PSA indicates prostate-specific antigen.
eTable 1. ICD-10, CPT, and HCPCS codes used in this study
eTable 2. Characteristics of enrollees eligible for screening in each month (January – July)
eTable 3. Screening rates per 100,000 enrollees by month
eTable 4. Screening rate change (2020 vs 2019, % change and 95% CI) per month by geographic region
eTable 5. Screening rates per 100,000 enrollees and % change (2020 vs 2019) per month by socioeconomic status index quartile
eTable 6. Multivariable modeling examining factors associated with cancer screening in 2020
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Chen RC, Haynes K, Du S, Barron J, Katz AJ. Association of Cancer Screening Deficit in the United States With the COVID-19 Pandemic. JAMA Oncol. 2021;7(6):878–884. doi:10.1001/jamaoncol.2021.0884
What was the association of the COVID-19 pandemic with cancer screening rates across the US?
This cohort study found that with sharp declines and subsequent recoveries of breast, colorectal, and prostate cancer monthly screening rates in 2020, there remained an estimated screening deficit of 9.4 million associated with the COVID-19 pandemic for the US population. Screening declines differed by geographic region and socioeconomic status index, and use of telehealth was associated with higher screening rates.
Public health efforts are needed to make up the large cancer screening deficit associated with the COVID-19 pandemic.
The COVID-19 pandemic led to sharp declines in cancer screening. However, the total deficit in screening in the US associated with the pandemic and the differential impact on individuals in different geographic regions and by socioeconomic status (SES) index have yet to be fully characterized.
To quantify the screening rates for breast, colorectal, and prostate cancers associated with the COVID-19 pandemic in different geographic regions and for individuals in different SES index quartiles and estimate the overall cancer screening deficit in 2020 across the US population.
Design, Setting, and Participants
This retrospective cohort study uses the HealthCore Integrated Research Database, which comprises single-payer administrative claims data and enrollment information covering approximately 60 million people in Medicare Advantage and commercial health plans from across geographically diverse regions of the US. Participants were individuals in the database in January through July of 2018, 2019, and 2020 without diagnosis of the cancer of interest prior to the analytic index month.
Analytic index month and year.
Main Outcomes and Measures
Receipt of breast, colorectal, or prostate cancer screening.
Screening for all 3 cancers declined sharply in March through May of 2020 compared with 2019, with the sharpest decline in April (breast, −90.8%; colorectal, −79.3%; prostate, −63.4%) and near complete recovery of monthly screening rates by July for breast and prostate cancers. The absolute deficit across the US population in screening associated with the COVID-19 pandemic was estimated to be 3.9 million (breast), 3.8 million (colorectal), and 1.6 million (prostate). Geographic differences were observed: the Northeast experienced the sharpest declines in screening, while the West had a slower recovery compared with the Midwest and South. For example, percentage change in breast cancer screening rate (2020 vs 2019) for the month of April ranged from −87.3% (95% CI, −87.9% to −86.7%) in the West to −94.5% (95% CI, −94.9% to −94.1%) in the Northeast (decline). For the month of July, it ranged from −0.3% (95% CI, −2.1% to 1.5%) in the Midwest to −10.6% (−12.6% to −8.4%) in the West (recovery). By SES, the largest screening decline was observed in individuals in the highest SES index quartile, leading to a narrowing in the disparity in cancer screening by SES in 2020. For example, prostate cancer screening rates per 100 000 enrollees for individuals in the lowest and highest SES index quartiles, respectively, were 3525 (95% CI, 3444 to 3607) and 4329 (95% CI, 4271 to 4386) in April 2019 compared with 1535 (95% CI, 1480 to 1589) and 1338 (95% CI, 1306 to 1370) in April 2020. Multivariable analysis showed that telehealth use was associated with higher cancer screening.
Conclusions and Relevance
Public health efforts are needed to address the large cancer screening deficit associated with the COVID-19 pandemic, including increased use of screening modalities that do not require a procedure.
