OBJECTIVE: To describe the population of pediatric patients readmitted to a children's hospital within 15 days of discharge.
PATIENTS AND METHODS: Medical records were reviewed to identify characteristics of patients and their hospitalizations for all children hospitalized during calendar years 2007–2008 who were readmitted up to and including 15 days after a previous discharge.
RESULTS: Of 30 188 total hospital admissions during the study period, 2546 (8.4%) were followed by a readmission within 15 days of discharge. The age groups with the greatest number of readmissions were infants (aged 31–364 days, 20.8% of readmissions) and patients aged >10 years (31.3% of readmissions). Most readmitted patients (78.0%) had an underlying chronic illness, and patients with malignancies were most likely to be readmitted, followed by newborns and patients with neurologic conditions. Patients with malignancies also experienced the greatest number of readmissions per patient (4.1). Most patients who were readmitted had only 1 readmission (71.5%), but the small subset of patients with 3 or more readmissions accounted for 43.7% of all 15-day readmissions. Disease recurrence and natural course of the original diagnosis were the most common reasons for readmission (44.9%), followed by planned readmissions (20.6%) and readmissions for a new, unrelated illness (7.7%).
CONCLUSIONS: This report is the first description of the epidemiology of all 15-day pediatric readmissions at a children's hospital. The results of this study serve as a basis for additional analysis to determine the extent to which readmissions in the pediatric population may or may not be preventable.
WHAT'S KNOWN ON THIS SUBJECT:
Hospital readmissions are common and costly occurrences for adult patients. Our understanding of the full scope of readmissions for pediatric patients is incomplete, particularly for readmissions that occur within a short time after hospital discharge.
WHAT THIS STUDY ADDS:
This study was the first investigation of the characteristics of all readmissions within 15 days of discharge of patients at a single children's hospital. The results serve as a basis for further analysis of the preventability of readmissions in the pediatric patient population.
Hospital readmissions are common in elderly patients, and readmissions are under scrutiny by payers and policy makers because of their associated medical care costs and their potential usefulness for assessment of quality of care.1 Officials of the Centers for Medicare and Medicaid Services have reported that three fourths of all hospital readmissions in the Medicare population are potentially avoidable,2 and other investigators have maintained that significant cost savings would result from reductions in unnecessary readmissions and payments for preventable readmissions.3,–,8
The information available on pediatric readmissions is relatively sparse and comes from studies in which investigators focused mainly on specific age groups or conditions, and the results give little insight into the preventability of readmissions.9,–,12 In a more extensive study of patients discharged from 38 children's hospitals in 2004, Feudtner et al found that 16.7% of patients aged 2 to 18 years were readmitted within 365 days of the initial hospitalization,13 a figure significantly less than the 19.6% and 34.0% of Medicare beneficiaries rehospitalized within 30 and 90 days, respectively.1 Pediatric readmissions were most strongly associated with the patient's specified primary payer, the number of previous admissions, the diagnosis of a complex chronic condition, and longer length of stay (LOS) during the initial admission.13 These authors excluded data for children aged <2 years, because patterns of hospitalization in this age group are significantly different from those of older children.
Given the absence of previous data on the full scope of readmissions to an individual children's hospital, we identified and examined all readmissions that occurred within 15 days after a hospital discharge in a single children's hospital and sought to answer the following questions:
What is the incidence of 15-day readmissions to our hospital?
Are patients who are readmitted more likely to have acute or chronic medical conditions?
What are the characteristics of the readmission hospitalizations?
