Effective Care Management for Children With Special Health Care Needs in the Era of Value-Based Payment

Abstract

Objectives.

To evaluate the impact of a pediatric-specific care coordination program for Medicaid children with special health care needs under a fully capitated payment model and assess whether sufficient savings can be achieved to offset the cost of the care coordination program.

Methods.

442 children with special health care needs, receiving health care under a Medicaid capitation payment program, were enrolled in a care coordination program. ED and inpatient utilization were measured for 1-year pre and post intervention. Use rates and costs for ED and inpatient services were evaluated using a Poisson random effect regression model.

Results.

There was a statistically significant reduction in ED utilization (31% reduction, P < .0001), inpatient admissions (38% reduction, P = .0002), and inpatient length of stay (34% reduction, P = .0112) comparing the pre and post intervention periods. Medical cost savings attributed to the reduction in ED and inpatient utilization was approximately 3 times the program costs.

Conclusions.

Enrolling children with special health care needs in a care management program was associated with a significant reduction in ED utilization, inpatient admissions, and hospital length of stay when compared with baseline expenditures. Under a fully capitated Medicaid model, the cost savings greatly exceeded the costs of the interventions. These results serve to highlight the efficacy of pediatric-specific care management programs for children with special health care needs, both clinically and economically. Such models can inform other interventions and contracting strategies to assure children receive the care they deserve in a sustainable cost model.

Introduction

In the evolving environment of accountable care and value-based payments, the importance of effective care coordination for children with special health care needs has escalated. As medical groups, hospital systems, and individual providers assume increasing levels of risk for the clinical and financial outcomes of their respective population of patients, the ability to effectively and efficiently coordinate evidence-based, high-quality care has become essential.

Under the traditional fee-for-service model, there are few incentives for care coordination programs that focus on reducing emergency department (ED) or inpatient utilization. In the fee-for-service environment, reductions in inpatient and ED utilization resulting from hospital or provider group investment in care coordination programs does not benefit the hospital or provider group who is investing in the program, but instead, it benefits the payor or managed care organization.

Alternative payment models (APMs) represent an emerging trend in which providers take different levels of “risk.” APMs have been promoted under Federal policy including the Medicare Quality Payment Program, which is focused on transitioning physicians from traditional fee-for-service payment modalities toward population-based payment or APMs such as capitation. Hospital-based payment models, like Diagnostic Related Groups and Bundled Payments, mandate the need to proactively coordinate care and care transitions. These payment arrangements may range from incremental payments for quality outcomes, to bundled payments for an episode of care or full risk capitation where a provider group cares for a population of patients for a fixed monthly fee. In the capitation model, the provider network retains any savings incurred but also carries the risk if the expenses exceed the monthly capitation. Under capitation, the investments in care coordination, if they result in higher quality and lower costs, provide an economic incentive to the provider group. However, little is known about the impact of these programs for pediatric patients in ED and inpatient utilization, and whether savings under a capitated model are sufficient to justify the incremental investment in care coordination.

Our hypothesis is that care management, for the appropriate population, can lead to high-quality, lower cost care in the pediatric population and that the cost of the care coordination program can be offset by the reduction in unnecessary hospital and ED utilization. The model represented here is an advanced phase of value-based payment (Medicare Category 4b—Population-based Payment), where the financial risk is borne by the medical group, hospital, and physicians. Intensive and standardized care management processes are necessary for a program to be clinically and financially successful in this payment paradigm. To support others considering developing such a model, we specifically outline staffing ratios and duties as well as the calculation of the costs and benefits of the program.

The literature is replete with examples of adult care management programs that demonstrate improved clinical and financial outcomes. Bodenheimer et al specifically focused on care coordination as one of the “10 Building Blocks of High-Performing Primary Care.”^1,2 The pediatric literature is less complete and focuses primarily on care coordination at the primary care practice level.

