Economic evaluation of a hybrid type 1 effectiveness-implementation trial comparing a community health worker palliative care intervention to enhanced standard of care in African American patients across four cancer centres in the USA: a study protocol

Abstract

Introduction

Despite increasing palliative care capabilities in the USA, utilisation rates remain low for patients with advanced cancer, particularly among African American patients. To address this gap, a theory-driven, stakeholder-informed community health worker (CHW) palliative care intervention for African American patients with advanced cancer and their informal caregivers is currently being assessed through a hybrid type 1 effectiveness-implementation trial at four cancer centres across the USA. To improve the quality and efficiency of palliative care delivery, inform resource allocation and guide broad-scale implementation, it is essential to generate evidence on the economic value of palliative care programmes. The objectives of this study are to evaluate the cost-effectiveness and estimate the social value of a CHW palliative care intervention for African American patients with advanced cancer and their caregivers.

Methods and analysis

We will conduct cost-effectiveness analyses (CEAs) and a social return-on-investment (SROI) analysis to assess the value of the CHW palliative care intervention compared with enhanced standard of care. Standard, extended and distributional CEAs will be performed from the perspectives of an adopting organisation or payer (eg, Medicaid), the US healthcare sector and society. An SROI analysis will also be conducted to assess the social value of the intervention. These analyses will focus on estimating the costs, health and distributional impacts of the intervention.

Ethics and dissemination

This study protocol was approved by the Johns Hopkins Medicine Institutional Review Board (IRB00372476). All methods will be carried out in accordance with relevant guidelines and regulations. Written informed consent will be obtained from all subjects prior to study participation. This manuscript does not contain participant-level data. The full protocol will be available from the corresponding author on reasonable request. The dissemination of findings from the clinical trial and accompanying economic evaluation outlined in this manuscript will be multifaceted to maximise reach and impact. Research findings will be presented at relevant scientific conferences, submitted for publication in peer-reviewed journals and shared with community stakeholders, including hospital leaders and administrators, providers, CHWs and patient advocacy groups.

Trial registration number

NCT05407844.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Extended and distributional cost-effectiveness analyses and a social return-on-investment analysis will be used to evaluate the community health worker palliative care intervention.

• Direct, indirect and intangible costs will be included to capture a comprehensive view of the cost-efficiency and distributional impacts of the intervention.

• The relatively short study horizon (6 months) may not fully capture long-term costs and benefits, and some intangible costs will remain difficult to quantify.

• The generalisability of findings may be limited by differences in healthcare settings, populations and payer structures.

Introduction

Guidelines recommend that patients with advanced cancer receive dedicated palliative care services early in the disease course, alongside active treatment.¹ This recommendation is based on multiple randomised controlled trials in patients with advanced cancer, which have shown that early integration of palliative care improves quality of life (QOL), symptom management and psychosocial well-being in patients, and decreases caregiver distress.²,⁶ Despite the benefits of palliative care services, uptake and utilisation in the USA remain low overall and within oncology.⁷,⁹ Furthermore, there are significant racial disparities in access to palliative care services among patients with advanced cancer, with African American patients facing lower utilisation rates compared with White patients.¹⁰,¹²

To address this disparity, we developed a community health worker (CHW) palliative care intervention, which uses CHWs as lay patient navigators in palliative care for African American patients with advanced cancer. CHWs are non-clinician public health workers who help patients from underserved communities navigate healthcare barriers by addressing social determinants of health (SDOH), providing health education and helping patients navigate their personal health and healthcare systems.¹³,¹⁷ The intervention is currently being assessed through a hybrid type 1 effectiveness-implementation trial at four cancer centres across the USA (NCT05407844).¹⁸

In addition to establishing the effectiveness of the intervention and understanding contextual factors that influence its implementation, decisions about intervention adoption may also be guided by the intervention’s value-for-money. Value-for-money of healthcare interventions is commonly assessed using traditional economic evaluation frameworks, such as return-on-investment (ROI) analysis, cost-benefit analysis (CBA), cost-effectiveness analysis (CEA) and cost-utility analysis (CUA).¹⁹,²² However, in practical applications, these frameworks often overlook the broader economic, environmental and social impacts of an intervention, potentially neglecting various dimensions of value, leading to biased estimates and suboptimal policy recommendations.^{23 24} Alternative value assessment frameworks, such as distributional CEA (DCEA), extended CEA (ECEA) and social ROI (SROI) analysis, have been proposed to explicitly account for additional dimensions of value generated by interventions—including multiple stakeholder perspectives—and supplement approaches like CBA and CEA.²⁴,²⁸ However, their application to the evaluation of palliative care interventions is limited.

Evidence on the economic value of palliative care programmes is essential to improve the quality and efficiency of palliative care delivery and inform resource allocation decisions. Yet, there is little robust evidence on the cost and benefit trade-offs of palliative care programmes for patients and caregivers, providers, payers and society. This is largely due to significant challenges with and variability in the economic evaluation of palliative care programmes with regard to the measurement and reporting of palliative care outcomes, cost types and perspectives, and time windows of analysis.²⁹ The objectives of this study are to assess the value-for-money of the CHW palliative care intervention for African American patients with advanced cancer and their informal caregivers, using CEAs and SROI methodology. We hypothesise that the intervention will be cost-effective, enhance health equity and generate significant social value for key stakeholders, which will support broader adoption and scale-up.

