OBJECTIVE

Characterize the prevalence of chronic physical illness types and mental illness and their comorbidity among adolescents and young adults (AYA) and assess the association of comorbidity on hospital utilization.

METHODS

This study features a population-level sample of 61 339 insurance-eligible AYA with an analytic sample of 49 089 AYA (aged 12–21) in Vermont’s 2018 all-payer database. We used multiple logistic regressions to examine the associations between physical illness types and comorbid mental illness and emergency department (ED) use and inpatient hospitalization.

RESULTS

The analytic sample was 50% female, 63% Medicaid, and 43% had ≥1 chronic illness. Mental illness was common (31%) and highly comorbid with multiple physical illnesses. Among AYA with pulmonary illness, those with comorbid mental illness had 1.74-times greater odds (95% confidence interval [CI]: 1.49–2.05, P ≤.0005) of ED use and 2.9-times greater odds (95% CI: 2.05–4.00, P ≤.0005) of hospitalization than those without mental illness. Similarly, comorbid endocrine and mental illness had 1.84-times greater odds of ED use (95% CI: 1.39–2.44, P ≤.0005) and 2.1-times greater odds of hospitalization (95% CI: 1.28–3.46, P = .003), comorbid neurologic and mental illness had 1.36-times greater odds of ED use (95% CI: 1.18–1.56, P ≤.0005) and 2.4-times greater odds of hospitalization (95% CI: 1.73–3.29, P ≤.0005), and comorbid musculoskeletal and mental illness had 1.38-times greater odds of ED use (95% CI: 1.02–1.86, P = .04) and 2.1-times greater odds of hospitalization (95% CI: 1.20–3.52, P = .01).

CONCLUSIONS

Comorbid physical and mental illness was common. Having a comorbid mental illness was associated with greater ED and inpatient hospital utilization across multiple physical illness types.

Adolescents and young adults (AYA) are known to have high hospital utilization, both with regard to the emergency department (ED)1,2  and inpatient units.3  For those suffering from mental illness or complex chronic conditions, utilization is particularly high.47 

Although AYA hospital use patterns have been studied for decades,1,4  few have taken a population-level approach to understanding the combined impact of physical and mental illnesses on hospital utilization. A number of common physical and mental illness comorbid combinations have been documented in the literature, including asthma and depression5,6  and anxiety,6  type 1 diabetes and anxiety7  and depressive disorders7,8  and suicidality,9  congenital heart disease and developmental and anxiety disorders,10,11  and epilepsy and attention deficit disorders12  and anxiety and depression.13  Studies have also revealed the disproportionate impact of comorbid mental illness on health care dollars spent during pediatric nonpsychiatric hospitalizations.14,15  Similar studies on ED utilization, however, remain sparse and are limited to those with Medicaid insurance16  or specific medical diagnoses (eg, asthma6  and diabetes8 ). Although disease- and setting-specific data are important, a population-based approach to understanding hospital utilization among AYA with multiple types of physical illness and comorbid mental illness can help identify systemic vulnerabilities across multiple pediatric subspecialty populations.

To attempt to address these limitations in the literature, we have 2 primary objectives in this study: (1) characterize the prevalence of chronic physical illness types and mental illness and the frequency of comorbidity among a population-level sample of AYA, and (2) among AYA with different types of chronic physical illness, assess the association of comorbid mental illness on hospital utilization. We hypothesize that AYA with comorbid physical and mental illness will have higher ED use and inpatient hospitalization than AYA with only physical illness.

The initial population-level cross-sectional sample consisted of 61 339 insurance-eligible AYA (ages 12–21) in Vermont’s 2018 all-payer database. We used 2018 health care claims data from the Vermont Health Care Uniform Reporting & Evaluation System for individuals with any health insurance plan (eg, commercial, Medicaid, Medicare). The database accounts for ∼75% of the Vermont population and excludes uninsured individuals (∼4%)17  and payers who did not opt in to sharing claims.18  For analyses, we excluded AYA without data indicating sex (n = 104), those without at least 1 International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) diagnosis code on a claim between 2016 and 2018 (n = 7340), and those without any valid (ie, paid as primary and not denied) medical claim in 2018 (n = 4910). The analytic sample thus consisted of 49 089 individuals. Using the pediatric medical complexity algorithm (PMCA), a subsample of AYA with chronic illness was also determined (n = 21 290).

We used the PMCA version 3.1 to categorize AYA into 3 medical complexity groups: nonchronic (including healthy), noncomplex chronic (illness affecting 1 body system that was neither progressive nor malignant), and complex chronic (illness involving ≥2 body systems or progressive and/or malignant illness).19  The subsample of interest (“chronic illness”) comprised the noncomplex chronic and complex chronic medical complexity groups. In consultation with a child and adolescent psychiatrist (DCR), we modified the PMCA 3.1 open-source code to include additional ICD-10-CM mental health and substance use diagnoses (eg, anxiety disorders and cannabis use disorder) that were not included in the PMCA (Supplemental Table 5). These diagnoses are known mental health conditions described in the Diagnostic and Statistical Manual of Mental Disorders20  and have been found to frequently lead to emergency and inpatient care. This addition was done with support from the author of the PMCA (personal communication).

