November 2016, VOLUME138 /ISSUE 5

Home Health Nursing Care and Hospital Use for Medically Complex Children

  1. James C. Gay, MD, MMHCa,b,
  2. Cary W. Thurm, PhDc,
  3. Matthew Hall, PhDc,
  4. Michael J. Fassino, MSd,
  5. Lisa Fowler, BAd,
  6. John V. Palusci, MBAd, and
  7. Jay G. Berry, MD, MPHe,f
  1. aDepartment of Pediatrics, Vanderbilt University School of Medicine, Nashville, Tennessee;
  2. bMonroe Carell Jr. Children's Hospital at Vanderbilt University, Nashville, Tennessee;
  3. cChildren’s Hospital Association, Overland Park, Kansas;
  4. dBAYADA Home Health Care, Moorestown, New Jersey;
  5. eDivision of General Pediatrics, Boston Children’s Hospital, Boston, Massachusetts; and
  6. fDepartment of Pediatrics, Harvard Medical School, Boston, Massachusetts
  1. Dr Gay participated in study concept and design, data analysis and interpretation, and orchestrated critical revision of the manuscript by all coauthors; Drs Thurm and Hall had full access to all data in the study, are responsible for the integrity of the data and the accuracy of the data analysis, and participated in study concept and design, data analysis and interpretation, and drafting and critical revision of the manuscript; and Mr Fassino, Ms Fowler, Mr Palusci, and Dr Berry participated in study concept and design, data analysis and interpretation, and drafting and critical revision of the manuscript.


BACKGROUND AND OBJECTIVE: Home health nursing care (HH) may be a valuable approach to long-term optimization of health for children, particularly those with medical complexity who are prone to frequent and lengthy hospitalizations. We sought to assess the relationship between HH services and hospital use in children.

METHODS: Retrospective, matched cohort study of 2783 hospitalized children receiving postdischarge HH services by BAYADA Home Health Care across 19 states and 7361 matched controls not discharged to HH services from the Children’s Hospital Association Case Mix database between January 2004 and September 2012. Subsequent hospitalizations, hospital days, readmissions, and costs of hospital care were assessed over the 12-month period after the initial hospitalization. Nonparametric Wilcoxon signed rank tests were used for comparisons between HH and non-HH users.

RESULTS: Although HH cases had a higher percentage of complex chronic conditions (68.5% vs 65.4%), technology assistance (40.5% vs 35.7%), and neurologic impairment (40.7% vs 37.3%) than matched controls (P ≤ .003 for all), 30-day readmission rates were lower in HH patients (18.3% vs 21.5%, P = .001). At 12 months after the index admission, HH patients averaged fewer admissions (0.8 vs 1.0, P < .001), fewer days in the hospital (6.4 vs 6.6, P < .001), and lower hospital costs ($22 511 vs $24 194, P < .001) compared with matched controls.

CONCLUSIONS: Children discharged to HH care experienced less hospital use than children with similar characteristics who did not use HH care. Further investigation is needed to understand how HH care affects the health and health services of children.

  • Abbreviations:
    complex chronic condition
    home health nursing care
    interquartile range
    private duty nursing
    skilled nursing visit
  • What’s Known on This Subject:

    There is increasing attention to the use of home health care for children with medical complexity. However, little is known about the extent to which home health services influence hospital resource use in this population of children.

    What This Study Adds:

    Hospitalized children discharged to home care services experienced fewer readmissions and subsequent hospitalizations than similar children not receiving such services. The use of postdischarge home health services may reduce subsequent use of hospital resources, particularly for children with medical complexity.

    The population of children with medical complexity is growing throughout the United States, accounting for a large and increasing proportion of pediatric care.1,2 These children have lifelong, complex chronic conditions (CCCs) that are often associated with significant functional impairment, myriad healthcare needs, and high resource use.35 Health care spending for these children is largely dominated by their hospital care6; minimizing it is a major focus of U.S. population-based cost containment efforts.7 It is hoped that the children’s transitions from the hospital to high quality outpatient, community, and home care management will optimize their health and contain healthcare costs.