Clinical trial background information

Study design

This study protocol outlines an economic evaluation for a multicentre, randomised, assessor-blind, parallel-group, pragmatic, hybrid type 1 effectiveness-implementation trial (NCT05407844). Participant recruitment started on 2 November 2023 and is ongoing. The study is expected to be completed by 1 September 2027. The Consolidated Framework for Implementation Research (CFIR) and Reach, Effectiveness, Adoption, Implementation and Maintenance (RE-AIM) are the theoretical frameworks underpinning trial design.^{30 31} The remainder of this section provides a brief overview of trial design; the protocol for the clinical trial has been published separately.¹⁸

Population and setting

Trial participants will be adult patients (≥18 years old) who self-identify as African American and have been diagnosed with advanced solid organ malignancy (American Joint Committee on Cancer stages III and IV) and their self-designated informal caregivers (patient-caregiver dyads). All participants will be English-speaking, have intact cognition and be able to provide informed consent. Patients and their caregivers will be recruited from four enrolment sites: Johns Hopkins University School of Medicine (Baltimore, MD), University of Alabama at Birmingham (Birmingham, AL), TidalHealth Peninsula Regional (Salisbury, MD) and Wake Forest University Health Sciences (Winston-Salem, NC). These hospitals represent diverse sociodemographic and cultural characteristics of the African American community and have established CHW programmes.

Comparators

Following trial enrolment, participants will be randomised to either the intervention arm or the control arm for a duration of 6 months. Participants in the intervention arm will receive the CHW palliative care intervention, with CHWs serving as lay patient navigators in palliative care. Baseline needs will be assessed using the 15-item Protocol for Responding to & Assessing Patients’ Assets, Risks & Experiences (PRAPARE) tool,³² which quantifies SDOH and barriers to care by summing the number risks present across four core domains: personal characteristics, family and home, money and resources, and social and emotional health. Each domain contributes one point if a risk is present (eg, lacking housing) and zero if absent (eg, having housing). Thus, total scores range from 0 (no reported risks) to 22 (all risks present). Following the baseline needs assessment, the CHW will provide ongoing support tailored to the individual needs of the patient and caregiver. This includes facilitating care team coordination, enhancing communication, helping patients access community resources, providing psychosocial support, educating patients and caregivers on the benefits of palliative care, and facilitating advance care planning (ACP) discussions. Participants randomised to the control arm will receive enhanced standard of care (eSOC). This will include usual oncologic care in addition to a palliative care brochure, outlining the definition and potential benefits of palliative care and its service offerings. Patients randomised to the eSOC arm may be referred to specialty palliative care at any time throughout the study period at the oncologist’s discretion.

Outcomes

Trial outcomes will be ascertained by trained research coordinators through researcher-administered surveys and electronic health record (EHR) review at baseline, 2 months and 6 months. The primary outcome is ACP, which includes documentation of a completed advance directive or a self-reported or documented discussion of care preferences between the patient and caregiver or healthcare team. The secondary outcome measures include QOL, assessed using the Functional Assessment of Chronic Illness Therapy-Palliative Care (FACIT-Pal) scale³³ and the 5-Level EuroQol 5-Dimensional (EQ-5D-5L) questionnaire³⁴,³⁷; quality of communication, assessed using the Quality of Communication (QOC) questionnaire³⁸; utilisation of hospice care within 14 days of death; and patient symptoms, assessed using the Edmonton Symptom Assessment System (ESAS)³⁹ and Centre for Epidemiologic Studies Depression Scale (CES-D).⁴⁰ An overview of all study outcomes can be found in table 1. These outcomes will be used in the economic analyses outlined below.

Table 1. Overview of clinical trial outcome measures.

Outcome type	Outcome measure	Description
Primary	Advance care planning	Defined as a self-reported or documented advance directive, such as a living will or durable power of attorney, or a documented discussion of care preferences between the patient and caregiver or healthcare team
Secondary	Quality of life	Functional Assessment of Chronic Illness Therapy-Palliative Care (FACIT-Pal) scale33 and the 5-Level EuroQol 5-Dimensional (EQ-5D-5L) questionnaire34,3748
Secondary	Quality of communication	Quality of Communication (QOC) questionnaire38
Secondary	Hospice care utilisation	Utilisation of hospice care within 14 days of death (Yes/No)
Secondary	Patient symptoms	Edmonton Symptom Assessment System (ESAS)39 and the Centre for Epidemiologic Studies Depression Scale (CES-D)40
Exploratory	Patient-physician communication	Princess Margaret Hospital Satisfaction with Doctor Questionnaire (PMH/PSQ-MD)91 92
Exploratory	Caregiver satisfaction	Family Satisfaction with Advanced Cancer Care (FAMCARE) scale93
Exploratory	Palliative and hospice care utilisation	Palliative care consulted (Yes/No), hospice care referral made (Yes/No), length of stay in hospice care (days), hospice care withdrawal (Yes/No)
Exploratory	Resource utilisation	Length of stay in hospital (days), length of stay in the intensive care unit (days), visits to the emergency department (number of visits), readmission(s) (Yes/No), timing of readmissions (days since discharge), time spent with CHW (hours), cost (dollars)
Exploratory	Social determinants of health	Protocol for Responding to & Assessing Patients’ Assets, Risks & Experiences (PRAPARE) tool32
Exploratory	Work productivity and activity impairment	Work Productivity and Activity Impairment (WPAI) questionnaire55,57

CES-D, Centre for Epidemiologic Studies Depression Scale; CHW, community health worker; EQ-5D-5L, 5-Level EuroQol 5-Dimensional; ESAS, Edmonton Symptom Assessment System; FACIT-Pal, Functional Assessment of Chronic Illness Therapy-Palliative Care; FAMCARE, Family Satisfaction with Advanced Cancer Care; PMH/PSQ-MD, Princess Margaret Hospital Satisfaction with Doctor Questionnaire; PRAPARE, Protocol for Responding to & Assessing Patients’ Assets, Risks & Experiences; QOC, Quality of Communication ; WPAI, Work Productivity and Activity Impairment .

