Including the method of detection for breast cancer in the Surveillance,Epidemiology,and End Results database is long overdue

Abstract

Debates about breast cancer screening have continued in part because the Surveillance, Epidemiology, and End Results database, which began in 1974, has never included the method of detection so that it has been impossible to determine the role that early detection has played in the major decline in deaths from breast cancer that we have seen in the US since 1990. Method of detection should be added to the Surveillance, Epidemiology, and End Results database as soon as possible.

Introduction

The Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute (NCI), the US national cancer database, began in 1974 to “collect complete and accurate data on all cancers diagnosed among residents of geographic areas covered by SEER cancer registries.” ¹ While including the size and stage of cancers along with histological and treatment information, and ultimately cancer death rates, SEER, nevertheless, lacks the fundamentally important information about how breast cancers are detected in the US. The collection of data on the method of detection (MOD) is long overdue.

Screening debate in the US

The randomized controlled trials (RCTs) of breast cancer screening proved that early detection saves lives for women aged 40–74. ² Despite the evidence for a mortality benefit beginning at age 40, in 2009 the United States Preventive Services Task Force (USPSTF) recommended that women delay routine screening until the age of 50. ³ Other national organizations, such as the American College of Radiology (ACR), citing RCT data, argued that screening should begin by the age of 40. After many years and debate, the USPSTF has recently upgraded their screening advice to include women starting at the age of 40, ⁴ but other groups continue to dispute this, and the conflicting recommendations continue to confuse women and providers.

Controversies about the onset, termination, and frequency of screening for breast cancer have persisted in the US for decades. A widely quoted paper in the New England Journal of Medicine, claiming to be based on SEER data, concluded that 70,000 breast cancers, diagnosed in 2008 alone, were not clinically relevant and represented massive “overdiagnosis” due to mammography screening. ⁵ The analysis and estimate of overdiagnosis due to screening failed to recognize a fundamental and critical fact: SEER data have never included the MOD for each patient and cannot provide any direct link between screening mammography and breast cancer diagnosis, or overdiagnosis, in the US. ⁶ These and other issues could have been resolved in a more timely fashion with individual data on MOD.

Correlation versus causation

Although women have attended screening across the US for decades, there is no centralized national screening program, and without MOD data we can only guess at participation and the consequences of participation. Observational incidence data from SEER suggest a correlation between increasing numbers of women at first attendance to screening and the start of a long prevalence peak in the mid-1980s. The SEER death rate, unchanged in the preceding 50 years, began to decline several years later, in 1990, and is now down by over 40%. These observations are correlative at best. Without collecting the single, specific, initial MOD for each patient with breast cancer in SEER, it is impossible for analysts to directly determine the cause and effect of what has driven the decline in deaths in the US. This uncertainty has produced unsubstantiated estimates, correlations without causation, and ongoing debates over the value of screening leading to conflicting recommendations from national healthcare organizations. This has caused uncertainty among women and their physicians and compromises the quality of healthcare. Despite the upgraded USPSTF screening guidelines to include women in their forties, we will have no way of directly assessing the effect of these expanded guidelines on screening attendance or patient outcomes without individual MOD data.

International models

The failure of the US health system to capture MOD in national databases suggests that the task is impossible. On the contrary, many nations and governments outside the US do just that. ⁷ Numerous centralized health service screening programs directly track and collect data on patient-specific screening and detection. The inclusion of these important and simple data allows other nations to ask and answer detailed questions about the direct impact of screening on many critical patient outcomes. Countries across Europe and Asia, as well as Canada, Australia, and New Zealand, with nationally organized screening programs routinely collect information on how cancers are detected. ⁸ A study of more than 500,000 women in Sweden has shown that the incidence of death has declined by more than 40%, mostly due to early detection. ⁹ Data from other nations, where MOD is included in national cancer registries, show that “overdiagnosis,” based on screening, is very low if it occurs at all. ¹⁰ Without national data, commentators in the US have been able to speculate freely on these issues and there is no way to know, with any certainty, why deaths from breast cancer have declined so dramatically.

MOD challenges and solutions

The fragmented healthcare system and lack of a national screening program have prevented contemporary and recurring assessment of the complete impact of screening and also pose the major challenge to including MOD in SEER and other US databases. However, the ACR, in partnership with the North American Association of Central Cancer Registries (NAACCR), the National Mammography Database (NMD), and the Breast Imaging Reporting and Data System (BI-RADS) committee of the ACR, have standardized and adopted a list of MOD categories. An active multi-site pilot program, designed by the ACR Screening and Emerging Technology Committee, asks radiologists to add MOD to diagnostic reports for registrars to find and abstract into cancer databases. Pilot data, presented at the annual meetings of NAACCR 2022, Society of Breast Imaging 2023, and National Association of Cancer Registries 2023, indicate that radiologists assign MOD in 92% of eligible cases, the assigned MOD is correct in 94% of cases and the MOD reaches the registry in 90% of cases.^11–13 The ACR has submitted a formal request to NAACCR to add MOD as an official data item for all registries.

Conclusion

The collection of the, fundamentally important, MOD for breast cancer is long overdue. The MOD should be added to the SEER database as soon as possible to clarify the value of early detection. The ACR has already developed methods for assigning the MOD in Breast Imaging Reporting, and trials at multiple sites have shown that this is practical and can facilitate the uptake of the information by tumor registries. As the half-century anniversary of SEER approaches, it is time for NCI to include this critical information in SEER.