    Home health nursing care (HH) is often an integral component of care for children with complex medical conditions and offers the potential to reduce hospital use. HH encompasses a wide range of health care services that can be given in a patient’s home for an illness or injury. For some patients, HH can be a less expensive, more convenient and equally effective alternative to care delivered in a hospital or postacute care facility.8 Readmission reductions are reported in studies of HH for postpartum women and older adults.912 However, little information is available on the influence of HH services for the pediatric population. Paul, et al12 found that HH was safe and effective after discharge from maternity and newborn care. Home visits by nursing personnel might be particularly beneficial for children with complex medical conditions, who are at high risk for recurrent use of hospital care over time.1,2

    Pediatric HH falls into 2 distinct types: “intermittent skilled nursing visits” (SNV) and “private duty nursing” (PDN). SNVs aim to teach and support families/caregivers toward the goal of independence at home. With SNV, a nurse may coordinate a home care plan with the family or health care team and perform laboratory work, infusion therapy, or blood glucose monitoring. SNVs are typically short, intermittent, and time-limited. PDN (also known as “extended hour care,” “shift care,” or “in-home” nursing) typically aims to replace the need for long-term hospitalization or institutionalization. For example, a PDN may implement the care plan for children with medical complexity (eg, tracheostomy/ventilator dependency) who otherwise could not live safely at home. PDN services are often provided on an extended basis (eg, 24 hours per day) for many months or years.

    Despite the growing use of HH services,7,13 little is known about the extent to which they influence hospital resource use in children. Therefore, the objective of this study was to assess the impact of postdischarge HH services on future hospital use. We hypothesized that children discharged to HH services would use fewer hospital resources over time compared with matched children not receiving HH services after discharge.


    Data Sources

    HH and hospital data were obtained from billing records of BAYADA Home Health Care (Moorestown, NJ) and the Case Mix database from the Children’s Hospital Association (Overland Park, KS), respectively. Case Mix is an inpatient administrative database from 97 hospitals in 37 states. For each admission, Case Mix includes information on demographics, cost, and International Classification of Diseases, Ninth Revision, Clinical Modification diagnosis and procedure codes. Deidentified, encrypted identifiers allow children to be tracked across hospitalizations at the same hospital. The BAYADA database includes administrative, demographic, clinical, and reimbursement information for children using BAYADA HH services from 19 states in all geographic regions. BAYADA’s standard operating model infuses consistency of care across regions and states.

    Study Population and Design

    Using birth date, sex, and zip code, we identified 3516 children discharged for the first time (ie, index hospitalization) from 52 Case Mix children’s hospitals to BAYADA HH between January 1, 2004 to September 30, 2012. Each child was matched with up to 3 controls within the same Case Mix hospital who were not discharged to HH (using discharge disposition codes) but who had the same (1) age in days (within ±10%); (2) number and type of CCCs3; (3) reason for and severity of hospitalization, determined with All-Patient Refined-Diagnostic Related Groups (3M, Health Information Systems, Wallingford, CT); and (4) race/ethnicity. CCCs, identified with International Classification of Diseases, Ninth Revision, Clinical Modification codes, include conditions expected to last >12 months and involve either several different organ systems or 1 organ system severely enough to require specialty pediatric care and hospitalization.3 CCCs include medical technology to maintain a child’s health status (eg, gastrostomy, tracheostomy).14,15

    Controls were identified for 2783 (79.2%) children within the same hospital; 2124 (76.3%) had 3 controls, 330 (11.9%) had 2, and 329 (11.8%) had 1. In the final cohort, these 2783 patients discharged to BAYADA HH were matched to 7361 control patients who were discharged from the hospital without HH services (Fig 1). Each patient had only 1 index hospitalization included for analysis; patients were represented only 1 time in the analytic dataset. We used matching to directly compare outcomes between cases and controls with similar demographic and clinical characteristics within the same hospital and to optimize statistical power by using multiple controls for each case when possible.

    FIGURE 1

    Flow diagram for merging BAYADA and Case Mix databases.

    Main Outcome Measures

    For 12 months after discharge from each index hospitalization between January 1, 2004 and September 30, 2012, we assessed the number of hospitalizations, hospital bed days, hospital cost, and all-cause hospital readmissions within 30 days for HH patients and matched controls. Hospital costs, inflated to 2015 dollars, were estimated in the Case Mix database by multiplying overall patient charges by hospital-specific cost to charge ratios.16

    Statistical Analysis

    All categorical variables were summarized using frequencies and percentages. Nonparametric Wilcoxon signed rank tests were used to determine statistical differences in the main outcomes between HH cases and controls. We also stratified this analysis by type of CCC and by neonates (<30 days old at index admission) versus nonneonates. SAS 9.3 (SAS Institute, Inc, Cary, NC) was used for all analyses. For demographic data in Table 1, population differences with P values less than a Bonferroni-adjusted significance level of 0.003 were considered statistically significant. For all other data shown, P < .05 was considered statistically significant. This study was not considered human subjects research in accordance with the policies of the Vanderbilt University institutional review board.