Across the US, the COVID-19 pandemic influenced health care rapidly and dramatically in 2020. As cases rose across the country, many states began to implement stay-at-home orders. In addition, the Centers for Disease Control and Prevention started to issue guidance to help individuals reduce exposure, and particular attention was directed toward older adults and those with comorbidities, who are at higher risk for developing severe complications from COVID-19. As a result, hospitals and clinics across the country dramatically reduced nonemergency clinical appointments. These changes likely had a direct and negative influence on cancer screening.1-5
Delays in cancer screenings owing to COVID-19 are suspected to lead to additional excess deaths that are directly attributable to the pandemic.6,7 The goal of the current study, using data from a large national commercial insurance carrier that provides coverage for both Medicare-aged and non-Medicare–aged individuals, is to assess the decline and subsequent recovery of monthly breast, colorectal, and prostate cancer screening rates. We also describe differences by geographic region in the US, as well as potential differential association of COVID-19 with cancer screening for individuals in different socioeconomic status (SES) index quartiles. We further compare screening rates in 2020 with 2019 and estimate a total deficit in cancer screening associated with the COVID-19 pandemic for the entire US population.
The HealthCore Integrated Research Database (HIRD) comprises single-payer administrative claims data and enrollment information covering approximately 60 million people in Medicare Advantage and commercial health plans from across geographically diverse regions of the US.8-10 The database tracks patient health plan enrollment, medical care, and prescription drug utilization data, making it a useful research resource that can be leveraged to identify current use of health care services, including cancer screening and telehealth. The University of Kansas Medical Center Institutional Review Board deemed the study exempt from review and patient informed consent requirements because deidentified data were used and the study did not involve human subjects research.
From the HIRD, we selected 3 distinct cohorts of enrollees to evaluate monthly screening rates for breast cancer (women only), colorectal cancer (men and women), and prostate cancer (men only). For each year (2018-2020) and month (January-July) in which we evaluated cancer screening, we first identified enrollees who were aged 50 to 79 years (for breast and colorectal cancer screening) or 50 to 69 years (for prostate cancer screening) who had at least 2 years of continuous enrollment prior to the first date of the month of interest (index month) and no medical claims with International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis or personal history codes for the cancer of interest (eTable 1 in the Supplement) during the 2 years preceding the index month. Excluding individuals with a history of the cancer of interest allowed us to more accurately assess cancer screening. The age ranges of the included enrollees for the current analysis are commonly considered to be eligible for screening.11-14 Enrollees meeting these criteria were eligible for inclusion in the evaluation of cancer screening for that index month.
The primary study end point was receipt of cancer screening, defined as the presence of 1 or more claims during the index month with an ICD-10-CM, Current Procedural Terminology, or Healthcare Common Procedure Coding System code specific for the type of cancer screening of interest (breast, colorectal, or prostate; see eTable 1 in the Supplement). Descriptive analyses were used to summarize the characteristics of analyzed cohorts and to calculate monthly rates for each type of cancer screening (per 100 000 enrollees). For each month analyzed, screening rates were compared between 2020 vs 2019 and described as a percentage change. The percentage change in monthly screening rates between 2020 vs 2019 were further analyzed by geographic region, because of differential timing and influence of COVID-19 on different regions, and by SES index quartile, to examine if the association of COVID-19 with cancer screening differed by SES indicators.
We then used these results to estimate the total deficit in cancer screening in 2020 for the entire US population. To accomplish this, we first estimated the age group–specific (50-59, 60-64, 65-69, and 70-79 years) US population eligible for screening each month. This involved calculating the prevalence of each cancer in each age group using HIRD data, taking 1 − prevalence to achieve the proportion of each group without the specific cancer, and therefore eligible for screening, and multiplying this proportion by the US population in the age group. We then multiplied 2019 screening rates and 2020 screening rates by the population eligible to approximate the number of screenings in each month for each year, took the difference between 2019 and 2020, and summed the differences across January through July for each age group. Finally, we summed January through July differences across age groups to get total deficits in screening for each cancer type for 2020 (compared with 2019).
For each month (January-July) in 2020, multivariable Poisson regression models were performed using a binary indicator of cancer screening as the dependent variable to identify associations between enrollee characteristics and risk of cancer screening. Separate models were constructed for each type of cancer screening. Covariates included in the regression models included age group, sex (for colorectal cancer screening only), geographic region, SES index quartile, Charlson comorbidity score,15 and the use of telehealth in the same month. The SES index score was calculated using 7 area-level social determinants of health variables, as developed by the Agency for Healthcare Research and Quality,16 and analyzed as quartiles. Use of telehealth was defined as the presence of 1 or more claims in the index month with 1 or more Current Procedural Terminology or Healthcare Common Procedure Coding System codes for telehealth services (eTable 1 in the Supplement). All analyses were conducted using SAS, version 9.4 (SAS Institute) and Excel, version 16.0 (Microsoft Corporation).
