Cost-Effectiveness and Implementation of Community Clinical Linkage Models to Increase Female Breast and Cervical Cancer Screening

1. Introduction

The United States (US) National Institutes of Health (NIH) reports that an estimated 321,910 women will be newly diagnosed with invasive breast cancer, and approximately 42,140 will die from this cancer in 2026 in the US [1]. In addition, around 13,490 new cases of invasive cervical cancer will be diagnosed, and an estimated 4,200 women will die from this cancer in 2026 [2]. Increasing screening rates and early detection of these cancers is critical to reducing these numbers. Women in the US still face significant barriers to screening and follow-up care, which include knowledge about cancer risks, logistical issues related to accessing services, and health system mistrust [3,4].

The US Centers for Disease Control and Prevention’s (CDC) National Breast and Cervical Cancer Early Detection Program (NBCCEDP), initiated in 1990, supports programs to increase breast and cervical cancer screening rates among low-income and medically underserved women [5]. In 2018, the NBCCEDP promoted Community Clinical Linkage (CCL) models to increase screening among underserved women. The models focus on establishing partnerships between community organizations involved in outreach and clinical providers to improve service coordination with the goal of eliminating barriers to screening for underserved women [6,7].

NBCCEDP funded state grantees who implemented the CCL model along with evidence-based outreach interventions (EBIs) identified by the United States (US) Community Preventive Services Task Force [8]. Evidence-based outreach includes one-on-one or group education, providing client reminders, and reducing out-of-pocket costs [9]. A recent systematic review of the EBI cost-effectiveness literature found substantial variation in cost-related outcomes, reflecting differences in intervention components and intensity [10]. The most common intervention was one-on-one education with CHWs, which averaged $421 (SD = $459.1; 2021 dollars) per additional woman screened, while the most expensive, a multi-component intervention, had a mean cost of $1625 (2021 dollars) [11].

The objective of this study is to assess the breast and cervical cancer screening outcomes and the cost-effectiveness of two CCL programs funded by the Colorado Department of Health and the Environment (CDPHE) under the NBCCEDP’s initiative in two sites, one rural and one urban. We analyze the impact of transitioning from a general outreach model to an evidence-based outreach model led by CHWs with links to a clinical provider. We also describe the evolution of costs and screening outcomes over time as programs mature and populations served, change.

2. Methods

2.1. Program Description

From 2016 to 2018, CDPHE funded general community outreach through the Targeted Outreach Program (TOP), which promoted breast and cervical cancer screening for hard-to-reach populations. Statewide, thirteen TOP organizations distributed over 31,600 evidence-based mailers, trained 775 community organizations and health clinics, sponsored 492 community events, and offered one-on-one or group education to more than 19,600 women over two years [12]. In total, organizations referred over 12,000 women for services, of which 48% were ultimately screened [12].

In 2019, Colorado transitioned from TOP to a CCL partnership model, known as The Community Outreach (TCO). Central to the model was reliance on community health workers (CHWs) who conducted outreach at locations such as food pantries and churches. They met with women to provide one-on-one education, to identify barriers to their accessing screening services, such as health literacy, transportation, and lack of insurance, and to offer direct assistance to address them.

2.2. Study Setting

Two study sites, one rural and one urban, were chosen for this study based on population and geographic diversity, the presence of staff with institutional knowledge of the program, and program longevity. Both sites were clinical providers: the urban site was part of a large health system, and the rural site was a federally qualified health center (FQHC). Each site had participated in TOP before transitioning to the TCO CCL model. While they had each conducted general outreach through community events, they primarily relied on patient navigators who identified women for screening via electronic health records (EHRs), then guided them through any subsequent treatment pathways [13,14]. Under TCO, sites shifted to CHWs, whereby the goal was to identify underserved women or those who were due for screening in the community.

