The impact of the COVID-19 pandemic on breast cancer screening and diagnosis in a Brazilian metropolitan area

Abstract

Objectives

To evaluate the performance of breast cancer screening and early diagnosis during the pandemic, compared to the pre-pandemic period.

Setting: The public referral centre for screening in Campinas, São Paulo State, Brazil.

Methods

This is an audit study of performance screening and diagnostic indicators. Two periods were analysed: 2019, the pre-COVID period, and 2020, the COVID period. All women who underwent mammography in these periods were included. Indicators were compared between periods, and the US Breast Cancer Surveillance Consortium benchmarks were used as a reference.

Results

A comparison between the periods shows a reduction of 57.4% in screening and 4.4% in diagnosis using mammography. Cancer detection rate per 1000 screening mammograms dropped from 4.62 to 2.83 (p  =  0.031), while it increased from 84.43 to 89.36 in diagnosis mammograms (p  =  0.701), higher than the reference (34.4, p < 0.001). With regard to diagnosis, the proportion of minimal cancers was reduced (p  =  0.005) and was lower than the reference (40.0%, p < 0.001), along with the proportion of node-negative invasive cancers (p < 0.001). The mean size of invasive tumours was similar in the two periods (32.50 mm and 33.40 mm, p  =  0.808) but larger than the reference value (16.50 mm, p < 0.001). Recall rate was lower in the COVID period (22.55% vs. 27.37%, p  =  0.015).

Conclusion

The COVID pandemic caused an overall decrease in breast screening and detection of breast cancer cases, although the reduction in number of diagnosis mammograms performed was minimal. Tumour mean size was large in both periods, the pandemic highlighting a previous profile of detection at an advanced stage.

Background

The COVID-19 pandemic brought diverse health service challenges, versus those directly related to the disease. In response to the pandemic in Brazil, a COVID-19 status announcement was made in March 2020, with Brazilian medical societies launching a halt to screening evaluations. For diagnostic tests in symptomatic women, the degree of clinical suspicion and individual risk-benefit were considered, with priority given to patients over 60 years.¹ In July 2020, there was an observation of transitory relief according to COVID-19 national indicators, as the Ministry of Health released a technical note recommending a gradual return of actions.²

Population-based breast cancer screening is not widely available in low- and middle-income countries. Although a national screening programme exists in Brazil, it lacks personal invitations.³ Screening and early diagnosis campaigns are encouraged, with gains in downstaging and less aggressive, costly treatment.^4,5 Access to care is influenced by inequality.^5–7 During the pandemic, social and economic vulnerability increased the disparity of access to screening and early diagnosis. The impact in low- and middle-income countries has not yet been determined.

Quality control and health programme audits offered data to policymakers to improve processes and identify weaknesses. The COVID-19 pandemic imposed more challenges for quality control in terms of screening performance.⁸ This study is an audit of a screening service for a Brazilian metropolitan area, Campinas. Validated indicators were used to evaluate the municipal programme during the pre-COVID and pandemic periods. The results of this study may reflect the impact of the pandemic in regions with similar conditions, generating support strategies that could overcome inequalities based on access to care.

Methods

This is an audit study of a screening service in the public health system (PHS) of Campinas, in São Paulo State, Brazil. Campinas is an urban city with 1.2 million inhabitants and a human development index of 0.805.⁹ Two periods were analysed: 2019, the pre-COVID period, and 2020, the pandemic period. All women undergoing mammography were included. In Brazil, the health system is universal, with free access to health services. In Campinas, 47.4% of women co-use the private health system, leaving a population of 55,497 women from 50 to 69 years exclusively relying on the PHS.^10,11 The screening service studied performs 60% of mammograms in the PHS of Campinas and makes about 90% of breast cancer diagnoses (an average of 200 breast cancer diagnoses per year).¹¹

The gateway to the PHS is primary health care. The national recommendation is to screen women using mammography from age 50 to 69 years every two years.³ Women under 50 with familial risk factors are prompted to be screened, but there is pressure from the media and medical societies to screen at 40, with compliance from women and health professionals.

Screening mammograms were requested for primary care at screening services, which invites women with abnormal results for further assessment. Symptomatic patients were admitted through primary care and referred to screening services for more investigation. All mammograms in the audited service are performed using digital mammography devices and analysed in an appropriate system by specialist radiologists. Diagnostic workups (core biopsy guided by ultrasound or stereotaxis) are managed at the same site. Histopathological analysis is referred to a central pathology unit. A breast surgeon evaluates all positive cancer cases and initial clinical staging. Women are referred to a tertiary service for treatment and follow-up.