In addition to categorizing by medical complexity, the PMCA also labels an individual’s affected body system(s). The PMCA includes 19 body systems; however, our sample had no patients in 2 body systems (Malignant, Otolaryngologic). The PMCA mental illness body system was indicated by having ≥1 ICD-10-CM chronic mental illness diagnosed at ≥2 health care claims within any 1 year from 2016 to 2018. We examined the other 16 body systems to explore associations with mental illness. Of note, body systems do not have mutually exclusive PMCA categorizations; an individual with both mental and physical illness is considered complex chronic (2 affected body systems), whereas an individual with either illness alone is considered noncomplex chronic (1 affected body system). Henceforth, we refer to the mental illness body system as “mental illness” and all other body systems as a type of “physical illness.”

The outcomes for 2018 hospital utilization were binary for all-cause ED visits (≥1 ED visit vs no ED visit) and inpatient hospitalization (≥1 hospitalization vs no hospitalization). ED visits were identified by using the “Yale definition.”21  Similar to the Centers for Medicare and Medicaid Research Data Assistance Center definition of an ED visit, the Yale definition uses revenue codes, health care common procedure coding system codes, and bill type. It also includes place of service (including critical care services rendered in the ED), and observational stays. Inpatient hospitalization was determined by having a hospital discharge date and length of stay ≥1 day. Demographic covariates included age (12–21 years), sex (female or male), and having any Medicaid insurance during the year as binary (Medicaid or non-Medicaid). The non-Medicaid group was largely commercially insured individuals; only 0.2% (n = 97) of AYA had Medicare.

To investigate our first objective, we conducted descriptive statistics to obtain frequencies of physical illness types and mental illness within the noncomplex chronic and complex chronic PMCA groups. To explore our second objective, we obtained frequencies for hospital utilization outcomes by type of physical illness both with and without comorbid mental illness. To maintain anonymity, we only analyzed hospital utilization outcomes for physical illness type subsamples with ≥500 AYA because of small inpatient hospitalization counts. For each of the different physical illness types, we ran separate multiple logistic regressions for both hospital utilization outcomes on comorbid mental illness, controlling for age, sex, and insurance type. Covariates were chosen a priori on the basis of previous study by our research team.22 

To examine whether associations were unique to physical illness types and comorbid mental illness or simply attributed to overall greater disease burden (eg, comorbid physical illnesses), we conducted sensitivity analyses looking at associations between each pairing of comorbid physical illness types. For each pairing of comorbid physical illness types, we tested the associations between the physical illness pairing and hospital utilization outcomes, controlling for age, sex, and insurance type. Stata SE version 15 (StataCorp LLC)23  statistical software was used. We considered associations statistically significant if P < .05. The university’s institutional review board approved this study.

In Table 1, we show the demographic characteristics of our 3 samples: (1) insurance-eligible AYA, (2) the analytic sample of those with health care claims in 2018, and (3) a subset of the analytic sample with ≥1 chronic illness. Among insurance-eligible AYA, we estimated that 35% (21 290/61 443) had a chronic illness (mental and/or physical). Within our analytic sample of those with health care claims in 2018, one-half were female and approximately two-thirds had Medicaid (Table 1). Forty-three percent of our analytic sample made up the chronic illness subsample.

TABLE 1

Demographic Characteristics of 3 Samples: Insurance-Eligible, Had Health Care Claims in 2018 (Analytic Sample), and Analytic Sample Subset With Chronic Illness

Insurance-Eligible (n = 61 443)Had Claims* in 2018 (Analytic Sample; n = 49 089)Had Chronic Illness (n = 21 290)
n%n%n%
Age group, y       
 12–14 18 161 29.6 15 584 31.8 6345 29.8 
 15–17 18 634 30.3 15 399 31.4 6981 32.8 
 18–21 24 648 40.1 18 106 35.9 7964 37.4 
Sex       
 Male 31 161 50.7 24 218 49.3 10 591 49.8 
 Female 30 178 49.1 24 871 50.7 10 699 50.3 
 Missing 104 0.2 — — — — 
Insurance       
 Medicaid 34 449 56.1 31 058 63.3 15 736 73.9 
 Non-Medicaid 26 994 43.9 18 031 36.7 5554 26.1 
Insurance-Eligible (n = 61 443)Had Claims* in 2018 (Analytic Sample; n = 49 089)Had Chronic Illness (n = 21 290)
n%n%n%
Age group, y       
 12–14 18 161 29.6 15 584 31.8 6345 29.8 
 15–17 18 634 30.3 15 399 31.4 6981 32.8 
 18–21 24 648 40.1 18 106 35.9 7964 37.4 
Sex       
 Male 31 161 50.7 24 218 49.3 10 591 49.8 
 Female 30 178 49.1 24 871 50.7 10 699 50.3 
 Missing 104 0.2 — — — — 
Insurance       
 Medicaid 34 449 56.1 31 058 63.3 15 736 73.9 
 Non-Medicaid 26 994 43.9 18 031 36.7 5554 26.1 
*

The analytic sample also excluded those with missing sex data. —, Individuals with missing data for sex were not included in analyses for the Analytic Sample. The Chronic Illness sample did not have any individuals with missing data for sex.

