The New Coronavirus Pandemic and the Trend of Breast Cancer Diagnosis according to Age-Group: Total and Interrupted Time Series Study

Breast cancer is one of the most common neoplasms in the world, with an estimated 2.3 million new cases per year, representing 11.7% of all cancer cases [1]. In Brazil, for each year of the triennium 2020–2022, it is estimated that about 625,000 new cancer cases occurred, with breast cancer being one of the most frequent (29.7% of malignant neoplasms in women) [2]. Early diagnosis of breast cancer is essential for reduction in mortality (7 deaths averted for every 1,000 women screened), reduction in the number of women diagnosed with late-stage cancer (29% reduction), increased survival at 5 years or more (98% in localized disease, 31% in advanced disease), and reduced costs in health systems [3].

About 83% of breast cancer occurs in women aged 50 years or older [3]. Therefore, many countries (including Brazil) perform breast cancer screening with biennial mammography in women aged between 50 and 69 [4]. Screening mammography in other age-groups is questionable due to the lack of evidence for reduced mortality and possibly increased false positive tests [5, 6]. In Brazil, mammography in other age-groups is performed according to the medical assessment of individual risks for breast cancer [4]. Breast biopsy is critical for a conclusive diagnosis of cancer, and its indication depends on the radiological findings of the mammogram (Breast Imaging Reporting and Data System, BI-RADS classification). Breast biopsy is indicated when the mammogram is classified as BI-RADS 4 (malignancy suspected) or 5 (highly suspected of malignancy), and eventually BI-RADS 3 (probably benign) [4].

In many countries, the COVID-19 pandemic has overloaded healthcare systems and negatively interfered with preventing, diagnosing, and treating other diseases, such as breast cancer [7‒9]. The Brazilian Ministry of Health decreed community transmission of COVID-19 in March 2020. Since then, there has been a reduction in tests and elective medical procedures in the country. This can promote a delay in the diagnosis of breast cancer and, consequently, an increase in advanced disease and mortality [7‒9].

There is a lack of studies evaluating the pandemic effect on breast cancer detection according to age-group. This information is essential for planning strategies to tackle breast cancer according to the women’s age-group in situations where healthcare services are overloaded by emerging diseases (as in a pandemic). This study aimed to assess the trend of mammograms, breast biopsies, and tumor stage at diagnosis according to age-group in the Brazilian public healthcare system during the COVID-19 pandemic.

This was an ecological time series study using an open-access dataset from the Brazilian Information System Department of the Public Healthcare System (DATASUS, in Portuguese) between January 2017 and July 2021 [10]. The DATASUS administrative dataset used in this study was the outpatient information system (SIASUS, in Portuguese), available from “https://datasus.saude.gov.br/acesso-a-informacao/producao-ambulatorial-sia-sus/,” the oncology panel dataset, available from “http://tabnet.datasus.gov.br/cgi/dhdat.exe?PAINEL_ONCO/PAINEL_ONCOLOGIABR.def,” and the resident population dataset, available from “http://tabnet.datasus.gov.br/cgi/deftohtm.exe?ibge/cnv/projpopuf.def” [10]. In 2021, Brazil’s estimated population was 213.3 million (108.2 million were women). About 32.4 million women were aged between 30 and 49, 21.6 million were between 50 and 69, and 7.7 million were 70 and over [11]. The sociodemographic data of the Brazilian population (gender and age-group) used in this study were from the resident population dataset. These data are available according to population projection studies based on the last census in the country (2010) [11]. In this study, the target population for screening was women aged between 50 and 69, and the non-screening population was women aged between 30 and 49 and women aged 70 and over [4].

