Hospital Readmission of Adolescents and Young Adults With Complex Chronic Disease

Key Points Question How do readmission rates vary across complex chronic disease for adolescents and young adults with increasing age? Findings In this cross-sectional study of 215 580 adolescents and young adults hospitalized for treatment of complex chronic diseases (cystic fibrosis, type 1 diabetes, inflammatory bowel disease, spina bifida, or sickle cell anemia), 30-day hospital readmission rates varied significantly across disease categories. As age increased from 15 to 30 years, unadjusted, 30-day, unplanned hospital readmission rates increased significantly for all complex chronic disease cohorts. Meaning Further attention is needed to hospital discharge care, self-management, and prevention of readmission in adolescents and young adults with complex chronic disease.


Introduction
Adolescents and young adults (AYA) with complex chronic disease (CCD) are a growing population who experience severe, acute health problems. [1][2][3] Examples of CCD include cystic fibrosis, type 1 diabetes, inflammatory bowel disease (IBD), sickle cell anemia, and spina bifida. Aging from adolescence to adulthood is an important process of AYA who have a CCD. As their age increases, many AYA who have a CCD may experience progression of disease, with new or worsening existing or coexisting illness and increased functional impairment. 4,5 This fragile health status may lead to acute health problems, including infections and chronic condition exacerbations.
Hospitalization of AYA who have a CCD may be required for treatment of acute problems and for support during recovery from illness or procedures. Hospital discharge can be a vulnerable time for AYA who have a CCD, depending on their age and stage of transition to adulthood. For example, AYA with newly acquired autonomy and/or less familial support as they age may not fully appreciate the importance of medication adherence or discharge care plan instructions. Additionally, transitions of care in the outpatient setting to new adult-based practitioners may complicate hospital discharge plans, including contingency plans for problems that could arise. These hospital discharge experiences may vary by the type of CCD experienced by AYA. For example, AYA with CCDs that can rapidly worsen in severity with insufficient care management (eg, ketoacidosis with type 1 diabetes) may be more apt to follow through with discharge plans.
Emerging evidence suggests that age is a prominent risk factor associated with the odds of unplanned hospital readmission in AYA. One prior study 6 reported a substantial increase in the odds of readmission for US adolescents as they aged into adulthood; their odds approximate the odds of readmission among elderly Medicare beneficiaries. As nationwide interest in reducing readmissions extends beyond elderly individuals, AYA-especially those with CCD-are an important, vulnerable population of hospitalized patients who deserve further attention. 7-10 Therefore, we conducted the current national study to improve knowledge about hospitalized AYA who have a CCD and their readmissions during their transition to adulthood. For AYA aged 15 to 30 years, we assessed variation in the likelihood of readmission with increasing age and across different CCDs. We hypothesized that the likelihood of readmission would increase with increasing age and would vary significantly across CCDs.

Study Population
Index admissions (defined as any admission during the study period with 30-day eligibility for   readmission measurement after discharge) for individuals aged 15 to 30 years at admission with   cystic fibrosis, type 1 diabetes, IBD, spina bifida, and/or sickle cell anemia between January 1, 2014,   and December 1, 2014, were identified using the AHRQ Chronic Condition Indicator and Clinical Classification System. 2,11 These 5 diseases were chosen for analysis because each represents a distinct organ system, has associations with complex medical needs, and requires strong selfmanagement skills to optimize well-being. We used the AHRQ Chronic Condition Indicator and Clinical Classification System International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes to distinguish the chronic diseases included in our study cohort. Index admission measurement ended on December 1, 2014, to allow for a full 30-day readmission period following all discharges. All-condition index admissions (ie, admission for any reason) were measured for individuals with the chronic diseases aside from admissions for pregnancy and childbirth, which were excluded using Major Diagnostic Category 14. Guided by methods used by the Centers for Medicare & Medicaid Services, we also excluded index admissions for patients who died, left against medical advice, were transferred to another acute-care hospital, or had a principal diagnosis of cancer (AHRQ Clinical Classification System groups 11-45). 12,13

Main Outcome Measure
The main outcome measure was 30-day, unplanned, all-cause readmission to any hospital following discharge from an index admission. Readmissions were defined using Centers for Medicare & Medicaid Services methodology. 13 Centers for Medicare & Medicaid Services excludes planned readmissions using AHRQ's Clinical Classification System, which uses principal ICD-9-CM diagnosis and procedure codes to identify admissions that are considered planned or potentially planned (eg, chemotherapy, labor and delivery).