In Table 2 we show medical complexity categorization and the percentage of patients with physical or mental illness in our 3 samples. Notably, more AYA were diagnosed with mental illness compared with all types of physical illness across our 3 samples; 26% had a mental illness in the insurance-eligible sample, 31% in the analytic sample, and 72% in the chronic illness subset of the analytic sample (Table 2). After mental illness, the 5 most physical illness types were neurologic, pulmonary, musculoskeletal, endocrine, and cardiac (Table 2). In Table 3 we show the 5 most common types of physical illness stratified by comorbid mental illness. Neurologic illness had the highest comorbidity with mental illness (45%), followed closely by endocrine (41%); all 5 physical illness types had comorbid mental illness >28% (Table 3).

TABLE 2

Medical Complexity and Affected Chronic Illness Body Systems of 3 Samples: Insurance-Eligible, Had Health Care Claims in 2018 (Analytic Sample), and Analytic Sample Subset With Chronic Illness

Insurance-Eligible (n = 61 443)Had Claims* in 2018 (Analytic Sample; n = 49 089)Had Chronic Illness (n = 21 290)
n%n%n%
Medical complexity       
 Nonchronic/healthy 39 346 64.0 27 799 56.6 — — 
 Noncomplex chronic 16 765 27.3 16 048 32.7 16 048 75.4 
 Complex chronic 5332 8.7 5242 10.7 5242 24.6 
Affected chronic illness body system       
 Mental health 15 875 25.8 15 307 31.2 15 307 71.9 
 Neurologic 3912 6.4 3807 7.8 3807 17.9 
 Pulmonary 3053 5.0 2975 6.1 2975 14.0 
 Musculoskeletal 1080 1.8 1056 2.2 1056 5.0 
 Endocrine 962 1.6 940 1.9 940 4.4 
 Cardiac 531 0.9 517 1.1 517 2.4 
 Ophthalmologic 320 0.5 308 0.6 308 1.5 
 Gastrointestinal 291 0.5 286 0.6 286 1.3 
 Otologic 242 0.4 238 0.5 238 1.1 
 Genetic 240 0.4 240 0.5 68 0.3 
 Metabolic 235 0.4 227 0.5 227 1.1 
 Renal 160 0.3 159 0.3 159 0.8 
 Immunologic 138 0.2 137 0.3 137 0.6 
 Genitourinary 68 0.1 68 0.1 240 1.1 
 Hematologic 55 0.1 54 0.1 54 0.3 
 Craniofacial 27 <0.1 27 0.1 27 0.1 
 Dermatologic 21 <0.1 21 <0.1 21 0.1 
Insurance-Eligible (n = 61 443)Had Claims* in 2018 (Analytic Sample; n = 49 089)Had Chronic Illness (n = 21 290)
n%n%n%
Medical complexity       
 Nonchronic/healthy 39 346 64.0 27 799 56.6 — — 
 Noncomplex chronic 16 765 27.3 16 048 32.7 16 048 75.4 
 Complex chronic 5332 8.7 5242 10.7 5242 24.6 
Affected chronic illness body system       
 Mental health 15 875 25.8 15 307 31.2 15 307 71.9 
 Neurologic 3912 6.4 3807 7.8 3807 17.9 
 Pulmonary 3053 5.0 2975 6.1 2975 14.0 
 Musculoskeletal 1080 1.8 1056 2.2 1056 5.0 
 Endocrine 962 1.6 940 1.9 940 4.4 
 Cardiac 531 0.9 517 1.1 517 2.4 
 Ophthalmologic 320 0.5 308 0.6 308 1.5 
 Gastrointestinal 291 0.5 286 0.6 286 1.3 
 Otologic 242 0.4 238 0.5 238 1.1 
 Genetic 240 0.4 240 0.5 68 0.3 
 Metabolic 235 0.4 227 0.5 227 1.1 
 Renal 160 0.3 159 0.3 159 0.8 
 Immunologic 138 0.2 137 0.3 137 0.6 
 Genitourinary 68 0.1 68 0.1 240 1.1 
 Hematologic 55 0.1 54 0.1 54 0.3 
 Craniofacial 27 <0.1 27 0.1 27 0.1 
 Dermatologic 21 <0.1 21 <0.1 21 0.1 
*

The analytic sample also excluded those with missing sex data. —, The Chronic Illness sample did not include individuals that were in the Nonchronic/healthy category.