The number of monthly mammograms (Bilateral Screening Mammogram) and breast biopsies (Breast Nodule Biopsy/Exeresis, Needle Aspiration Punctures, Thick Needle Breast Puncture) performed by age-group in the Brazilian public healthcare system were from the SIASUS dataset [11]. All medical tests and medical procedures performed and financed in Brazil’s public healthcare system are available in the SIASUS dataset. The monthly breast cancer diagnosis and stage at presentation (noninvasive or localized cancer – stage 0 to II; advanced disease – stage III and IV) by age-group were from the oncology panel dataset. In this dataset, the stage at diagnosis is available if there was treatment with chemotherapy, hormone therapy, or radiotherapy. On the other hand, if the treatment was only surgical, information on breast cancer stage at diagnosis is not included in this dataset. Finally, ductal carcinoma in situ treated with surgery and radiotherapy is in tumor stage 0. Lobular carcinoma in situ is not in the group stage 0 to II [4, 10].

The total and interrupted time series is a nonexperimental resource used to test hypotheses about factors that modify the behavior over time of interest to health measures, such as a pandemic [12, 13]. We used the Joinpoint Regression Program (JR), version 4.9.0.0 of 2021, made available by the National Cancer Institute (NCI), for data analysis in total and interrupted time series model. This program identifies significant changes in the trend of a dependent variable over time through the Poisson regression model. The JR program tests if multiple timeline segments (with multiple Joinpoint or inflection points) better explain a trend in time than a single line, defining the model that best represents the time series of the dependent variable [12, 13]. In this study, in the interrupted time series, the number of inflection points ranged from 0 to 3 or 1 to 4 segments, respectively, each with a lower and upper endpoint. The JR program automatically defines the location and the number of inflection points in the interrupted time series. The JR program analyzed the dependent variable in the total time series considering zero inflection points or one segment. Once the model is defined, the program calculates each segment’s Monthly Percent Change (MPC) or Annual Percent Change. It allows us to describe and quantify the trend by segment (stationary, increasing, or decreasing) and assess if it is statistically significant. The null hypothesis is MPC or Annual Percent Change equal to zero; that is, the dependent variable is neither increasing nor decreasing, considering the confidence intervals of 95% and significance level of 5% [12, 13].

In this study, we preferentially use the independent variable “month” to obtain a more significant number of points in the time series and improve the accuracy of the results [12, 13]. The dependent variables evaluated in the total and interrupted time series were the monthly trend of mammograms rate per 10,000 women by age-group, the monthly trend of breast biopsy rate per 100,000 women by age-group, the monthly trend of breast cancer diagnosed at stages 0 to II (noninvasive or localized cancer) per 100,000 women by age-group, and the monthly trend of breast cancer diagnosed at stages III and IV (locally advanced or advanced disease) per 100,000 women by age-group. To calculate the dependent variables, we use the number of women estimated by age-group in the respective year (between 2017 and 2021) [10, 11]. In this study, the pre-pandemic was between January 2017 and February 2020, and the pandemic was between March 2020 (declaration of the COVID-19 pandemic by the World Health Organization) and July 2021 [7‒9]. We analyzed the trend of the dependent variables in the pre-pandemic (interrupted time series) and the pandemic effect on the total time series. In other words, we evaluated whether changes in the trend of the dependent variables during the pandemic were significant to change the total time series trend relative to the pre-pandemic trend.

We used Microsoft^® Excel^® (Office 365 MSO) for descriptive statistical analysis of sociodemographic variables and the Minitab^® 19.2020.1 to build time series plots. We obtained data for this study from 10/30/2021 to 11/30/2021. This study used a public database and was conducted by the relevant research guidelines/regulations of research.

In the pre-pandemic years (2017–2019), there was an annual average of 4,345,259 mammograms (1,313,656 in women aged between 30 and 49, 2,750,534 in women aged between 50 and 69, and 281,069 in women aged 70 and over). In the year the COVID-19 pandemic was declared (2020), there were 2,565,168 mammograms (762,578 in women aged between 30 and 49, 1,633,515 in women aged between 50 and 69, and 169,075 in women aged 70 and over). In the pre-pandemic years, there was an annual average of 61,792 breast biopsies (30,384 in women aged between 30 and 49, 25,651 in women aged between 50 and 69, and 5,756 in women aged 70 and over). In 2020, there were 58,332 breast biopsies (28,041 in women aged between 30 and 49, 24,663 in women aged between 50 and 69, and 5,628 in women aged 70 and over). Thus, compared to pre-pandemic years, in 2020, there was a 40.97% decrease in mammograms and a 5.6% decrease in breast biopsies. In addition, the monthly average breast biopsies ratio per mammogram in the pre-pandemic (between January 2017 and February 2020) has changed from 1.45% (2.36% in women aged between 30 and 49, 0.95% in women aged between 50 and 69, and 2.07% in women aged 70 and over) to 2.67% (4.10% in women aged between 30 and 49, 1.83% in women aged between 50 and 69, and 4.02% in women aged 70 and over) during the pandemic (between March 2020 and July 2021).