We obtained data from the CDPHE on the TOP and TCO programs for fiscal years (FY) 2018 through 2021. As the fiscal year spans July 1 through June 30, the third TCO year, 2021, coincided with the SARS-CoV-2 (COVID-19) pandemic. We thus discuss costs and outcomes for this year separately. We used a micro-costing approach to estimate TOP and TCO costs from the CCL’s perspective, based on itemized invoiced expenses submitted to the CDPHE. Cost data for TOP were available for the second half of 2018, but not for prior years. We estimated annual TOP costs for 2018 by multiplying the second-half values by two. Bi-annual TCO cost data were available for 2019 through 2021 for each site.

We grouped expenditures into the following categories: assessment and planning, personnel, outreach and education, office supplies, staff travel, and travel vouchers provided to women participants. As the invoice forms varied slightly from year to year, we consulted with CDPHE staff to ensure comparability in classifications over time. All values were discounted at 3% to reflect the timing of expenditures, then converted to 2021 dollars to adjust for inflation [15]. We did not include the cost of screening, as the focus was the outreach program, nor the time cost associated with women receiving direct assistance.

Each TCO participant completed a Worksheet (see Supplementary Materials) with the CHW, who later documented whether the woman ultimately received a screen. CDPHE provided de-identified Participant Worksheet data in Excel files, including final screening status by cancer type. Some data were unavailable for women served under TOP due to differences in reporting between the programs. The primary outcome was the percentage of women who completed screens among those receiving direct assistance for both cancer types.

We assessed average costs per woman screened for both breast and cervical cancer as program costs could not be separated by cancer type. We calculated the cost-effectiveness of program expansion over time using incremental cost-effectiveness ratios (ICERs), which indicate the cost of serving one additional woman compared to the prior year.

The ICER is defined by the change in the average cost per woman screened divided by the change in the screening rate per woman from the current year to the prior year. Following the literature’s convention, the ICER denominator is the percentage-point (pp) difference in screening rates, expressed as a decimal [10]. For example, a 26% pp change in screening rates is interpreted as 26/100 (people) and thus 0.26 per person. The ICER is expressed as.: [Average Cost per Woman Screened_T+1 – Average Cost per Woman Screened_T]/[(Percent Women Screened_T+1 – Percent Women Screened_T)/100], where t refers to the period for FY 2018-2021 [16]. We assessed the ICER for each year of the TCO period, and we compared the second year of TCO, post-transition but not affected by the pandemic, to TOP.

3. Results

3.1. Urban Site Case Study

The urban CCL site is a large, inner-city hospital system with satellite clinics and a mobile van that enables screenings to be completed at community locations. CHWs focused outreach on minority women, particularly non-white women of Hispanic descent. CHWs primarily partnered with community centers and food pantries to conduct outreach, with over half of the activities occurring at these two locations in fiscal years 2019 and 2020. Other sites included health fairs and homeless shelters. (See Supplement 2) The COVID-19 pandemic significantly impacted outreach modalities, prompting the TCO to resume reviewing EHRs to identify women who had never been screened and those who were five years overdue for screening.

Table 1 presents data on race/ethnicity as self-reported by participants and the insurance status of women who received direct assistance from CHWs. The percentage of Black non-Hispanic women served varied over the three years, ranging from 15.1% to 21.4%, while Hispanic women constituted an increasingly larger share, rising from 35.8% in 2019 to 58.3% in 2021. The percent of white non-Hispanic women receiving direct assistance fell from 33.2% to 16.2% during the pandemic.

Initially, under TCO, 23.5% of women served were privately insured, with the percent declining each year to a low of 5.6% in 2021. The percent of women insured by Medicaid or Medicare rose initially from 2019 to 2020, then declined in 2021 to 44.6%, below the 2019 proportion (57.3%). The percentage of uninsured women remained steady at roughly 20% through 2019, but more than doubled in 2021 to 49.8%, the year most affected by COVID-19.