Study data were collected through the screening service's information system, which uses electronic data captured via REDCap^®, hosted by the institution that maintains the unit. The data form contains information about age, symptoms, menopausal status, hormone replacement therapy, and personal/family history of breast cancer. It records the date of the tests, including histopathological results. Clinical staging is also recorded, while forms are filled in by a nurse or physician, collecting information from electronic medical records.

Screening mammography is a routine test for women without signs/symptoms of breast cancer. Diagnostic mammography is used to investigate suspicious breast lesions, as ordered by a physician post-clinical exam. Mammograms are categorised according to the ACR BI-RADS^® Atlas 5th edition. Brazil does not have its own standard in mammography quality control. The Breast Cancer Surveillance Consortium (BCSC) is a collaborative network of American mammography registries of the American College of Radiology whose benchmark guidelines are used for audit.^12,13 Although the breast cancer rate in Brazil is lower than in the US (61.9 vs. 90.3 per 100,000 women),¹⁴ given the absence of a national standard, 50th percentile indicators were adopted from the BCSC as references. Results of the comparison with BCSC, therefore, have to be interpreted with caution.

True-positive mammography (TP) is defined as a positive test with a diagnosis of cancer after further evaluation. True-negative mammography (TN) is a negative test without a diagnosis of breast cancer. False-positive mammography (FP) is a test interpreted as positive without breast cancer diagnosed in the evaluation. False-negative mammography (FN) is a negative test followed by a diagnosis of breast cancer after further evaluation. Cancer detection rate is defined as TP tests divided by a total of screening mammograms (TP/Total). The false-negative rate is defined as the number divided by the total number of screening mammograms (FN/Total). Specificity involves TN divided by total cancer-free tests (TN/TN  +  FP). Positive predictive value (PPV) calculations were made with BI-RADS methodology: PPV1 (cancer probability after initial assessment: 0, 3, 4, or 5) or PPV2 (cancer probability after final assessment: 4 or 5).

X² or Fisher tests were used for comparison between periods, except for the tumour's mean size, with the Mann-Whitney test used due to the absence of normal distribution of variables. The level of significance adopted for the statistical tests was 5%, or p < 0.05. The Research and Ethics Committee of the University of Campinas approved the study's population-based assessment of breast cancer screening, diagnosis, and death according to age, under CAAE 89399018.2.0000.5404. The Committee was able to waive consent.

Results

During the pre-COVID period (2019), 20,897 mammograms were performed, 19,914 of which were screening mammograms and 983 diagnostic mammograms. As a result, 175 women were diagnosed with breast cancer. During the pandemic period (2020), 9418 mammograms were performed: 8478 screening and 940 diagnostic (a reduction of 57.4% and 4.4%, respectively). As a result, 108 women were diagnosed with breast cancer (a decrease of 38.3%) (Figure 1).

Figure 1.

Effects of COVID pandemic on breast cancer screening/diagnosis in Campinas, Brazil.

Tables 1 and 2 show screening and diagnostic mammogram indicators, respectively, comparing the two periods and the reference (50th percentile, BCSC). From the screening analysis (Table 1), the cancer detection rate was significantly lower during the pandemic period (2.83 vs. 4.62 per 1000 mammograms, p  =  0.031) and was also lower than the reference (p  =  0.016). The proportion of cancers in stages 0/1 (or ductal carcinoma in situ) was around the reference in both periods (not significantly different, p > 0.05). The mean size of invasive tumours was 19.10 mm pre-COVID and 17.70 in the pandemic period (p  =  0.954). It was higher than the reference of 15.10 mm (p  =  0.005, pre-COVID; p  =  0.291, pandemic). The recall rate was lower during the pandemic than pre-COVID (9.63% vs. 11.10%, p < 0.001) or the reference (10.80%, p < 0.001). In comparison to the reference, PPV1 was lower, PPV2 was higher, and specificity was higher (p < 0.05). Specificity was lower pre-COVID than during the pandemic (89.36% vs. 90.63%, p < 0.001).

Table 1.

Evaluation of quality indicators for breast cancer screening mammograms in Campinas.