Pordes et al³ examined pediatric-specific care coordination models and identified 3 approaches that lead to positive clinical outcomes: (1) primary care-centric, (2) co-management-centered, and (3) episode-based. Early reviews of pediatric-specific Accountable Care Organizations (ACOs) specifically emphasize the ability to “coordinate and oversee the clinical provision of care across the continuum of health care services” as a foundational success factor.⁴ Karpook and colleagues⁵ identify complex care coordination as a “key component to successfully caring for children with complex medical needs. In a Commonwealth Fund White Paper from 2009, Antonelli et al⁶ describe care coordination as a “critical component of the pediatric health system.”

The American Academy of Pediatrics (AAP) has endorsed the need for effective pediatric-specific care coordination in multiple publications and policy statements. Specifically, the AAP states: “The pediatric community’s vast experience with care coordination (is) a critical success factor for pediatric ACOs.” Additionally, care coordination has been described as “an essential element of a transformed health care delivery system that encompasses optimal quality and cost outcomes.”^7–9

Multiple authors have pointed to optimal pediatric care coordination as a means to achieving the Institute for Healthcare Improvements’ “Triple Aim” of improved health, improved patient experience, and lower costs. Cady et al specifically recognized pediatric care coordination as “an effective IHI Triple Aim strategy.”¹⁰ Cooley et al¹¹ identified care coordination as a key component of an overall strategy to reduce unnecessary hospital admissions. Coller and colleagues¹² recognized that “patients receiving home visits, care coordination, chronic care management, and cross-continuum care had fewer preventable hospitalizations.”

Additional work has been done attempting to assess the cost-to-benefit ratio of care coordination and the potential savings as a result of this investment. Wong et al¹³ cite the challenge with this calculation: “Children’s health care benefits may take years to decades to realize a financial or health effect … thus the goal of identifying short- and medium-term effects that are associated with improved health over the long term.” Mosquera and colleagues¹⁴ confirmed a “comprehensive care” approach, with care coordination, reduced serious illness, and medical costs in high-risk children.” The AAP Policy, “Principles of Child Healthcare Financing” also speaks to “provision of specific medical home functions such as case management, care coordination, etc.”¹⁵ Antonelli and associates were able to quantify the amount of time and the associated cost of care coordination in the primary care office setting.^16,17

Demonstrating the clinical and financial success of care management programs is a foundational benchmark for pediatric value-based payment models. Few programs have demonstrated tangible quality and fiscal improvement in the pediatric population. We present one example of such a program.

Methods

Patient Cohort

The pediatric population represented in this study resides in Orange County, CA, where the overall rate of uninsured children is 3.4%, slightly lower than the national average of 4.8%. A disproportionate number of Hispanic children (4.3%) are uninsured compared with Caucasian (2.6%), Asian (3.0%), and African American (1.5%) youth. Additionally, an estimated 10.8% of Orange County’s children do not have a usual source of care to access when they are sick or need health advice. Nineteen percent of Orange County youth participate in food subsidies.¹⁸ There are nearly 730 000 children, 0 to 18 years of age, in Orange County, CA, with over 300 000, or 41%, covered by the single-payer, Medicaid County Operated Healthcare System (COHS).

Payment Model

Children’s Hospital of Orange County (CHOC), in partnership with the CHOC Physician’s Network (CPN) Independent Practice Association, represents over 800 primary care and specialty care physicians. CPN participates in a global capitation (professional and facility fees) population-based payment program where the COHS pays a fixed per member-per month (pmpm) fee to the physician-hospital consortium to provide inpatient and ambulatory, primary, and specialty care for 150 000 Medicaid children enrolled in the CHOC network. Within this population of 150 000 children in the CHOC network exists a higher risk cohort defined as “Seniors and Persons with Disabilities.” We do not care for adult patients and they are not included in this study. These children are identified by certain clinical conditions, including, but not limited to, autistic spectrum disorders, congenital syndromes, neurodevelopmental delay, cerebral palsy, mental health conditions, hematology-oncology conditions, prematurity, and complex cardiac disease (Table 1). These 4514 eligible children categorized as high risk, were stratified by the COHS to determine the order of completion of the health risk assessment (HRA). The 442 subjects in our study were the first to complete the HRA.

