Introduction

Since the first months of the COVID-19 pandemic, several publications worldwide have described the impact of this health crisis on cancer care. Decreases in diagnostic tests [1], in diagnosed cases [1, 2], interruption of screening programs [3,4,5], and reduction in surgeries, consultations, and hospital admissions for cancer patients [6] were identified in different countries as a consequence of adaptations and/or disruptions of the health services in this new scenario. Furthermore, the fear of contamination by the new disease kept patients away from health services, even when faced with symptoms that should be immediately investigated [7].

In Brazil, cancer-related hospital admissions for surgical purposes reduced by 28% from March to July 2020 compared to the same period of the previous year. This decline occurred in the entire country but was more significant in some regions, such as the South and Southeast [6]. There was a sustained decrease in hospitalizations due to oral and oropharyngeal cancer in Brazil that began in the first months of 2020 and continued until August 2021 – the last month documented [8]. In the first quarter of the pandemic, the number of surgeries for patients with lung cancer reduced by around 22% compared to the previous quarter, reaching a 51.5% reduction in the Northern region [9].

These and similar findings have concerned the medical and scientific community since the beginning of the pandemic, as delays in diagnosis and cancer treatment represent a risk for the progression of this disease and could lead to an increase in the number of avoidable deaths [10, 11]. Several statistical modeling studies worldwide projected that mortality from some types of cancer would increase as a result of the disorganization in the screening, diagnosis, and treatment services caused by the COVID-19 pandemic [12,13,14,15]. When death registries and information systems began to release mortality data, still during the pandemic, some studies investigating the effects of the collapse on the actual number of deaths started to outline the scenario. Until now, most results have disagreed with predictions, showing decrease in cancer deaths [16,17,18].

A study that analyzed the impact of the pandemic on mortality from neoplasms and cardiovascular diseases in Brazil in 2020 identified that, from March to December, the number of deaths with cancer as the underlying cause was 10% lower than expected [19]. Another study from the same country identified that deaths from neoplasms decreased in the age group over 40 years old in 2020 compared to the period between 2015 and 2019, while mortality from ill-defined causes increased [20]. However, these studies had assessed the preliminary version of the data from 2020 of the Mortality Information System of Brazil (SIM), and the source’s updates may have impacted the results.

The impact of the COVID-19 pandemic on cancer mortality is not fully understood. Cancer mortality is not an outcome that occurs immediately, and analyses with more extended follow-up periods, which are not yet available, are necessary. Furthermore, understanding which characteristics of the health system were associated with COVID-19 mortality in cancer patients can provide insights to mitigate the damage caused by this and future health crises in this population. This study aims to investigate the impact of the COVID-19 pandemic on the mortality of cancer patients in Brazil by analyzing the variation in observed and expected death rates from all types of cancer and the five most incident types and through the analysis of mortality from COVID-19 in cancer patients. Additionally, it aims to analyze the relationship between the density of hospital beds and mortality from COVID-19 in cancer patients in Brazil’s 133 Intermediate Geographic Regions (RGI).

Methods

This ecological study analyzed the mortality from trachea, bronchus, and lung (TBL), colorectal, stomach, female breast, and prostate cancer and all types of cancer in the pre-pandemic (2017–2019) and pandemic (2020–2022) periods and the mortality from COVID-19 in individuals who had cancer as a contributing cause of death. The Brazilian Mortality Information System (SIM, for the acronym in Portuguese) was the data source on deaths. This study used the SIM databases from 2017 to 2021 in final versions and the 2022 database in a preliminary version, last updated in September 2023 and available on the “openDataSUS” platform of the Brazilian Ministry of Health.

For cancer mortality, the SIM provided the number of deaths for which the primary cause was cancer of the TBL (ICD-10 codes C33 and C34), colon and rectum (C18-C21), stomach (C16), female breast (C50, in women), prostate (C61) and cancer in general (all “C” codes), by sex, age group (every five years), month, and municipality of residence in Brazil. A few deaths without the information of sex, age, or sex plus age were redistributed according to the respective distribution in the 26 Brazilian states and the Federal District; deaths without information on the municipality of residence were distributed to the respective municipality of occurrence. The deaths, by sex and age group, were then summed for Brazil’s 133 RGI [21]. To correct cancer mortality considering the underreporting due to ill-defined underlying causes, which can mask the actual number of deaths due to specific causes [22], a statistical correction (i.e., redistribution) was performed. To this end, the SIM also provided the total number of deaths with known causes and the number of deaths that the underlying cause was a code from chapter XVIII of ICD-10 (“Symptoms, signs, and abnormal clinical and laboratory findings, not elsewhere classified”; the “R” codes), by sex, age group, month, and RGI.

