Since its establishment, NPCR has played a lead role in advancing optimal approaches and standards to improve cancer registry data collection through efforts, such as the Electronic Pathology (ePath) Implementation Project, by which CDC has facilitated electronic reporting of anatomic pathology reports from national and regional laboratories to central cancer registries (Jones, Alimi, Pordell, Tangka, Blumenthal, et al., 2021; Pollack, Jones, Blumenthal, Alimi, Jones, et al., 2020). Accurate, timely cancer incidence data are critical to informing local and national cancer prevention and control efforts (Weir, Thompson, Soman, Møller, Leadbetter, 2015; White, Babcock, Hayes, Mariotto, Wong, et al., 2017). Innovative data transfer and processing approaches could increase timeliness of data collection and reduce the labor resources required to process manually collected data. As part of CDC’s broader Data Modernization Initiative (DMI), NPCR is focused on adoption of feasible data modernization approaches, such as electronic reporting and automation (CDC’s DMI, 2023; NPCR Data Modernization, 2023). As part of this effort, CDC is testing its Cancer Surveillance Cloud-based Computing Platform (CSCBCP) with selected central cancer registries, intending to make it more widely available to NPCR-funded registries in 2024 (NPCR CSCBCP, 2023).

Although prior studies have systematically evaluated the economics of cancer registries, these studies did not specifically assess innovative data transfer and processing approaches and their impact on cancer registry costs and resource allocation (Cole Beebe, Subramanian, Tangka, Weir, Babcock, et al., 2018; Subramanian, Tangka, Cole Beebe, Trebino, Weir, et al., 2016; Tangka, Subramanian, Cole Beebe, Weir, Trebino, et al., 2015). Formative research has provided insight into barriers and facilitators to adoption of electronic data reporting but has not assessed how resource use may vary across registries (Tangka, Edwards, Pordell, Wilson, Blumenthal, et al., 2021). This information is required to inform data collection and identify better practices, which may provide key resource use metrics to monitor implementation. These metrics are especially important as innovative approaches become more established and are adopted into the general core operations of cancer registries.

CDC initiated the current study in 2019 to assess the resource requirements of the process steps and activities, which led to the efficient collection of high-quality cancer incidence data, as electronic reporting evolves. This article provides a detailed overview of the methodological approach and shares findings on population-based cancer registry activity-based costs, which were explored by using prospective and retrospective data collection approaches. The information provided will support CDC in evaluating the efficiency of cancer registry operations and planning the implementation of data modernization approaches, such as electronic reporting and automation. Other stakeholders, such as the cancer registries, evaluators of similar programs or those operationalizing them, and those establishing new programs, may also benefit from the findings of this study. For example, findings from prior evaluations have informed efforts by the World Health Organization to support the establishment and growth of registries in low- and middle-income countries (de Vries, Pardo, Arias, Bravo, Navarro, et al., 2016).

We conducted a comprehensive evaluation of the cancer registry activities and associated resource requirements for 21 of the 46 state-based registries in NPCR. Registries completed data collection from October 2021 to September 2022 and reported retrospective costing data for the period July 1, 2020, through June 30, 2021. The participating registries were purposively selected to ensure a balanced representation of registry attributes across several dimensions, including the following: case volume; region; size of area served; rurality; sources of additional (non-NPCR) funding, such as state or National Cancer Institute (NCI) Surveillance, Epidemiology, and End Results (SEER) Program funding; and data quality attainment. Participating registries reported on resource use in terms of labor (from time spent on each activity by staff member) and other costs associated with registry activities.

Among our sample, we identified registry locations based on the four US census regions (Midwest, Northeast, South, and West). We defined registries based on one of three case volume categories: low volume (registries reporting fewer than 10,455 cases per year), medium volume (registries reporting 10,455 to 26,558 cases per year), or high volume (registries reporting more than 26,558 cases per year). Registries were identified as fitting into one of three sizes of area served: small (less than 44,453 square miles), medium (44,453 to 70,684 square miles), or large (more than 70,684 square miles). We identified registries as fitting into one of three categories that are based on the presence of rural areas within the area served: low (more than 147 residents per square mile), medium (58 to 147 residents per square mile), or high (fewer than 58 residents per square mile). We determined that registries were in one of three quality attainment categories by using the most recently available data at the start of the study: always (registries that met the 12-month standard for all years [2014–2017]), sometimes (registries that met the 12-month standard at least once in 2014–2017 and met 24-month standards for all years [2014–2017]), or seldom/never (registries that did not meet 24-month standards for all years [2014–2017]).