PATIENTS AND METHODS
The Monroe Carell Jr Children's Hospital at Vanderbilt (MCJCHV) is a tertiary children's medical center associated with Vanderbilt University Medical Center (Nashville, TN). During the study period, the hospital had a total of 222 inpatient beds (including 36 for critical care and 78 for neonatal intensive care), and a newborn nursery. For the purpose of this analysis, pediatric patients admitted to the newborn nursery and those admitted under “observation status” were included as patient admissions to MCJCHV, but those admitted for inpatient psychiatric treatment at Vanderbilt University Medical Center, rather than MCJCHV, were not. In addition to providing care to patients in all pediatric medical and surgical subspecialties, MCJCHV is also the major hospital in the community to which practicing pediatricians admit patients who are not newborns, although other hospitals in the area have newborn nurseries and several have NICUs. Thus, MCJCHV serves a broad mix of acute and chronic medical and surgical patients and thus provides an overview of hospitalizations and readmissions among pediatric patients.
As the first phase of a quality improvement initiative, we identified all readmissions to MCJCHV in a 2-year period (2007–2008) that occurred within 15 days of a previous admission (designated as the “index admission”). The initial database of readmissions was generated from internal Medipac administrative data collected on all MCJCHV inpatients. There were 30 188 total admissions for the study period, including 5545 newborns born at Vanderbilt University Medical Center. Data elements for each admission included medical record number, dates of admission and discharge, and all diagnostic and procedure codes of the International Classification of Diseases, 9th Edition, Clinical Modification (up to 25 codes for each admission). All children admitted to MCJCHV between the dates of January 1, 2007, and December 31, 2008, were eligible for inclusion, and those who were readmitted to MCJCHV up to and including 15 days after discharge from a previous hospitalization were identified for chart review. Readmissions that occurred after December 31, 2008, were included if the index admission had occurred on or before that date. The 15-day interval was selected instead of a 30-day interval because although the longer interval would have enabled us to capture data on more patients who were readmitted to the hospital, a 30-day interval would also have led to the inclusion of more data on readmissions with no relation to care received during the first hospitalization. Use of the 15-day interval correlated with a future goal to examine the role of the first hospitalization in the potential preventability of the readmission. In addition, an interval of 15 days provided the opportunity to examine and compare the subset of readmissions within a 7-day window, and 15 days has been a standard used in a number of other evaluations of readmission rates.6,8 Each readmission was linked to its index admission for comparisons because we considered each index admission to have the greatest potential impact on the subsequent readmission.
Diagnostic and procedure codes of the International Classification of Diseases, 9th Edition, Clinical Modification and the electronic medical record were reviewed for each admission and its associated readmission by 1 of the authors (Dr Gay). Clinical data were abstracted and patients were assigned to mutually exclusive patient diagnostic categories on the basis of their most prominent acute or chronic underlying condition (eg, oncology, neurology, neurosurgery, congenital heart disease, asthma, bronchiolitis) to track readmission characteristics among different inpatient populations. In the process of assignment of patients to a chronic illness category, we took into account the list of complex chronic conditions developed by Feudtner et al14 and also included other significant chronic illnesses such as diabetes and asthma. Relatively minor chronic conditions such as seasonal allergies and eczema were not included. Also noted was the most prominent reason for each readmission as given in the medical record (eg, planned admission for procedures such as chemotherapy or elective surgery, or return for treatment of postsurgical infection, a continuation of the initial disease process, or central venous line infection).
This study was approved by the institutional review board of Vanderbilt University.
For continuous measures that were generally skewed (eg, age, LOS), medians and quartiles are reported instead of means and SDs. Statistical comparisons were done by using Pearson's χ2 (categorical outcomes) and the Wilcoxon rank-sum test (continuous outcomes) using a significance level of .05. Because of the large sample size, we focused more on the magnitude of differences and related confidence intervals (if applicable) rather than the P-values for many comparisons. Individual hospitalizations were treated as independent observations, although individual patients often had >1 hospitalization.
Between January 1, 2007, and December 31, 2008, a total of 2546 patients who were admitted to MCJCHV were readmitted within 15 days of discharge. These readmissions were distributed among 1435 unique patients, such that the average number of 15-day readmissions for this patient group was 1.8. Overall, these 2546 readmissions accounted for 8.4% of all admissions (30 188 total) to the inpatient units of MCJCHV for the 2-year period.
