Evaluating the cost-effectiveness of pediatric concurrent versus standard hospice care

Abstract

Using a sample of 18,152 pediatric hospice patients, this study assessed the cost-effectiveness of concurrent care over standard hospice care. Analysis of incremental cost-effectiveness ratios with bootstrapping simulations showed that concurrent care was more effective but at a higher cost.

The 2010 Patient Protection and Affordable Care Act (ACA) fundamentally altered the delivery of pediatric end-of-life care. Section 2302 of the ACA enabled pediatric Medicaid beneficiaries under 21 years to enroll in concurrent hospice care rather than standard hospice. Under concurrent care, children with a prognosis of six months or less to live can receive medical treatments, medications, and therapies for their health condition, while simultaneously enrolling in hospice care (Lindley, 2011). Concurrent care went into effect upon the signing of the ACA in March 2010 and has been available nationally for over a decade (Laird et al., 2020). With the guidance of the pediatric care team, children and their families may choose to elect concurrent care or standard hospice care. Standard hospice care is a model of end-of-life care delivered to patients with a prognosis of 6 months or less to live and who waive all disease-directed care for their terminal illness (Lindley, 2011).

Although there is emerging evidence about the impact of concurrent hospice care on improving patient care, little is known about the cost of this care. Standard hospice care for children is frequently fragmented and discontinuous, with almost a third of the children disenrolling from standard hospice because they seek improved pain and symptom management or more aggressive treatment of their conditions in the inpatient setting (Lindley & Keim Malpass, 2017). However, a recent study showed that pediatric concurrent hospice care was more effective than standard care at improving care continuity by reducing these live discharges from hospice, but did not show the differences in costs between these two models of care (Lindley et al., 2021) One study identified the cost components relevant for economic analysis of pediatric concurrent hospice care with data from a hospice provider (Lindley, Richar, Hoit, & Steinhorn, 2021). The authors reviewed 15 hospice cost articles from which they categorized costs as personnel costs, supplies/equipment, and other costs. Using these categories and data from a regional hospice provider, the authors found that concurrent care compared to standard hospice care, had slightly lower personnel costs ($68 and $79 per day, respectively), twice higher costs of supply and equipment ($50 and $22 per day, respectively), and more than three times lower other hospice costs ($48 and $155 per day, respectively). A second study identified the costs of concurrent care compared to standard hospice care from a Medicaid perspective and estimated the average incremental Medicaid cost per patient per month (PPPM; Cozad et al., 2022). These authors found that for all lengths of stay, concurrent hospice reduced the total cost of care by $14 PPPM and inpatient costs by $11 PPPM but increased outpatient costs by $7 PPPM and costs of prescription drugs by $5 PPPM. However, for the lengths of stay of 15 days or more, the costs of concurrent care were higher across all types of costs, from as low as $29 PPPM for outpatient costs to as high as $1,904 PPPM for the total costs.

These studies provided preliminary evidence about the differences in costs between concurrent and standard hospice care for children but did not assess the cost-effectiveness of concurrent care. Cost-effectiveness, which compares the relative costs and outcomes of healthcare interventions, enables nurses, clinicians, policymakers, and key stakeholders to evaluate the performance and costs of an intervention such as concurrent hospice care against standard care. In particular, there is a lack of knowledge of the cost-effectiveness of concurrent care over the standard hospice in reducing patients’ hospice live discharges and hospitalizations, the two primary challenges of hospice care. Live discharges, meaning a discharge from hospice prior to the patient’s death, can be especially traumatic for families and caregivers, who may feel abandoned and without resources for ongoing care needs (Wladkowski, 2017). Early hospitalization may lead to the utilization of more aggressive treatment with more chaotic care, poor coordination of healthcare services, and lack of psychosocial support of family members (Dickens, 2010). As a result, early live hospice discharge with the following hospitalization comes with significantly higher costs of care (Mor et al., 2019). Economic studies of concurrent hospice care are timely and relevant given that this model of care has been mandated by the ACA for over a decade with sparse knowledge about the costs of care. Therefore, the purpose of this study was to evaluate the cost-effectiveness of pediatric concurrent hospice care versus standard hospice care.