Using multiple modalities, we collected activity and costing data, which facilitated a comprehensive understanding of resource use for each registry. Prior studies have found that registry engagement is critical to ensuring the collection of high-quality data (Cole Beebe, Subramanian, Tangka, Weir, Babcock, and Trebino, 2018). To that end, we engaged participating registries early in the research process and throughout to ensure that data collection tools were appropriately tailored to each registry while retaining sufficient standardization to allow for meaningful comparisons across registries.

Building on prior studies, we refined our study approach, developed a draft list of registry activities, and developed draft data collection tools (Cole Beebe, Subramanian, Tangka, Weir, Babcock, et al., 2018; Subramanian, Tangka, Cole Beebe, Trebino, Weir, et al., 2016; Tangka, Subramanian, Cole Beebe, Weir, Trebino, et al., 2015). Figure 1 provides a high-level illustration of our approach to engaging registries and collecting data. Our data collection modalities included registry interviews, prospective staffing surveys, and retrospective costing data collection. We held conference calls with each registry to provide an overview of the study goals and methodology and a draft activities list. These calls also included interviews during which registries shared insights about their data collection processes, data sources, as well as barriers and facilitators of quality case reporting. As a next step, we collaborated iteratively with each registry to finalize a tailored list of registry activities. Using these lists, our team refined both data collection instruments (i.e., prospective staffing survey and retrospective costing tool).

This study is unique because the timing of data collection overlapped with the COVID-19 public health emergency (PHE). Our approach to collaborating with registries during study design and throughout data collection allowed us to identify early a new activity category related to COVID-19 research and support activities.

We reviewed each data submission and worked iteratively with registry staff to update or clarify data reported, as necessary. Our goal was to capture data that accurately reflected the operations of each individual registry while maintaining the ability to roll data up to standardized categories to ensure meaningful cross-registry comparisons. Registries varied in their use of the open comment fields within the data collection tools, which often provided insight that enriched our understanding of registry operations and barriers or facilitators to high-quality cancer data collection.

After finalizing each registry’s data submissions, we analyzed data for each prospective staffing snapshot and compared resource allocation with the retrospective costing data. We shared summarized and registry-specific data with each participant for their review. Given the cyclical nature of some registry activities and the variation in timing of snapshots across registries, we performed a quarterly analysis based on timing of each snapshot submission. We did not find substantial variation in activity patterns across quarters and therefore opted to analyze prospective staffing data collectively and report aggregate results. Building on findings from prior analyses showing that case volume is a major driver of registry cost, we present total registry funding and activity-based costs by case volume (Tangka, Subramanian, Cole Beebe, Weir, Trebino, et al., 2015).

Case volume continues to drive registry spending overall and among registry activities associated with variable costs (typically labor-intensive activities), suggesting opportunities for cost savings with the adoption of data modernization approaches. Manual data collection and validation costs vary with case volume. While all registries receive at least some data electronically, variation exists in the quality of data reported electronically. Increasing availability and accessibility of standardized reporting and common data standards are necessary. Advances in automation of data collection and validation through electronic reporting and data checks and moving to the cloud will reduce the impact of case volume on costs. Our findings also indicate that retrospective costing data collection using a tailored approach can yield accurate information because we were able to validate our retrospective data collection with prospective data obtained for selected periods. This finding will inform future studies, which may reduce burden for study participants if prospective data collection is eliminated.

Active registry engagement is a critical component of priority NPCR evaluation activities. This study has benefitted from input from registry staff throughout the process. Many contextual factors, such as quality of reporting sources and historic processes, impact registry operations, and feedback from registry staff is essential to understanding the impact of these factors on activity-based cost. Any future work in this space, or similar work with other disease registries or related entities, should consider incorporating feedback and knowledge sharing from the registries or similar entities.

The analysis presented here is based on a subset of central cancer registries that were selected to capture a variety of registry characteristics and does not consider potential interactions among characteristics. The small sample size is more sensitive to outliers, which can shift calculated averages. Data reported prospectively represented 2-week snapshots, which may result in some cyclical registry activities not being captured in any prospective staffing survey. Retrospective costing data were reported for a prior year-long period and are subject to recall bias. We designed and planned this study before the COVID-19 PHE, which was in effect during the data collection period. Many registries had to shift staffing resources to support COVID-19 response, potentially disrupting normal registry operations. Without registry input and feedback, the impact of the unprecedented COVID-19 PHE on operations of some registries may have been underestimated. Our study results may not be illustrative of non-PHE times. Further, we have yet to understand the persistent effects of the pandemic on how we will work going forward, so although the data may not be entirely reflective of historical (pre-PHE) experience, that experience may also look different from post-PHE experience.