    TABLE 1

    Demographic and Clinical Characteristics of Hospitalized Children Discharged to HH Services and Their Matched Controls


    Demographic and Clinical Characteristics

    Median age at index admission was 2.7 years (interquartile range [IQR], 0.2–11 years) for HH cases and 2.2 years (IQR, 0.1–10 years) for controls. HH cases had a smaller percentage of neonates (age <30 days) than controls (20.6% vs 24.5%, P < .001). Race/ethnicity did not vary significantly between children discharged to HH and controls (P = .9); both HH cases and controls were predominately non-Hispanic white (61.6% vs 62.3%). HH cases had a higher percentage of CCCs (68.5% vs 65.4%, P = .003), technology assistance (40.5% vs 35.7%, P < .001) and neurologic impairment (40.7% vs 37.3%, P = .002). Neuromuscular and gastrointestinal CCCs were the most prevalent among HH cases and controls. A greater percentage of HH cases used private insurance (53.9% vs 43.4%, P < .001) (Table 1).

    Index Hospitalization Characteristics

    Conditions affecting newborns and nervous, respiratory, and musculoskeletal system problems were the most common reasons for hospitalization. These problems accounted for 57.9% and 58.4% of index hospitalizations for HH cases and controls, respectively (Table 2). Pneumonia/bronchiolitis, seizure, and hip and femur procedures were among the most common specific reasons for index admission in HH cases and controls. Compared with controls, HH cases had longer index hospitalization lengths of stay (median 4 days [IQR, 2–11 days] vs 3 days [IQR, 2–10 days], P < .001] with higher costs (median $9436 [IQR, $3756–$32 502] vs $6541 [IQR, $2770–$21 815], P < .001) (Table 2).

    TABLE 2

    Index Hospitalization Characteristics of Hospitalized Children Discharged to HH Services and Their Matched Controls

    Home Health Nursing Characteristics

    Of HH cases, 92.0% (n = 2561) used PDN and 8.0% (n = 222) used exclusively SNV. In the 12 months after index hospitalization, HH cases received a median of 2.0 hours (IQR, 1.0–33.2 hours) per week for a median of 4.0 weeks (IQR, 2.0–23.3 weeks). The median total HH payment per patient was $3816 ($488–$44 006). The highest and lowest median payments were observed in children with a respiratory CCC (median, $82 833 [IQR, $12 474–$177 292) and a malignancy CCC (median, $3789 [IQR, $250–$28 406]), respectively.

    Exposure to HH and Subsequent Hospital Resource Use

    During the study period, the 2783 HH cases experienced 2311 subsequent hospitalizations within 12 months of discharge; the 7361 controls experienced 7365 subsequent hospitalizations.


    For the entire cohort, the 30-day readmission rate after discharge from the index hospitalization was lower in HH cases than in controls (18.3% vs 21.5%, P = .001) (Table 3). HH cases also had a lower 15-day readmission rate (data not shown). Findings by CCC revealed that some of the largest differences in 30-day readmission rates for HH cases versus controls occurred in children with a respiratory CCC (30.6% vs 41.1%, P < .001) or a neuromuscular CCC (22.4% vs 29.9%, P < .001) (Table 3). When stratifying the readmission analyses by neonates versus nonneonates, a lower readmission rate was observed for nonneonates only (Fig 2).

    TABLE 3

    One-Year Hospital Resource Use After Index Hospitalization Discharge Using HH Care and Matched Controls by Type of CCC

    FIGURE 2

    Readmissions after index hospitalization discharge for HH and matched controls by type of CCC. Shown are the 30-day all-cause readmission rates after index hospitalization for HH cases (in black) and controls (in gray). The bars labeled “overall” refer to a child with any CCC. Neonates are infants <30 days of age at index admission. The asterisks on the bars indicate statistically significant differences in readmission rate between HH cases and controls (*P < .001; **P value range, .001–.005).