Table 1.

Demographics of Study Population.

Variable	Category	Mean/Frequency	SD/Percent
Age (years)		4.02	3.2
Gender	Female	158	35.8
	Male	284	64.3
Primary language	English	195	44.1
	Spanish	210	47.5
	Vietnamese	19	4.3
	Other	18	4.1
Ethnicity	Caucasian	218	49.3
	Hispanic	169	38.2
	Vietnamese	13	2.9
	African American	7	1.6
	Other	35	7.9
Primary diagnosis	Autism	163	36.9
	Congenital syndrome/chromosome abnormality	59	13.4
	Neurodevelopmental delay	46	10.4
	Cerebral palsy	44	10.0
	Mental health	25	5.7
	Oncology/hematology	22	5.0
	Prematurity	22	5.0
	Complex cardiac disease	16	3.6
	Other	45	10.0

The COHS currently pays a pmpm care coordination fee to the physician-hospital consortium to offset some of the care coordination resource requirements.

Care Coordination Program Description

Patients are identified as eligible participants in the program based on select high-risk diagnoses (Table 1). The patient/guardian is initially contacted by a representative from the COHS, telephonically, to complete a validated, pediatric-specific HRA. Three telephonic attempts are made to complete the assessment and then, if unsuccessful, a HRA is mailed to the home of the patient. Once completed, this self-reported HRA is sent to our Health Network and we assign a Patient Care Coordinator (PCC) to the family. Each PCC is responsible for 600 families. The PCC obtains the patient’s medical records from their providers and confirms the most recent well care visit and immunizations history. The PCC also contacts the patient/family telephonically to complete an Introductory Screening Tool (IST), which expands on answers from the HRA to identify any potential issues regarding access to care and other social determinants of health. Based on this secondary screen, the patient is placed into 1 of 3 categories for further care: Basic Case Management, Care Coordination, or Complex Case Management. Patients stratified into Basic Care Management receive ongoing concierge services facilitating access to care. For example, if the patient is due for well care, the PCC will assist in scheduling that appointment. Those stratified into Care Coordination or Complex Case Management receive the same concierge services enhanced by an Individual Care Plan (ICP) and Interdisciplinary Care Team (ICT). For all categories, the PCC remains the contact for the family and serves as the entry point for parent/caregiver inquiries. The PCCs are not licensed clinicians and do not independently provide medical advice.

A registered nurse (RN) then reviews the medical records, HRA, and IST to develop a preliminary ICP for each patient. The ICP includes the patient’s medical and behavioral health conditions, a general case summary, community benefits/programs the patient has accessed, perceived concerns or barriers to care, and goals and interventions, including preventative services required. The PCC then schedules an ICT meeting that takes place in our care coordination offices. Invitees include the patient/family, RN care coordinator, social worker, primary care provider (PCP) or representative, specialty care providers, behavioral health providers and home health, other community program staff, school representatives, and certified interpreters. We encourage the patient’s primary care physician to attend the ICT. If the patient’s PCP is unable to appear, our Pediatric Medical Director attends on their behalf. Telephonic attendance is offered for all potential attendees. Prior to the meeting, the PCC sends the preliminary ICP to all invited attendees.

The ICT is an opportunity for all invitees to discuss and address any concerns or barriers to receiving coordinated, comprehensive care. The ICP is updated after the meeting to include goals, interventions, and target dates, and identification of the responsible individual for completing these goals. A Provider Summary version of the ICP is sent to the patient’s PCP and Specialty Physicians, as appropriate, and a patient-friendly version is sent to the patient/family. The PCP is asked to review the ICP, sign the document, and return it to the PCC. The ICP is then placed in the electronic or paper medical record of the patient for reference. After the meeting, the PCC continues to serve as a resource for the patient/family/providers to assist in scheduling appointments, facilitating access to needed care or equipment, and to connect patients to community programs, behavioral health, or other resources. The RN Care Coordinator continues to provide clinical advice, education, and connections to required specialty services, as needed. If the child is admitted to the hospital or undergoes a major event that could potentially alter the plan of care, a review of the ICP is undertaken with edits and appropriate revisions disseminated to the extended care team and the family. The HRA and ICP are updated annually. Staffing for the program is described in Table 2, and may vary based on specific acuity and geographic availability of specialty and community social services.

