Impact of Longitudinal Care Coordination on Medically Complex Infant Outcomes in a Level IV Neonatal Intensive Care Unit Open Access

Due to major advances in neonatal care, more medically complex infants are surviving to hospital discharge and beyond.^1–3 Families of these infants cope with a unique set of challenges in the neonatal intensive care unit (NICU) and during the transition to home.^4–6 Increased psychosocial stressors including financial instability and limited health literacy can make caring for medically complex infants even more challenging.^7,8 These infants, especially those requiring advanced medical technologies, experience high rates of readmission, emergency department (ED) visits, and death in the first 2 years after hospital discharge.^9,10 

As the population of medically complex infants continues to grow, management challenges must be addressed.^2,11 Inpatient programs have been developed to support infants during NICU hospitalization and the transition to home.^12,13 After discharge, tertiary care center–based structured clinical programs at children’s hospitals have emerged to meet the needs of medically complex children by providing care coordination.^14–18 These programs strive to fulfill multiple health service needs not comprehensively met in the primary care setting.¹⁹ The Agency for Healthcare Research and Quality defines care coordination as “deliberately organizing patient care activities and sharing information among all the participants concerned with a patient’s care to achieve safer and more effective care.” (https://ahrq.gov/ncepcr/care/coordination.html).

Although care coordination programs already exist, few are targeted to meet the needs of medically complex neonates and their families, starting during the NICU stay through the first years of life.¹³ In a prior report, we evaluated the development and implementation of our longitudinal care coordination program (LCCP) that spans NICU admission, discharge, and outpatient care to help families of infants with medical complexity navigate the health care system.²⁰ Here, we further evaluate the program’s impact with respect to resource utilization. We compare patient and family demographics as well as inpatient and outpatient clinical courses of LCCP-enrolled patients with a cohort of preprogram NICU patients. We hypothesize that LCCP-enrolled infants will demonstrate improved utilization of resources over time, as well as improved discharge outcomes of survival and outpatient follow-up. The objective is to evaluate the program’s impact on key NICU resource utilization outcomes, specifically, length of hospitalization and program efficiency, to effectively identify, engage, and ultimately serve the most medically complex patients.

In late 2017, a care coordination program for medically complex NICU patients was conceptualized by the medical team at Ann & Robert H. Lurie Children’s Hospital of Chicago. The goal was to provide longitudinal care coordination during admission and transition to home and through the first years of life for the most medically complex NICU patients and families. Philanthropic support was provided by a parent-inspired entity named the “Little Heroes League.” This program was launched August 2018 at the level IV NICU where all the patients are outborn. Patients were enrolled by inpatient care coordinators who were recruited and trained at the discretion of the program’s inpatient medical directors in collaboration with NICU nursing and hospital leadership staff. Each family was assigned an inpatient and an outpatient care coordinator. The inpatient care coordinator acts as a central point of contact, coordinates family care conferences with the medical team, provides education and guidance in navigating care including setting weekly learning goals and discharge planning, and removes barriers that may delay discharge. The outpatient care coordinator performs home visits, accompanies families to pediatric and other subspecialty appointments as needed, helps coordinate appointments, identifies transportation resources, connects family with community resources, and provides education and social and emotional support. Details were previously described in a published qualitative analysis of the LCCP.²⁰ The program employs up to 2 inpatient and 2 outpatient coordinators with a ratio of 6 to 8 families per coordinator. At most, the program could accommodate 32 enrolled NICU patients at any one time. Specific components of the program include (1) weekly screening using strict enrollment criteria including prioritization of infants with social needs; (2) a defined process for triaging, intake, and exit interviews; (3) establishing and maintaining clear roles for inpatient and outpatient care coordinators; and (4) regular meetings between inpatient and outpatient care coordinators.

The study includes patients enrolled between August 2018 and September 2021. Criteria for enrollment were stringent, and patients were screened and identified regularly through a formal process by the program’s medical directors, coordinators, and nursing leaders whenever an open position became available. Enrollment criteria included at least 1 of the following: (1) a complex diagnosis involving at least 3 organ systems and/or 3 or more subspecialty consultants; (2) anticipated need for complex equipment upon discharge (eg, tracheostomy, home ventilation, nasogastric or gastric tube feedings); (3) anticipated need for substantial discharge teaching and assistance attending multiple outpatient appointments; and (4) families with substantial social needs including language barriers and low medical literacy.