In the interrupted time series, between January 2017 and February 2020, the monthly trend of the mammography rate per 10,000 women was decreased in the age-group between 30 and 49 (MPC: –0.3%; p = 0.017) and stationary in other age-groups (Table 1; Fig. 1). The changes in the monthly trend of mammograms during the COVID-19 pandemic in women aged between 50 and 69 and women aged 70 and over has changed the stationary trend in the pre-pandemic to a decreasing trend in the total time series (MPC: –1.4% and -1.6%, respectively). In addition, in women aged between 30 and 49, the decreasing trend of mammograms in the pre-pandemic (MPC: –0.3%; p = 0.017) worsened after the COVID-19 pandemic, as identified in the total time series (MPC: –1.4%; p < 0.001). In the interrupted time series, between January 2017 and January 2020 or February 2020, the monthly trend of breast biopsies ratio per 100,000 women increased in all age-groups (MPC between 1.0% and 1.1%). The changes in the monthly trend of breast biopsies during the COVID-19 pandemic has changed the increasing trend identified in the pre-pandemic to a stationary trend in the total time series in all age-groups (Table 1; Fig. 1).

In the interrupted time series, the monthly trend of breast cancer diagnosed at stages 0 to II per 100,000 women increased in all age-groups in the pre-pandemic (MPC between 0.3% and 0.4%). The changes in the monthly trend of breast cancer diagnosed at stages 0 to II during the COVID-19 pandemic has changed the increasing trend identified in the pre-pandemic in all age-groups to a stationary trend (women aged between 30 and 49 or women aged 70 and over) or decreasing trend in women aged between 50 and 69 (MPC: –0.2%; p = 0.031) in the total time series (Table 2; Fig. 2). Finally, the monthly trend of breast cancer diagnosed at stages III and IV per 100,000 women increased in all age-groups in the pre-pandemic (MPC 0.5%). The changes in the monthly trend of breast cancer diagnosed at stages III and IV during the COVID-19 pandemic, despite the reduction of the MPC (between 0.2% and 0.4%), did not change the increasing trend in the total time series in all age-groups, compared to the pre-pandemic (Table 2; Fig. 2).

This study showed that the COVID-19 pandemic has changed the stationary trend to a decreasing trend of mammogram rate per 10,000 women in all age-groups. Also, the pandemic has reversed the increasing trend in breast biopsy rate per 100,000 women to a stationary trend in all age-groups. The pandemic has reversed the increasing trend of breast cancer diagnosed at stages 0 to II per 100,000 women to a decreasing or stationary trend. On the other hand, the pandemic has not influenced the increasing trend of breast cancer diagnosed at stages III and IV per 100,000 women in all age-groups.

Breast cancer screening aims to promote the diagnosis of noninvasive or localized disease (stages 0 to II), reducing the incidence of advanced disease (stages III and IV), consequently increasing disease-free survival and reducing mortality. In Brazil, breast cancer screening aims to perform bilateral mammography every 2 years in 70% of women aged between 50 and 69 [4]. A time series study in Brazil shows that breast cancer screening coverage remained stagnant between 2013 and 2017 in the public healthcare system and was insufficient to guarantee the population’s demand [14]. Our study showed that the trend of mammogram rate per 10,000 women in the target population for screening was stationary in the pre-pandemic. Furthermore, in the pre-pandemic, our study showed an increasing trend of breast cancer diagnosed at stages III and IV. This could be associated with a failure to screen the target population, even before the COVID-19 pandemic.