Table 2.

Program Staffing (Core Team Members).

Position	Education and Certification	Full-Time Equivalent	Role
Patient Care Coordinator (PCC)	High school diploma; AA Preferred	1.0 per 600 patients	Patient/family liaison. Facilitates care and communicates with family on an ongoing basis. Conducts the Introduction Screening Tool (secondary screen), leads ICT meetings.
RN Care Manager	Pediatric RN	1.0 per 9 PCCs	Supervises and provides education to PCCs, offers medical advice to families (when appropriate), prepares ICPs, interacts with specialty case managers and physicians.
Medical Director	Board-certified pediatrician with case management and quality improvement experience	0.25 per 9 PCCs	Final approval of all ICPs, attends ICT meetings, and coordinates care with primary and specialty care physicians.
Licensed Clinical Social Worker	Certified licensed clinical social worker	0.5 per 9 PCCs	Attends ICT meetings, addresses social determinants of health, family safety, and connects families to community resources.

Abbreviations: ICT, interdisciplinary care team; RN, registered nurse.

Results

Statistical Methods

The study population consisted of Medicaid children who were stratified to Care Coordination or Complex Case Management levels and had an ICT meeting in 2016 or 2017. The summary of demographics was provided as means (standard deviation; SD) for continuous variables or frequency (proportions) for categorical variables. A Poisson random-effect regression model, using claims data, was used to compare the rate of ED visits and inpatient admissions for the cohort of 442 children before and after ICT meeting completion with demographics (except main diagnosis) as covariates and a random intercept to account for within subject correlation. We further examined the effect modification of whether or not the parent attended the ICT meeting by a likelihood ratio test. As a post hoc analysis, a mixed regression model was used to compare the difference of average length of stay per inpatient admission before and after the ICT meeting. Time period (pre/post) and the interaction term between time period and number of inpatient admissions were treated as fixed effect and subject was fit as a random effect. Analyses were performed using SAS 9.4.

Baseline Characteristics

A total of 442 children who were both enrolled in the care management program and had an ICT meeting in 2016 or 2017 were included in the study. The mean of eligible months before and after the ICT was 10.48 months with a SD = 3.43, and 10.86 months, with SD = 3.72, respectively. The summary demographics of the patients are provided in Table 1. The mean age of the cohort was 4 years with SD = 3.17. There were 158 (35.75%) female patients and 284 (64.25%) male patients. Approximately 40% of the ICT meetings were attended by the parent(s).

Pre-Post Comparisons on ED Visits and Inpatient Admissions for the Cohort of 442 Children

Overall, the rate of ED visits and inpatient admissions were 114 and 169 per 1000 member months prior to ICT, and 78 and 110 per 1000 member months after ICT, respectively. In the Poisson random effect regression model, the ED visit rate was 31% lower in the post-ICT period than prior with Incidence rate ratio (IRR) of pre versus post = 0.69 (95% confidence interval [CI] = 0.60–0.79; P < .0001; Table 3A) after controlling for age, gender, language, ethnicity, and parent attendance and accounting for within subject correlation. The inpatient admission rate was 38% lower in post-ICT period than prior with IRR of pre versus post = 0.62 (95% CI = 0.48–0.79; P = .0002; Table 3B). Over-dispersion was not observed in the 2 models (scale parameter = 0.86 for emergency room [ER] visits and 0.55 for inpatient admissions). We further examined effect modification of parent ICT attendance by a likelihood ratio test. Parent attendance at the ICT failed to modify the effect of ICT on either ER visits or inpatient admissions (P = .8642 for ER visits and .4817 for inpatient admissions).