Table 2 presents information on reported barriers to screening and the types of direct CHW assistance provided. Many women reported multiple barriers. In 2019, unstable housing was the most commonly reported barrier, affecting nearly 40% of women, while financial issues, work requirements, and transportation were reported by 20% to 25% of women. This pattern was similar in 2020. A notable change was the growth in language and health literacy barriers, likely reflecting the increase in Hispanic women being served by this TCO. The share of respondents citing lack of insurance as a barrier increased from ~10% to nearly 30% between 2019 and 2021.

CHWs provided one-on-one education, appointment scheduling assistance, and appointment reminders, and successfully scheduled appointments for over 95% of women. The provision of translation services increased dramatically from 2019 to 2021, as did insurance support, due to the growing proportion of uninsured and Hispanic women served.

Table 3 presents costs by spending category. Overall, the percent spent on CHWs rose substantially from 34% under TOP to over 50% every year of the TCO. Funding for travel vouchers increased sixfold from 2018 to 2021, addressing a central barrier to screening access.

Table 4 shows the number of women receiving direct assistance from CHWs, the number of completed screens, and screening rates for both cancer types. Program reach increased from 204 women under TOP to 333 in 2019, reflecting, in part, continued EHR use to identify women already connected to the health care system as community outreach was phased in. Screening rates initially fell during the transition from TOP to TCO, given the large increase in women receiving direct assistance. As EHR identification was phased out, 30 fewer women received direct assistance, and screening rates rose by 5 pp. In the COVID-19 Year, screening rates increased to 76% as EHR identification of women resumed amid pandemic restrictions.

Table 5 summarizes trends in the average cost per woman screened, screening rates, and incremental cost-effectiveness ratios, which show the cost per additional woman screened as TCO evolved and compares the second year of TCO with TOP. During the transition year from TOP to TCO, the average cost per woman screened declined by $529 from $1,778 in 2018 to $1,249 in 2019, while the screening rate fell by 9 pp. In the second TCO year, the average cost per woman screened rose by $187and the screening rate rose by 5 pp, with fewer women receiving direct assistance, yielding an ICER of $3740, the cost to screen an additional woman as the program focused on the underserved. In 2021, the average cost per woman screened declined, while the screening rate rose, reflecting efficiencies gained from women returning to using the EHR during the COVID-19 year.

We compared TOP to the second year of TCO to allow for a TCO transition year and to avoid the pandemic effect on the program. Under TCO, the average cost per woman screened was lower than under TOP (-342$) due to budget reductions and a larger number of women screened. However, screening rates were 4% lower, reflecting, in part, use of CHWs and the focus on underserved women under the TCO.

3.2. Rural Site Case Study

The rural site serves residents in three frontier counties in western Colorado, where the population density is 6 or fewer people per square mile. Cervical cancer screenings can be performed on site, but women must generally travel significant distances to obtain mammography services. As an FQHC, the site accepts women with Medicaid and those without insurance; citizenship is generally not checked.

CHWs conducted a third of outreach activities at food banks, a percentage that grew over time. They continued limited media outreach for the first year, resulting in 22.4% of referrals (See Supplement 2). In TCO year two, 28.2% of women came to the TCO through personal referrals. With the onset of COVID-19, outreach shifted to community centers, and the rural CCL used its EHR only to identify women for possible screenings.

Over half of the women served in the rural site were white, non-Hispanic, with the percent rising from 51.9% to 92.3% in 2020, then falling to ~80% in 2021 (Table 1). The percent of Hispanic women receiving direct assistance varied from ~5% to ~13%. In contrast, no non-Hispanic Black women were served, reflecting the makeup of the surrounding population, where 0.4% of the population is Black [17].

In the first year of TCO, women insured by Medicaid, privately, or uninsured each constituted 31% of those receiving direct assistance, while the remainder were insured by Medicare (Table 1).

In 2020, the percent of women insured by Medicaid fell, while the percent of uninsured women increased to 41.0%. With the onset of COVID-19, the insurance composition changed: privately insured women made up the largest share of the total population at 38%, followed by women insured by Medicare, and the uninsured rate fell by half, likely reflecting a return to EHR-based identification for screening.

