Introduction

Community health worker (CHW) programs have demonstrated improvements in health outcomes, especially for low-income or otherwise marginalized populations who face significant barriers to receiving quality healthcare services [1,2,3]. CHWs are generally people from the same community as a patient, who provide critical social and emotional support, health education, and/or navigation assistance that supplements the work of other healthcare providers to assist in improving access to or quality of healthcare services for a patient [4]. CHW program evaluations have included randomized controlled trials among diverse populations, showing improvements of health outcomes and healthcare costs, particularly among Medicaid beneficiaries [5, 6].

CHW programs have faced considerable financing challenges despite the evidence of their benefits to health outcomes and healthcare costs. Many CHW programs have been financed under temporary grants, which has posed challenges for their long-term viability [7]. Recently, CHW programs have been included among states’ Medicaid 1115 waivers and related demonstration programs that aim to direct Medicaid dollars to CHW services [8]. CHW programs funded by Medicaid financing are either paid through fee-for-service (FFS) payments for visits between CHWs and patients, or capitated per-member-per-month (PMPM) payments to CHWs caring for a population panel of patients as part of a Medicaid managed care organization or a healthcare practice that has entered into value-based contracting. What remains unclear is how much financing should be allotted in FFS or PMPM payments to enable CHW programs to be financially sustainable. CHW programs, unlike standard medical care in clinics or hospitals, often require considerable transportation time, mobile equipment, and associated expenses, alongside flexible systems for supervision and support to ensure safety of CHWs and effectiveness in delivering roving, within-community services.

In the current study, we seek to establish and apply a rigorous approach to estimating the levels of Medicaid FFS or PMPM payments necessary to support the financial viability of CHW programs across all 50 US states, Washington D.C. and Puerto Rico. We develop and implement a microsimulation model that synthesizes data on CHW salaries, supervision, equipment, transportation, benefits and related costs across the US. We use the model to estimate the levels of financing needed to cover CHW program costs across different geographies, while maintaining patient volume or panel sizes that are consistent with the randomized trials showing benefits of CHWs on patient outcomes and healthcare costs.

Methods

Methods Overview

We constructed a microsimulation model (Fig. 1), which is a type of simulation that accounts for the covariation among input variables to estimate variations in output variables across locations (e.g., capturing how wages or transportation costs may be higher, and therefore subsequent payments may need to be higher in a given location). In our prespecified economic analysis plan (see “Data Availability”), we chose microsimulation because wages and costs related to CHW programs are expected to be highly variable across geographies, and hence we judged that capturing variations through a microsimulation, rather than only using national average estimates for cost, would be important to ensure the financial viability of CHW programs that are typically susceptible to local variations in labor and overhead expenditure. Our methods and reporting corresponded to the Consolidated Health Economic Evaluation Reporting Standards (see Appendix for checklist) [9].

Fig. 1
figure 1

Microsimulation model diagram. The diagram depicts how input variables were synthesized into an estimate of the outcome variables

Full size image

Input Variables and Data Sources

Input variables and data sources for the model are summarized in Table 1, and further detailed in the Appendix by state (including Washington D.C. and Puerto Rico) and by metropolitan or non-metropolitan statistical area within states. The study population was defined as US-based CHW programs serving Medicaid patients.

Table 1 Input variables and data sources for the model
Full size table

The first set of input variables were labor costs (Fig. 1). CHW and CHW supervisor wages from the Bureau of Labor Statistics Occupational Employment and Wage Statistics (OEWS) database, which provides wage estimates from 82 million people (62% of total national employment) working at 1.1 million non-farm employer establishments surveyed through six semiannual panels collected over a 3-year period (2018–2021) [10]. In the main analysis, we used CHW supervisor wages for Registered Nurse (RN) care managers, to correspond to a common supervision structure reported in systematic reviews, accounting for a typical 1:8 ratio of RN supervisors to CHWs; in a sensitivity analysis (detailed further below), we varied the CHW supervisor to correspond to a Master’s-level Medical Social Worker (MSW) with a 1:6 ratio of MSW to CHW, to address another commonly-reported supervisor profile [11, 12]. Both the mean and standard deviations around the mean for CHW and CHW supervisor salaries were obtained from the OEWS database, for each state and each metropolitan or non-metropolitan statistical area within states (see Appendix).