TABLE 3

Frequency of ED and Inpatient Hospitalizations Patients in Our Analytic Sample With 5 Physical Illness Types With and Without Comorbid Mental Illness

ComorbidityOutcome 1: EDOutcome 2: Inpatient Hospitalization
Physical illness typen%Yes ED% Yes EDYes Inpatient% Yes Inpatient
Pulmonary 2975 100.0 1141 38.4 177 5.9 
Comorbid mental illness 1135 38.2 568 50.0 120 10.6 
Without mental illness 1840 61.9 573 31.1 57 3.1 
Endocrine 940 100.0 337 35.9 74 7.9 
Comorbid mental illness 381 40.5 175 45.9 45 11.8 
Without mental illness 559 59.5 162 29.0 29 5.2 
Neurologic 3807 100.0 1177 30.9 177 4.6 
Comorbid mental illness 1700 44.7 590 34.7 116 6.8 
Without mental illness 2107 55.4 587 27.9 61 2.9 
Musculoskeletal 1056 100.0 282 26.7 59 5.6 
Comorbid mental illness 301 28.5 101 33.6 28 9.3 
Without mental illness 755 71.5 181 24.0 31 4.1 
Cardiac 517 100.0 189 36.6 34 6.6 
Comorbid mental illness 194 37.5 82 42.3 11 5.7 
Without mental illness 323 62.5 107 33.1 23 7.1 
ComorbidityOutcome 1: EDOutcome 2: Inpatient Hospitalization
Physical illness typen%Yes ED% Yes EDYes Inpatient% Yes Inpatient
Pulmonary 2975 100.0 1141 38.4 177 5.9 
Comorbid mental illness 1135 38.2 568 50.0 120 10.6 
Without mental illness 1840 61.9 573 31.1 57 3.1 
Endocrine 940 100.0 337 35.9 74 7.9 
Comorbid mental illness 381 40.5 175 45.9 45 11.8 
Without mental illness 559 59.5 162 29.0 29 5.2 
Neurologic 3807 100.0 1177 30.9 177 4.6 
Comorbid mental illness 1700 44.7 590 34.7 116 6.8 
Without mental illness 2107 55.4 587 27.9 61 2.9 
Musculoskeletal 1056 100.0 282 26.7 59 5.6 
Comorbid mental illness 301 28.5 101 33.6 28 9.3 
Without mental illness 755 71.5 181 24.0 31 4.1 
Cardiac 517 100.0 189 36.6 34 6.6 
Comorbid mental illness 194 37.5 82 42.3 11 5.7 
Without mental illness 323 62.5 107 33.1 23 7.1 

Only these 5 physical illness types are shown because they have at least 500 AYA overall.

Among AYA with a mental illness in our analytic sample (n = 15 307), 34% (n = 5241) went to the ED at least once in 2018 for any reason and 5.4% (n = 831) had at least 1 inpatient hospitalization in 2018. The unadjusted percentages of AYA hospital utilization for the 5 most physical illness types are shown in Table 3. The highest frequency of ED use (50%) was noted among AYA with comorbid pulmonary and mental illness. AYA with comorbid endocrine and mental illness had the highest frequency of inpatient hospitalizations (12%).

In Table 4 we show 5 physical illness type subsamples with the adjusted association between having a comorbid mental illness and hospital utilization, controlling for age, sex, and insurance. For 4 of these subsamples (pulmonary, endocrine, neurologic, and musculoskeletal), comorbid mental illness was associated with significantly greater odds of an ED visit or hospitalization with adjusted odds ratios between 1.38 and 2.87. Mental illness comorbidity was not associated with hospitalization for AYA with cardiac conditions (Table 4).

TABLE 4

The Association Between Having a Comorbid Mental Illness and Hospital Utilization (Emergency Department and Inpatient Hospitalization) Within 5 Different Physical Illness Types

Outcome 1: Emergency DepartmentOutcome 2: Inpatient Hospitalization
Physical illness typeaORPLower 95% CIUpper 95% CIaORPLower 95% CIUpper 95% CI
Pulmonary 1.74 <.001 1.49 2.05 2.87 <.001 2.05 4.00 
Endocrine 1.84 <.001 1.39 2.44 2.11 .003 1.28 3.46 
Neurologic 1.36 <.001 1.18 1.56 2.39 <.001 1.73 3.29 
Musculoskeletal 1.38 .04 1.02 1.86 2.05 .01 1.20 3.52 
Cardiac 1.38 .10 0.94 2.02 0.73 .41 0.34 1.55 
Outcome 1: Emergency DepartmentOutcome 2: Inpatient Hospitalization
Physical illness typeaORPLower 95% CIUpper 95% CIaORPLower 95% CIUpper 95% CI
Pulmonary 1.74 <.001 1.49 2.05 2.87 <.001 2.05 4.00 
Endocrine 1.84 <.001 1.39 2.44 2.11 .003 1.28 3.46 
Neurologic 1.36 <.001 1.18 1.56 2.39 <.001 1.73 3.29 
Musculoskeletal 1.38 .04 1.02 1.86 2.05 .01 1.20 3.52 
Cardiac 1.38 .10 0.94 2.02 0.73 .41 0.34 1.55 

aOR, adjusted odds ratio; CI, confidence interval

Only these 5 physical illness types are shown because they have at least 500 AYA overall. Each physical illness type included only those with that physical illness in a multiple logistic regression model of comorbid mental health (yes vs no) predicting hospital utilization outcomes controlling for the potential confounding effects of age, sex, and insurance.