Screening women younger than 50 or aged 70 or older is questionable because there is no evidence of reduced breast cancer mortality in these age-groups [5, 6]. Therefore, mammography is not routinely performed in these age-groups, but according to the medical assessment of individual risks for breast cancer [4]. This may explain a higher breast biopsy ratio per mammogram in these age-groups, as our study shows. On the other hand, the increase in the trend of breast cancer diagnosed at stages III and IV in these age-groups worries, especially in young women. In other words, this may mean a failure to identify women at risk for breast cancer in the non-screening population, even before the COVID-19 pandemic.

The negative impact of the COVID-19 pandemic on mammograms and breast biopsies has also occurred in other countries [12, 13, 15, 16]. For example, in an Italian study, the percentage decrease in mammograms in the year the pandemic began was 37.6% in women aged between 50 and 69, similar to the result of our study (a reduction of about 40% in women aged between 50 and 69 and in other age-groups) [17]. Similar to our study, most countries identified a point of inflection or change in the trend of mammograms or breast biopsies in March 2020, when the World Health Organization declared the COVID-19 pandemic [7‒9]. Since March 2020, there has been a reduction or interruption in breast cancer screening in many countries lasting between 1 and 6 months [15, 16]. We found no interruption, but a decrease in mammograms and breast biopsies, lasting between 3 and 4 months, in women aged between 50 and 69 and in other age-groups. Another study showed that the greatest percentage reduction in mammograms (lowest inflection point in the time series) was 85.1% and 40.9% in breast biopsies in women aged 18 and older [16]. Our study showed a similar result, with the greatest percentage reduction in mammograms of about 80% and breast biopsies of 40% in women aged between 50 and 69 and in other age-groups.

The subsequent recovery period in the number of mammograms and breast biopsies is essential to identify the efficiency of the healthcare system in the early detection of breast cancer. In other words, whether the healthcare system can resume mammograms and breast biopsies not performed during the pandemic to achieve the same trend in the shortest time observed in the pre-pandemic. An ineffective recovery period may justify a prolonged failure to detect breast cancer [12, 13, 15, 16]. For example, in the North Carolina study, after a 6-months recovery period, the trend of mammograms and breast biopsies was already similar to the pre-pandemic trend [16]. On the other hand, in our study, the subsequent recovery period was not enough to maintain the same trend of mammograms and breast biopsies observed in the pre-pandemic in women aged between 50 and 69 and in other age-groups.

Some time series studies have projected that interrupting or reducing breast cancer screening during the COVID-19 pandemic could increase the prevalence of advanced disease and mortality [15, 18]. For example, in the United Kingdom, a study found that there would be an increase in 5-year mortality, from 6.3% to 22.3%, in the case of screening interruption at 3 and 6 months, respectively [18]. Furthermore, in 5 years, there would be an increase in advanced disease (stage III and IV) of 30% and 109% if there was an interruption of 3 and 6 months, respectively [18]. Our study showed that the pandemic has reversed the increasing trend of breast cancer diagnosed at stages 0 to II in all age-groups. On the other hand, the pandemic has not altered the increasing trend of breast cancer diagnosed at stages III and IV in all age-groups. Likewise, another time series study that used the same Brazilian dataset (SIASUS) showed that the increasing trend of direct costs with chemotherapy in breast cancer at stages III and IV did not change during the COVID-19 pandemic. On the other hand, the pandemic has reversed the increasing trend of direct costs of treating breast cancer at stages I and II [19]. Therefore, this result suggests a failure to detect noninvasive or localized breast cancer (stage 0 to II) during the pandemic. In addition, this could be associated with reducing mammograms and breast biopsies in the same period.

According to another study, the impact of the COVID-19 pandemic was lower on breast biopsies than mammograms in all age-groups [16]. In our study, during the pandemic, the lowest inflection point of mammograms in the time series was about two times greater than breast biopsies in all age-groups. This was similar to the results of another study [16]. Our study did not assess whether there was a greater tendency for women at higher risk for breast cancer to undergo mammography during the COVID-19 pandemic. However, there were probably more women with the BI-RADS IV and V radiological classification in this period, justifying a higher breast biopsies ratio per mammogram in all age-groups. Even so, this was not enough to reduce the advanced or locally advanced breast cancer diagnoses (stage III and IV) during the pandemic in all age-groups.