    Number of Hospitalizations

    At 12 months, HH patients averaged 17% fewer admissions than their matched controls (0.83 [SD, 1.67] vs 1.00 [SD, 1.82], P < .001]. HH cases also had fewer admissions within 6 months (data not shown). Findings by CCC revealed that some of the largest differences in the number of admissions for HH cases versus controls occurred in children with a respiratory CCC (1.5 vs 2.1, P < .001) or a neuromuscular CCC (1.0 vs 1.4, P < .001) (Table 3). When stratifying the number of admissions within 12 months by neonates versus nonneonates, fewer admissions were observed in nonneonates only (Supplemental Tables 4 and 5).

    Hospital Bed Days and Cost

    HH cases had fewer hospital days than controls in the 12-month period after discharge from index hospitalization (mean, 6.4 days [SD 20.7] vs 6.6 [SD 18.6] days, P < .001] (Table 3). HH cases also had a lower mean cost per patient for subsequent hospital care over 12 months (mean, $22 511 [SD $85 095] vs $24 194 [SD $76 143], P < .001] Findings by CCC revealed that some of the largest differences in hospital days and costs for HH cases versus controls occurred in children with a respiratory or neuromuscular CCC (Table 3). When stratifying the hospital days and costs by neonates versus nonneonates, less hospital resource use was observed in nonneonates only (Supplemental Tables 4 and 5).


    This retrospective matched cohort study suggests that use of HH after hospitalization for children is associated with less hospital resource use over time. When compared with matched controls, children discharged to HH had lower readmission rates, hospital days, and hospital costs over the following year. This finding occurred despite characteristics suggesting that HH cases may have been more complex or severe; HH cases had a higher prevalence of CCCs and technology assistance. When stratifying analyses by neonates versus nonneonates, different findings emerged. In neonates, hospital resource use was similar in HH cases and controls; in nonneonates, hospital resource use was lower in HH cases.

    In the current study, 30-day readmission rates were high for children using HH and their matched controls (18% and 21%, respectively). These high rates likely indicate the high medical acuity and fragility of both patient groups. Similar readmission rates have been reported in children with CCCs, including those with sickle cell disease and malignancies.17 Our finding of fewer readmissions in children using HH contrasts with Feudtner et al,18 who found that HH was not associated with decreased risk of readmission within 1 year. There are several reasons to explain the difference in findings. The study by Feudtner et al18 did not compare HH with matched controls and it was not restricted to 1 HH provider. By reducing the variation across HH providers, our focus on a single entity (ie, BAYADA) was positioned differently to assess HH benefits.

    Further investigation is needed to determine why HH was associated with lower hospital resource use over time. HH may provide both anticipatory guidance and early direct treatment of health decline for children with CCCs, thus alleviating their need for hospitalization. Of course, these children could receive this type of care in a variety of different settings with other providers (eg, primary care physician in a medical home or case manager in the community) that we could not measure in this study. Future studies may wish to explore how quickly HH is equipped to respond to the children’s health needs when compared with primary and other types of care. Children with CCCs are particularly vulnerable to rapidly developing, severe manifestations of acute and chronic illnesses that, without prompt attention, often result in recurrent, lengthy hospitalizations.19 Therefore, HH may be particularly effective at limiting hospital use for these children.

    It is believed that a small percentage (<2%) of children with CCCs receive HH.3 Population health initiatives with shared savings among providers across the care continuum should assess whether cost savings from a modest reduction in hospital use could support expansion of HH.3 Although our findings of reduced hospital resource use with HH support this possibility, they should be interpreted with caution. The current study is not positioned to assess the impact of HH on total healthcare spending. The databases used do not contain the outpatient, community, emergency department, or other health services data necessary to assess this impact. Moreover, the payments made for HH reported in the current study are not equivalent to the reported costs for hospital care. Future studies should strive to compare the same type of financial data (eg, cost or payment) across the care continuum when assessing the impact of HH in children.

    There are several reasons that might help explain why the relationship between HH and hospital resource use differed between neonates and nonneonates. Perhaps the clinical trajectory of the neonates’ CCCs influenced their need for hospital care beyond the capability of HH to affect it. For example, regardless of whether HH was used, some of the neonates may have required subsequent hospital care for treatments to maintain their health and functioning. It is possible that the neonatal period of time for infants with CCCs is associated with a less “stable” clinical status that, for some, tends to settle as the infants advance in age. Subsequent studies may wish to assess how much this suspected clinical stability plays a role in the impact of HH on children with CCCs.