Our sensitivity analyses examining pairings of comorbid physical illness types and associations with hospital utilization (Supplemental Table 6) were mixed. Of the 20 comorbid physical illness type pairings, only 5 pairings revealed that having a second type of physical illness was associated with greater ED use (Supplemental Table 6). For inpatient hospitalization, 15 of the 20 comorbid physical illness pairings revealed that having a second type of physical illness was associated with greater hospitalization (Supplemental Table 6).

We investigated the prevalence of multiple types of physical illness and mental illness and their comorbidity among AYA and found that mental illness was the most common chronic condition, surpassing all types of physical illness combined. Among AYA with neurologic, pulmonary, endocrine, musculoskeletal, and cardiac illnesses, mental illness was the most common comorbid illness. In line with our hypothesis, we found that AYA with neurologic, pulmonary, endocrine, and musculoskeletal physical illnesses who also had a comorbid mental illness were associated with greater odds of ED use and inpatient hospitalization than AYA with only physical illness. The association between physical illness and comorbid mental illness and ED use was largest among those with pulmonary and endocrine conditions, whereas the association between comorbid mental illness and inpatient hospitalization was largest among AYA with pulmonary and neurologic conditions. For those with pulmonary conditions, a cooccurring mental illness was associated with nearly 3-fold greater odds of inpatient hospitalization.

Forty years ago, the idea of the “hidden morbidity” of mental illness was introduced.24,25  Recent efforts to illuminate this morbidity revealed 40% of adolescents (13–17 years) had a diagnosable mental illness in a 12-month period.26  Our population-level estimates for mental illness prevalence among AYA in a slightly broader age range (12–21 years) support this finding with 26% prevalence among our insurance-eligible sample and 31% among AYA accessing health care in 2018. Our findings are consistent with other estimates of the baseline prevalence of youth mental illness in the population.27  In addition, 35% of AYA aged 12 to 21 years who were eligible for insurance had a chronic illness (mental and/or physical), which aligns with a report from the Data Resource Center for Child & Adolescent Health revealing that, among a narrower group of 12- to 17-year-olds, nearly 25% of adolescents had ≥1 chronic illness.

Among the 5 most common types of physical illness, we found that >28% of AYA also had a comorbid mental illness. AYA comorbidity literature at the population level is limited; previous work focuses on specific types of physical and mental conditions. Looking only at mood and anxiety disorders,28  the authors of one study reported nearly 20% lifetime comorbidity prevalence for females and 7% to 34% for males. Our study supports the mental illness comorbidity prevalence reported in smaller studies of youth with asthma (25%),6  epilepsy (25%),13  type 1 diabetes (33% to 42%),7  and congenital heart disease (18%).10  Moreover, we are the first group of researchers to our knowledge to report on a study including both ED and inpatient hospitalization rates for AYA with comorbid mental illness across a variety of physical illness types.

Our results not only support previous findings within specific physical illness types but go 1 step further to expand the literature by including both ED use and inpatient hospitalization in the same study. Although our results support evidence for increased ED visits among AYA with comorbid pulmonary conditions, like asthma and mental illness,6  it also revealed greater inpatient hospitalization among this population. For AYA with an endocrine illness, like type 1 diabetes, we confirmed previous findings that comorbid mental illness is associated with greater ED use8  and found that comorbid mental illness was associated with greater inpatient hospitalization for all AYA, not just females.8  For neurologic conditions, one adult study revealed greater ED and inpatient hospitalization with comorbid epilepsy and mental illness;29  our study suggests these findings may also be replicable among AYA. Taken altogether, our findings highlight that when considering possible predictors of hospital utilization (eg, sex, insurance payor, etc.), having a comorbid mental illness may also play an important role in understanding hospital use in both the ED and inpatient setting.

Interestingly, for cardiac conditions, the association of mental illness comorbidity with ED use was the same magnitude as the endocrine and musculoskeletal subsamples but was statistically nonsignificant, likely because of the smaller sample size. Although studies of adolescents and adults with congenital heart disease confirm our finding of high rates of comorbid mental illness,10,11  there is a paucity of research on ED use and mental illness comorbidity, specifically among AYA cardiac patients. Regarding hospitalization, we found no association between comorbidity and inpatient hospitalization in our cardiac sample. This finding diverges from other researchers who reported that adults with congenital heart disease and comorbid mental illness had more frequent cardiac and extracardiac comorbidities, as well as greater all-cause inpatient hospitalization.30  Of note, our cardiac subsample, in addition to being small, included other cardiac pathologies (eg, arrhythmias, infections, hypertension) in addition to congenital heart disease.