Table 3.

Results of Poisson Random Effect Regression Model Analysis.

(A) Inpatient Admissions
Variable	Category	IRR (95% Cl)	P
ICT	Completed	0.62 (0.48–0.79)	.0002
Age (years)		0.95 (0.88–1.02)	.1665
Gender	Female	Reference
	Male	1.07 (0.68–1.70)	.7673
Language	English	Reference
	Spanish	0.89 (0.54–1.45)	.6384
	Vietnamese	2.43 (0.79–7.48)	.1218
	Other	1.21 (0.36–4.04)	.7616
Ethnicity	Caucasian	Reference
	Hispanic	1.29 (0.79–2.10)	.3157
	Vietnamese	0.31 (0.05–1.84)	.1995
	African American	4.83 (1.22–19.21)	.0257
	Other	1.22 (0.50–3.01)	.6633
Parent attended ICT meeting	No	Reference
	Yes	0.87 (0.55–1.37)	.5404
(B) Emergency Department Visits
Variable	Category	IRR (95% Cl)	P
ICT	Completed	0.69 (0.60–0.79)	<.0001
Age (years)		0.94 (0.90–0.98)	.0032
Gender	Female	Reference
	Male	0.81 (0.63–1.03)	.0893
Language	English	0.9041	.9041
	Spanish	1.02 (0.78–1.32)	.9041
	Vietnamese	1.40 (0.71–2.75)	.3276
	Other	0.86 (0.42–1.75)	.6715
Ethnicity	Caucasian	Reference
	Hispanic	1.15 (0.89–1.50)	.2876
	Vietnamese	0.38 (0.15–0.96)	.0425
	African-American	0.89 (0.32–2.48)	.8248
	Other	0.83 (0.50–1.40)	.4925
Parent attended ICT meeting	No	Reference
	Yes	0.93 (0.73–1.19)	.5819

Abbreviations: CI, confidence interval; ICT, Interdisciplinary Care Team; IRR, Incidence Rate Ratio.

Post Hoc Analysis on Bed Days

There were 100 subjects who required inpatient admissions prior or subsequent to the ICT meeting. Among these subjects, the average length of stay per inpatient admission was found to be significantly longer before the ICT than after the ICT meeting (P = .0112). The average length of stay per inpatient admission prior to the ICT was 9.59 days, and the average length of stay after the ICT was 6.32 days, a 34% reduction in length of stay post-ICT.

The overall base cost of the program is approximately $13.76 per member per month (Table 4). However, the demonstrated savings in ED and inpatient utilization, in our population-based payment system, is approximately 3 times the program costs, $42.04 per member per month, allowing the opportunity to reinvest in additional quality improvement activities. The calculation of savings is derived from the reductions in ED and inpatient stays and the associated cost aversion related to those services in our population-based payment system.

Table 4.

Annual Program Costs.

Average monthly membership	4514
Team FTE count	10
Labor costs	$543 619
Overhead	$201 874
Total cost	$745 493
Cost pmpm (for the 4515 children)	$13.76

Abbreviations: FTE, full-time equivalent; pmpm, per member-per month.

Discussion

In implementing this intervention, we applied the strategic tenets of care coordination to a population of children with special health care needs and demonstrated, as a proxy for quality, significant reductions in ED and inpatient utilization. This program was created in the context of a population-based payment environment requiring demonstrable cost efficiency.

We have shown that a care coordination program focused on a relatively small cohort of high-risk children was associated with a significant reduction in ED and inpatient utilization across the entire population studied. Additionally, if the children were hospitalized, inpatient length of stay was significantly shorter than prior to the intervention. While the costs of the program exceeded the amount budgeted within a total capitation funds flow, the financial benefits from cost savings more than adequately covered the program costs.