The average age of patients at the time of their index admissions was 6.5 ± 6.7 years. The age distribution of these patients is shown in Table 1. Patients who were neonates (aged 0–30 days) at the time of their first admission during the 2-year study period accounted for ∼21% of patients with 1 or more readmissions, but neonates made up a smaller proportion of total readmissions (12.7%). In contrast, 26.5% of patients with readmissions were aged >10 years during the study period, and these individuals accounted for an even greater proportion of total index readmissions (31.3%). Of patients with 15-day readmissions, 54.5% were boys and 45.5% were girls.
The clinical patient categories listed in Table 2 show the types of illnesses for which patients were readmitted, along with counts and percentages according to clinical category and subgroups for individual patients and for total readmissions. The percentages differ considerably because patients with illnesses in certain clinical categories tended to have only 1 readmission, whereas patients with other illnesses frequently had multiple readmissions. For example, patients on the oncology service were most likely to have 15-day readmissions (13.9% of readmitted patients) followed by patients with acute infectious diseases, neonates, and patients with neurologic conditions. A substantial subset of this latter group (6.0%) had severe neurologic compromise. These patients often had significant compromise of other organ systems as well (eg, recurrent pneumonias, chronic constipation, musculoskeletal problems) in addition to their neurologic condition, and these other conditions often led to their readmissions (see below).
Patients with malignancies had the greatest number of readmissions per patient (4.1), followed by patients with short bowel syndrome who were dependent on total parenteral nutrition (3.9 readmissions per patient) and patients with biliary atresia (3.8 readmissions per patient). Other patient subgroups with an average of >2 15-day readmissions per patient included patients with brain tumors with ventricular shunts, Down syndrome with congenital heart disease, and bronchopulmonary dysplasia and patients who had undergone stem cell transplantation. One of the largest patient groups, those with acute infectious conditions, usually had only 1 readmission per patient, and newborns who had 1 readmission for hyperbilirubinemia rarely had a second readmission. The latter 2 groups probably had fewer readmissions because these patients were generally healthy, with acute conditions only.
The majority of patients (71.5%) with 15-day readmissions had only 1 such readmission, 14.2% had 2 readmissions, and 14.4% had 3 or more readmissions. Patients with 1 or 2 readmissions accounted for 40.3% and 16.0% of all readmissions, respectively. Interestingly, although they accounted for just 14.4% of total patients, patients with 3 or more readmissions accounted for 43.7% of all 15-day readmissions.
Characteristics of Initial Admissions and Readmissions
Although oncology patients made up only 13.9% of the readmitted patients, they had more than twice that proportion of readmissions (31.8%), which reflects the high number of readmissions per patient (Table 2). Patients with neurologic conditions had the next highest number of readmissions, accounting for 9.5% of all 15-day readmissions. The index admission occurred in the first week of life in 255 patients, including 209 patients whose index admission was for delivery at Vanderbilt University Medical Center and who were readmitted after their NICU or newborn nursery course (representing 7.5% of all readmissions). The largest single subset of newborn readmissions was for treatment of hyperbilirubinemia.
Most readmissions (77.1% of the total) occurred after initial admissions with an LOS of ≤7 days, with more than half occurring after index admissions with an LOS of <4 days. However, a significant subset of patients had much longer stays such that the average LOS for the initial admission for all patients who were readmitted was 7.6 ± 14.0 days (range: 0–192 days).
Planned Versus Unplanned Readmissions
Planned readmissions accounted for 20.6% of all readmissions, and 74.7% of these were for patients with malignancies who were readmitted for scheduled therapy or surgery (Table 3). A much smaller percentage of planned readmissions were for patients with congenital heart disease, general neurology including seizures, and severe neurocompromise (3.8%, 3.1%, and 2.9%, respectively). The remaining 79.4% of readmissions were unplanned and occurred in patients who were much younger and had more acute illnesses than patients who had planned readmissions (Table 3). As with patients with planned readmissions, oncology patients also had the highest number of unplanned readmissions but constituted a much smaller percentage of the total for this subset (20.7%). The most common reasons for unplanned readmissions involved care for the same or related condition as in the index admission (44.0% of all admissions) followed by complications of surgery or other procedure (11.4%), unrelated conditions (7.7%), and device complications (6.5%), which included central venous catheter infections (4.9%) (Table 4).