It is important to briefly consider the possible mechanisms underlying the associations between comorbid physical and mental illness and greater hospital utilization. Pain and fear associated with a physical illness could affect one’s mood and level of anxiety, potentially unmasking a diagnosable psychiatric disorder, whereas various mental health disorders may be associated with the occurrence or exacerbation of physical illness. This bidirectionality underlies the complex interplay of physical processes and mental illness, the focus of emerging and accumulating evidence regarding the “gut brain axis.”31  Also, increased ED visits among AYA, regardless of their physical illness type, could be associated with psychiatric disorders like attention deficit hyperactivity disorder, which has been linked to increased numbers of accidental injuries.32  In addition, these complexities present challenges for the evaluating clinician when trying to address the etiology of somatic complaints in the presence of mental illness symptoms, such as anxiety.33  Ultimately, the dynamic relationship between mental and physical illness is likely to affect hospital utilization in multiple ways.

AYA with chronic physical illness treated in subspecialty care may benefit from more integrated mental health support. Recent efforts to improve the integration of physical and mental health care have increased overall access to mental health services,34  but the bulk of these initiatives have been within primary care settings that are already accustomed to working across different organ systems and illness domains. Additionally, the anticipated decrease in ED visits has not necessarily been borne out yet.34  Although specialists tend to focus more on their specific domain, our results suggest that there is significant comorbidity, particularly with mental health conditions, that is impacting important outcomes of care for their patient populations. Thus, further study may consider extending integrated models beyond primary care to include the subspecialty care setting.

There is evidence revealing that including mental health screening as part of pediatric subspecialty clinic visits is feasible and effective at identifying youth with chronic illness who may benefit from further mental health evaluation.35  In addition, integrated behavioral health programs with embedded child and adolescent psychiatrists and behavioral health staff in subspecialty care settings, such as pediatric palliative care and pain clinics, have shown promising results in addressing symptomatology and reducing health care utilization.36 

Although comprehensive, preventive outpatient care is the goal, interventions in the inpatient setting have been shown to address some of the sequelae of mental illness. The early inclusion of pediatric psychiatry consultation liaison services can reduce the length of stay and overall hospital costs for youth with mental health concerns who are hospitalized for physical reasons.37  Our work provides an important foundation for understanding AYA hospital utilization outcomes associated with comorbid mental and physical illness, but more work is needed on how to consistently recognize mental illness among AYA with chronic physical illness, as well as how to empower pediatric subspecialty clinics to support treatment of comorbid mental illnesses.

Sensitivity analyses lend some support to our hypothesis that mental illness would be associated with increased hospital utilization over and above general increases in disease burden; however, these findings are preliminary. Most of our findings for pairings of comorbid physical illnesses revealed no association with increased ED use and, thus, do not support the interpretation that overall disease burden is responsible for the results found with mental illness comorbidity and ED utilization. For inpatient hospitalization, however, we did observe greater odds of hospitalization among three-quarters of our pairings of comorbid physical illnesses. Not all physical illness comorbidities are the same; the most consistent finding in these sensitivity analyses was that pulmonary illness comorbidity with all 4 physical illnesses (endocrine, musculoskeletal, neurologic, and cardiac) was associated with greater inpatient hospitalization. These findings suggest that inpatient hospitalization may be more sensitive to the overall disease burden for patients with pulmonary illness compared with ED use. However, it is important to note that these sensitivity analyses have smaller sample sizes for comorbid physical illness types, and therefore, some instances in which the magnitude of the adjusted odds ratios are similar to those for mental and physical comorbidity, but are not statistically significant, may be due to not having enough statistical power to detect differences. Therefore, larger sample sizes are needed in future studies to draw conclusions from these sensitivity analyses.

We acknowledge multiple study limitations. This was a cross-sectional study; we cannot draw causal inferences from associations. In addition, our findings from a statewide, rural population of AYA may not be as generalizable to urban areas. However, we provide a methodologic framework for analyzing all-payer claims data, regardless of the location. Our sample did not identify patients in all 19 PMCA body systems, which may be due to our small state, limited algorithm diagnoses, or the database having 75% of the state population. Additionally, all-payer claims data lack potential confounders like race, ethnicity, and socioeconomic data. Some AYA may not disclose mental health issues to health care providers; thus, our sample may be a conservative estimate of comorbid mental illness. Given our population-level approach, illness type was categorized broadly by using the PMCA; disease type, severity of disease, and functional status were not assessed. Another limitation of all large administrative datasets is the accuracy of claim coding, whether it be hospital billing, the insurer, or the administrative governing body. We also did not investigate the primary diagnosis or severity of the ED or inpatient hospitalization but rather used hospital utilization as a binary variable. There is a growing body of research focusing on nonurgent or preventable ED visits,2  which would be an important future extension of this study. Lastly, this study focused on the total number of ED or inpatient admissions and did not account for repeat presentations per individual. It is likely that, among those with complex chronic conditions, there is a smaller population of high health care utilizers. Although we were limited by sample size in the current study, it would be interesting to dive deeper into such utilization differences in future work with larger datasets.

Mental illness was common among AYA and highly comorbid with multiple types of physical illness. Mental illness comorbidity is associated with greater odds of ED use among AYA with physical illness, especially pulmonary and endocrine illnesses, and greater odds of inpatient hospitalization across multiple physical illness types. Pediatric subspecialists might consider future efforts to identify and treat mental health disorders within subspecialty clinics to help reduce total symptom burden and overall hospital utilization among their AYA patient populations. However, more research is needed to advance various methods of integrating mental health care within subspecialty clinics and to measure and test ways in which these methods might potentially decrease morbidity, lower costs, and improve the overall quality of life for AYA, especially those with chronic illness.