Sample size

The sample size was calculated to detect clinically meaningful differences in both ACP (primary outcome) and QOL (secondary outcome). To detect a 30% difference in ACP at 6 months (α=0.05, power=90%), we estimated 70 participants per arm after adjusting for 20% attrition (140 participants total). To detect a 0.5 standardised mean difference in QOL (α=0.05, power=80%), we estimated 80 participants per arm after adjusting for 20% attrition (160 participants total). To ensure power for both outcomes, we plan to enrol 160 participants. Whenever possible, we will aim to enrol a caregiver alongside each patient as part of a patient–caregiver dyad; however, patient participation will not be contingent on caregiver involvement. We will therefore recruit up to 160 patient participants and, where applicable, their caregivers, for a total sample of up to 320 participants.

Methods and analysis

The analyses will be conducted and findings reported in accordance with existing cost and economic analysis guidelines,^{19 20} and the 2022 Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines (online supplemental file 1).^{20 41}

Health economic evaluations

Overview of economic evaluations

We will conduct cost-effectiveness and SROI analyses to assess the value of the intervention compared with the eSOC. The CEAs will be undertaken from the perspectives of a payer or adopting organisation (eg., Medicaid), as well as from the US healthcare sector and societal perspectives.

Base case analyses will be conducted from the perspective of Medicaid, the primary public payer for the low-income population represented in the study. Three types of CEAs will be conducted: standard CEA, ECEA and DCEA. Standard CEA will assess whether the intervention achieves better health outcomes at an acceptable incremental cost compared with usual care. ECEA will extend standard CEA by incorporating non-health benefits and costs of the intervention, such as avoided illness-related impoverishment or financial risk protection (FRP), as well as the distribution of these outcomes by SDOH risk category (ie, low, moderate and high risk), and potentially by sex (men vs women) and across different income quintiles.²⁶ DCEA will help evaluate the health equity impacts of the intervention across subpopulations (SDOH risk category), as well as potential trade-offs between the objectives of allocative efficiency—measured by the incremental cost-effectiveness ratio (ICER) in CEA and CUA—and equity, which will be explicitly incorporated into economic evaluations through ECEA and DCEA.^{25 28} By design, DCEA will be conducted from the US societal perspective, as there are no incentives for private payers to deviate from the utilitarian distribution of health benefits without a normative stance on the fairness of alternative outcome distributions.^{25 28} In addition to CEAs, an SROI analysis will be conducted. SROI is a social impact accounting and valuation framework, also described as a localised, pragmatic form of CBA.²⁴ It incorporates the social and environmental impacts of the intervention for all key stakeholders, estimating its broader value beyond standard conceptions of value.⁴² Together, these evaluations will provide a comprehensive assessment of the intervention’s economic value, efficiency, equity implications and broader societal benefits.⁴³

In all analyses, we will measure the costs and consequences of the intervention using an intention-to-treat approach.^{44 45} All costs will be reported in 2024 (or the latest year permissible) US dollars. The costs and consequences of the intervention will be measured over a time horizon of 6 months, the duration of the trial. In our base case analyses, both costs and consequences will be discounted at a 3% annual discount rate,⁴⁶ with sensitivity analyses over the range of 1%–7% based on recent White House revisions of the 2% rate,⁴⁷ to account for the time value of money.

Identification and measurement of effectiveness outcomes

Health impacts will be measured in natural units and will be derived from the primary and secondary effectiveness outcomes (outlined above). Health impacts will then be used to estimate health utilities, which will be used to calculate cost-effectiveness and the SROI ratio.

Health utilities

We will use the EQ-5D-5L and FACIT-Pal to measure health-related QOL (HRQoL) scores, which will then be mapped into utility scores. Briefly, the EQ-5D-5L is the second version of the EuroQol 5-Dimensional (EQ-5D) instrument, one of the most widely used generic multi-attribute utility instruments, with applications in research and clinical practice.³⁴,³⁷⁴⁸ The EQ-5D-5L augments the first 3-level version of the EQ-5D instrument (EQ-5D-3L) to include two additional levels in each of the five dimensions of the EQ-5D (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) and a Visual Analogue Scale (VAS). Thus, each dimension has five levels, ranging from 1 (no problems) to 5 (extreme problems), describing 3125 distinct health states.³⁴,³⁷⁴⁸ The VAS score ranges from 0 (worst health imaginable) to 100 (best health imaginable).³⁶ Respondents’ EQ-5D-5L profiles can be converted into US-specific utility scores using validated country-specific value sets, such as the one developed by Pickard et al,⁴⁸ to support economic evaluations and inform decision making within the US context.

The Functional Assessment of Chronic Illness Therapy (FACIT) is a collection of HRQoL questionnaires designed to assess multidimensional health status in individuals with various chronic illnesses and includes multiple disease and symptom-specific subscales.⁴⁹ The FACIT-Pal is a 19-item palliative care subscale pertaining to persons with life-limiting illness.³³ HRQoL scores derived from the FACIT-Pal will be mapped onto EQ-5D-5L utility scores using validated and theoretically grounded algorithms.⁵⁰

The primary analysis will use EQ-5D-5L-derived Quality-Adjusted Life Years (QALYs). The FACIT-Pal mapping will be used to assess the robustness of the results, particularly in cases where EQ-5D completion is limited.

Identification, measurement and valuation of resource uses

Costing will focus on estimating the implementation costs, intervention costs and downstream costs (figure 1). All research-related costs will be excluded from the cost analyses.

Figure 1. Schematic of the economic consequences of implementing the CHW palliative care intervention. CHW, community health worker; CEA, cost-effectiveness analysis; DCEA, distributional cost-effectiveness analysis; ECEA, extended cost-effectiveness analysis; EHR, electronic health record; EDE-INHB, equally distributed equivalent level of incremental net health benefit; FRP, financial risk protection; ICER, incremental cost-effectiveness ratio; INHB, incremental net health benefit; INMB, incremental net monetary benefit; NPV, net present value; OOP, out-of-pocket; SROI, social return-on-investment; WTP, willingness-to-pay.