Characteristics of analyzed individuals are summarized in eTable 2 in the Supplement. Approximately 20% of individuals were in Medicare plans and 80% were in commercial plans. The 4 regions of the country were well represented.
While monthly screening rates were similar between 2018 and 2019 (Figure 1), breast, colorectal, and prostate cancer screening in 2020 decreased sharply in March through May compared with the same months in 2019, with near complete recovery of monthly screening rates by July for breast and prostate cancers (Figure 1; eTable 3 in the Supplement). Colorectal cancer screening rates remained 13.1% lower in July 2020 compared with 2019. The month of April 2020 had the sharpest drop in screening rates, by 90.8% for breast cancer (decline from 4287 per 100 000 [95% CI, 4258 to 4316] to 394 per 100 000 [95% CI, 385 to 403]), 79.3% for colorectal cancer (decline from 2073 per 100 000 [95% CI, 2058 to 2087] to 430 per 100 000 [95% CI, 423 to 436]), and 63.4% for prostate cancer (decline from 4025 per 100 000 [95% CI, 3993 to 4057] to 1474 per 100 000 [95% CI, 1454 to 1493]), compared with the same month in 2019.
Despite recovery in monthly screening rates, there remained a deficit in total screening from January through July 2020 compared with 2019. This translated to an estimated absolute deficit across the entire US population in screening for 3.9 million women (breast), 3.8 million men and women (colorectal), and 1.6 million men (prostate) for a total deficit of 9.4 million in 2020 compared with 2019 (Table).
The magnitudes of decline and subsequent recovery in cancer screening differed by geographic region (Figure 2; eTable 4 in the Supplement). For all 3 cancers, the Northeast region experienced the sharpest decline in March, April, and May; on the other hand, recovery in screening during June and July was slower for the West compared with the Midwest and South.
Figure 3 and eTable 5 in the Supplement show screening rates by SES index quartiles. For all cancers before the pandemic (in 2019 and January/February 2020), screening rates were higher for individuals in the highest 2 SES index quartiles and lower for individuals in the lowest 2 SES index quartiles. When we compared the average monthly screening rates from the 7-month period of January through July of 2020 vs the same period in 2019, we found that the decline in screening for all 3 cancers in 2020 was largest in magnitude for individuals in the highest SES quartile (eTable 5 in the Supplement, January-July column). This had the effect of narrowing the gap in screening across the 4 SES quartiles in 2020 compared with 2019.
Multivariable modeling examining covariates associated with screening during each month of 2020 is summarized in eTable 6 in the Supplement. For each cancer, increasing age was associated with less screening. In April, for breast and colorectal cancer screening, individuals in higher SES index quartiles actually had lower screening rates than those in lower SES index quartiles. An additional finding was that individuals who used telehealth had higher rates of screening; this finding was consistent across all 3 cancers.
In this analysis of administrative claims data that includes approximately 60 million people in Medicare Advantage and commercial health plans, we found an association of sharp declines in breast, colorectal, and prostate cancer screening rates with the COVID-19 pandemic, especially during March through May 2020, with near complete recovery of monthly screening rates by July. Our results are consistent with recent reports,1,2,4,5,17 while revealing additional insights.
First, to our best knowledge, this is the first study to demonstrate that declines in screening differed by geographic region. For all 3 cancers studied, the Northeast region experienced the biggest declines in March through May of 2020 compared with the same period in 2019, while the South and West regions experienced lower magnitudes of decline. This corresponds to the differential timing of COVID-19 rates across different regions of the country, with New York and other Northeastern states reporting early and higher surges of COVID-19 cases.
Second, we quantified cancer screening rates by SES index. As expected, prior to the COVID-19 pandemic, there was a measurable disparity in cancer screening rates. For all 3 cancers, screening from January through July 2019 was lower for individuals in the lower 2 SES index quartiles and higher for individuals in the higher 2 SES index quartiles. However, our data suggest that the COVID-19 pandemic had a bigger negative association with cancer screening for individuals in the highest SES index quartile. Specifically, for January through July of 2020 as a whole compared with the same 7-month period in 2019, the declines in cancer screening for the highest SES quartile vs lowest SES quartile were as follows: breast cancer (−30.0% vs −26.1%), colorectal cancer (−30.2% vs −25.6%), and prostate cancer (−18.9% vs −11.5%). This had the effect of closing the gap in cancer screening by SES index. For January through July 2020, the monthly average of cancer screening rates per 100 000 enrollees was nearly identical for individuals in the highest vs lowest SES index quartile for 2 of the 3 studied cancers: breast (2916 vs 2922) and colorectal (1432 vs 1419).