The authors would like to thank Jonathan N. Flyer, MD, for providing pediatric cardiology expertise and reviewing a draft of the manuscript. The analyses, conclusions, and recommendations from the Vermont Health Care Uniform Reporting and Evaluation System data are solely those of the study authors and are not necessarily those of the Green Mountain Care Board. The Green Mountain Care Board had no input into the study design, implementation, or interpretation of the findings.

Dr Holland participated in study design, conducted statistical analyses, interpreted results, and drafted the manuscript; Dr Rettew helped interpret findings and helped draft the introduction and discussion; Dr Varni set up data systems, acquired the data from the claims database, and helped draft the methods section; Dr Harder designed the study, guided the statistical analyses, helped interpret results, and contributed to all sections of the manuscript; and all authors reviewed and revised the manuscript, approved the final manuscript as submitted, and agree to be accountable for all aspects of the work.

FUNDING: Dr Holland was supported as a Medical Student Research Fellow through the University of Vermont Robert Larner, MD College of Medicine. Drs Harder and Varni were supported in part by the Agency for Healthcare Research and Quality (grant R03HS024575).

CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no potential conflicts of interest relevant to this article to disclose. Study data have been previously presented at a national conference as a poster.

1
Ziv
A
,
Boulet
JR
,
Slap
GB
.
Emergency department utilization by adolescents in the United States
.
Pediatrics
.
1998
;
101
(
6
):
987
994
2
Weiss
AL
,
D’Angelo
LJ
,
Rucker
AC
.
Adolescent use of the emergency department instead of the primary care provider: who, why, and how urgent?
J Adolesc Health
.
2014
;
54
(
4
):
416
420
3
National Research Council (US) and Institute of Medicine (US) Committee on Adolescent Health Care Services and Models of Care for Treatment, Prevention, and Healthy Development
. In
Lawrence
RS
,
Appleton Gootman
J
,
Sim
LJ
, eds.
Adolescent Health Services: Missing Opportunities
.
Washington, DC
:
National Academies Press
;
2009
4
Christakis
DA
,
Mell
L
,
Koepsell
TD
, et al
.
Association of lower continuity of care with greater risk of emergency department use and hospitalization in children
.
Pediatrics
.
2001
;
107
(
3
):
524
529
5
Merikangas
KR
,
Calkins
ME
,
Burstein
M
, et al
.
Comorbidity of physical and mental disorders in the neurodevelopmental genomics cohort study
.
Pediatrics
.
2015
;
135
(
4
):
e927
e938
6
Bardach
NS
,
Neel
C
,
Kleinman
LC
, et al
.
Depression, anxiety, and emergency department use for asthma
.
Pediatrics
.
2019
;
144
(
4
):
e20190856
7
Rechenberg
K
,
Whittemore
R
,
Grey
M
.
Anxiety in youth with type 1 diabetes
.
J Pediatr Nurs
.
2017
;
32
:
64
71
8
Lawrence
JM
,
Standiford
DA
,
Loots
B
, et al
;
SEARCH for Diabetes in Youth Study
.
Prevalence and correlates of depressed mood among youth with diabetes: the SEARCH for Diabetes in Youth study
.
Pediatrics
.
2006
;
117
(
4
):
1348
1358
9
Robinson
ME
,
Simard
M
,
Larocque
I
, et al
.
Risk of psychiatric disorders and suicide attempts in emerging adults with diabetes
.
Diabetes Care
.
2020
;
43
(
2
):
484
486
10
Gonzalez
VJ
,
Kimbro
RT
,
Cutitta
KE
, et al
.
Mental health disorders in children with congenital heart disease
.
Pediatrics
.
2021
;
147
(
2
):
e20201693
11
Khanna
AD
,
Duca
LM
,
Kay
JD
, et al
.
Prevalence of mental illness in adolescents and adults with congenital heart disease from the Colorado Congenital Heart Defect Surveillance System
.
Am J Cardiol
.
2019
;
124
(
4
):
618
626
12
Russ
SA
,
Larson
K
,
Halfon
N
.
A national profile of childhood epilepsy and seizure disorder
.
Pediatrics
.
2012
;
129
(
2
):
256
264
13
LaGrant
B
,
Marquis
BO
,
Berg
AT
,
Grinspan
ZM
.
Depression and anxiety in children with epilepsy and other chronic health conditions: national estimates of prevalence and risk factors
.
Epilepsy Behav
.
2020
;
103
(
Pt A
):
106828
14
Berry
JG
,
Ash
AS
,
Cohen
E
, et al
.
Contributions of children with multiple chronic conditions to pediatric hospitalizations in the United States: a retrospective cohort analysis