Index Admission Demographic and Clinical Characteristics
We assessed sex, payer (Medicare, Medicaid, private, self-pay, no charge, and other), length of stay, and discharge disposition (home with self-care, postacute care, home with home nursing services).
We assessed the reason for each index admission using 3M Health Information System's All Patient Refined Diagnosis Related Groups. 14, 15 We also assessed the quarter of the year for hospital discharge, with January through March composing the first quarter; April through June, the second; July through September, the third; and October through December, the fourth. Race/ethnicity was not assessed because it was not included in the NRD.
To count and describe patients' coexisting chronic conditions, we used the AHRQ Chronic Condition Indicator, 11 a tool that identifies conditions that are expected to last 12 months or longer and meet 1 or both of the following criteria: (1) the condition places limitations on self-care, independent living, and social interactions and (2) the condition results in the need for ongoing intervention with medical products, services, and special equipment.

Hospital Characteristics
We assessed the teaching status and location of index and readmitting hospitals (metropolitan teaching, metropolitan nonteaching, and nonmetropolitan) as available in the NRD. A hospital was considered a teaching hospital if it had an American Medical Association-approved residency program, was a member of the Council of Teaching Hospitals, or had a 0.25 or higher ratio of full-time equivalent interns and residents to hospital beds. Metropolitan hospitals were those in small and large metropolitan areas defined by the US Department of Agriculture Urban Influence Codes. 16

Statistical Analysis
Using NRD weighting methods, we estimated the total number of index admissions and 30-day unplanned readmissions for all individuals aged 15 to 30 years at admission with each of the 5 CCD cohorts. The weighing method estimates findings from the NRD sample (n = 2006 hospitals) to the universe of all US hospitals and their discharges. 17 We calculated discharge weights using poststratification on the sample hospital characteristics (US Census region, urban or rural location, teaching status, bed size, and hospital control) and patient's sex and age. The target universe of inpatient discharges across all hospitals in the United States was determined for each stratum (defined by the hospital and patient characteristics listed) using AHRQ's 47 State Inpatient Databases, which include 95% of all US hospital discharges, and American Hospital Association hospital discharge counts for hospitals not reported in the State Inpatient Databases. Within each stratum, each NRD inpatient admission received a discharge weight that was equal to the total number of US inpatient discharges it represented.
In bivariable analysis, comparisons of index admission characteristics between patients with and without a readmission were made using χ 2 tests for categorical variables and Wilcoxon rank sum tests for continuous variables. In multivariable analysis, a logistic regression model was derived for individuals with each chronic disease (5 total models) to estimate the adjusted odds of readmission for age at admission using fixed effects to control for confounding variables known to influence the odds of readmission, including the number of chronic conditions, sex, payer, length of stay, discharge disposition, and hospital type. 14,[18][19][20] In the model, age was entered in 2-year increments with age 15 to 16 years as the reference. All analyses were performed using SAS statistical software version 9.4 (SAS Institute). Two-sided P values less than .001 were considered statistically significant owing to the large sample size.  Table 1). There was a female predominance for all chronic diseases; this finding persisted when excluding admissions for pregnancy and childbirth. There was significant variation in Medicaid enrollment across disease categories (27.1% for IBD, 33.1% for cystic fibrosis, 41.6% for type 1 diabetes, 47.7% for spina bifida, and 58.5% for sickle cell anemia).

Characteristics of the Study Population
Regarding clinical characteristics, most hospitalized individuals had multiple coexisting conditions and many were assisted with medical technology ( Table 2). The percentages of hospitalized AYA with 4 or more coexisting conditions were 33.4% for IBD, 34.2% for sickle cell anemia, 44.5% for diabetes, 65.2% for cystic fibrosis, and 74.2% for spina bifida. Among all 5 studied diseases, depression was one of the most common coexisting conditions (18.8% for sickle cell anemia, 21.0% for IBD, 24.0% for spina bifida, 28.4% for diabetes, and 34.2% for cystic fibrosis).
Participants with spina bifida had more diverse reasons for index admission, the most common of which was kidney and urinary tract infection (11.3%).

30-Day Unplanned Hospital Readmissions
Hospital readmission rates varied among the studied diseases (19.8% for IBD, 20.2% for cystic fibrosis, 20.4% for spina bifida, 22.5% for diabetes, and 34.6% for sickle cell anemia) and increased with increasing age (Figure 1). The smallest increase in readmission rate (31.3%) was observed with

Multivariable Analysis of Patient Characteristics and 30-Day Unplanned Hospital Readmission
After controlling for patients' other demographic and clinical characteristics, use of public insurance was associated with higher odds of readmission for all studied diseases; odds ratios (ORs) for readmission in AYA with public vs other types of insurance ranged from 1. There was variation across diseases in the association between increasing age and the adjusted odds of readmission (Figure 2). For example, the odds of readmission for AYA with cystic fibrosis  (Figure 2).