Implementation costs

These are the costs associated with the development and execution of the implementation strategy, including the integration of the CHW palliative care intervention into clinical care. They will cover CHW recruitment and training, administrative costs and general supplies. These costs will be calculated using programme data from the trial and will generally be considered direct non-medical costs (defined below).

Intervention and downstream costs

We will measure two types of intervention-related costs: intervention costs and downstream costs. Intervention costs are those that are borne as a direct consequence of implementing the CHW palliative care intervention. These costs arise from changes in intervention-related behaviours by stakeholders such as patients and caregivers, providers, payers and the healthcare system. At the same time, the intervention may also result in other downstream costs and consequences. Downstream costs are similar to intervention costs but differ in that they occur later in the sequence following implementation (ie, they occur after the intervention). These costs may change (intentionally or unintentionally) due to the implementation strategy and intervention. Therefore, it is critical to carefully account for both types of cost and avoid double-counting.

Estimation of these costs will focus on both healthcare and non-healthcare resource use among trial participants, by tracking all major and relevant resource use items pertaining to the intervention, as well as their associated unit costs.^{19 20} Specifically, costing will focus on estimating the incremental costs of the intervention in the following cost categories: (1) direct medical and non-medical costs (paid for by patients out-of-pocket and by third-party payers, depending on the perspective adopted); (2) indirect costs and (3) intangible costs.

Direct medical costs

These are defined as the opportunity costs of formal healthcare goods and services, such as CHW home visits, palliative care specialist consultations, outpatient visits, hospitalisations, chemotherapy and radiation therapy, nursing home care, hospice care, home-based palliative care equipment, and medications such as those for pain management. Direct medical costs represent the expenses directly linked to the provision of medical care, services and procedures within the formal healthcare sector.

Direct non-medical costs

These are patient and family expenses directly related to treatment. In this study, direct non-medical costs will include, for example, programme implementation costs (as described above), CHW travel time for home visits, patient time, patient and caregiver transportation, care coordination and social support services, and housing modifications for home-based care.

Indirect costs

These costs arise from resource use related to the loss or reduction of work productivity due to cancer-related morbidity and mortality (eg, patient and caregiver productivity costs), caregiving, as well as resource use associated with receiving social services as a result of the intervention.⁵¹,⁵⁴ Consistent with economic tradition, we will consider positive net productivity as a net reduction of resource utilisation, and thus value it at pre-tax wage rate plus fringe benefits. Therefore, if the intervention increases productivity or prevents productivity loss, it will result in fewer resources being consumed for treatment or care. Responses to the Work Productivity and Activity Impairment (WPAI) questionnaire for caregivers will be used to produce estimates of presenteeism and absenteeism.⁵⁵,⁵⁷

Intangible costs

These quantify the non-monetary consequences of the intervention, and include costs like pain and suffering, disability, poor or bad communication, lack of care coordination, patient and provider satisfaction, and provider morale. At the patient level, these non-monetary burdens are already incorporated in the EQ-5D-based utility weights and the resulting Quality-Adjusted Life Year (QALY) estimates. Hence, to avoid double-counting, these will be included exclusively in the denominator of the ICER in the CEAs. Additionally, we will analyse qualitative data from participant feedback and qualitative interviews with patients, caregivers and providers to contextualise the quantitative findings and enhance the interpretation of intangible cost impacts.

Total costs

We will calculate the total costs in each cost category by multiplying the units of resource use with their unit cost for each intervention arm. We will estimate formal healthcare sector costs during the period of the trial by analysing patient-level and provider-level clinical, administrative and financial data from EHR data and billing records, as well as published fee schedules and cost-to-charge ratios.⁵⁸ We will develop econometric models to estimate these costs and will include costs related to physician visits, laboratory tests, medications and hospitalisations. Depending on the perspective considered, not all costs will be included in the calculations. While all costs will be captured from the societal perspective, only formal healthcare costs will be included from the perspectives of the US healthcare sector and payer. Given the trial-based nature of the proposed economic evaluations, we will only measure costs borne during the trial period. Thus, future-related and unrelated medical costs beyond the trial period will be omitted.^{19 19 59}

Data source

Cost components and data sources are summarised in table 2. Costs will be estimated using trial programme data, EHR, patient and caregiver surveys, and stakeholder interviews (a sample interview guide is available in online supplemental file 2). These sources will inform estimates of direct medical costs, including clinical outcomes, utilisation and charges. We will use EHR encounter charges and published cost-to-charge ratios to derive estimates of societal costs for each patient. While this approach is imperfect, it is commonly used in health economic applications when precise cost data are not available.

Table 2. Cost components and data sources.

Cost category	Component	Data source	Perspective
			Payer and healthcare sector	Societal
Formal healthcare sector
Direct medical costs	CHW home visit time costs Physician/nurse palliative care consultations Outpatient visits Hospitalisations Visits for chemotherapy, radiation therapy, etc. Pain management medications Hospice care Nursing home care Home-based palliative care equipment	Programme data (visit frequency, visit duration, CHW pre-tax wage rate plus fringe benefits) EHR (visits, charges) EHR (visits, charges) EHR (LOS, admissions, readmissions, charges) EHR (visits, charges) EHR (prescriptions), PCS (OOP expenses), FS (VA price, WAC) PCS PCS PCS (OOP expenses), FS (VA price, WAC)	Included	Included
Direct non-medical costs	Implementation costs CHW travel costs Patient transportation and time costs Caregiver transportation and time costs Care coordination and social support services Housing modifications for home-based care	Programme data (administrative, CHW recruitment and training, general supplies) Programme data (travel time, CHW pre-tax wage rate plus fringe benefits) PCS (time, marginal post-tax wage rate plus fringe benefits) PCS (time, marginal pre-tax wage rate plus fringe benefits) PCS PCS	Included*	Included
Informal healthcare sector
Indirect costs	Caregiver time spent on patient care Productivity costs	PCS (hours of care, marginal pre-tax wage rate plus fringe benefits) PCS and WPAI	Excluded	Included
Other sectors
Non-healthcare sector costs	Employer costs due to absenteeism† ACP Funeral and bereavement costs	PCS and WPAI PCS PCS	Excluded†	Included

^*Direct non-medical costs will only be included in analyses from the healthcare sector’s perspective, where some payers such as Medicaid may cover these costs (eg, transportation costs).