Third, we estimated the deficit in cancer screening for the US population associated with the COVID-19 pandemic. A recent editorial7 from the National Cancer Institute director clearly defined a concern for increased cancer mortality caused by delayed screening. To inform modeling studies that estimate the magnitude of this mortality impact and to inform health care policies to mitigate this potential increased mortality, accurate estimates on the association of COVID-19 with screening for specific cancers, and in specific populations, are needed. This study provides data to inform these efforts. The current study improves on earlier publications by including a large US population comprising both Medicare-aged and non-Medicare–aged individuals, excluding individuals with prevalent cancers, and analyzing data through the months of screening decline and subsequent recovery. We also restricted the analysis to individuals in age groups that are appropriate for cancer screening. Our results further revealed that the screening deficit was larger for breast and colorectal cancers, which require procedures (commonly, mammography and colonoscopy), compared with prostate cancer screening, which commonly involves a blood test (ie, prostate-specific antigen testing). However, further study is needed to ascertain the reasons for the different screening deficits. While monthly breast and prostate cancer screening rates recovered fully by July 2020, colorectal cancer screening remained 13% below 2019 rates. From a public health perspective, increasing the use of alternatives to colonoscopy for colorectal cancer screening may help mitigate this screening deficit.
Fourth, we found a novel association between patient use of telehealth and receipt of cancer screening even after adjusting for other covariates, including SES index score. This suggests a potential benefit of telehealth: while the COVID-19 pandemic led to broad cancellations of nonemergency in-person health care appointments, telehealth appointments might have enabled individuals to still receive medical consultation and advice, as well as a plan for rescheduling screening tests.
There are several limitations to this study, including analysis of only insured individuals, which may skew our population-level estimation of cancer screening deficit associated with the COVID-19 pandemic. Specifically, the analysis may not be representative of underinsured and/or publicly insured individuals and could underestimate the association of SES. Another limitation is the lack of race/ethnicity information for analysis. In addition, while many of the codes used in the analysis were specific for screening tests, some codes did not distinguish between screening vs a test for another clinical indication. Our inclusion of individuals without a history of the cancer of interest partially mitigated the limitation of inaccurately counting a test performed for nonscreening purposes.
Breast, colorectal, and prostate cancer screening declined sharply in March through May 2020 throughout the US, but most dramatically in the Northeast region and for individuals in the highest SES index quartile. The estimated screening deficit associated with the COVID-19 pandemic across these 3 cancers for the US population was 9.4 million. This may be a temporary delay; public health efforts are needed to address the large cancer screening deficit, including increased use of screening modalities that do not require a procedure.
Accepted for Publication: March 9, 2021.
Published Online: April 29, 2021. doi:10.1001/jamaoncol.2021.0884
Corresponding Author: Ronald Chen, MD, MPH, Department of Radiation Oncology, University of Kansas, Kansas City, KS 66160 (firstname.lastname@example.org).
Author Contributions: Drs Chen and Katz had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: Chen, Haynes, Du, Katz.
Drafting of the manuscript: Chen, Katz.
Critical revision of the manuscript for important intellectual content: Haynes, Du, Barron, Katz.
Statistical analysis: Katz.
Administrative, technical, or material support: Chen, Haynes.
Supervision: Chen, Barron.
Conflict of Interest Disclosures: Dr Chen reported receiving personal fees from AbbVie, Myovant, Bayer, Blue Earth Diagnostics, and Accuray outside the submitted work. Dr Haynes reported being employed by Anthem during the conduct of the study; receiving grants from Patient-Centered Outcomes Research Institute and contracts from the US Food and Drug Administration outside the submitted work. Dr Du reported receiving personal fees (salary) from HealthCore during the conduct of the study. Dr Barron reported being an Anthem employee and stock shareholder. Dr Katz reported receiving personal fees from Kite Pharma and Atara Biotherapeutics outside the submitted work.