    This study has several other limitations. The cases receiving home health services were identified by their first encounter with BAYADA HH. Some patients could have received previous HH from another provider. Also, patients in the control group may have received HH after hospital discharge that we could not detect using the discharge disposition information available in the Case Mix database. We were unable to distinguish the true reasons for HH use, including the need for respiratory, enteral, or parenteral therapies and treatments. Readmissions to a different hospital are not included in the Case Mix database, which could have led to undercounting of readmissions for cases and controls. Absent the availability of patient name and medical record number, BAYADA patients were identified in the Case-Mix database using sex, birthdate, and zip code; this identification process may have been imprecise for some patients.

    Despite a robust attempt to match HH cases and controls for complexity and severity, there could have been unaccounted differences. For example, the type of insurance varied for HH cases and controls. In a post hoc analysis, we stratified the relationship of HH with hospital resource use by payor. We found that HH was associated with lower resource use in both children with public and private insurance (data not shown). HH cases and controls were not matched specifically on the exact timing of the index admission because too few matches would have been found. Nevertheless, most cases and controls were matched during a similar time period (eg, recent HH cases were matched with recent controls). Although we matched HH cases and controls within the same hospital, there could still be practice variation across regions that could affect the findings. HH cases and controls were not explicitly matched by age of onset of CCC because the administrative data were not designed to distinguish this. However, the cardiovascular CCC category, comprised mostly of patients with congenital heart disease, likely had an onset during gestation. Matching HH cases and controls by age for children with a cardiovascular CCC helped compare children at their onset of disease (ie, closer to birth) or later in life (ie, farther out from birth) depending on when their index admission occurred.

    These limitations provide motivation for further studies to assess the impact of HH services. The findings from the current study suggest that HH may influence hospital use of children with CCC, leading to a reduction in hospital readmissions, admissions, total hospital days, and costs. Shifting care to nonhospital environments may have the potential for significant cost savings for hospitals and integrated medical systems. For the family, care for the child with medical complexity in the home environment may provide greater comfort and satisfaction. Fewer missed days of work, reductions in travel time and expense, and the reassurance of familiar surroundings are all potential benefits for patients and/or family members.

    Future investigations should explore prospective analyses with rigorous designs (eg, randomization or pre–post test designs with a control group) to truly assess the impact of HH on hospital use for children. Additionally, although the majority of BAYADA patients in this study received PDN, some patients also benefited from episodic SNV, and the impact of these specific services should be studied in more detail. Because patients with medical complexity have variable health care needs often over the span of many years, it would also be valuable to follow patients beyond a 12-month time frame. Furthermore, an all-encounter dataset, such as Medicaid claims data, would allow much needed assessment of the true impact of HH across the entire spectrum of care, including outpatient visits for primary and subspecialty care as well as emergency department and inpatient care.


    Hospitalized children with CCCs discharged to HH experienced fewer short-term readmissions, subsequent hospitalizations, and lower hospital costs over a 12-month period than matched controls of children with similar attributes who were not discharged to HH. The use of postdischarge HH may help to limit subsequent use of hospital resources, particularly for children with complex medical conditions.


    The authors thank Margaret O’Neill, BS, for her valuable assistance in preparing the figures.


      • Accepted August 23, 2016.
    • Address correspondence to James C. Gay, MD, MMHC, Department of Pediatrics, Vanderbilt University School of Medicine, Monroe Carell Jr. Children’s Hospital at Vanderbilt, 11204 Doctor’s Office Tower, 2200 Children’s Way, Nashville, TN 37232. E mail:{at}
    • FINANCIAL DISCLOSURE: Ms Fowler, Mr Fassino, and Mr Palusci are all full-time employees of BAYADA Home Health Care. The other authors have indicated they have no financial relationships relevant to this article to disclose.

    • FUNDING: No external funding.

    • POTENTIAL CONFLICT OF INTEREST: Ms Fowler, Mr Fassino and Mr Palusci are all full-time employees of BAYADA Home Health Care. The other authors have indicated they have no potential conflicts of interest to disclose.