid conditions. The fourth comparison was designed to estimate the risk of mortality associated with the addition of ASD in persons with other morbid conditions; the reference group consisted of persons with any neurologic or other mental/behavioral disorder but no ASD. The mental/behavioral or neurologic morbid conditions were considered as a group and, in addition, one by one.

Our study was approved by the Danish Data Protection Agency, and data were received from the Danish National Serum Institute. Informed consent was not required according to Danish law governing registry-based research studies with no participant contact.

By the end of follow-up, 20 492 members (1.1%) of the study cohort received a diagnosis of ASD, including 5902 (0.3%) with childhood autism. Compared with persons without ASD, persons with ASD were more likely to be male, preterm, have low or high birth weight, and have older parents or parents with a history of psychiatric diagnoses; they were also more likely to have a history of other mental/behavioral or neurologic disorders (Table 1).

As shown in Table 2, 68 persons with ASD died at a median age of 19.0 years with a median time from diagnosis to death of 5.5 years, and 16 persons with childhood autism died at a median age of 18.6 years with a median time from diagnosis to death of 5.1 years. Most of the persons with ASD who died were male, born before the mid-1990s, or received a diagnosis of ASD after 9 years of age. In unadjusted analyses, there was about a 2-fold significant increased risk of death among persons with ASD overall (or with childhood autism) compared with persons without ASD (Table 2; Figure), as well as persons who died of natural or external causes (data on cause of death were available for 57 of the 68 persons with ASD who died). Only 3 specific categories of cause of death included at least 5 persons with ASD who died: there was a 4.1-fold risk of death (95% CI, 2.0-8.8) from neurologic causes among persons with ASD, a 4.6-fold risk (95% CI, 2.7-8.0) from intentional self-harm among persons with ASD, and a 40% increased risk from accidents among persons with ASD (HR, 1.4 [95% CI, 0.8-2.3]) that was not statistically significant. In adjusted analyses, there remained a significant 2-fold increased risk for death among persons with ASD overall; the relative risk for males (aHR, 1.7 [95% CI, 1.2-2.6]) was half that for females (aHR, 3.5 [95% CI, 1.7-7.0]).

Comorbid mental/behavioral and neurologic disorders were common among persons with ASD who died; 57 of 68 of persons with ASD who died (83.8%) had at least one of these other disorders, and 22 of 68 (32.4%) had both mental/behavioral and neurologic comorbid conditions. Two-thirds of persons with ASD who died of external causes had mental/behavioral comorbid conditions only, whereas half of the persons with ASD who died of natural causes had both mental/behavioral and neurologic comorbid conditions.

The results of the 4 comparisons of mortality risk considering combinations of ASD and neurologic or mental/behavioral morbid conditions are shown in Table 3 (see eTable 2 in the Supplement for results when considering each neurologic or mental/behavioral disorder one by one and eTables 3 and 4 in the Supplement for covariate results). Compared with persons with no ASD and no other morbid conditions, the risk of death for persons with neurologic or mental/behavioral disorders but no ASD was significantly increased (with aHRs of 10.4 and 3.3, respectively); the risk of death for persons with ASD in combination with neurologic or mental/behavioral disorders was also significantly increased (with aHRs of 7.6 and 2.6, respectively); the risk of death for persons with ASD but no neurologic or other mental/behavioral disorders was lower (with aHRs of 2.0 and 1.7, respectively); and only the mortality risk for persons with ASD in the absence of neurologic disorders was significantly increased. Finally, compared with persons with any neurologic or other mental/behavioral disorder but no ASD, the risk of death for persons with ASD, in addition to any neurologic or other mental/behavioral disorder, was low (with aHRs of 0.7 and 0.8, respectively) but not statistically significantly lower.

Although the risk of premature death among persons with ASD is 2-fold higher than among persons without ASD, mortality is fortunately quite rare in youth and young adults, affecting only 0.3% of persons with diagnosed ASD by the end of the follow-up in our study. The overall increased risk for mortality that we observed is consistent with a previous systematic review of ASD-associated mortality studies.¹¹

We report a significant 4.6-fold risk of death among persons with ASD due to intentional self-harm. Fully 22.8% (13 of 57) of persons with ASD died of self-inflicted injuries (ie, intentional self-harm was listed as the cause of death); 26.3% (15 of 57) of persons with ASD died in accidents, but the risk of death due to accidents among persons with ASD was not significantly increased. The majority of individuals with ASD whose death was due to either intentional self-harm or an accident had mental/behavioral comorbid conditions.