		Pre-COVID period (2019)		Pandemic period (2020)		P COVID 2019  ×  2020	p50 BCSC	P BCSC x 2019	P BCSC x 2020
Benchmarks for screening	Units	Value	Figures	Value	Figures		Value
Cancer detection rate	/1000 MMG	4.62	92/19914	2.83	24/8478	0.031	4.60	0.967	0.016
Minimal cancer (<1.0 cm/DCIS)	%	48.91	45/92	58.33	14/24	0.411	54.8	0.257	0.728
Mean size (invasive)	mm (-/ + SD)	19.10 (10.10)		17.70 (8.00)		0.904	15.10	0.005	0.291
Node-negative (invasive)	%	82.14	46/56	75.00	9/12	0.687	81.00	0.827	0.596
Recall rate	%	11.10	2210/19914	9.63	816/8478	<0.001	10.80	0.176	<0.001
PPV1	%	4.06	91/2241	2.90	24/829	0.131	4.40	0.434	0.035
PPV2	%	29.15	65/223	42.31	22/52	0.066	26.10	0.300	0.008
Specificity	%	89.36	17703/19822	90.63	7662/8454	<0.001	89.70	0.071	0.005

p50 BCSC: 50^th percentile in the Breast Cancer Surveillance Consortium; P COVID: p-value comparing the two periods, by X² or Fisher test, except for mean size where Mann-Whitney test was used; P BCSC: p-value compared to reference, by Student's t-test; DCIS: ductal carcinoma in situ; MMG: mammograms; SD: standard deviation; Recall rate: proportion of MMG BI-RADS 0, 4 or 5; PPV1: rate of cancer detection in MMG BI-RADS 0, 3, 4, or 5; PPV2: rate of cancer detection MMG BI-RADS 4 or 5; Specificity: proportion of true negative MMG from total number of MMG without cancer (true negative plus false positive MMG).

Table 2.

Evaluation of quality indicators for breast cancer diagnostic mammograms in Campinas.

		Pre-COVID period (2019)		COVID period (2020)		P COVID 2019   ×  2020	p50 BCSC	P BCSC x 2019	P BCSC x 2020
Benchmarks for diagnosis	Units	Value	Figures	Value	Figures		Value
Cancer detection rate	/1000 MMG	84.43	83/983	89.36	84/940	0.701	34.40	<0.001	<0.001
Minimal cancer (<1.0 cm/DCIS)	%	7.23	14/83	3.57	3/84	0.005	40.0	<0.001	<0.001
Mean size (invasive)	mm (-/ + SD)	32.50 (20.8)		33.40 (17.3)		0.275	16.50	<0.001	<0.001
Node-negative (invasive)	%	53.66	44/82	55.56	45/81	0.808	69.6	<0.001	0.006
Recall rate	%	27.37	269/983	22.55	212/940	0.015	11.6	<0.001	<0.001
PPV2	%	75.33	58/77	76.00	57/75	0.923	31.6	<0.001	<0.001
Specificity	%	78.78	709/900	84.46	723/856	0.002	91.6	<0.001	<0.001

p50 BCSC: 50^th percentile in the Breast Cancer Surveillance Consortium; P COVID: p-value comparing the two periods, by X² or Fisher test, except for mean size where Mann-Whitney test was used; P BCSC: p-value compared to reference, by Student's t-test; DCIS: ductal carcinoma in situ; MMG: mammograms; SD: standard deviation; Recall rate: proportion of MMG BI-RADS 0, 4 or 5; PPV1: rate of cancer detection in MMG BI-RADS 0, 3, 4, or 5; PPV2: rate of cancer detection MMG BI-RADS 4 or 5; Specificity: proportion of true negative MMG from total number of MMG without cancer (true negative plus false positive MMG).

Evaluation of indicators for diagnostic mammograms (Table 2) showed similar cancer detection rates in the pre-COVID and pandemic periods: 84.43 vs. 89.36 per 1000 mammograms (p  =  0.701). These rates were higher than the reference (34.40, p < 0.001), with the proportion of minimal cancer reduced significantly from pre-COVID to pandemic (7.23% to 3.57%, p  =  0.005). Those proportions are lower than the reference (40.00%, p < 0.001), given the proportion of node-negative invasive cancers (p < 0.001). The mean size of invasive tumours was similar in the pre-COVID and pandemic periods (32.50 mm and 33.40 mm, p  =  0.808) but higher than the reference (16.50 mm, p < 0.001). The recall rate was lower in the pandemic vs. pre-COVID (22.55% vs. 27.37%, p  =  0.015), and higher when compared to the reference (11.60%, P < 0.001). PPV2 was higher than the reference in both periods (p < 0.001). Specificity increased from pre-COVID to the pandemic period (78.78% vs. 84.46%, p  =  0.002) but was lower than the reference (91.60, p < 0.001).