The second set of input variables were overhead expenditures (Fig. 1). Based on prior accounting efforts to estimate costs among CHW programs, the overhead expenditure categories we considered in the model included equipment, software, transportation, space, and benefits. For equipment, we gathered the cost of annual mobile phone plan with unrestricted text, call and data including mobile Wi-Fi hotspot; the annualized cost of a Chromebook laptop over a typical four-year lifecycle; and costs of ongoing Information Technology (IT) support [13,14,15]. For software, we incorporated costs of CHW and CHW supervisor accounts on the most popular patient care management software systems incorporating records management, panel management, telehealth functionalities and billing [16]. For transportation, we gathered the typical time and associated mileage and gasoline expenses required for transportation from a CHW home to a patient home or CHW workspace, based on commute times by location [17]. The transportation time was included to account for both the costs of transportation itself (fuel costs, maintenance, insurance and depreciation) and opportunity costs (the wage costs to support CHWs for their time in transit, which would need to be indirectly compensated by visit-based FFS or panel-based PMPM payments) [18, 19]. For space, we computed the average cost of office space per person in the locale from a national real estate database [20]. For benefits, we assessed the overhead costs of initial and continuing CHW education based on a review of state education requirements associated with CHW certification, along with provision and administration of employer-sponsored healthcare insurance coverage (medical, dental, and vision) with a health maintenance organization or preferred-provider organization plan based on the average benefit expenditure rate reported among healthcare organizations in a national employer-sponsored benefits survey [21, 22].

Outcome Definitions

We calculated two outcomes, estimating the threshold value for each outcome to cover the total annual expenses for a CHW, after incorporating the above costs. The two outcomes were: (i) the Medicaid FFS payment rate per a 30-min visit between a CHW and an individual patient (typically Medicaid billing code 98960, which most states use as the basis for CHW payment [21]), either in person or virtually (telephone or video); or (ii) the Medicaid PMPM payment rate for every patient on the CHW’s panel at a given time.

For both outcomes, we computed the payment rates for a full-time CHW with 2080 h of work per year, with a mean patient panel of 65 patients at a given moment, and a mean 6-month duration of service per patient (varied in sensitivity analyses detailed below) [12]. All costs were summed from a CHW service provider perspective (accounting for labor and overhead including opportunity cost) over a one-year time horizon, and updated to undiscounted 2023 US Dollars using the Consumer Price Index to account for inflation [23].

Computation Methods

We estimated each outcome separately, assuming each CHW would be paid under either a Medicaid FFS or capitated PMPM contract. For each outcome, we first computed the total monthly cost to support each CHW each month by summing the total costs listed above and detailed in Table 1, scaling the estimate to a cost per CHW per month including overhead costs associated with that CHW. We then estimated the time left for visits and the associated transportation time to estimate the typical visit rate per CHW per month, with a ratio of 1:3 of in-person to virtual (phone or video) visits [24]. Finally, we estimated each of the outcome variables by estimating what threshold for payment would be the minimum amount to meet the total monthly costs given the visit rate or typical patient panel size per CHW (65 patients per CHW, each enrolled for 6 months) [12]. We completed the calculations for each state and for each sub-state metropolitan or non-metropolitan area. For validation, we compared our estimates to the program costs available from CHW program costs in two states (Virginia and Washington state) for which independent and complete wage and overhead expenditure data were available [25].

Sensitivity and Uncertainty Analysis

For each outcome, we sampled 10,000 times with replacement from normal distributions constructed around the mean and standard deviation of each input variable (Appendix) to estimate the mean and 95% credible interval (CI) around each outcome estimate [26].

We performed two sensitivity analyses. In the first sensitivity analysis, we swapped out the salary and overhead expenditures for an RN care manager supervisor for each team of eight CHWs for the salary and overhead expenditures for an MSW supervisor for each team of six CHWs [11, 12]. In the second sensitivity analysis, we varied the size of the patient panel associated with each CHW from a low of 15 to a high of 100 to understand the implications of widely varying panel size among CHW programs.

