Association of Care Models and Hospital Reutilization for Children With High-Intensity Neurologic Impairment Free

Children with medical complexity (CMC) have been characterized by the presence of high health care needs, multiple chronic conditions, and functional limitations. Those with high-intensity neurological impairment (HINI), a subset of CMC, account for a disproportionate amount of hospital-based health care, including readmissions and emergency department (ED) visits.^1,2 Children with HINI may be identified through International Statistical Classification of Diseases, Tenth Revision codes indicating neurological diagnoses expected to last 12 months or longer and that result in significant systemic and/or multisystem physiologic impairment requiring subspecialty care.^3,4 Children with HINI have functional impairments often requiring assistance with daily living and reliance on durable medical equipment that add to their medical complexity and can result in more-frequent hospitalizations and longer admissions.⁵ As such, they have been identified as a subset of patients warranting dedicated clinical care and coordination.⁶ 

Multiple CMC-dedicated care models have been implemented with various degrees of success across US health care systems. Pordes et al classified these into 3 main groups: primary care–centered models, which operate on an outpatient basis; episode-based models, which often operate on an inpatient basis; and consultative- or comanagement-centered models, which may span both inpatient and outpatient settings.⁷ Studies evaluating these models have largely looked at patient, caregiver, and provider satisfaction, with several single-site randomized controlled trials evaluating health care quality and cost as primary outcomes.^8–11 Mosquera et al examined the effect of a comprehensive outpatient care model on hospital-based use. They demonstrated decreased ED visits, hospitalizations, admissions to the intensive care unit (ICU), and hospital length of stay associated with their intervention, although they did not evaluate outcomes for inpatient-based models.¹² Cohen et al similarly randomized CMC to structured complex care clinics in the outpatient setting across multiple sites, although they found no difference in care coordination among health care providers or between health care providers and families.¹³ Despite these investigations around specific care models, limited evidence exists for their effectiveness regarding reutilization, generalizability, or reproducibility.^14–16 To our knowledge, there have been no studies examining the relationship between the availability and composition of different CMC-dedicated care models within a health care system and health care reutilization for children with HINI.

Our study objective was to explore the association between presence and type of CMC-dedicated care models and health care reutilization for children with HINI. We hypothesized that children discharged from hospital systems with CMC-dedicated care models spanning both inpatient and outpatient services would experience lower reutilization, including both ED visits and rehospitalizations, within 14 and 30 days of discharge.

This was a retrospective cohort study of acute care hospitalizations (inpatient and observation) at US children’s hospitals from June 1, 2021, to June 1, 2022. Two data sources were used: (1) exposure variables were extracted from a hospital survey of CMC care model presence and composition,¹⁷ and (2) outcome variables for children with HINI were extracted from the Pediatric Health Information System (PHIS) database (Children’s Hospital Association [CHA], Lenexa, KS). The PHIS administrative database contains clinical and resource use data from tertiary children’s hospitals in the US. Data quality is ensured through a joint effort between the CHA and participating hospitals. This study was approved by the hospital system’s institutional review board.

Discharges for children aged 1 to 18 years with an HINI diagnosis⁴ were included if they occurred at a hospital that responded to the survey.¹⁷ We excluded discharges from an ICU, discharges that were not to home, and discharges of children with a concurrent transplant- and/or malignancy-related condition identified with the Complex Chronic Conditions (CCC) classification system.¹⁸ Children with transplant- and/or malignancy-related CCCs were thought likely to be cared for by a subspecialty team, as opposed to a general pediatric or complex care team. A subanalysis was conducted with technology-dependent HINI patients (identified with the CCC system).

Data about availability, total number, and type(s) of CMC-dedicated care models and discharge practices across hospitals was extracted from a previously published survey.¹⁷ This electronic survey was distributed in the summer of 2022 to hospital medicine division directors at institutions in the PHIS database. Respondents were asked to identify the types of CMC-dedicated care models available at their hospitals and could choose from various common care model types. Each institution could select multiple options with additional free text space to describe any details not fitting within the predefined categories.

Hospitals were grouped as those without any dedicated complex care model, those with an inpatient-only model, those with an outpatient-only model, and those with both an inpatient and outpatient model. Hospitals were defined as having an inpatient care model if they had 1 or more of the following: a complex care service that serves as the primary team for hospitalized patients, a consult team for medically complex patients admitted to various services, or an inpatient continuity provider who follows the patient during each admission. Hospitals with outpatient care models were defined as those with outpatient complex care services only.