^†When the payer is the employer, presenteeism and absenteeism costs should be included.

ACP, advance care planning; CHW, community health worker; EHR, electronic health record; FS, fee schedule; LOS, length of stay; OOP, out-of-pocket; PCS, patient and caregiver surveys; VA, Veterans Affairs; WAC, wholesale acquisition cost; WPAI, Work Productivity and Activity Impairment.

For analyses from the payer’s perspective, paid amounts will be used as cost estimates. For medications, we will use prescription information from the EHR, patient-reported out-of-pocket expenditures and unit prices from fee schedules. US Veterans Affairs prices will be used for the societal perspective, while published average wholesale prices will be used for third-party payers. While the latter is subject to bias, it remains the best readily available estimate, as the true price paid by a payer results from complex and confidential negotiations with the manufacturer.

We will value time costs using the marginal pre-tax and post-tax wage rate plus fringe benefits, consistent with the replacement cost and opportunity cost methods.^{54 60} Time costs for patients and caregivers represent real changes in the use of resources by the patient and society. Patient time (eg, time spent in treatment) is generally assumed to be taken from leisure time rather than productive work time, and is therefore valued at the marginal post-tax wage rate plus fringe benefits. Unpaid caregiver time is considered a productive activity and is, therefore, valued at the marginal pre-tax wage rate plus fringe benefits. Accordingly, patient and caregiver time costs will be valued using pre-tax or post-tax marginal wage rates plus fringe benefits, consistent with replacement cost and opportunity cost methods. We note that on average, fringe benefits in the US accounted for 31.1% of the total compensation in 2024.⁶¹

Cost-effectiveness analyses

We will test the hypothesis that the intervention is a more efficient strategy compared to eSOC. The rationale is that a CHW palliative care model, which uses CHWs trained to provide advocacy, support, motivation, empowerment and education on palliative care to patients and their caregivers, will increase the use of palliative care services in underserved populations.

Standard CEA

Using the cost and effectiveness estimates calculated above, we will conduct a CEA to evaluate the trade-offs between cost and health benefits (eg, relative efficiency) of implementing the intervention compared with eSOC. This analysis will be performed using standard analytical methods recommended by the Second Panel on Cost-Effectiveness Analysis,¹⁹ from the perspectives of a payer, an adopting organisation (eg, Medicaid), the US healthcare sector and society.

Value will be measured in terms of the ICER and the incremental net monetary benefit (INMB) of the intervention relative to eSOC. Measurements will be performed at different willingness-to-pay (WTP) thresholds, which represent the ‘opportunity cost’ of each unit of effectiveness. The ICER will be calculated as the ratio of the intervention’s incremental costs to its incremental effectiveness, expressed as dollars per unit of effectiveness. The ICER will be compared with the decision-maker’s WTP threshold (eg, US$150 000/QALY in a societal perspective analysis)²⁰⁶²,⁶⁴ to determine the intervention’s acceptability in terms of its efficiency in generating health benefits. An ICER below the WTP threshold will be interpreted as indicative of good value-for-money. We will calculate the INMB, the delta between the monetary value of the incremental health effects (product of the ICER and the WTP) and incremental costs. A positive INMB will indicate that the intervention’s monetised health benefits exceed its opportunity costs.⁶⁵

While standard CEA is valuable, it has been criticised for its limited focus on aggregate costs and health benefits, neglecting the social distribution of these consequences.^{25 26 28 66 67} However, many health interventions also yield non-health benefits, such as preventing impoverishment and providing FRP, especially for advanced-stage cancer patients facing significant financial toxicity due to treatment costs, with notable disparities in burden.^{68 69} To address these critiques of standard CEA, we will employ ECEA and DCEA methodologies, also rooted in extra-welfarism like standard CEA.^{25 26 28 70 71} We note that rather than replacing standard CEA, these methods only extend and supplement it with equity and FRP considerations. We will test the hypothesis that, compared with eSOC, the CHW palliative care intervention will reduce inequities in cancer care outcomes across different SDOH risk categories among African Americans. We will classify patients into different risk categories at baseline and follow-up based on their PRAPARE-derived risk scores: high risk (score ≥10), moderate risk (score 5–9) or low risk (score <5).^{72 73} We will also explore the feasibility of conducting similar analyses by sex and income quintiles.

Together, these approaches will allow us to assess potential trade-offs between costs, health benefits and equity impacts from different perspectives. Findings will inform implementation within trial sites and contribute to a generalisable framework for adapting the intervention across diverse settings.

Extended CEA

ECEA builds on standard CEA by evaluating both the health and financial consequences of an intervention across four domains: (1) health gains; (2) FRP benefits; (3) total costs to decision-makers and (4) distributional benefits.^{25 26 28} The results can help decision-makers consider multiple criteria and weigh trade-offs among competing priorities.