The fact that most of the persons with ASD who died in our study were males but that females with ASD had a higher relative risk of death is consistent with prior reports.^5-8,10-12 Notably, there was a small but not significantly increased risk of death due to ASD among children younger than 18 years of age in our cohort, whereas Pickett et al⁷ reported increased standard mortality ratios in this age group among persons with ASD in the California Developmental Disability System, 1998-2002.

We tested the potential modifying effects of comorbidity on mortality risk for persons with ASD into young adulthood, with a focus on comorbid mental, behavioral, and neurologic disorders. The majority of deaths among persons with ASD were among persons with ASD and comorbid mental, behavioral, or neurologic disorders, and the mortality risk was significantly increased for persons with ASD and these other morbid conditions compared with persons without any of these disorders. We demonstrated for the first time, however, a marked difference in the risk of death from ASD depending on the presence or absence of other mental/behavioral and neurologic morbid conditions. That is, for the mortality risk specific to ASD, we observed a 1.7- to 2.0-fold increased mortality risk for persons with ASD but without any other mental/behavioral or neurologic disorders, whereas for persons with 1 or more of these conditions, the co-occurrence of ASD added no additional mortality risk. However, the number of persons with ASD but no other mental/behavioral or neurologic disorders who died is small, and they may have other conditions contributing to mortality risk. Further work is needed to clarify the range of specific comorbidity effects in ASD-associated mortality.

Differences in mortality risk specific to ASD, depending on the presence or absence of neurologic or mental/behavioral comorbid conditions, suggest that the mechanisms underlying mortality risk associated with ASD, in part, may be shared with these other disorders. What might be the specific shared mechanisms underlying mortality cannot be determined with these data. As shown in our data and in previous reports, generally there is increased mortality among persons with other psychiatric illnesses^20,21 and neurologic disorders, including epilepsy²² and cerebral palsy.²³ The proposed reasons for the increased mortality across these other conditions are varied, and some may potentially be modifiable, including poor lifestyle behaviors and physical health (eg, increased smoking, an unbalanced diet, or a sedentary lifestyle), medication effects (eg, leading to weight gain), negative social conditions (eg, isolation or unemployment), enhanced suicidal tendencies, and underlying neurologic problems (eg, seizures or poor neuromotor function). Poor medical monitoring of young adults, which could lead to undetected health problems, could be a contributing factor, possibly exacerbated in ASD by poor social communication skills, resistance to new experiences, and altered perception rendering them less likely to seek health care. Because many of these mortality risk factors are shared across the disorders, these common factors could underlie the apparent shared risk for mortality between other psychiatric and neurologic disorders and ASD.

An investigation with larger data sets is needed to determine whether the apparent shared mortality risk is limited to a subset of conditions in the total pool of mental/behavioral and neurologic disorders (eg, schizophrenia, affective disorders, epilepsy, and cerebral palsy). There is much debate as to whether the major psychiatric conditions are truly distinct disorders or have shared etiologies and underlying neurobiological mechanisms. Genetic studies have revealed the overlap of copy number variations in ASD with epilepsy, intellectual disability, schizophrenia, and bipolar disorder.^24,25 A recent multidisorder genome-wide association study and cross-disorder polygenic score analysis identified specific single-nucleotide polymorphisms associated with schizophrenia, bipolar disorder, major depression, ASD, and attention-deficit/hyperactivity disorder and a cross-disorder genetic association between ASD and schizophrenia.²⁶ It has also been hypothesized that, based on the array of specific genetic syndromes, genes, and copy number variations associated with both ASD and epilepsy, neurodevelopmental processes leading to alterations in synaptic plasticity may be an underlying mechanism linking the 2 disorders, as well as linking the 2 disorders with intellectual disability.^27,28 Motor impairment is common in ASD and intellectual disability, as well as in genetic conditions linked to ASD,^29,30 and increased frequencies of ASD have been observed in hypotonic cerebral palsy.³¹

In summary, the mortality patterns observed in our study may support the hypotheses that ASD and the foregoing conditions are linked by common underlying neurobiological features despite disparate clinical diagnostic profiles. Although the origins of these common features may rest in early neurodevelopment, as seen in our data, the effects of the common features may be lifelong and may even contribute to common phenotypes that shape mortality risk. To test this hypothesis, additional data on the underlying causes and associated neuropathology of ASD and these disorders and their associations with mortality are needed.