The primary outcomes were the adjusted 14- and 30-day hospital reutilization rates (ED visits and rehospitalizations) for children with HINI discharged from an acute care service. Notably, our unit of analysis was discharges as opposed to individual patients, although patient-level identifiers were available to confirm that most of the discharge encounters reflected separate patients. Of the patients included in the study, 78% had 1 discharge, 14% had 2 discharges, and 8% had 3 or more discharges during the specified time period.

Model covariates included patient demographics (age, payer status), residential setting (urban/rural), Child Opportunity Index (COI),¹⁹ number of CCCs, and Hospitalization Resource Intensity Scores for Kids (H-RISK).²⁰ The COI measures neighborhood factors that support children’s well-being, showing differences in opportunity across US neighborhoods over time. H-RISK is a measure of relative severity of illness for the hospitalized pediatric population and can be used for risk adjustment of an assortment of hospital-level outcomes.

Categorical variables among hospital- and discharge-level characteristics were summarized using frequencies and percentages. Comparisons were made across hospital groups using chi-square tests. Continuous variables were summarized with median and IQRs and compared across groups with Wilcoxon rank-sum tests. Generalized estimating equations were used to evaluate the association between type of CMC care model and hospital reutilization outcomes while accounting for study covariates. A subgroup analysis for children with technology dependence was performed given the potential for higher reutilization in this population. A separate analysis was performed examining the association between total number of CMC care models available at each participant hospital, regardless of composition, and hospital reutilization. Variability in CMC health care use by race and ethnicity has been documented previously and was included in our analysis as a critical initial step toward revealing and ultimately addressing inequities in care access and delivery.²¹ All analyses were performed using SAS v.9.4 (SAS Institute, Cary, NC), and P < .05 was considered statistically significant.

Thirty-one hospitals were included in this study based on survey completion and PHIS data availability (Table 1). Of these, 24 (77%) had some type of CMC-dedicated care model available at their institution, whereas 7 (23%) had none. For hospitals with CMC-dedicated care models, 10 (32%) had only an outpatient-based CMC care model, 4 (13%) had only an inpatient-based CMC care model, and 10 (32%) had both inpatient- and outpatient-based CMC care models.¹⁷ To provide contextual information about the hospitals included in this study, median numbers of total discharges per year and HINI discharges per year are included. These were drawn from the total number of discharges after exclusions were applied. Neither the median number of total discharges per year nor the number of discharges of children with HINI were statistically different between hospitals with and without dedicated care models.

In total, 81 705 discharge encounters met inclusion criteria for this study. After exclusions, 66 560 remained (Figure 1). Most of these discharges (82.4%) occurred at hospitals with at least 1 CMC-dedicated care model present. Discharges from hospitals with CMC-dedicated care models differed from those hospitals without dedicated care models with respect to the following patient characteristics: age, race/ethnicity, primary insurance, rural vs urban zip code, need for ICU level care, COI, technology dependence, number of CCCs, H-RISK, and length of hospital stay. Compared with hospitals with no care model present, hospitals with any type of CMC-dedicated care model available discharged a higher percentage of non-Hispanic white patients and patients with private insurance. Conversely, hospitals with no CMC-dedicated care models discharged non-Hispanic Black and Hispanic patients at a higher frequency. ICU care was more common among those patient discharges from hospitals with no CMC care model present (28.9%) vs those with inpatient CMC care models present (20.5%). There was no statistically significant difference across sex (Table 1).

Unadjusted outcomes for children with HINI by care model presence are shown in Table 2. Children discharged from hospitals with no CMC-dedicated care model had the highest rates of reutilization at 13.5% and 20.9% at 14 and 30 days postdischarge, respectively. Conversely, those discharged from hospitals with only an inpatient CMC care model had the lowest rates of reutilization at 9.9% and 16.3% at 14 and 30 days postdischarge, respectively.

We explored the independent association of inpatient, outpatient, or combined inpatient and outpatient CMC care models with respect to hospital reutilization (Figure 2). When compared with children discharged from hospitals with no CMC care models available, children discharged from hospitals with only an inpatient CMC care model had lower odds of any reutilization within 14 days (adjusted odds ratio [aOR], 0.69; 95% CI, 0.51–0.94) and 30 days (aOR: 0.72; 95% CI 0.56–0.93). When comparing children discharged from hospitals with an outpatient CMC care model with those discharged from hospitals without any CMC-dedicated care models available, lower odds ratios for any reutilization were not statistically significant within either 14 or 30 days of discharge. The same pattern was seen when comparing discharges from hospitals with a combination of both inpatient and outpatient care models to those without any model.