Results

Descriptive Statistics on the Input Variables

CHW wages estimated in the Bureau of Labor Statistics OEWS database had a mean national hourly wage of $23.51 (varying from $15.90 in Puerto Rico to $31.61 in Rhode Island; Fig. 2), with wages averaging $22.54 in metropolitan areas and $21.13 in non-metro areas. CHW supervisor wages when including RN care managers as supervisors had a mean national hourly wage of $39.88 (varying from $17.96 in Puerto Rico to $64.10 in California, and averaging $40.87 in metro and $38.93 in non-metro areas), where MSW supervisors had a mean national hourly wage of $29.37 (varying from $16.22 in Puerto Rico to $42.49 in California; Fig. 2).

Fig. 2
figure 2

State variations in wages among community health workers (CHW). Wages are expressed in dollars per hour of work in 2023 US Dollars using the Consumer Price Index to account for inflation [23]

Full size image

Overhead expenditures (equipment, software, transportation, space, and benefits) per CHW work hour averaged $43.65 nationwide (varying from $19.21 in North Dakota to $196.41 in New York, and averaging $43.84 in metro and $43.46 in non-metro areas). Among the varied overhead expenditures, the highest sub-cost was universally for transportation, which constituted a mean of 65.1% of the overhead expenditure nationally (varying from 21.0% in Hawaii to 94.2% in Arkansas). In our face validity checks, wage and overhead expenditure data did not differ between model estimates and independent program data available from two states (Appendix).

Fee-for-Service Payment Rate Minimums

The minimum Medicaid fee-for-service payment rate for a 30-min CHW visit (Medicaid billing code 98960), after considering the wages and overhead costs for CHWs, was estimated to be a mean of $53.24 nationwide (95% CI $24.80, $91.11). Among states, the minimum varied from $40.44 in South Dakota (95% CI $17.65, $71.09) to $70.89 in Washington D.C. (95% CI $32.08, $120.25) due to the combination of wages and overhead costs (Table 2, Fig. 3).

Table 2 Estimated minimum payment threshold rates for sustaining community health worker (CHW) programs across states
Full size table
Fig. 3
figure 3

State variations in the payment threshold rates for sustaining community health worker (CHW) programs. The A minimum fee-for-service payment rate was computed for a 30-min CHW visit (Medicaid billing code 98960), and the B minimum capitated payment rate per member per month (PMPM), after considering the wages and overhead costs for CHW programs

Full size image

Among sub-state metropolitan or non-metropolitan areas, the minimum fee-for-service payment rate for a 30-min CHW visit varied from $40.24 in the Las Cruces, New Mexico area (95% CI $17.60, $69.89) to $78.08 in the Napa, California area (95% CI $33.85, $136.93; see Appendix). Fee-for-service minimums were typically higher in metropolitan than in non-metropolitan areas but with overlapping credible intervals ($52.38 in metro, 95% CI $24.34, $89.77; $49.82 in non-metro, 95% CI $23.29, $81.15).

Capitated Payment Rate Minimums

The minimum Medicaid capitated payment rate per member per month (PMPM), after considering the same overhead costs as above, was estimated to be a mean of $140.18 nationwide (95% CI $105.94, $260.90). Among states, the minimum varied from $113.55 in South Dakota (95% CI $84.58, $229.03) to $176.58 in Washington D.C. (95% CI $132.83, $298.64; Table 2, Fig. 3).

Among sub-state metropolitan or non-metropolitan areas, the minimum PMPM rate varied from $111.76 in the Las Cruces, New Mexico area (95% CI $83.37, $223.62) to $196.08 in the Napa, California area (95% CI $156.88, $326.14). Capitated minimums were typically higher in metropolitan than in non-metropolitan areas but with overlapping credible intervals ($138.41 in metro, 95% CI $104.51, $258.02; $132.78 in non-metro, 95% CI $99.54, $250.80).

Sensitivity Analyses

In the first sensitivity analysis, when we swapped out the expenditures for an RN care manager supervisor (1 for each 8 CHWs) for an MSW supervisor (1 for each six CHWs), the minimum fee-for-service payment rate for a 30-min CHW visit was estimated as a mean of $52.80 nationwide (95% CI $24.78, $90.08), while the minimum capitated PMPM rate was estimated as a mean of $139.02 nationwide (95% CI $104.87, $258.13; see Appendix for data across states).