In our analysis, we will measure both costs and health benefits, as in standard CEA, from the perspective of a societal decision-maker. As noted, ECEA and DCEA-like analyses are often more appropriate for societal decision-makers aiming to maximise social welfare from an extra-welfarist approach. Non-health benefits, such as FRP benefits and equity impacts, will be estimated as follows. First, we will estimate the average illness-related financial consequences at the subgroup level (ie, by SDOH risk category, and potentially by sex and/or income quintile) for participants assigned to either eSOC or the CHW palliative care intervention. These consequences will include out-of-pocket direct medical payments (eg, insurance co-pays), direct non-medical costs (eg, transportation costs to seek care, patient and caregiver time costs), as well as caregiver productivity losses (ie, absenteeism and presenteeism), which can be translated into forgone wages and income.^{25 26 28} We will define the financial loss protection (FLP) of the intervention as the private (patient and caregiver out-of-pocket) expenditures ‘crowded-out’ by the intervention. This will be calculated as the difference in subgroup-level private expenditures between the two experimental arms, scaled by the subgroup-level average disposable income. Scaling by disposable income accounts for the varying impact of financial loss on individuals with different income levels or social vulnerability. Using the estimated FLP, we will calculate FRP within each population subgroup based on prespecified income thresholds. The FRP metrics will include: (1) cases of catastrophic health expenditures averted; (2) cases of poverty averted and (3) the money-metric value of insurance. Finally, we will define and calculate the financial protection ICER (ICER_FRP). This ratio represents the incremental cost to the decision-maker compared with the aggregate financial protection incremental benefits (eg, FRP metric) of implementing the intervention. To interpret the resulting ICER_FRP estimate, we will use benchmark WTP thresholds established by authoritative sources.

Distributional CEA

DCEA enables the explicit integration of health inequality concerns into CEA, allowing for the identification of strategies that are both efficient and equitable.^{25 28} We will conduct a DCEA to quantify the equity impacts of implementing the intervention and to assess potential trade-offs between its adoption costs, health benefits and equity impacts (by SDOH risk category and, if feasible, by sex and across income quintile).²⁸ We will calculate the equally distributed equivalent (EDE) level of health, measured in QALYs, produced by the CHW palliative care intervention. The EDE represents the level of health that, if equally distributed across all population subgroups, would yield the same aggregate social welfare as the actual distribution. We will present results in the equity impact plane, which is analogous to the CEA plane in standard CEA. This captures potential trade-offs between the net equity impact and efficiency, with efficiency measured by the intervention’s INMB relative to eSOC. Using this framework, we will trace the equity trade-off curve, which represents the equity-weighted population health impact of the intervention. The curve illustrates the intervention’s value to the payer or decision-maker under varying levels of inequality aversion. We will examine the sensitivity of the inequality measures and equity-efficiency trade-offs under alternative interventions to different inequality indices (eg, Atkinson and Kolm) and varying levels of the decision-maker aversion to the relevant dimension of social inequality.⁷⁴,⁷⁹

SROI analysis

We will also conduct an SROI analysis, a pragmatic, localised form of CBA that assesses the broader value of interventions by incorporating social and environmental impacts for all key stakeholders.^{24 42 80 81} According to standard guidance for SROI analysis, as outlined by the SROI Network (now Social Value International), stakeholder involvement is essential at every stage. In fact, it is not just recommended but considered integral to conducting a robust SROI analysis, including the development of the theory of change (ToC).

We will use financial proxies to translate the social, environmental and economic outcomes of the intervention into monetary values, which will be represented in the SROI ratio as the intervention’s net benefits relative to its costs. For example, a 4:1 SROI ratio indicates that each dollar invested in the programme generates four dollars’ worth of social value.^{24 42 82} The SROI ratio provides a clear representation of value, helping stakeholders make informed decisions about palliative care resource allocation.

We will conduct an evaluative-type SROI analysis, which will include the following six steps:

Stage 1: identifying key stakeholders and defining the purpose and scope of the SROI analysis

We will begin by identifying key stakeholders involved in implementing the CHW palliative care intervention. Based on preliminary work, the following stakeholders have been identified for inclusion in the study and SROI measurement: patients and caregivers, oncologists, palliative and hospice care providers, cancer centre and hospital leaders, and payers. Each stakeholder will be classified in a stakeholder analysis matrix, considering their level of influence and importance across the continuum of palliative care. Consistent with best practices, we will engage these stakeholders (ie, patients and caregivers, providers, payers and healthcare system representatives) through qualitative interviews to help define the purpose and scope of the analysis.

Stage 2: mapping outcomes (development of the ToC)

In this stage, we will develop a ToC to outline how and why the CHW intervention is expected to lead to specific hypothesised outcomes and impacts. This will involve mapping the causal relationships between the intervention’s activities, outcomes and impacts, using both quantitative and qualitative metrics that extend beyond financial outcomes. Consistent with best practices, stakeholders will be engaged to help identify additional inputs, activities, outputs and outcomes not captured in the illustrative ToC (table 3).

Table 3. Illustrative theory of change (ToC) for the SROI analysis of the CHW palliative care intervention.