Rural women reported multiple barriers to accessing services (Table 2). Transportation was reported by 69.0% in the first year, closely followed by financial concerns at 67.2%. Over 40% of women reported barriers related to literacy, insurance, and a lack of support. In 2020, reported barriers remained consistent, with the percentage reporting financial challenges rising by 12 percentage points, likely reflecting the increase in uninsured women. Cultural beliefs (70%) and transportation (92%) became the primary barriers reported during 2021.

The TCO focused on one-on-one education by CHWs, which over 90% of participants received over the three years. Nearly three-quarters of women received assistance with appointment scheduling and reminders in 2019, with the percent rising to 90% and above in year 2. In the first two years, approximately one-fifth of participants received a referral for insurance counseling. Transportation vouchers were provided to 62.1% of participants in 2019 and 53.8% in 2020. The nature of CHW assistance changed in 2021, with fewer women (-20 pp) receiving assistance with appointment scheduling, while the number of women receiving travel vouchers increased to 92.0%.

3.3. Program Cost-Effectiveness

Table 3 shows that during the transition from TOP to TCO, program spending decreased by 24% ($94,781 to $71, 588), with most of the decrease attributable to reduced hours for a patient navigator who had assisted with EHR-identified outreach during TOP. Budget expenditures increased by 16.8% the following year to fund a CHW. The patient navigator position was eliminated in 2020 but was reinstated during the COVID-19 pandemic when EHR-identified outreach resumed on a limited basis.

As shown in Table 4, following the shift from broad-based outreach to TCO, the number of women receiving assistance declined precipitously, from 94 under TOP to 58 under the first year of TCO. The decline continued the following year, reflecting the challenge of connecting with ’hard-to-reach’ women. Numbers rebounded, to a degree, under COVID-19 as the site returned to EHR identification of women for outreach.

Table 5 provides cost-effectiveness results. From 2018 to 2019, the average cost per woman screened rose by $307, from $1,529 to $1,836, primarily because fewer women were screened while the screening rate remained constant. In 2020, program performance improved markedly, with average costs per woman declining by $155 to $ 1,681, while screening rates rose sharply to 95%, up 28 pp from 67%. However, with the onset of COVID-19, average costs per woman screened nearly doubled to $3,170, as screening rates dropped to 50%, a decrease of 45 pp. Comparing the second year of TCO to TOP, the average cost per woman screened increased by $152, while screening rates increased by 28 pp, for an ICER of $543.

4. Discussion

This study evaluated the implementation, average cost, and incremental cost-effectiveness of two Community Clinical Linkage (CCL) outreach models, one urban, one rural, designed to increase breast and cervical cancer screening rates among underserved women. The findings offer valuable insights into the contextual, structural, and programmatic factors that influence screening outcomes and the associated costs of cancer screening outreach to underserved women in diverse geographic settings.

Trends varied by site during the transition from TOP to TCO. The urban site experienced a sizable increase in the number of women receiving direct assistance and a reduction in the screening rate. Fewer women received direct assistance from the rural site, and the screening rate remained unchanged. Sites relinquished use of EHR-identification of women for screening at different speeds, with the process moving more slowly at the urban site. Both sites spent the first TCO year building community partnerships and relationships to connect hard-to-reach populations with screening services, a labor-intensive, complex effort that depended on the woman’s circumstances. In the second TCO year, FY 2020, screening rates increased substantially in both sites, to 5 and 28 pp in the urban and rural sites, respectively, as community connection grew and resources were focused on fewer women.

By the second year of operation, traditionally underserved women made up a larger percent of the populations being served. In the urban site, women were twice as likely to be uninsured and of Black or Hispanic race, and over three times more likely to be on Medicaid than women in the surrounding community [18,19]. The rural site made successful inroads in the uninsured population, as rural site participants were nearly three times more likely to be uninsured, and breast cancer screening rates were nearly a third higher than in the surrounding area [20].