Methods

Study Design

Economic analysis is used to systematically highlight important healthcare trade-offs between interventions, treatments, or care delivery approaches when circumstances are uncertain (Paulden, 2020). Cost-effectiveness analysis (CEA) was performed to compare different pediatric end-of-life care models resulting from the decision to enroll in concurrent or standard hospice care. A key feature of CEA is how the effects are summarized in an incremental cost-effectiveness ratio (ICER; Paulden, 2020). It is calculated as follows:

$$\mathit{\text{ICER}}=(C_{\mathit{1}}-C_{\mathit{2}})/(E_{\mathit{1}}-E_{\mathit{2}})$$

Where C₁ equals the cost of concurrent hospice care, C₂ equals the cost of standard hospice care with corresponding outcome measures (E₁ and E₂).

Decision trees provide a framework to depict the sequential nature of decision-making and the uncertainty of the clinical environment through a series of branches that lead to different outcomes (Figure 1). A decision tree for children with serious illness at the end of life was constructed with an emphasis on differential clinical outcomes that occur with a decision to enroll in concurrent or standard hospice care. First, children and their families, along with their care teams, must decide which model of hospice care fits the child’s goals. For example, for children and families who want to focus on symptom management, standard home-based hospice care would be the most appropriate. For children, who have established relationships with their treatment team and want to keep them while receiving an additional layer of hospice care, concurrent hospice care would be the most suitable. In contrast, for children, without goals of care, the family may strive for the “best interests of the child” for the given circumstances. These parents may opt for early termination of hospice services and receiving care provided in inpatient and emergency department settings. Although it may resolve the short-term health crises, but may be chaotic and traumatizing for both children and their parents. At the end of life, the goal is often to avoid early live discharges and hospitalizations because they ultimately diminish the quality of care for the child (Mooney-Doyle et al., 2019). Thus, the decision of concurrent care or standard hospice care and the resulting potential clinical outcomes, which are shown by different branches in the decision tree, form different alternative end-of-life care delivery models.

Figure 1.

Decision tree

Note: (+) indicates standard hospice care subtree collapsed

The models focus on three outcomes 1) hospice live discharges avoided, 2) hospice live discharges and hospitalizations avoided, and 3) hospitalizations avoided. The CEA compared concurrent versus standard hospice care delivery by estimating how much it costs to gain a unit of the clinical outcome transformed into incremental units of cost representing Medicaid dollars per hospice live discharge avoided, hospitalization avoided, or both discharge and hospitalization avoided. All costs were expressed in 2013 U.S. dollars using the medical care services component of the Consumer Price Index to adjust for inflation.

Data Sources, Variables, and Sample

The study used retrospective administrative Medicaid data of pediatric decedents for the years 2011 to 2013, provided by the Centers for Medicare and Medicaid Services. Medicaid data were selected because they are one of the only national sources of pediatric hospice data. It included person-level data on the type of hospice enrollment, itemized costs of care, and demographic characteristics of children serviced in all 50 states and the District of Columbia (Ruttner et al., 2015). The sample consisted of 18,147 children, with 34% receiving concurrent hospice and 66% receiving standard hospice.

The variable definitions, base case values, and ranges are listed in Table 1. The live hospice discharge rate was the number of times a child was disenrolled and re-enrolled in hospice care (Lindley et al., 2021). Specifically, the child left hospice care alive and reenrolled into hospice at a later date. The effectiveness measure was that live hospice discharges were avoided. Hospitalization was defined as admission to inpatient hospital care (Lindley et al., 2021). The measure of effectiveness was hospitalizations avoided. The combined measure of hospice live discharges and hospitalizations indicated whether a child avoided both outcomes. The cost parameters were derived from 2011 to 2013 Medicaid data (Cozad et al., 2022).

Table 1.