Discussion

This audit study evaluated the impact of the COVID pandemic on breast cancer screening in Campinas, observing a reduction in detected cancer cases. The mean size of invasive cancers was prominent. During the pandemic, the proportion of minimal cancers found in diagnosis mammograms was reduced. The recall rate was also reduced, with increased specificity for screening and diagnostic mammograms.

The current analysis shows an overall reduction of 55% in mammograms, and a reduction of 38% in breast cancer cases. The drop was greater in screening mammograms (57%), and the cancer detection rate was also reduced from 4.62 to 2.83 per 1000 mammograms (p  =  0.031). The rate of diagnostic mammograms was similar before and during the pandemic, with data suggesting that although screening activities dropped significantly, symptomatic users could still access health services.

COVID-19 affected the rate of diagnosis and cancer detection even in high-income countries, with notable decreases compared to the period prior to the pandemic.¹⁵ In China, fewer diagnoses, surgeries, and significant early-stage delays were reported.¹⁶ In Italy, data showed an increase in the proportion of stage III-positive lymph node cases after a two-month interruption of screening.¹⁷ In Canada, a three-month interruption projected an increase in cases at advanced stages, including deaths from 2020 to 2029.¹⁸ The effect of a reduction in screening tests may be more noticeable in the medium- and long-term, as they identify pre-malignant, initial, and asymptomatic malignant lesions, with earlier intervention for cancer. We noted that cancer detection rates in our programme were lower than in the reference, the BCSC, but this must be at least partly explained by the lower breast cancer incidence rate in Brazil compared to the US. The important issue here is the effect of COVID rather than the difference from the standard employed. The higher detection rates from diagnostic mammograms and larger tumour sizes reveal an extended time from clinical suspicion to diagnosis. This was observed even in the pre-pandemic period, and may reflect low awareness and barriers to access for early investigation of suspected cases. Fewer cancers were detected by screening, and the pandemic may have increased this trend.

In low- and middle-income countries, most breast cancers are self-detected, as seen in a Peruvian study (92.9% detected by the patient), demonstrating an extended average time from onset of symptoms to treatment.¹⁹ In Brazil, one study showed that women relying on the PHS were more often diagnosed as symptomatic (76.9%) than patients with private insurance (47.0%). More stage III cases were detected in the public (33.5%) than private sector (14.7%).⁵

The ability to detect a mammographic abnormality is related to the quality of the test High-quality tests yield a positive predictive value (PPV) of 85%-90% in women over 50, enabling detection of initial tumours before lymph node involvement.²⁰ The most relevant standardisation recommendations for quality control were suggested by the European Commission (with related guidelines)²¹ and the American College of Radiology²²; these include positioning and exposure criteria. Initiatives such as the BCSC quantify performance monitoring and identification of acceptable levels for mammogram interpretation.^12,13,23

This study shows a high cancer probability after final assessment for BI-RADS of 4 or 5 (PPV2) from diagnostic tests in both periods, possibly due to cases that present more suspicious and advanced issues because of delay in referral for evaluation. Specificity was significantly lower than the reference, which may reflect a need for earlier investigation to avoid delayed diagnosis. There was no difference compared to the reference in how mammograms were recorded or read, so this was related to a substantial decline in tests performed, as well as the profile of women attending.

Recall rates were much like reference screening, but very high for diagnostic mammograms. A national cohort of women at standard risk, with more than a thousand mammograms/year, had a frequency of BI-RADS of 0 of 6.7% of the target population screened, and a frequency of BI-RADS^® of 4 or 5, or 0.5%, for screening mammograms.^24,25 The recall rate decreased in the pandemic period, which may reflect that women with screening were more prone to self-care. Recall rates are lower for consecutive mammograms.²⁶ For diagnostic mammograms, the absence of previous tests, augmented by a higher prevalence of suspicious lesions, increased positive pre-test probability and a need for diagnostic supplements. The recall rate was higher during the pandemic, which may reflect a greater need to carefully investigate these cases.

This study benefits the female population relying on the PHS of a Brazilian metropolitan area, with regard to breast cancer diagnosis. Quality of data is a challenge in low- and middle-income countries, yet the high quality of data in this study enabled a refined analysis of indicators. The main limitation was the absence of national benchmarks for comparison, identifying the need for new quality parameters.

Conclusion

The COVID pandemic caused a downturn in the clear diagnosis of breast cancer cases in the audited service. The large tumour mean size is possibly an indication that symptomatic patients still had access to the service. It also appears that the pandemic highlighted a previous profile of detection at an advanced stage. Medium- and long-term studies will be necessary to fully investigate its impact.