The 2010 and 2022 demographic censuses and the intercensal projections of the Brazilian Institute of Geography and Statistics (IBGE) provided the Brazilian population by sex, age group, year, and RGI – the population for the years 2017 to 2022 was projected by a linear trend, based on the 2010 Census and intercensal projections from 2011 to 2012, by sex, age group, and RGI, and adjusted for the number of inhabitants from the 2022 Census by RGI, from which population distributions by sex and age group were not available until the conclusion of this article.

The rate calculation observed the following steps: (1) correction of the number of deaths for each type of cancer and for all cancers to consider deaths with an ill-defined underlying cause by sex, age group, month, and RGI, using the technique proposed by França et al. [22]. This technique recommends that the amount of cancer deaths should receive a proportion of deaths with ill-defined causes, and this proportion should be ½ the percentage that cancer occupies in the total number of deaths with a known cause. This adjustment was made for each type of cancer by sex, age, and RGI and the corrected number of deaths was summed by the five Brazilian macro-regions – North (N), Northeast (NE), Southeast (SE), South (S), and Midwest (MW); and (2) calculation of the rate (corrected deaths divided by the respective population) for every 100 thousand inhabitants, as adjusted for sex and age group (five years), by the direct method, using the WHO World Standard population distribution [23]. Final mortality rates for cancer and types of cancer were monthly obtained for each macro-region and the whole country. Prostate cancer and female breast cancer mortality exclusively refer to the sex-specific populations groups.

Rates from the pre-pandemic triennium (2017–2019) were used as a reference for predicting the expected rates for 2020, 2021, and 2022, which were calculated using generalized linear regression – and checked/validated using the Prais Winsten method. A rate ratio (RR) was the measure used to compare the rates obtained from the SIM (called “observed rates”) with the predicted (or expected) rates. An RR > 1 indicated observed rates greater than predicted, an RR < 1 indicated observed rates lower than predicted, and an RR = 1 indicated observed rates equal to predicted.

For mortality from COVID-19 in individuals who had cancer as a contributing cause of death, the SIM provided the number of deaths for which the primary cause was COVID-19 (ICD-10 B34.2) and which had some cancer (any code “C”) as a contributing cause. The rates were calculated per 100 thousand inhabitants by RGI, by macro-region, and for Brazil, for all months from 2020 to 2022, and adjusted for sex and age group (five years), using the direct method, with the WHO World Standard population distribution as the reference [23].

Finally, an association analysis was carried out between mortality rates from COVID-19 in individuals who had cancer and the density of hospital beds in the 133 RGIs in Brazil, adjusted by the Human Development Index (HDI). The National Registry of Health Establishments (CNES) provided the total number of inpatient hospital beds and the number of inpatient hospital beds linked to the SUS from 2020 to 2022 in each RGI – the CNES provides this data in a monthly format; the number of beds in December of each year represented the respective year. The bed density was then calculated as a yearly rate for every 100 thousand inhabitants, for which the denominator was the annual population of each RGI. To calculate the HDI for each RGI, the Human Development Atlas in Brazil provided the HDI for each municipality in 2010. The municipal HDI was then aggregated by RGI using a weighted average, whose weight was the number of inhabitants of each municipality according to the 2022 Census. Linear regression analyses were used to evaluate the association, with models for Brazil and for each macro-region, by type of hospitalization bed (total or beds linked to the SUS). Results with p < 0.05 were considered statistically significant.

All analyses used Stata/IC 15.1 software. The delimitation of the “waves of the pandemic” (i.e., periods of recrudescence), which was used to represent some results graphically, was based on the number of weekly deaths from COVID-19 in the Brazilian population obtained from the World Health Organization observatory (https://www.who.int/data/gho).