Model baseline estimates

Variables	Baseline Estimate	Range
Hospice live discharge probabilities
Hospice live discharge – concurrent	.076	0–1
Hospice live discharge – standard	.286	0–1
Hospitalization probabilities
Probability hospitalization given hospice live discharge	.276	0–1
Probability hospitalization given no hospice live discharge	.157	0–1
Cost parameters
Mean monthly Medicaid total cost (PPPM) – concurrent	$6371	$22–$250,380
Mean monthly Medicaid total cost (PPPM) – standard	$4545	$5–$373,855

Note. Costs are displayed in 2013 dollars. PPPM = per-patient-per-month

The time horizon of the study is the last year of life for each child, and it was chosen to accurately capture viable end-of-life delivery models for both concurrent and standard hospice care. Standard hospice care is defined as end-of-life care delivered to patients with a prognosis of 6 months or less to live and who waive all disease-directed care for their terminal illness. The definition of concurrent hospice care is medical care delivered simultaneously with hospice care for pediatric patients under 21 years with a 6-months-to-live prognosis. Based on these definitions, the target population for the study was U.S. pediatric Medicaid decedents under 21 years enrolled in hospice care. The study was guided by the Consolidated Health Economic Evaluation Reporting Standards statement (Husereau et al., 2022) and was approved by the University of Tennessee, Knoxville Institutional Review Board.

Analyses

Costs data are not normally distributed; therefore, we bootstrapped costs to produce confidence intervals (CI). Bootstrap simulation with 1,000 replications was performed to estimate the statistical uncertainty of each ICER. The results of the bootstrapped ICERs are presented in cost-effectiveness planes (Paulden, 2020). These planes show the differences in costs against clinical outcomes and visualize whether concurrent hospice care has better outcomes at higher costs (trade-off quadrant), better clinical outcomes at lower costs (new dominates quadrant), worse outcomes at higher costs (old dominates quadrant), or worse outcomes at lower costs (trade-off quadrant) than standard hospice. Data analysis and graphical illustrations were performed using Stata 15.0 (Stata Corp, LLC, 2017). All costs were reported as costs per month enrolled in Medicaid per person (per-patient-per-month, PPPM). Statistical analyses were evaluated on a 5% significance level.

Several key assumptions were made. First, we assumed that any break in hospice enrollment of 1 day or more followed by a re-enrollment was a live discharge from hospice care. Given the scarcity of pediatric hospice providers in the U.S., it was therefore assumed that patients were admitted back to their original hospice care provider. The second assumption was that cost parameters reflected total Medicaid costs in the last year of life for the child. The third assumption was that pediatric concurrent hospice care patients received care in accordance with the 2010 Patient Protection and Affordable Care Act, section 2302. Cost-effectiveness would be the same for all pediatric age groups was the final assumption.

A number of key variables (i.e., hospitalization rate; cost parameters) were subject to a one-way sensitivity analysis. In a one-way analysis, an input variable is allowed to vary from the minimum to maximum value of its range while holding all other factors constant.

Results

Table 2 displays the results of the three different clinical outcomes across concurrent and standard hospice care scenarios. The total Medicaid cost of care for concurrent care was $6,371 PPPM compared to standard care at $4,545 PPPM. The incremental Medicaid cost difference between hospice types was $1,826 PPPM. The magnitude of this cost is not substantial when considering the potential cost incurred based on the Medicaid system. The annual net cost from concurrent care translates to an additional $21,912 per patient in total Medicaid costs (i.e., $1826*12months).

Table 2.

Summary of 3 base case models

Outcome measure	Hospice Care scenario	Total cost (PPPM), $	Incremental costs, $	Effectiveness
1. Hospice live discharge avoided	Concurrent hospice	6,371.00	1,826.00	.924
	Standard hospice	4,545.00		.714
2. Hospice live discharge and	Concurrent hospice	6,371.00	1,826.00	.876
Hospitalization avoided	Standard hospice	4,545.00		.734
3. Hospitalization avoided	Concurrent hospice	6,371.00	1,826.00	.891
	Standard hospice	4,545.00		.900

Effectiveness of concurrent care was greatest for hospice live discharge avoided, along with hospice live discharge and hospitalization avoided (incremental effectiveness ranged from .714 to .924); however, this effect results in higher costs. As displayed in Figure 2 (Graph A & B), the ICER was $8,695 (95%CI, $6173, $15,411) per hospice live discharge avoided, which increased to $12,859 (95%CI, $7959, $41,095) for hospice live discharge and hospitalization avoided. Because these outcomes resulted in higher costs, the ICERs were in the trade-off quadrants. Concurrent care was not as effective as standard care at hospitalizations avoided (incremental effectiveness ranged from .891 to .900) with an ICER of -$11,033 (95%CI, $18,967, -$17,670) (Figure 2, Graph C). Because concurrent care was more costly and less effective at hospitalizations avoided than standard care, this ICER was in the old dominates quadrant.