Acute Versus Chronic Conditions
Readmissions were also examined according to the presence or absence of a significant chronic condition. Most 15-day readmissions occurred in patients with an underlying chronic illness (78.0% of total readmissions), the most common subgroup being patients with malignancies (Table 5). The patients with acute illnesses were much younger on average and had a much shorter LOS in both the index admission and the readmission. More than half of readmissions for acute illnesses were for patients with acute infections (31.3%) or newborns with hyperbilirubinemia, fever, or apparent life-threatening events (29.0%) (Table 5).
The reasons for the readmission also varied between patients with acute and chronic illnesses. Although both groups had a significant number of readmissions for the same condition (55.3% of patients with acute illness and 42.0% of patients with chronic illness), there were more readmissions unrelated to the initial admission and very few planned readmissions or complications of medical devices among the patients with acute illness (see Table 4).
Comparison of Readmissions at 0–7, 8–15, and 0–15 Days
By examining all readmissions within 15 days of discharge, we were able to compare the subset of readmissions within 7 days with readmissions that occurred during other intervals within the 15-day period; 59.5% of readmissions occurred ≤7 days from discharge from the index admission, the fewest occurred on the same day or the day after discharge (9.0%), and the largest number occurred 2 and 3 days from discharge (19.3%). Fewer patients were readmitted from days 8 to 15 after discharge from an index admission, but these readmissions still made up a significant proportion of the total (40.5%).
When compared with readmissions that occurred from days 8 to 15 after discharge, readmissions within 0 to 7 days were more likely to be for acute illnesses, less likely to be planned, and less likely to be for conditions that were new or unrelated to the index admission (Table 6). Patients with malignancies dominated all readmission timeframes, but at 0 to 7 days from discharge there were somewhat more readmissions for patients with acute infections and newborns. Planned readmissions made up a similar proportion of readmissions at 0 to 7 days and 8 to 15 days from discharge from a previous admission (51.8% vs 48.2% of all planned readmissions, respectively).
We examined the most common reasons for readmissions in the patient groups we observed to be most frequently readmitted to the hospital. Among readmissions for patients with malignancies, 52.0% of readmissions were planned (mostly for chemotherapy), whereas 33.8% were for conditions related to treatment or the disease process itself (mostly empiric antibiotic treatment for febrile neutropenia) and 8.4% were for infections related to central venous catheters. Almost half (46.8%) of our newborns who were readmitted returned for treatment of hyperbilirubinemia, and most of the other newborns were readmitted for suspected or actual infections or apparent life-threatening events not related to their index admission in the newborn nursery. Patients who were severely neurocompromised were readmitted with additional exacerbations of the same disease process (35.8%); with a new, unrelated disease process (9.9%); or with complications from surgery or other procedures (8.6%). In another 9.3% of these patients, readmission was for another acute condition present during the index admission but not completely resolved at discharge, which led to a readmission within a very short time period.
To our knowledge, this study was the first performed to examine all readmissions to a single children's hospital within 15 days of discharge from a previous admission. Such readmissions account for 8.4% of all hospitalizations within the study period, and unplanned readmissions make up 6.7% of all admissions. Although few data on readmissions for adults or children in other children's hospitals were available for us to compare directly with ours, our readmission rate seemed to be less than that reported for elderly patients, among whom ∼20% of Medicare beneficiaries discharged from a hospital were rehospitalized within 30 days, and 34.0% were rehospitalized within 90 days.1 In our patients, both planned and unplanned readmissions occurred predominantly in patients with chronic illnesses, and patients with malignancies represented the largest subgroup for each type of readmission. Readmissions within a 7-day window were generally similar to those that occurred later in regard to patient subgroups and reasons for readmission, although within the shorter time frame there were somewhat more readmissions for patients with acute conditions such as acute infections and newborns with hyperbilirubinemia.