Inputs*	Activities†	Outputs‡	Outcomes§	Impacts¶
Financial resources for programme implementation Palliative care infrastructure (staff, technology, care coordination tools)	CHW training and implementation CHWs conduct patient and caregiver assessments to identify palliative care needs; provide education, care coordination and emotional support; facilitate referrals to palliative and hospice care services; collaborate with oncologists, palliative care providers and social workers to improve care integration Stakeholder engagement (patients, caregivers, providers, payers, healthcare administrators) to inform programme implementation and continuous quality improvement	Number of patients receiving palliative care services Number of CHW-patient interactions (education, referrals, care coordination) Number of provider interactions with CHWs regarding patient care Number of facilitated palliative care referrals Number of stakeholder meetings and engagements	Patients and caregivers Increased awareness and knowledge of palliative care Improved access to palliative care services Enhanced care coordination and patient navigation Reduced unmet care needs (eg, symptom management, psychosocial support) Improved QOL and patient satisfaction Providers and healthcare system Improved integration of CHWs in interdisciplinary care teams Enhanced communication between oncologists, palliative care providers and CHWs Increased provider satisfaction with care coordination support Payers and health system administrators Reduced avoidable ED visits and hospitalisations Increased cost-efficiency of palliative care services delivery Improved resource allocation for palliative care Evidence to inform policy and funding decisions for CHW integration in palliative care	Improved health and well-being of patients with advanced cancer Enhanced caregiver support, reduced caregiver burden and improved caregiver experiences Strengthened palliative care workforce and service delivery models Cost savings from reduced hospitalisations and improved end-of-life care planning

This ToC provides a structured framework for the planned SROI analysis. It ensures that stakeholder-driven insights inform the evaluation of the CHW palliative care intervention’s impact and value creation. We will assess the impact of the CHW intervention using both quantitative (eg, healthcare utilisation, cost savings, HRQoL scores) and qualitative (eg, patient and provider experiences) measures. Financial proxies will be used to translate social, economic and health outcomes into monetary values and to calculate the SROI ratio, which will guide decision-making on palliative care resource allocation. The ToC assumes that CHWs are effectively trained and integrated into care teams. It also assumes that patients and caregivers engage with CHWs and accept their support, and that providers recognise and use CHWs to enhance care coordination. The ToC also assumes that sufficient funding and resources are available for programme sustainability, and that stakeholders involved in the co-development of the SROI provide accurate and meaningful data for SROI calculations. Therefore, there are potential risks to validity, including variability in CHW effectiveness across different settings; resistance from healthcare providers or administrators to integrate CHWs; challenges in quantifying intangible benefits (eg, patient empowerment); and insufficient long-term funding to sustain the intervention over a longer period.

^*Resources and investments.

^†Intervention strategies.

^‡Immediate deliverables.

^§Short- to medium-term changes.

^¶Long-term social and economic benefits.

CHW, community health worker; ED, emergency department; QOL, quality of life; SROI, social return-on-investment.

Stage 3: evidencing and valuating outcomes

In this stage, we will collect evidence and data to substantiate the identified outcomes and quantify both tangible and intangible impacts of the intervention. The same tools used in CEAs to measure costs will be employed. Indicators will be defined, and the magnitude of change (eg, number of stakeholders who experienced the outcome) will be estimated for each outcome (eg, ACP, improvements in QOL and QOC, utilisation of hospice care, improved symptoms, improved SDOH risk, etc). Stakeholder data and perspectives will also be gathered to support outcome measurement and valuation.

Financial proxies from our costing analyses will be assigned to each outcome to estimate unit costs associated with resource use.^{19 20} For outcomes lacking trial or market data, or those that are too intangible for direct valuation (such as ACP, QOC, improved SDOH risk), we will apply valuation methods from existing SROI and economic studies. This will include directly eliciting values from stakeholders through surveys. We will also monetise changes in HRQoL utility scores using standard approaches, such as WTP thresholds (eg, US$150 000/QALY gained from a societal perspective) and the value of statistical life.^{19 20 83}

Stage 4: establishing impact

We will convert the quantified outcomes into a common monetary unit to estimate their impact. Gross impact for each outcome will be calculated by multiplying the estimated magnitude of change by the corresponding financial proxy. Net impact will be determined by adjusting the gross impact for key counterfactuals, including: (1) deadweight, which measures the proportion of the outcome that would have occurred regardless of the intervention; (2) displacement, which measures the extent to which positive outcomes generated by the intervention were offset by negative outcomes elsewhere (eg, shifting benefits without actual improvement); (3) attribution, which captures the proportion of the outcome attributable to the intervention, rather than other factors and (4) drop-off, which measures the rate at which outcome effects diminish over time. While there are no universally fixed values for these adjustments, we will use the following as starting points based on best practices in health economic evaluations and SROI studies: deadweight (10%–15%)^{84 85}; displacement (0%–30%); attribution (20%–70%)⁸⁴ and drop-off (10%–30%).⁸⁶ We will consult with experts and stakeholders to reach a consensus on the appropriate ranges for these estimates.

Stage 5: calculating the SROI ratio

In this step, we will calculate the SROI ratio by dividing the value of the impact generated by the intervention by its implementation, intervention and downstream costs borne by the implementing organisation. Both gross and net SROI ratios will be computed, representing the intervention’s gross and net impacts relative to the total investment. The SROI ratio offers a quantitative measure of the relationship between investments in the CHW palliative care intervention and the monetary value of its social and economic benefits.

Stage 6: reporting, using and embedding the SROI results into decision-making

Finally, we will report our findings in ways that are most relevant and meaningful to stakeholders. This will include convening a stakeholder meeting to discuss the results, as well as disseminating findings through websites and in peer-reviewed publications. In addition to the quantitative SROI ratios, we will prepare summary reports that highlight both qualitative and quantitative aspects of the analysis to provide a more comprehensive view of the intervention’s social value. The SROI analysis will help potential adopters, such as organisations or payers, assess the intervention’s expected impact on revenues and operating costs and guide adaptations to optimise both outcomes and financial performance. The peer-review process and stakeholder engagement will ensure the analysis aligns with SROI best practices and that the report’s data and findings are robust. This final step supports informed action, ensuring the insights gained from the economic evaluation inform decision-making and guide future implementation and scale-up.^{24 42 80}

Sensitivity and uncertainty analyses

One-way and multivariate sensitivity analyses will be conducted to identify the most influential drivers of the estimated value of the intervention. We will vary each cost and effectiveness component across its observed empirical or assumed range. For example, the discount rate will be varied between 0% and 7%. We will also assess the joint impact of variability in intervention cost components and effectiveness measures through probabilistic sensitivity analyses. The bootstrap method will be used to quantify the effect of variability in cost and effectiveness or impact measures on the results, and to construct 95% credible CIs, in order to assess the robustness of the findings.