Our evaluation design was a pre/post program cohort comparison. We conducted an observational study collecting program data prospectively and historical data retrospectively. Comprehensive data were collected retrospectively (in the preprogram cohort) and prospectively (in the LCCP-enrolled cohort) in PICO format: (1) patients/population was defined as the 95 NICU patients enrolled in the LCCP program from 2018 to 2022; (2) intervention was defined as LCCP enrollment; (3) comparison group included the 96 preprogram NICU patients admitted in 2017; and (4) outcomes (primary) were length of hospital stay, time to program enrollment, and degree of medical complexity. Secondary outcomes included rates of postdischarge survival and outpatient follow-up.

This study was determined to be exempt by the Institutional Review Board of Lurie Children’s Hospital. Clinical data were collected on all enrolled patients and included baseline demographics, clinical characteristics, longitudinal details of the hospital course, and postdischarge outcomes. Specifically, outcome measures included details on hospital course and program benchmarks, which were calculated from date of birth, age at admission, age at enrollment, length of stay (LOS)/transfer, and discharge, as defined in Table 1. Details surrounding infant deaths, including primary diagnoses and organ system involvement (number and type), were tracked prospectively. Postdischarge data were reported by inpatient and outpatient care coordinators and included disposition (home, transitional, or chronic care facility), respiratory support at discharge, feeding support at discharge, readmission rate, ED visits, completed outpatient appointments, no-show rates, and loss to follow-up. Process measures included LOS during initial admission, enrollment of families with complex social needs, non-English primary language, health literary concerns, enrollment efficiency, satisfaction, and sustainability over time. Implementation measures included rates of total infants screened, rate of enrollment, time to enrollment, and rate of MyChart activation (family access to electronic medical record). Health literacy scores from admission and at time of program enrollment,²¹ race, and ethnicity were recorded.

In addition to the above prospective data collection, a retrospective review was conducted to identify a historical control cohort of NICU patients born the year prior to program initiation (January 1, 2017, to December 31, 2017). Infants with similar medical complexity were identified using the same criteria for LCCP enrollment, specifically, diagnoses with more than 3 organ systems involved and/or more than 3 subspecialists consulted. Social determinants of health (SDOH) were not used to select this cohort, given the retrospective nature. A total of 96 such infants were identified. For this preprogram historical cohort, complex care activities such as multidisciplinary family meetings, caregiver teaching/education, surgical and procedure scheduling, discharge planning, and outpatient follow-up were coordinated to varying degrees and consistency by resident physicians, advanced practice nurses, social workers, discharge coordinators, and case managers.

In the early stages of program planning and launch, we followed a nonrandomized program evaluation framework with the goal to justify and guide program expansion and continued philanthropic support (https://dspace.mit.edu/bitstream/handle/1721.1/152547/res-14-001-spring-2009/contents/index.htm?sequence=9&isAllowed=y). The overarching key question was whether program initiation would improve quality of life through better resource utilization outcomes. Although quality-of-life measures were not actively collected, these were implicit in the outcomes studied such as mortality, LOS, readmission rates, ED visits, and completion of outpatient appointments. We completed data collection in 2023, 2 years after the last study patient was enrolled. We subsequently analyzed the data to address the key question, evaluating the primary, secondary, and additional outcomes that were not originally anticipated, with the intent to identify areas of improvement and implement interval changes toward future success of the program. To better conceptualize how primary/secondary drivers and change ideas were implemented and balanced to pursue the major goals of the program, a key driver diagram is shown in Figure 1.

Clinical data were extracted from the electronic medical record and maintained on the hospital’s secure database. Data were summarized using frequencies and percentages for categorical variables, means and SDs for normally distributed continuous data, and medians and IQRs for nonnormally distributed continuous variables. LCCP data were compared with the historical control group. To evaluate changes in enrollment efficiency over time, we also stratified the LCCP cohort into three 1-year epochs starting with the official start date of the program to define year 1 (August 1, 2018, to July 31, 2019), year 2 (August 1, 2019, to July 31, 2020), and year 3 (August 1, 2020, to September 30, 2021). To evaluate variations in clinical covariates and outcomes between pre- and postprogram initiation cohorts and over time, we used the Kruskal-Wallis test, chi-square test of independence, or Fisher exact test, depending on suitability. Multiple logistic regression models were constructed to calculate odds ratios (ORs) and 95% CIs, using the historical control cohort as the referent group. Models were adjusted for gestational age (weeks), birth weight (grams), infant sex, maternal race, and number of organ systems involved. Continuous data were log-transformed to approximate normal distribution. Bonferroni-adjusted P threshold of .05/3 = .01667 was used to take into account multiple comparisons among the 3 epochs. All analyses were performed using RStudio (version 1.2.1335).