Figure 2.

Study Incremental Cost-Effectiveness Ratio planes

Note: Reps = Incremental Cost-Effectiveness Ratio point; PE_line = Incremental Cost-Effectiveness Ratio Line; ll = lower limit; ul= upper limit

In the sensitivity analysis, the results of the base case were sensitive to changes in hospitalization and hospice care costs. For instance, when hospitalizations decreased by 10%, standard hospice care was no longer the dominant strategy for hospitalizations avoided. When concurrent hospice care costs decreased by 31%, concurrent care became the dominant strategy for live discharges avoided, and live discharges and hospitalization avoided.

Discussion

As one of the first studies to examine the cost-effectiveness of pediatric concurrent hospice care, this model of care delivery improved clinical outcomes yet was more expensive than standard hospice care. Concurrent hospice care was more effective at avoiding live discharges from hospice and avoiding live discharges from hospice combined with hospitalization. However, these improvements in clinical outcomes resulted in higher overall Medicaid costs. This study’s findings are consistent with adult hospice research, which has reported that standard hospice care is a cost-effective approach to care delivery among patients with cancer (Huo et al., 2014). The expectation from prior policy analyses of concurrent care was that its benefit would improve continuity of care as children are able to obtain curative medical care along with hospice care which provides continued communication and bonds with the medical team while receiving hospice care (Lindley, 2011; Mack et al., 2005). Using medical care adds expense to the cost structure. In standard hospice care, most patients waive medical care services for their terminal illness. Therefore, by definition, concurrent care children would have higher expenditures.

The findings from this study suggest that policymakers need to assess their goals when considering the tradeoffs associated with concurrent hospice care. For state Medicaid programs focused on improved patient outcomes, concurrent hospice care provides an additional layer of quality through improved continuity of care. Conversely, for those programs focused on cost reduction, concurrent care might be a future target when cutting costs is necessary or may experience cost shifting in its implementation to avoid additional expenses. Because pediatric concurrent hospice care is a mandated federal/state service, Medicaid beneficiaries who are eligible are entitled to the benefit regardless of the cost to Medicaid. The sensitivity analysis suggested that reducing hospitalization by 10% or overall costs by approximately a third would shift concurrent care to a cost-effective approach to end-of-life care. Future research might examine how care coordination strategies targeted to this population might reduce costs while ensuring high-quality care.

Limitations to this study include limited generalizability because the analysis was conducted under a single healthcare payer perspective. The findings reflect pediatric Medicaid beneficiaries and not children covered under other concurrent hospice care payers such as Tricare or private insurance. Given that Medicaid is the predominant healthcare payer for children at the end of life, the results provide important benchmarks. Second, this study assessed only the overall cost-effectiveness of concurrent care over standard care. Future studies may look at specific health conditions. Third, this study tested limited outcomes. Although a decision tree is an appropriate way to compare cost-effectiveness and enable the comparison of live hospice discharge and hospitalization, there are other clinical, patient/family, and health services outcomes that could be modeled in the future. Finally, the data from this study were from 2011–2013. Events such as the COVID-19 pandemic and other shifts in the health care system (e.g., Medicaid expansion) have occurred since these data that might impact results. However, these were the most current data available from the Centers for Medicare & Medicaid Service at the onset of the project, and it represents findings from the early stages of concurrent hospice care implementation (ACA, Section 2302). Given that the data captured the early period of concurrent hospice care, maturation of pediatric concurrent care might now have a different set of outcomes in terms of cost and cost-effectiveness with newer data, which presents an opportunity for future research.

Conclusion

The findings from the study demonstrated that pediatric concurrent hospice care improved clinical outcomes with fewer hospice live discharges but resulted in higher costs when compared to standard hospice care. Reductions in hospitalizations and reduced overall cost would shift concurrent care to the dominant, cost-effective approach to delivering end-of-life care for children. Future research might examine the role of care coordination in improving cost-effectiveness.