We will also depict variability using cost-effectiveness acceptability curves, which illustrate the probability that the CHW intervention is considered acceptable across a range of assumed WTP and SROI ratio thresholds. In addition, we will use expected loss curves to present, in a single figure, the expected foregone benefits of implementing the intervention, the optimal strategy based on expected value, and the potential value of future research to reduce uncertainty.⁸⁷ While the methods described above characterise uncertainty by estimating variability in incremental costs and effects, they do not retain information about the uncertainty of the costs and effects for the two comparison groups. This distinction may be important for informing risk-averse decision-makers, as the standard methods described above do not explicitly incorporate the decision-maker’s degree of risk-aversion.

To address these limitations, we will estimate the risk-aversion curves, which explicitly incorporate varying levels of risk-aversion into the analysis and can therefore inform decisions by risk-averse stakeholders.^{88 89} We will calculate the risk-adjusted expected net monetary benefit (NMB) for both the CHW palliative care intervention and eSOC. This approach penalises the expected NMB—a metric that retains the uncertainty of individual programmes—for downside risk (eg, bad risk), enabling the rank ordering of interventions.⁹⁰ Using these risk-adjusted metrics, we will compute net benefit-to-risk ratios across a range of SROI acceptability and WTP threshold values.⁸⁸ Finally, we will demonstrate how these measures can be jointly applied to support decision-making and policy development related to palliative care.

Patient and public involvement

Patient and public involvement has been integral to both the design and conduct of the clinical trial subject to this outlined economic evaluation. At the outset of the trial, we established a community advisory board (CAB) composed of patients, caregivers and members of the public to provide input on the research design and ensure that the study reflects the needs and priorities of the target population. The CAB has remained involved throughout all phases of the research, offering feedback on the study protocol, recruitment strategies and data collection methods. Regular meetings with the CAB continue to ensure ongoing engagement and guidance as the study progresses. In addition, our dissemination plans include the active involvement of patients and public. This will involve coproducing plain language summaries of the study results and organising public-facing events to share findings in accessible formats.

Discussion

A lack of information on the cost-efficiency, sustainability and financial viability of palliative care interventions has been identified as a barrier to implementing innovative palliative care models.²⁹ To address this gap, this study presents an approach for evaluating the CHW palliative care intervention currently being formally assessed through a multicentre, randomised, assessor-blind, parallel-group, pragmatic, hybrid type 1 effectiveness-implementation trial at four cancer centres across the USA.

We outline our approach to measuring implementation, intervention and downstream costs, followed by a series of economic evaluations, including novel methods designed to incorporate equity considerations and capture the broader social value of the intervention. These innovative approaches may be applied to a wide range of interventions and contexts and have important implications for both research and practice.

Our findings are most directly applicable to Medicaid programmes, given the payer perspective adopted in the base case analysis and the demographic and socioeconomic characteristics of the study population. However, the implications may extend to other public and private payers, particularly in contexts where similar populations are served or where coverage policies for the intervention are under consideration. Future studies could explore cost-effectiveness across a broader range of payer perspectives, accounting for variation in reimbursement rates and benefit structures.

Strengths and limitations

This study presents a rigorous, transparent plan for the economic evaluation of a CHW palliative care intervention, integrating standard, ECEA and DCEAs to assess costs, health benefits and FRP associated with the intervention. By accounting for direct, indirect and intangible costs, the study aims to provide a comprehensive assessment of the intervention’s cost-efficiency and its distributional impacts. The use of multiple data sources—including trial programme data, EHR data and stakeholder surveys—also strengthens the comprehensiveness of cost estimation. Additionally, an evaluative SROI analysis captures the broader social and economic value of the intervention, incorporating stakeholder input and financial proxies to enhance real-world applicability.

However, we acknowledge several limitations. First, the 6-month study horizon may not fully capture long-term costs and benefits, particularly those related to financial hardship, lost productivity and caregiver burden. While decision-analytical modelling could extend these estimates, doing so would diverge from a trial-based economic evaluation. Second, some intangible costs, such as caregiver burden and psychological distress, are difficult to quantify and may not be reliably captured. Third, the generalisability of the findings may be limited, as costs and financial outcomes can vary by healthcare setting, population and payer structure. In addition, reliance on administrative data and cost-to-charge ratios may not accurately reflect actual healthcare costs. Fourth, although ECEA and DCEA help address equity considerations, unmeasured SDOH may still influence outcomes. Fifth, while the SROI analysis offers valuable insights, it depends on the validity of financial proxies and assumptions about counterfactual scenarios. These proxies may not fully reflect societal values and could introduce uncertainty or bias. Finally, although stakeholder engagement enhances the relevance of the analysis, it may inadvertently introduce bias, as the perspectives of participating stakeholders may not fully represent broader societal priorities. Nonetheless, this approach remains valuable for providing a localised view of the intervention’s social value.

Ethics and dissemination

This study protocol was approved by the Institutional Review Board (IRB) of the Johns Hopkins University School of Medicine (IRB00372476). All methods will be carried out in accordance with relevant guidelines and regulations. Informed consent will be obtained from all subjects prior to study participation. This manuscript does not contain participant-level data. The full protocol will be available from the corresponding author on reasonable request. The dissemination of findings from the clinical trial and accompanying economic evaluation outlined in this manuscript will be multifaceted to maximise reach and impact. Research findings will be presented at relevant scientific conferences, submitted for publication in peer-reviewed journals and shared with community stakeholders, including hospital leaders and administrators, providers, CHWs and patient advocacy groups.