Between August 1, 2018, and September 30, 2021, N = 95 infants were enrolled during initial NICU admission. Baseline differences between the historical cohort (N = 96) and LCCP patients are summarized in Table 2. Mean gestational age was lower (P = .03) and birth weights trended lower (P = .05). Fifty-six (59%) were male, with a higher percentage of infants born to Black mothers (40% vs 17%, LCCP vs Control; P = .01). Demographics of LCCP infants were not significantly different over the course of the 3-year enrollment period.

Table 1 summarizes the primary diagnosis (defined as the main admitting diagnosis at hospital discharge or death), which differed between cohorts (P = .006). The majority of LCCP patients had a primary diagnosis of prematurity with related complications including bronchopulmonary dysplasia, pulmonary hypertension, necrotizing enterocolitis, and intraventricular hemorrhage (Figure 2). Organ systems most involved included pulmonary (75%), cardiovascular (79%), gastrointestinal (80%), and neurologic systems (78%) (Table 3; Figure 2). Comparison with the historical cohort revealed a higher median number of organ systems involved with LCCP patients (P = .002) and a higher number with more than 5 organ systems involved (P = .012), which was reflected across most all organ systems and increased over time (Table 3). After adjustment for gestational age, birth weight, infant sex, and maternal race, LCCP-enrolled infants were more likely to have multiorgan involvement of more than 5 systems (OR, 2.0; 95% CI, 1.1–3.9). This complexity increased over time to an OR of 4.9 (95% CI, 2.0–12.2) (P < .001) by year 3 (epoch 3, 2020–2021) (Supplemental Table 1). As for hospital course (Table 1), more LCCP infants were transferred to transitional or chronic care facilities. More than 30% of LCCP patients were discharged with a tracheostomy and were ventilator dependent, 65% required gastrostomy tubes, and 12% required nasogastric feeding tubes.

Reducing barriers to health care among medically complex patients was a major goal of the LCCP. Table 2 shows the distribution of primary language and health literacy scores. Almost 20% enrolled were from families whose primary language was not English. Almost 70% had some to significant health literacy concerns based on initial assessment at NICU admission (P < .001).

As shown in Table 1, the median age at LCCP enrollment/entry decreased from 116 days in year 1 to 65 days in year 2, and to 93 days in year 3 (P < .05). Accordingly, there was a significant decrease in median days from NICU admission to LCCP entry between year 1 and year 2 (102 to 54 days) and between year 1 and year 3 (102 to 54 days).

The median age at NICU admission was 11 days for enrolled patients and 4.5 days for the control group (all were outborn, Table 1). Total LOS (including transfers to other units within the children’s hospital) was longer compared with historical controls (206 days vs 103.5 days, respectively, P < .001). The majority of LCCP infants were discharged directly from the NICU (76.8%), and 67.4% were discharged directly to home. After adjustment for gestational age, birth weight, infant sex, maternal race, and number of organ systems involved, LCCP-enrolled infants were more likely to have longer LOS, defined as above the median LOS (>107 days) from NICU admission to discharge for the entire cohort: OR, 5.6; 95% CI, 2.7–11.6; P < .001 (Supplemental Table 1). However, comparison of the three 1-year epochs with the historical control group revealed a trend toward decreased odds of prolonged LOS over time from an OR of 10.1 in year 1 to an OR of 3.2 by year 3 (Supplemental Table 1).

As shown in Table 1, 2-year survival rates were higher among LCCP (86% vs 79%; P < .001). Table 4 shows that among the 86 enrollees who survived to hospital discharge, more were readmitted within the first month (31% vs 15.6%, respectively, P = .01). In the 3 months following discharge, the median number of follow-up appointments completed was 23 per patient; there was no difference between LCCP-enrolled and the control group, nor was there a significant difference in no-show rates. However, no patients in care coordination were lost to follow-up (0% vs 13%; P = .001).

We evaluated the longitudinal impact of a unique level IV NICU care coordination program after the first 3 years in operation. Using protocols aimed at enrolling NICU patients with the highest medical complexity based on multiple organ system involvement, multiple subspecialty consultations, anticipated technology dependence, and complex follow-up care at discharge, we were able to sustain the program over a 3-year period. We were able to track the program’s success over time, demonstrating enrollment of infants with increasing medical complexity accompanied by a trend of decreasing odds of prolonged hospitalization. Remarkably, enrollees had significantly higher survival rates after discharge and no loss to follow-up. These findings support that the described LCCP is a transferable model that could benefit other sites caring for neonatal patients with high degrees of medical complexity and complex social needs.