In the second sensitivity analysis, we varied the size of the patient panel associated with each CHW from a low of 15 to a high of 100. At the lowest panel size of 15, the minimum fee-for-service payment rate rose to a mean of $230.72 (95% CI $107.45, $394.83), and capitated PMPM to a mean of $607.45 (95% CI $459.09, $1130.55; see Appendix). At the highest panel size of 100, the minimum fee-for-service payment rate declined to a mean of $34.61 (95% CI $16.12, $59.22) and capitated PMPM to a mean of $91.12 (95%CI $68.88, $169.58; see Appendix).

Discussion

As state Medicaid agencies seek to provide sustainable funding for CHW programs, we established and applied a rigorous approach to estimating the levels of Medicaid FFS or PMPM payments to support the financial viability of CHW programs across all 50 US states, Washington D.C. and Puerto Rico. In the past, payments for services outside of face-to-face medical encounters in hospitals or clinic settings have lacked a systematic approach to establishing their basis, and CHW programs particularly face numerous challenges that differ from traditional medical visits—such as potentially meaningful transportation time to patient locations.

We found that the threshold minimum levels of both Medicaid FFS and PMPM payments estimated through our microsimulation model to sustain CHW programs were typically much higher than those currently publicly disclosed by Medicaid officials. For example, whereas the estimates in our main analysis indicated a minimum FFS rate of $53.24 nationally for a 30-min Medicaid billing code 98960, publicly-accessible Medicaid state plan amendments and waivers lack clarity on how their rates were established and indicate variations from $9.60 in Indiana to $30.89 in South Dakota [27, 28]—all much lower than the threshold to achieve financial sustainability, per our estimates. Similarly, while our minimum capitated rate estimate was $140.18 PMPM, capitated rates that are publicly disclosed appear lower, on the order of $1.65 in Maine and $9.65 in Idaho [29, 30], though such data are not systematically publicly disclosed per recent reviews of the subject [31]. Additionally, our findings indicate that meaningful accounting of overhead expenditure and local wage rates is important to account for when computing the funding levels necessary to sustain a CHW program. Overall overhead expenditures did not differ markedly between metro and non-metro areas, as transport time was higher in non-metro areas but was counterbalanced by higher employee benefit and space costs in metro areas.

As with any study based on simulations, our study is subject to limitations associated with model assumptions. First, we considered two common types of supervisor types for CHWs, but did not consider all possibilities; CHW programs vary widely in their regulation and implementation. Secondly, we consciously focused on a suite of expenditures including standard employee benefits at the overhead rates seen in the healthcare sector generally, although some CHW programs do not offer such employee benefits to CHWs. We chose to include benefit overhead expenditures intentionally, as we did not want to compute minimum thresholds for payment that did not consider the need for sustainable, long-term career potential for CHWs, and particularly wanted to ensure consideration of providing healthcare coverage to employees who themselves are providing healthcare services. Third, our study revealed that the threshold estimates for Medicaid fee-for-service and capitated rates could be sensitive to the panel size adopted by CHWs; while we adopted panel sizes from published research articles reporting successful CHW programs, it is possible that higher and lower panel sizes may be successful among varied patient populations, and associated payments would need to be adjusted accordingly. Finally, while our study focused on patients receiving Medicaid, for whom CHW programs have been carefully studied, CHW services are increasingly available to uninsured, Medicare, and commercially-insured patients who have a different burden of illness and potentially different cost structure for their infrastructure.

We believe future research on the subject of CHW payment would benefit from the broader disclosure of payment levels, as publicly-available listings for CHW payment rates were limited to a few states with Medicaid state plan amendments or waivers requiring such disclosure. Additionally, further study may be required into the intangible opportunity costs facing CHWs and affecting program sustainability, such as the availability and affordability of childcare, and of tools and data technologies needed to ensure safety when performing community-based home visits.

While such opportunities for further research may be present in coming years, our current study synthesized available data on CHW salaries, supervision, equipment, transportation, benefits and related costs across the US to provide estimates of the levels of financing needed to cover CHW program costs across different geographies. We found that to maintain patient volume or panel sizes that are consistent with the randomized trials showing benefits of CHWs on patient outcomes and healthcare costs, levels of fee-for-service and capitated program payment for CHW programs would likely need to be considerably higher than current publicly-available data sources suggest are offered for Medicaid-covered CHW services.