As the largest cancer-screening program in the United States, the National Breast and Cervical Cancer Early Detection Program (NBCCEDP) serves as a learning laboratory for other national cancer and non-cancer prevention and control programs. The NBCCEDP’s organizational structure is complex; CDC funds states, tribes/tribal organizations, and US territories across the US and each grantee’s program has established a unique screening delivery system to serve their population of eligible women. Each of these programs uses one of the following three different service delivery mechanisms in their respective jurisdictions: (1) centralized, (2) decentralized, and (3) mixed [1]. Centralized programs offer clinical services in satellite offices, which may include local health departments that are not independent entities from the state health departments, but all other program activities are performed centrally (e.g., tracking and case management service). Decentralized programs, on the other hand, contract with local and regional health departments, primary care clinics (e.g., community health centers), private hospitals, or other healthcare facilities to fully manage and provide screening and diagnostic services and other program activities.

Programs with mixed service delivery structures have both centralized and decentralized features [2.] There is a great deal of variation among states with a mixed delivery structures and this variation depends upon the needs and capacity at the health department. Various activities of the program may be managed at the state health department while other activities are managed at district level or even in the providers’ offices. For example, some state health departments do their program enrollment through the provider offices, but have patient navigation and the billing controlled centrally at the state health department. Other programs contract with providers directly for clinical services such as Federally Qualified Health Centers.

NBCCEDP grantees selected their program structures for specific reasons and the delivery structure was not arbitrary or randomly assigned. The delivery structure for each grantee is based on the infrastructure of their health department system. States that function through health districts were more likely to have a decentralized program whereas states with internal infrastructure may centralize the entire program in the state health department.

In recent years, attention has been given to the estimated costs and effects of these delivery structures for cancer screening, diagnostic follow-up, and other program services. Because program resources reach only a fraction of eligible women, all other things equal, service delivery mechanisms that provide efficient allocation of resources are clearly desirable. Such service delivery mechanisms should enable programs to screen and serve optimal numbers of eligible populations. Currently, there is no information describing how the three service delivery structures used by NBCCEDP grantees impact costs and effectiveness. In this study, we assessed the costs and effectiveness of various delivery structures used to deliver screening, diagnostic, and other program services in the NBCCEDP and examined how different types of service delivery structures affect the number of women served.

Across the 51 state programs (including Washington DC), ten were centralized, 17 were decentralized, and 24 were mixed (Fig. 1). Figure 2 reports median costs per woman served for each activity by program structure. Screening and diagnostic services had the highest median costs per woman served (from $131.84 for mixed programs to $152.57 for centralized programs). The next highest median costs per woman served was for program management (from $14.98 for mixed programs to $32.99 for decentralized programs) and patient support/case management (from $10.88 for centralized programs to $30.71 decentralized programs). Differences in median costs per woman served were largest for these activities and were significantly different from screening and diagnostic services (p = 0.012) and program management (p = 0.012). Decentralized programs spent the most on program management. Mixed programs’ median screening and diagnostic service costs per woman served were approximately $20 lower than centralized and decentralized programs.

Table 1 presents the results of the pseudo-incremental cost-effectiveness analysis. Centralized programs are reported first because they served the smallest number of women, followed by the decentralized programs, and then the mixed programs. Compared with a base of centralized programs, decentralized programs incurred costs of $281 for each additional woman served for breast cancer and $284 for each additional woman served for cervical cancer. Compared with decentralized programs, mixed programs added an additional $109 for each additional woman served for breast cancer but saved $1,777 for each additional woman served for cervical cancer. In both sensitivity analyses (excluding the 2007–2008 program-years and including the program-years in which total costs differed by more than 10% from adjusted annual funding) the results for average and incremental cost effectiveness were nearly identical to those reported Table 1, with estimates within a few dollars of the original analytic sample (available upon request).

Our results indicate that program delivery structure was associated with the cost and effectiveness of services provided by NBCCEDP grantee programs. Mixed program structures were the most common among NBCCEDP grantees and they had the lowest screening and diagnostic costs per woman served. This is important, as programs were required to spend at least 60% of their funds on these activities during the time of this study. Screening and diagnostic costs were approximately five times higher than costs per woman served for the next highest cost activity (program management).

Programs with mixed structures served the most number of women annually. For breast cancer screening services, these program s also had the highest total costs. However, for cervical cancer screening services, programs with mixed structures actually served more women than decentralized programs at a lower cost.

These results could reflect the fact that a mixed program structure allows programs to specialize on specific activities for which they are the most efficient. M any mixed programs do not have local health departments that can deliver screening services. Thus, the most common activities to be outsourced are direct screening and follow-up services and billing support. There is heterogeneity across programs in the decision on which activities to keep within the program and which to decentralize.

We excluded tribes and territories from our analysis. Previous cost analyses of the NBCCEDP have also excluded cost data from tribes and territories for several reasons [1, 2, 4]. First, there are significant demographic and cultural differences between states, tribes, and territories [5]. Second, organizational factors that impact delivery of screening services are different among tribe and territory grantees of the NBCCEDP as compared to the states [6]. Lastly, including data from smaller tribe and territory grantees created instability in national estimates.

Our results give some understanding of the relative costs of these program structures. However, we have a small number of programs for some structures (e.g., only ten centralized program s). Furthermore, costs and the number of women served depend on many factors not included in this analysis. These include the population of eligible women in the catchment area, availability and costs of clinical and program staff, and other characteristics specific to the program ‘s target population [2, 7–9].

In addition, NBCCEDP grantees selected their program structures for specific reasons, and the delivery structure was not arbitrary or randomly assigned. The choice of delivery structure may be associated with other factors that affect cost effectiveness, creating confounding of the measured association. For example, the NBCCEDP has been shown to exhibit economies of scale, in which the average cost per woman served decreases as the number of women served increases [10]. If larger programs tend to choose a mixed delivery structure, this could lead to lower costs per woman served. Therefore, caution should be used in interpreting the results of this study.

A key decision for disease prevention and control programs is how to structure the delivery of health services. In the NBCCEDP, delivery structure was associated with average and incremental cost effectiveness of the state programs. In particular, a mixed delivery structure was associated with lower costs per woman served. The results are suggestive that a mixed structure, in which programs perform activities in which they have a comparative advantage and outsource all other activities, may be a promising approach to improve the efficiency of the programs. Programs would need to ensure that such a structure is feasible and appropriate given their own context.