Children discharged from hospitals with only inpatient CMC care models experienced lower odds of rehospitalization within 14 days of discharge (aOR, 0.64; 95% CI, 0.43–0.95) when compared with children discharged from hospitals with no care model present. A similar association was seen within 30 days of discharge. Odds ratios for rehospitalization at 14 and 30 days were not statistically lower for children discharged from hospitals with only outpatient care models or for those with combination models when comparing each category with the absence of any CMC care model. Children discharged from hospitals with strictly inpatient CMC care models experienced fewer ED return visits within 14 days of discharge (aOR, 0.85; 95% CI, 0.72–1). Children discharged from hospitals with either an outpatient CMC care model or a combination model did not experience lower ED return visits within either 14 or 30 days of discharge when compared with those hospitals with no care model present.

When examining the association between the number of CMC-dedicated care models available (regardless of type) and outcomes (Supplemental), children discharged from hospitals with 1 care model in comparison with none had a lower odds of reutilization within 14 days of discharge (aOR, 0.75; 95% CI, 0.56–0.99). Having multiple care models (ie, ≥2 CMC care models available) was not associated with statistically lower reutilization rates.

Of the 66 560 discharges that were included in the analysis, 25 688 (38.6%) involved a patient with technology dependence (Supplemental). The adjusted outcomes for this subgroup by care model type and availability are shown in Supplemental. When compared with children discharged from hospitals with no CMC-dedicated care models, children discharged from hospitals with inpatient-only CMC-dedicated care models experienced lower odds of any reutilization within 14 days (aOR, 0.78; 95% CI, 0.63–0.96) and 30 days (aOR, 0.82; 95% CI, 0.68–1) of discharge. In contrast, children discharged from hospitals with only outpatient CMC-dedicated care models did not have a statistically lower aOR for any reutilization, rehospitalization, or ED return. Children discharged from hospitals with both inpatient and outpatient care models experienced fewer ED return visits (aOR, 0.79; 95% CI, 0.65–0.96) within 14 days of discharge than those cared for at hospitals with no CMC-dedicated care models. Otherwise, no statistically significant associations were detected for this category.

Children’s hospitals nationwide are caring for children with HINI under a variety of care models. Additionally, different hospitals caring for similarly complex patients experience wide variation in resource use.^22–24 This study highlights the potential impact of having a CMC-dedicated care model at children’s hospitals on reutilization. Specifically, we found that children with HINI were less likely to experience any reutilization when discharged from a hospital with an inpatient CMC-dedicated care model within both 14 and 30 days of discharge. When adjusting for patient-level factors, children with HINI discharged from hospitals with any dedicated CMC care model, regardless of type, experienced reduced health care reutilization. Contrary to our hypothesis, a higher number of care models (≥2) was not associated with lower reutilization. Children discharged from hospitals with only outpatient care models experienced less-profound differences in reutilization when compared with those discharged from hospitals with no care model present. Compared with hospitals with no CMC-dedicated care models, children discharged from hospitals with combination care models that span the inpatient and outpatient setting did not experience reduced reutilization at either 14 or 30 days. We suspect this finding is secondary to the lower number of total hospitals being studied.

We noted a wide variation in presence and type of care models, which is consistent with prior literature.²⁵ We also discovered differences in patient population when comparing hospitals with no CMC-dedicated care models to those with at least 1. Hospitals with dedicated CMC care models were more likely to serve patients who had higher COIs, fewer patients of racial and ethnic minorities, and fewer patients with public insurance. Although little difference in clinical outcomes has been noted when comparing CMC based on insurance status, our finding suggests that hospitals serving populations in need of higher resources may be more likely to have created a dedicated care model for CMC.²⁶ 

The biggest differences in reutilization with dedicated care models were noted within 14 days of discharge. This association was more pronounced for the subgroup of technology-dependent patients. Hospital discharge communication is less likely to occur for this subpopulation,²⁷ suggesting that they could benefit from higher-level care coordination when in the hospital setting and at time of transition to home. Specialized care models for CMC may be better equipped to provide complex care coordination services needed to reduce preventable readmissions, particularly for patients who are technology dependent.²⁸ Readmission within a longer time frame may be more likely related to a new illness and possibly less modifiable in this population.^25,29 

It is interesting to note that children discharged from hospitals with only inpatient CMC-dedicated care models experienced lower reutilization than those discharged from hospitals with only outpatient CMC-dedicated care models when each type was independently compared with the absence of any care model. Most studies of care coordination programs are based on an outpatient-centered medical home model.^10,12,30 Inpatient complex care teams may be better positioned to provide the specific hospital-based care transition services shown to reduce reutilization, such as facilitating care coordination between multiple subspecialty teams and working with families to develop contingency plans postdischarge.³¹ 