.
Hosp Pediatr
.
2017
;
7
(
7
):
365
372
15
Doupnik
SK
,
Rodean
J
,
Feinstein
J
, et al
.
Health care utilization and spending for children with mental health conditions in Medicaid
.
Acad Pediatr
.
2020
;
20
(
5
):
678
686
16
Berry
JG
,
Rodean
J
,
Hall
M
, et al
.
Impact of chronic conditions on emergency department visits of children using Medicaid
.
J Pediatr
.
2017
;
182
:
267
274
17
Kaiser Family Foundation
.
Health insurance coverage of the total population 2019
.
18
Gobeille v., Liberty Mut. Ins. Co
.
S. Ct
. Vol
136
:
Supreme Court
;
2016
:
936
19
Simon
TD
,
Haaland
W
,
Hawley
K
, et al
.
Development and validation of the pediatric medical complexity algorithm (PMCA) Version 3.0
.
Acad Pediatr
.
2018
;
18
(
5
):
577
580
20
American Psychiatric Association
.
Diagnostic and Statistical Manual of Mental Disorders: DSM-5
.
Washington, DC
:
American Psychiatric Association
;
2013
21
Venkatesh
AK
,
Mei
H
,
Kocher
KE
, et al
.
Identification of emergency department visits in Medicare administrative claims: approaches and implications
.
Acad Emerg Med
.
2017
;
24
(
4
):
422
431
22
Holland
JE
,
Varni
SE
,
Pulcini
CD
, et al
.
Assessing the relationship between well-care visit and emergency department utilization among adolescents and young adults
.
J Adolesc Health
.
2022
;
70
(
1
):
64
69
23
StataCorp
.
Stata Statistical Software
.
Release 15
.
College Station, TX
:
StataCorp LP
;
2017
:
Software
24
Skuse
D
,
Williams
P
.
Screening for psychiatric disorder in general practice
.
Psychol Med
.
1984
;
14
(
2
):
365
377
25
Costello
EJ
,
Edelbrock
C
,
Costello
AJ
, et al
.
Psychopathology in pediatric primary care: the new hidden morbidity
.
Pediatrics
.
1988
;
82
(
3 Pt 2
):
415
424
26
Kessler
RC
,
Avenevoli
S
,
Costello
EJ
, et al
.
Prevalence, persistence, and sociodemographic correlates of DSM-IV disorders in the National Comorbidity Survey Replication Adolescent Supplement
.
Arch Gen Psychiatry
.
2012
;
69
(
4
):
372
380
27
Whitney
DG
,
Peterson
MD
.
US national and state-level prevalence of mental health disorders and disparities of mental health care use in children
.
JAMA Pediatr
.
2019
;
173
(
4
):
389
391
28
Ferro
MA
.
Major depressive disorder, suicidal behaviour, bipolar disorder, and generalised anxiety disorder among emerging adults with and without chronic health conditions
.
Epidemiol Psychiatr Sci
.
2016
;
25
(
5
):
462
474
29
Sajatovic
M
,
Welter
E
,
Tatsuoka
C
, et al
.
Electronic medical record analysis of emergency room visits and hospitalizations in individuals with epilepsy and mental illness comorbidity
.
Epilepsy Behav
.
2015
;
50
:
55
60
30
Desai
R
,
Patel
K
,
Dave
H
, et al
.
Nationwide frequency, sequential trends, and impact of co-morbid mental health disorders on hospitalizations, outcomes, and healthcare resource utilization in adult congenital heart disease
.
Am J Cardiol
.
2020
;
125
(
8
):
1256
1262
31
Cussotto
S
,
Sandhu
KV
,
Dinan
TG
,
Cryan
JF
.
The neuroendocrinology of the microbiota-gut-brain axis: a behavioural perspective
.
Front Neuroendocrinol
.
2018
;
51
:
80
101
32
Chang
Z
,
Quinn
PD
,
Hur
K
, et al
.
Association between medication use for attention-deficit/hyperactivity disorder and risk of motor vehicle crashes
.
JAMA Psychiatry
.
2017
;
74
(
6
):
597
603
33
Pao
M
,
Bosk
A
.
Anxiety in medically ill children/adolescents
.
Depress Anxiety
.
2011
;
28
(
1
):
40
49
34
Walter
HJ
,
Vernacchio
L
,
Trudell
EK
, et al
.
Five-year outcomes of behavioral health integration in pediatric primary care
.
Pediatrics
.
2019
;
144
(
1
):
144
35
Iturralde
E
,
Adams
RN
,
Barley
RC
, et al
.
Implementation of depression screening and global health assessment in pediatric subspecialty clinics
.
J Adolesc Health
.
2017
;
61
(
5
):
591
598
36
Samsel
C
,
Ribeiro
M
,
Ibeziako
P
,
DeMaso
DR
.
Integrated behavioral health care in pediatric subspecialty clinics
.
Child Adolesc Psychiatr Clin N Am
.
2017
;
26
(
4
):
785
794
37
Bujoreanu
S
,
White
MT
,
Gerber
B
,
Ibeziako
P
.
Effect of timing of psychiatry consultation on length of pediatric hospitalization and hospital charges
.
Hosp Pediatr
.
2015
;
5
(
5
):
269
275

Supplementary data