Discussion
This article describes the characteristics of hospitalized AYA who have a CCD and shows that this population experienced high rates of 30-day unplanned hospital readmission that increased with age. The hospitalized population of AYA who have CCDs is remarkable for its female predominance  The graph shows rates of 30-day, unplanned hospital readmission after index hospitalization for individuals with 5 chronic diseases. The rates are presented with increasing age in years. and its prevalence of comorbid conditions, particularly mental health diagnoses. Of the CCDs studied here, AYA who have sickle cell anemia had the highest unadjusted readmission rates. Adolescents and young adults who have cystic fibrosis experienced the largest increase in the adjusted odds of hospital readmission with increasing age. In contrast, AYA who have diabetes experienced an increase in the odds of readmission-peaking at 23 years-and a subsequent decrease with age.
Across all CCDs, public insurance and multiple coexisting conditions were associated with higher odds of readmission.
Regarding patient characteristics, hospitalized AYA who have CCDs were predominantly female, with female patients accounting for three-fifths of admissions. Prior studies 21-25 also report a  The graph shows ORs with 95% confidence intervals of 30-day, unplanned hospital readmission by age in 2-year epochs. The ORs were adjusted for age, sex, payer, and number of chronic conditions. predominance of female admissions across the CCDs studied. Sex differences in incidence of the studied CCDs do not explain this finding. For example, type 1 diabetes has a male predominance of indicence. 26 Reproductive health care needs might be hypothesized to help explain this finding.
However, admissions for pregnancy and childbirth were excluded from our analysis. Although not assessed in the current study, the female predominance in hospitalization may be related to prior investigations that show potential differences in self-management (eg, medication management), 27,28 use of outpatient care (eg, for chronic disease management), 29,30 disparate severity of disease (eg, earlier mortality in women with cystic fibrosis), 22,31 and perceptions of symptoms that influence the decision to seek and provide inpatient care. 32 Further investigation is thus needed to understand the causes and significance of the sex differences in hospital use for AYA who have CCDs.
In addition to the finding on sex, our study revealed a high prevalence of coexisting conditions Relatedly, our study also found a high prevalence of depression in hospitalized AYA who have CCDs. Mental health diagnoses may compound the difficulty of CCD management by impairing patients' abilities to adhere to discharge treatment plans. For example, in patients with type 1 diabetes, depression is associated with higher values of glycated hemoglobin. 33 Prior studies also report an association between reduced lung function and depression in individuals with cystic fibrosis. 12,30,34,35 Depression has also been associated with impaired disease control for IBD. 36 It remains difficult to distinguish whether coexisting depression leads to worsened health or whether depression develops as a result of worsened health. 37 Regardless, emerging mental health interventions for individuals with chronic disease show promise to optimize disease control. 36,38 For example, intensive psychotherapy has been shown to decrease admissions for diabetic ketoacidosis in patients with diabetes. 38 Insufficient access to and inconsistent use of these interventions complicate this proposition, especially during discharge planning. 39,40 Therefore, the presence of a mental health diagnosis may be another indicator of the need for enhanced discharge planning for AYA who have a CCD.
Nearly one-fourth of hospitalized AYA who have a CCD in the current study experienced 30-day hospital readmission. This rate is 3-fold higher than the general population in this age group and also higher than Medicare beneficiaries older than 65 years. 6  Reliance on diagnostic coding created additional limitations. The inpatient diagnosis billing codes did not convey information about the severity of the coexisting chronic conditions. Furthermore, it is possible that some conditions may be undercoded in the inpatient setting. The NRD does not contain data on inpatient clinicians, including their training and field. Therefore, we could neither cluster data by clinician nor assess whether pediatric or adult practitioners discharged the AYA.

Conclusions
The examinations from the current study underscore the vulnerability of health after discharge for AYA who have a CCD. Adolescents and young adults who have a CCD have 30-day unplanned hospital readmission rates that increase with age and that are 3 times higher than the general population of AYA. Increased attention to hospital readmissions in AYA who have a CCD is necessary to optimize their health and safety at hospital discharge. Future studies should assess how self-management of CCD and transfer of care to adult health care practitioners influences the likelihood of hospital readmission. Future investigations might also include CCDs beyond the 5 we selected such that overarching trends in readmission rates across age can be identified.
In the interim, hospital and outpatient practitioners may find the results from the current study useful during discharge planning. Helping AYA who have a CCD understand the elevated risk of hospital readmission and how this risk increases with age may be important. Paying particular attention to processes of discharge care that AYA might experience differently in context of pediatric vs adult care may be important; examples of such processes include (1) assessing readiness for hospital discharge, (2) creating contingency plans for problems that might arise after discharge, and (3) making appointments for follow-up with outpatient and community practitioners after discharge.
Those discharge care activities might be especially pertinent for particular types of AYA, including those with multimorbidity and mental health diagnoses.