Based on our results, most of the reutilization reduction was due to reduced rehospitalizations rather than a reduction in ED return visits. Inpatient complex care teams may be particularly well suited to reduce readmissions because providers familiar with the patient may be able to physically evaluate them in the ED and provide contingency planning to avoid admission. This function may be more difficult to coordinate with outpatient-focused services. Lastly, hospital reutilization may not be the highest-priority outcome to study for this population of HINI patients who, by virtue of being higher-risk patients, may require more-frequent ED visits and readmissions regardless of the care models and services provided. Other outcomes to continue examining include patient and caregiver satisfaction, quality of life, and number of safety events, to name a few.

Our study has a few limitations. First and foremost, it is critical to acknowledge the threat of ecological fallacy. The children who experienced lower reutilization may not have necessarily been the same as those who were exposed to certain CMC-dedicated care models. While our unit of analysis did reflect individual patients, we were not able to guarantee that each of these patients consistently received care under their respective inpatient, outpatient, or combined care models. This is, in part, because of the nature of a large database study. Moving forward, it would be useful to consider multicenter observational studies of HINI children who receive these specialized services in comparison with those who do not.

Second, the survey asked about the presence of a CMC-dedicated care model but did not ask when the care model was established or whether the model remained stable throughout the study period as a proxy for function or efficiency. The survey did not differentiate the care model categories by staffing and resources available or inclusion criteria, among other more granular characteristics that may differ across the same type of care model. Furthermore, the survey asked specifically about hospital medicine-dedicated services and did not universally include discharge services for subspecialty teams (ie, discharge practices on the pulmonary team for patients with cystic fibrosis or on the hematology team for patients with sickle cell disease).

Third, our unit of analysis was discharges rather than individual patients. This allowed for the treatment of each discharge encounter as a unique episode of care. As stated in the methods, most of the discharge encounters did represent individual patients, with less than 30% of the patients included having 2 or more discharges during the study period. Even so, patients who experienced more than 1 discharge during the study period were thought to have received the same exposure.

Fourth, it is difficult to elucidate from PHIS data which patients are cared for by a subspecialist and which are cared for by a hospitalist and potentially an inpatient CMC care model. Similarly, for patients within the study population that were transferred out of the pediatric ICU to the acute care floor, there is no way to determine how long they remained on the acute care floor prior to discharge. This is important to note because some of the patients will have been on the acute care floor and presumably under the care of a hospitalist for at least a few days prior to discharge such that the impact of the discharge services provided by a CMC-dedicated care model would have direct impact on their transition home and on subsequent health care reutilization. On the other hand, it is also possible that some CMC care models start working on discharge care coordination while the patient is still in the ICU.

Lastly, we acknowledge that our data were collected during the midst of the COVID-19 pandemic, which was shown to be associated with a decrease in hospitalizations for children with HINI and a longer length of stay for technology-dependent patients.^32,33 This suggests the need to reexamine the effects of various CMC-dedicated care models on hospital reutilization in the postpandemic era.

Children with HINI discharged from hospitals with a CMC-dedicated care model present, regardless of composition, experience lower hospital reutilization rates when compared with those discharged from hospitals with no specific care model. The presence of an inpatient CMC-dedicated care model appears to be the most highly associated with lower rehospitalizations for children with HINI within both 14 and 30 days of discharge. These results have implications for children’s health care systems nationwide interested in concentrating efforts to provide care dedicated to CMC. Further studies are needed to clarify whether inpatient-based CMC care models result in a greater reduction of hospital reutilization over outpatient-based models.

Our study also highlights the need for further inquiry into inequities in care for CMC who are of minoritized racial and ethnic backgrounds because they are more likely to be served by hospital systems without dedicated care models, potentially resulting in higher reutilization rates. Future directions might also include exploring the association of specific CMC-discharge practices with hospital reutilization, such as medication teaching and follow-up phone calls, to provide more granular guidance to programs across the country.

aOR

adjusted odds ratio

CCC

complex chronic condition

CHA

Children’s Hospital Association

CMC

children with medical complexity

COI

Child Opportunity Index

ED

emergency department

HINI

high-intensity neurologic impairment

H-RISK

Hospitalization Resource Intensity Scores for Kids

ICU

intensive care unit

PHIS

Pediatric Health Information System

We wish to acknowledge Dr Joanna Thomson for her guidance and expertise on our study population and for her review of the manuscript.