National claims data analysis of breast cancer diagnosis and treatment before versus during the COVID-19 pandemic

ABSTRACT

Background: The COVID-19 pandemic imposed great burden on the healthcare system and has required patients and their physicians to make unprecedented choices about cancer care. Hospital-based retrospective reviews have suggested changes in breast cancer management during 2020 compared to previous years, including greater use of preoperative therapy. We used insurance claims data to understand the impact of the pandemic on breast cancer diagnosis and treatment at a national level. Methods: We identified new diagnoses of breast cancer from 2017-2020 in the Optum Clinformatics claims data set, consisting of claims records linked to electronic health records. The overall population (enrolled in Optum for at least 6 months with at least one diagnosis of any condition and no prior breast cancer diagnosis) included an average of 8 million adult Americans per year. A new breast cancer diagnosis was defined as a first-ever ICD code for breast cancer with a breast diagnostic biopsy procedure code (considered the cancer diagnosis date) within 6 months before to 3 months after that ICD code. Each year's cohort of breast cancer cases was limited to those diagnosed between February 1 and May 30, with follow-up through June 30 of the diagnosis year. First treatment after diagnosis was classified as either endocrine therapy, chemotherapy, or surgery. Geographic area was defined by the 9 Census Bureau regions. We used a Poisson regression to compare the rate of breast cancer diagnosis in 2020 compared to 2017-2019 and a Chi-squared test to compare the distribution of first treatment in 2020 compared to 2017-2019. To investigate differences in the impact of the pandemic on rate of diagnosis (Poissonregression) or use of preoperative therapy (logistic regression) by race/ethnicity, income, or geographic area, we included each of these covariates as well as its interaction with year (2020 vs 2017-2019) in separate models. Results: There were 2, 841 breast cancer diagnoses February-May 2020 (0.037% of overall population), compared to 3, 880 in 2019 (0.045%), 3, 509 in 2018 (0.043%), and 2, 999 in 2017 (0.041%). In 2020 compared to 2017-2019, new breast cancer diagnoses decreased by 12.3% (95% CI 8.6%-15.9%;p < 0.0001). No significant differences were observed in this reduction in diagnoses by race/ethnicity, income level, or geographic area. Median date of diagnosis was earlier in 2020 (March 11) compared to 2017-2019 (March 29, April 1, and April 1 respectively), a result of a larger drop in diagnoses later in the time interval in 2020. Among patients who received treatment during follow-up (83.1% in 2017-2019 vs 86.2% in 2020, a difference likely reflecting this shift in diagnosis date), there was a marked reduction in surgery as first treatment in 2020 compared to previous years (88.7% in 2017-2019 vs 69.3% in 2020), while both preoperative chemotherapy (6.1% in 2017-2019 vs 10.7% in 2020) and preoperative endocrine therapy (5.2% in 2017-2019 vs 20.1% in 2020) increased (p < 0.0001). There were no differences in the shift toward preoperative therapy by race/ethnicity or income, but there was a significant difference by geographic area (p=0.0003) the Mountain region had least change in use of preoperative therapy (odds ratio 2.46 [95% CI 1.75-3.47] of preoperative therapy during vs before the pandmic) while the Middle Atlantic region had the greatest (odds ratio 5.64 [95% CI 3.79-8.38]). Conclusions: Among insured U.S. patients, new breast cancer diagnoses decreased by 12.3% during February-May 2020 compared to the same period in the previous three years, and use of preoperative therapy, largely endocrine, increased by 2.7-fold. The impact of the pandemic on choice of first treatment differed by geographic area, but not by race/ethnicity or income in this insured population. We will monitor with continued follow-up of claims data to assess the longer-term impact of these pandemic-related changes on treatment patterns, cost, and patient outcomes.