The strengths of our study include the large population-based cohort design and unbiased, prospective medical follow-up in Denmark, where the provision of health care is universal and free of charge. The validity of a registry-reported diagnosis of childhood autism is high.³² The validity of an ASD diagnosis is unknown, but ascertainment appears to be fairly complete because the prevalence of ASD in Denmark (1.1% in these data) is consistent with US surveillance estimates.⁴ The adoption of the ICD-10 from 1994 and the inclusion of outpatient discharge diagnoses from 1995 contributed to substantial increases in registry-reported ASD diagnoses,³³ although it is unknown how these changes affected reporting of other morbid conditions. Our analyses controlled for changes in ASD prevalence over time by incorporating separate baseline diagnostic rates in strata of 3-year birth-year groups in the Cox regression models. Errors or imprecision in cause of death data may arise owing to medical and reporting practices, but there is no way to determine whether such misclassification is differential with respect to an ASD diagnosis. Despite the large sample size and the longitudinal follow-up, the study cohort is fairly young. A longer follow-up into later adulthood and also the use of larger data sets to confirm the mortality patterns are needed. The relatively youthful study cohort members may also underlie the relatively high absolute frequency of both accidents and intentional self-harm as causes of death; as the cohort ages, the contribution to mortality from chronic, adult-onset medical conditions may increase.

The recognition that the sources of risk for premature death among persons with ASD may be mediated through or shared, in substantial part, with other mental/behavioral and neurologic morbid conditions provides a basis for future research into the mechanisms underlying these premature deaths among persons with ASD. Shared mortality risk among these serious disorders also lends a novel perspective from which to examine their potential neurobiological links. Meanwhile, it is important for persons with ASD, their families, and health care professionals to recognize the important role these comorbid conditions play in ASD-associated deaths and to consider preventive measures against associated mortality risk, including risk for intentional self-harm.

Accepted for Publication: October 8, 2015.

Corresponding Author: Diana E. Schendel, PhD, Department of Public Health, Section for Epidemiology, Aarhus University, Bartholins Allé 2, Bldg 1260, 8000 Aarhus C, Denmark (diana.schendel@ph.au.dk).

Published Online: January 11, 2016. doi:10.1001/jamapediatrics.2015.3935.

Author Contributions: Dr Schendel had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Schendel, Christensen, Hjort, Vestergaard, Parner.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Schendel, Parner.

Critical revision of the manuscript for important intellectual content: Schendel, Overgaard, Christensen, Hjort, Jørgensen, Vestergaard.

Statistical analysis: Overgaard, Christensen, Parner.

Obtained funding: Schendel.

Administrative, technical, or material support: Schendel.

Study supervision: Schendel, Vestergaard, Parner.

Conflict of Interest Disclosures: Dr Christensen reported receiving honoraria for serving on the scientific advisory boards of UCB Nordic and Eisai AB; receiving lecture honoraria from UCB Nordic and Eisai AB; being involved in clinical trials initiated by UCB Nordic, Eisai, Pfizer, and Novartis; and receiving travel funding from UCB Nordic. No other disclosures are reported.

Funding/Support: The study was supported in part by an unrestricted grant from the Lundbeck Foundation (Mental Health in Primary Care and Lundbeck Foundation Initiative for Integrative Psychiatric Research) and by the Danish Epilepsy Association.

Role of the Funder/Sponsor: The funders of the study had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank Christine Warburg, MSc, Department of Public Health, Aarhus University, for performing the literature search while employed as a research assistant to Dr Schendel. Support for Ms Warburg was not provided through the funding sources.