This is an IRB-approved retrospective cohort analysis of hospitalizations experienced by patients with a history of utilizing a structured complex-care clinical program for children with medical complexity within one of four freestanding children's hospitals (). These programs were chosen as a convenience sample for evaluation based on their participation in a quality improvement forum for children with medical complexity in April 2008.(14
) Although these programs are situated in geographically diverse sections of the country, they are not intended to be representative of all complex-care clinical programs for children with medical complexity.
Characteristics of Structured Clinical Programs for Children with Medical Complexity
The characteristics of each program are described in . The programs developed independently, but all share the common goal of optimizing the health of children with medical complexity and minimizing their inpatient resource utilization through care coordination among a child's hospital, outpatient and community providers. Rainbow-Boston is the only program that provides primary care for their patients; this characteristic distinguishes the Rainbow-Boston program from the Complex Care Service-Boston program. All programs describe multiple chronic illnesses or involvement with multiple pediatric specialists as patient characteristics that are typically associated with program enrollment. Only one service (Complex Care Service – Boston) describes neurodevelopmental disability as a highlighted characteristic for selection. Patients are referred to the programs by local primary, specialty and hospital providers. The Arkansas and Milwaukee programs are unique in that they receive many referrals directly from their hospital's neonatal and medical intensive care units, respectively. Rainbow-Boston accepted healthy siblings of complex patients, but they are not included in this study.
Hospitalization data for program patients were obtained from the Pediatric Health Information System (PHIS), an administrative database of inpatient admissions for children within freestanding children's hospitals.(16
) All hospitals are affiliated with the Child Health Corporation of America (CHCA, Shawnee Mission, KS), a business alliance of 42 children's hospitals. Data quality and reliability are maintained through a joint effort between CHCA and Thompson Healthcare (New York, NY). Program patients were linked with PHIS hospitalization data using their medical record number and PHIS unique identifier.
Study inclusion criteria were children with medical complexities who experienced (1
) at least one health encounter with a structured complex-care clinical program during their life; and (2
) one or more hospitalizations between July 2006 and June 2008. For each patient, all hospitalizations within the study period, affiliated with any medical or surgical service, were included for analysis.
Demographic characteristics analyzed across complex-care clinical program patients were age, sex, race/ethnicity (Caucasian non-Hispanic, black, Hispanic, other), and insurance type (public, private, self-pay) as available within PHIS.(16
Diagnosis characteristics abstracted were the number of diagnoses encountered by each child during a hospitalization and the name of each diagnosis. PHIS contains up to 21 individual International Classification of Diseases, Ninth Revision, Clinical Modification
(ICD-9-CM) diagnosis codes for each hospitalization. We categorized individual diagnosis codes into three clinical categories based on previous studies: Feudtner et al (2000)'s Complex Chronic Conditions, neurologic impairment and technology assistance.(5
Neurologic impairment and technology-assistance were chosen as proxies of functional impairment and because of their presence within the patient selection characteristics of two of the five programs.(20
) Neurologic impairment (NI) was defined as static and progressive, central and/or peripheral neurologic diagnoses associated with chronic functional and or/intellectual impairment.(22
) The NI category was an extension of Feudtner's neuromuscular diagnoses described above. Example diagnoses included encephalopathy and peripheral nervous system disorders. Technology assistance was defined as a medical device used to maintain a child's health status.(25
) Examples of technologies included gastrostomy, tracheostomy, cerebrospinal fluid ventricular shunt, and permanent indwelling catheter.5, 19–21
Hospitalization characteristics included the number of hospitalizations (and intensive care hospitalizations), length of stay per hospitalization, readmission within 30 days of a previous admission, and total charges per hospitalization. We also analyzed the principal diagnosis and procedure ICD-9-CM code for each admission as an indicator of the primary reason for admission.
We compared nominal patient characteristics (race/ethnicity, insurance type, sex, and diagnosis clinical categories) among each service using chi-square tests. We compared continuous characteristics (age, number of diagnoses) using wilcoxon rank sum and t-tests based on normality. Statistical Analysis Software (SAS Institute, Inc, Cary, NC) version 9.1.3 was used for all analyses.