The number of women receiving direct assistance varied over the period, as did the cost per woman screened. Changes in screening rates primarily drove changes in the average cost per woman screened. Over time, the cost-effectiveness of both programs varied with the stage of development, the level of direct assistance provided, the transition to or use of EHR-based identification of women for screening, and the percentage of underserved women in the outreach population. The data here underscores that cost-effectiveness assessments should be viewed in context and results over short timeframes should be viewed with caution.

Further, as the goal of the TCO was to focus on underserved women, a reasonable expectation is that the TCO, compared to the TOP, would be more labor-intensive and thus likely more expensive per woman; fewer women would be served, as they are relatively more difficult to identify, and screening rates would be lower. TCO is thus likely to be less cost-effective than TOP, although for the rural site, by TCO year 2, the ICER was within the range of other screening outreach programs [10].

Pandemic-related experiences differed significantly between the CCLs. The urban site adapted quickly and returned to leveraging the EHR system. In contrast, the rural TCO encountered persistent challenges related to geographic dispersion and travel barriers, including mountainous terrain, limited healthcare infrastructure, and the population’s socioeconomic characteristics. Rural programs may require enhanced infrastructure investments and tailored interventions to more directly address geographic and transportation challenges. The urban TCO’s evolving focus on Hispanic women and the resulting increase in translation and insurance support services point to the importance of cultural and linguistic tailoring in improving engagement and outcomes.

These site-specific differences highlight key considerations for future implementation and the scaling-up of CCL programs. The performance of programs is closely tied to local resources, staffing continuity, and the ability to adapt outreach strategies to changing circumstances and the period in which they are evaluated, related to implementation.

5. Conclusions

This study finds that Community Clinical Linkage (CCL) models have the potential to improve the screening rates among underserved women by providing intensive direct assistance and may become more cost-effective over time as community relationships grow. On the other hand, more resources may be required to reach increasingly hard-to-reach women. Overall, this evaluation reinforces that, while CCL models can improve access to screening for underserved women, their success depends on context-specific factors including organizational capacity, infrastructure, population characteristics, and responsiveness to external shocks. These findings can inform future funding decisions and program design strategies to promote equitable and cost-effective cancer screening across diverse settings.

5.1. Limitations

Our cost estimates reflect certain limitations. Values used in our cost calculations are derived from site invoices submitted to CDPHE. We do not know to what extent, if any, sites received subsidies from their health systems or centers; if so, costs may be underestimated. Historical data on the TOP program were limited. Thus, we do not have comparable information on TOP program costs over multiple years prior to TCO. We do not include participant costs at either site, even though travel time was clearly greater at the rural site.

Also, we could not separately evaluate the cost-effectiveness of breast and cervical cancer, which may have different ICER values, as we could not separate costs, and most outreach activities provided information on both screening types. We do not know what would have happened in the third year of implementation in the absence of the pandemic, but at a minimum, we can speculate that the rural site would likely have fared better than the results here show. The urban site may have continued to increase efficiency, given its increased leveraging of the health system’s mobile van.

As we evaluated two CCL models implemented in distinct sites in Colorado, we do not know to what degree our results are generalizable to other geographic settings. However, our results support the benefits of operating in a health system with greater resources, such as the urban site, compared to the lower-resourced rural setting. We also have no way to assess the quality of the community clinical linkages initiated by health providers or the range of partnerships available in each setting. We were also unable to investigate why women scheduled for screening appointments did not complete them.

5.2. Implications for Policy and Practice

• Cost-effectiveness evolves over program implementation stages rather than remaining static.
• Short-term increases in cost per woman screened may reflect the increased resources needed to reach underserved populations.
• Organizational infrastructure, resources, and adaptive capacity strongly influence program sustainability and ability to withstand negative external changes over time.
• Rural outreach programs face distinct structural challenges requiring tailored investments.
• Equity-focused interpretation of economic outcomes is essential when evaluating community-based outreach programs.