The implication of this study was to demonstrate that the overload of healthcare services due to emerging diseases (for example, a pandemic) can negatively interfere with tackling other diseases (such as breast cancer) in a public healthcare system. Therefore, all actions are necessary to avoid interrupting or reducing mammograms and breast biopsies in situations of overloading healthcare systems due to other diseases. If this is not possible, strategies to promote an efficient recovery period after the interruption or reduction of mammograms and breast biopsies can reduce the impact of the pandemic on failure to diagnose early breast cancer. Therefore, the healthcare system must guarantee a sustainable and sufficient structure to meet the population’s demand, even in the most remote regions, with the possibility of expanding service providers whenever necessary. In addition, indicators must be efficient management, promoting quantitative and qualitative control of mammograms and breast biopsies performed. Finally, there must be actions to encourage the population’s participation in breast cancer screening and to promote the active search by healthcare services for people at risk for the disease.

Our study had some limitations. The epidemiological data of the Brazilian population are available according to projections studies from the last census (2010). The country’s private healthcare system serves about 25% of the Brazilian population [10], and the database used in this study (DATASUS) does not include data from this private healthcare system. All tests, procedures, and treatments performed in the Brazilian public healthcare system are available on DATASUS; despite the importance of this dataset, some limitations may occur, such as delay, absence, or error in recording information in this system. The dataset used in this study did not provide information on breast cancer stage at diagnosis when the treatment performed was exclusively by surgery. Finally, the observation time during the pandemic may be insufficient to demonstrate the trends of some variables. For example, a more extended analysis period of breast cancer diagnosis at stages III and IV may be necessary to identify an enhanced or reversed trend.

In Brazil’s public healthcare system, the pandemic has changed the stationary or increasing trend to a decreasing or stationary trend of mammograms, breast biopsies, and tumors at stages 0 to II but has not influenced the increasing trend of tumors at stages III and IV in all age-groups. In addition, during the pandemic, the reduction in mammograms was greater than breast biopsies across all age-groups. Finally, during the pandemic, the recovery period of mammograms and breast biopsies after the lowest inflection point in the time series was insufficient to achieve the same trend observed in the pre-pandemic in all age-groups. A new study with a longer observation time and the analysis of other variables (for example, breast cancer mortality) could contribute to better evaluation of the impact of the COVID-19 pandemic on breast cancer detection.

This study exclusively used a public database. All analysis was performed by relevant research guidelines/regulations in research. The Ethics Committee of the Health Sciences Sector of the Federal University of Paraná approved this research (CAAE 51438521.2.0000.0102).

The authors have no conflicts of interest to declare.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Adriano Hyeda, Élide Sbardellotto Mariano da Costa, and Sérgio Cândido Kowalski: conception, study design, execution, acquisition of data, analysis, and interpretation; Adriano Hyeda, Élide Sbardellotto Mariano da Costa, and Sérgio Cândido Kowalski: writing and reviewing. Adriano Hyeda, Élide Sbardellotto Mariano da Costa, and Sérgio Cândido Kowalski: reviewing at all stages of preparation of this manuscript.

The data supporting this study’s findings are available from the Brazilian Information System Department of the Public Healthcare System (DATASUS, in Portuguese). The DATASUS administrative dataset used in this study was the outpatient information system (SIASUS, in Portuguese), available from “https://datasus.saude.gov.br/acesso-a-informacao/producao-ambulatorial-sia-sus/,” the oncology panel dataset, available from “http://tabnet.datasus.gov.br/cgi/dhdat.exe?PAINEL_ONCO/PAINEL_ONCOLOGIABR.def,” and the resident population dataset, available from “http://tabnet.datasus.gov.br/cgi/deftohtm.exe?ibge/cnv/projpopuf.def.”