A limitation of the present study was that it was undertaken at a single tertiary children's hospital. With the inclusion of newborns who were not treated in the NICU and many patients of pediatricians who practiced in the community, the data we collected for the broad mix of patients who were admitted to our hospital with acute and chronic medical and surgical conditions provide an overview of readmissions over the entire pediatric age range (except for mental health conditions that required inpatient psychiatric care). Our results, therefore, may be reasonably representative of readmissions to children's hospitals within large medical centers, although they may differ somewhat from those for many freestanding children's hospitals. Furthermore, these results may not be applicable to non–tertiary-care hospitals (including community hospitals and general hospitals) that admit children.
The present study was also limited by incomplete data regarding patient admissions to other hospitals, and the actual number of readmissions in our data may have been higher if these had been taken into account. We recognize that all neonates aged ≤15 days who were admitted to MCJCHV were 15-day readmissions or transfers from some hospital if not our own. However, our goal in this study was to understand our own patients and services as a prelude to evaluating our patient care processes and eventually discovering ways to prevent unnecessary readmissions of patients treated at our institution. Furthermore, a large portion of the patients in our study with chronic diseases and special needs (who constituted 78% of our 15-day readmissions) received all or the majority of their inpatient care at MCJCHV, so the impact of admissions to other hospitals was likely to be small outside of the neonatal period. Nevertheless, a comprehensive study of pediatric hospital readmissions should include admissions to all inpatient facilities, and potentially emergency department visits as well.
Having described our 15-day readmissions, in the next phase of our study we will try to determine the extent to which readmissions differ by comparing the characteristics of patient readmissions to those of all other patient admissions to our institution. This comparison will allow us to target select subgroups of readmissions for additional scrutiny to ascertain their potential preventability. We encourage replication of this descriptive work at other institutions to test the extent to which our results can be generalized.
This study was the first to investigate all 15-day readmissions to a large children's hospital. The majority of patients were readmitted for treatment of chronic conditions, and a substantial number of these readmissions were planned. This study serves as a basis for additional analysis of differences between readmissions and other hospitalizations in the pediatric patient population and the extent to which readmissions may be preventable.
Dr Gay's work was supported in part by a contract with the National Association of Children's Hospitals and Related Institutions (NACHRI) to serve as a medical consultant for many of its classification research projects, including NACHRI research on the nature and patterns of hospital readmissions for children, which was instrumental to the development of the current study.
The authors express sincere appreciation to Wenli Wang, MS, for statistical analyses.
- Accepted February 17, 2011.
- Address correspondence to James C. Gay, MD, Division of General Pediatrics, Department of Pediatrics, Vanderbilt University School of Medicine, 11204 Doctor's Office Tower, Monroe Carell Jr Children's Hospital at Vanderbilt, 2200 Children's Way, Nashville, TN 37232-9760. E-mail:
FINANCIAL DISCLOSURE: Dr Gay serves as a medical consultant to the Classification Research division of the National Association of Children's Hospitals and Related Institutions and receives funding support for his activities in this role; and Dr Hain, Mr Grantham, and Dr Saville have indicated they have no financial relationships relevant to this article to disclose.
- LOS =
- length of stay •
- MCJCHV =
- Monroe Carell Jr Children's Hospital at Vanderbilt
- Frizzera C
- Heimler R,
- Shekhawat P,
- Hoffman RG,
- Chetty VK,
- Sasidharan P
- Feudtner C,
- Levin JE,
- Srivastava R,
- et al
- Feudtner C,
- Christakis DA,
- Connell FA
- Copyright © 2011 by the American Academy of Pediatrics