The engagement of organizational leadership, who supported our approach in the paradigm of the Triple Aim, was essential to development and implementation of this program. Instrumental to the demonstrated success was the ability to accurately attribute patients to specific primary and specialty care providers to ensure care continuity. Bodenheimer et al reference “empanelment” as a key to population health care coordination and we confirmed this as a key driver.1 The other key attributes of the program include the following:

• The ability to risk stratify the patients to guide appropriate resource allocation and interventions.

• Specific assignment of nonlicensed, lay-person staff to assist the families in navigating the complex health care system (PCCs) and the creation of ongoing relationships.

• The patient and family engagement facilitated by the HRA and IST completion, as well as the ICT Meeting and ongoing dialogue with our team.

• Our overall team-based approach with clinical and nonclinical personnel collaborating on multiple challenges facing these families: medical, social determinants, access, and so on.

• The ability to review timely, patient-specific data, including inpatient and ED utilization, HEDIS quality metrics, well child care compliance, and immunization completion rates to guide our individualized interventions.

It bears further mention that the relationship component of programs like ours is often under-emphasized but actually creates the greatest opportunity for success. The families who came to our office to attend ICT Meetings were extremely grateful to have a consolidated team of professionals addressing their entire spectrum of needs. The trust developed in these sessions, whether in-person or telephonic, created an environment where the families felt comfortable and remained engaged and proactive.

Interestingly, we demonstrated that the physical or telephonic attendance of the parent or guardian at the ICT was not a predictor of positive outcomes. One could hypothesize that the act of gathering the care team serves a positive function, as does the completion of the HRA and pre-ICT interview where parental concerns are recorded. Even if the parent failed to attend the ICT, the PCC continued to serve in the concierge role for the family.

Areas for further refinement of the program certainly exist. We acknowledge that having families physically attend the ICT meetings, while empirically beneficial, places undue financial and logistic burden on them. As a result, we are piloting a telehealth approach whereby patients can access the ICT via video and audio smartphone utilizing HIPPA-compliant technology. We also recognize that while having the actual primary care physician attend the ICT is ideal, it is unrealistic to believe that busy providers will be able to leave the office to attend. This is another opportunity for telehealth enhancements. Last, although we received tremendous subjective positive feedback on the program from patients, families, and clinicians, we did not conduct a pre and post formal, objective satisfaction survey. This has now been incorporated for future study.

Study Limitations

As with many analyses of care management efficacy, there were limitations to this study. The use of reduced ED visits and inpatient admissions to represent quality of care, although beneficial, could be challenged. Because there is inconsistency in health plan continuous eligibility in this population, our comparison is at the member-eligible months level rather than a continuous period of time. Although the study is a year-over-year comparison of the same cohort, serving as its own control, there remains the possibility that seasonal or year-to-year variation in disease prevalence (ie, respiratory syncytial virus, influenza) could play a role in frequency of ED visits and inpatient admission, independent of our intervention. Utilization and expenses may also regress to the mean over time as was demonstrated by the incidence rate of ED visits and inpatient admissions decreasing about 5% for every advancing year of age based on the results of Poisson models (Table 3). The children were, obviously, getting older between the pre and post ICT period; therefore, a small portion of rate reduction of pre-post ICT on ED visits and inpatient admissions could be attributed to the increase in age. Additionally, the population included here is a subset (442) of the overall population of patients eligible for the program (4514). There is the possibility of biased selection for inclusion of those families who are more highly engaged, as demonstrated by their completion of the initial HRA.

Conclusion

As health care payment reform continues to evolve and greater numbers of patients are covered by population-based payments, the ability to coordinate care, improve quality, and lower costs will become more and more vital. Programs like ours, where quality outcomes, as measured by reduced needs for ED and inpatient care, are accompanied by a tangible positive economic outcome, can further inform pediatricians regarding the components necessary to be successful in this changing paradigm. Models, like the one described, can inform other clinical interventions and contracting strategies to assure children receive appropriate and effective care coordination. If we can demonstrate the value proposition for payers, both governmental and private sector, of investing in care management programs, the ability to bring pediatric-specific care coordination to our most vulnerable children will be greatly enhanced.