The majority of patients enrolled were premature infants with multiorgan system complications, newborns with multiple congenital anomalies, genetic conditions, or congenital heart disease, and patients with technology dependence. Medically complex infants often require prolonged NICU hospitalizations and, after discharge, multiple readmissions and ED visits. Despite the increased medical complexity of patients enrolled as compared with the preprogram cohort, there was no increase in ED visits after hospital discharge, but there was an increase in readmission rates, likely reflecting a more medically complex cohort (Table 4). No-show rates were driven to 0%, likely due to the enhanced outpatient care coordinator involvement that was initiated before NICU discharge.

Another program goal was to identify and reduce barriers to health care. Families with low health literacy, defined as “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions,”²² face even more difficulty. One study found that even after adjustment for health and socioeconomic status, low health literacy is associated with increased mortality.²³ The majority of enrolled families reported concerns with health literacy. Low health literacy often stems from limited English proficiency.²⁴ Thus, the LCCP also prioritized enrollment of families that did not speak English as their primary language. Care coordinators facilitated the use of interpreter services and screened for SDOH including food insecurity, transportation needs, housing instability, caregiver education/work status, safety, and environment to identify patients who would most benefit from LCCP support.

The concept of care coordination in the NICU is not entirely new.²⁵ Historically, numerous reports demonstrate that such programs are likely to improve health outcomes of ICU patients.^26,27 How they are structured to screen patients, recruit and train coordinators, and provide longitudinal follow-up will vary according to the health care setting that each program serves. Certain aspects of our program are unique and may be generalizable to other centers: criteria for enrollment, coordination of procedures, discharge teaching and preparedness, and highest risk for loss to follow-up. These criteria may be tailored to each NICU and the patient population it serves. The focus on barriers to health care and identifying SDOH was a unique priority of this particular program, enhanced by the growing recognition of health barriers, particularly in Chicago.

There were several limitations of our study that precluded us from completely evaluating the full impact of the program. Given the inherent nature of a busy level IV NICU serving a large catchment area and Chicagoland’s diverse population, comparing the patient demographics, levels of medical complexity, and associated outcomes before and after program launch were not definitive measures of the program’s success. For example, our attempts to compare the cohort with a “historical control group” of patients admitted to this NICU in the year immediately prior to program launch revealed overall shorter hospitalizations during a preprogram epoch. The differences in groups can be explained by the lack of SDOH information available in the control group to match that of the LCCP cohort. In addition, we speculate that several changes in the level IV NICU that may have overlapped with the program launch and 3-year course (eg, introduction of a chronic lung referral program and growth of the fetal surgery program) may have altered referral patterns and the acuity/complexity of NICU patients and thus could account for substantial differences in the pre- and postprogram cohorts. Thus, it was necessary to also compare the outcomes over time by 3 epochs. After adjustment for key covariates, including infant sex and maternal race, we were able to identify trends of increasing medical complexity that coincided with decreasing odds of prolonged hospitalization over the 3-year period. These and other indicators of enrollment and discharge efficiency remained steady despite the program’s initiation just prior to and during the COVID-19 pandemic—which was another unanticipated factor that prompted exploratory analysis by the epochs. Regardless of the above challenges, the overall striking increase in postdischarge survival in the LCCP group is a major finding that supports our hypothesis that LCCP may have a positive impact on resource utilization overall, despite the increased medical complexity of patients enrolled.

Other limitations included the small sample size, and the lack of data to quantify benefits to individuals or society (social return on investment), to determine cost-effectiveness of the program or even cost per outcome. Because the program could only enroll and observe 16 patients at a time on a rolling basis, not all eligible NICU patients were enrolled, and thus, the patient sample is not representative of all medically complex neonates. In other words, the limiting factor to consistent implementation is the limited number of care coordinators and required funding to support the care coordinators, which heavily relied on philanthropy. Thus, this study requires further expansion to assess the generalizability of our findings.

In conclusion, the NICU LCCP described in this evaluation is a transferrable model that could benefit similar sites caring for neonatal patients with high degrees of medical complexity. Future studies include assessment of LOS and other NICU outcomes according to statewide and national benchmarks, and evaluation of the impact of more specific care coordination activities (eg, time to procedures, clinician-patient ratios, caregiver education, and family empowerment and engagement strategies) and application of novel strategies to inform the development of cost-effective coordinated care for medically complex neonates, which will ultimately improve future childhood health.

ED

emergency department

LCCP

longitudinal care coordination program

LOS

length of stay

NICU

neonatal intensive care unit

OR

odds ratio

SDOH

social determinants of health

We are grateful to the Ulaszek and Etue Families for their tremendous inspiration and vision. We thank Megan Colwell, BSN, Hannah Lyons, BSN, Jenny Keane, MSW, and Lisa Robison, BSN, CPN, for their dedication to the Little Heroes League program and their endless compassionate care for our NICU patients and families.