Results

Mortality from all cancers was 122.93 deaths per 100 thousand inhabitants in Brazil, 2020 – the first pandemic year. Based on the three pre-pandemic years, the expected rate for 2020 was 129.87/100,000. Among the types of cancer included in this study, prostate cancer had the highest mortality (20.43/100,000), followed by female breast (17.10/100,000), TBL (16.00/100,000), colon and rectum (10.92/100,000) and stomach cancer (8.68/100,000). In 2020, the observed rate was lower than predicted for all types of cancer in Brazil and its regions, except in four cases: colon and rectum in the North (observed: 7.20/100,000, predicted: 6.96/100,000) and prostate in the North (observed: 19.91/100,000, predicted: 19.72/100,000), Northeast (observed: 21.60/100,000, predicted: 21.35/100,000), and Midwest (observed: 22.12/100,000, predicted: 22.07/100,000) – Table 1.

Table 1 Observed (O) and predicted (P) mortality (rate per 100,000 inhabitants) from tracheal, bronchus, and lung, colorectal, stomach, female breast, prostate, and all types of cancer, and observed mortality from COVID-19 in individuals with cancer, between 2020 and 2022, for Brazil and Brazilian macro-regions
Full size table

In 2021 and 2022, cancer mortality increased gradually compared to 2020 in all regions of Brazil. Like in 2020, in these two years, mortality observed was lower than predicted for the country for all cancers and each type analyzed. Regional exceptions, in which the observed rate was higher than expected, were more frequent in 2022 than in 2021. The types of cancer that presented the highest mortality in 2021 and 2022 remained the same as in 2020 (Table 1).

Figure 1 shows that, for all regions of Brazil, the cancer mortality curve had an increasing trend from 2017 to 2019, and in 2020, the first year of the pandemic, this pattern was inverted. The increasing trend was resumed in 2021; however, the North, Northeast, and Southeast regions remained with rates lower than those of 2019. In 2022, the increasing direction became more evident, and all regions of Brazil showed higher cancer mortality in 2022 than in 2019. From 2017 to 2022, the South region had the highest cancer death rates in Brazil, followed by the Southeast; the lowest rates were always in the North. In 2020, mortality from all cancers in Brazil reduced by 2.61% compared to 2019, and in 2021, it reduced by 0.24%. The variation changed direction in 2022: mortality increased by 2.96% compared to the reference year – results not shown.

Fig. 1
figure 1

Mortality (rate per 100,000 inhabitants) from COVID-19 in individuals with cancer per month, between 2020 and 2022, in Brazil. Note: 1st, 2nd and 3rd waves refer to periods of worsening of the COVID-19 pandemic in Brazil, considering the number of deaths from this disease

Full size image

Mortality from COVID-19 in cancer patients in Brazil in 2021 was the highest among the pandemic years analyzed: 5.98/100,000. This year, the Midwest and South regions had the highest rates – respectively, 7.44 and 7.21/100,000; however, in 2020, the Southeast had the most elevated rate, 5.69/100,000, being the only region with a rate above the national for the respective year (4.67/100,000) – Table 1. Figure 2 allows us to identify that mortality from COVID-19 among cancer patients was higher in the so-called waves of the pandemic, that is, in periods when this health crisis was most intense in the country, considering the number of deaths from COVID-19 in the general population. The highest monthly rates were, respectively, those of March and April 2021, which were part of the second wave of the pandemic in the country, the most extended and lethal one. A similar pattern was identified for the regions separately – results not shown.

Fig. 2
figure 2

Mortality (rate per 100,000 inhabitants) from all types of cancer between 2017 and 2022 for Brazilian macro-regions. Note: N – North, NE – Northeast, SE – Southeast, S – South, MW – Midwest

Full size image

In 2020, mortality from all cancers in Brazil was lower than expected, with a ratio of 0.95 between the observed and predicted rates; in 2021 and 2022, this relationship remained similar, with RRs of 0.94 and 0.95, respectively. The largest difference between observed and expected rates in Brazil was found for stomach cancer, with RRs of 0.89 in 2020 and 2021 and 0.88 in 2022. Prostate cancer showed the smallest difference, with observed rates close to those expected in the three pandemic years (RRs: 0.98, 0.99, and 0.97, respectively). Some results stood out in the regional comparison due to the low RRs: in the Northeast, an RR of 0.84 between the rates of stomach cancer in 2021 and 2022, 0.90 and 0.91 for lung cancer in 2021 and 2022, respectively, and 0.90 for breast cancer in 2022; in the Southeast, an RR of 0.91, 0.91, and 0.92 in the three pandemic years, respectively, for colorectal cancer, and 0.90 for TBL in 2022; and, in the North, an RR of 0.91 for TBL in 2021 (Table 2).

Table 2 Ratio between observed and predicted mortality rates from tracheal, bronchus, and lung, colorectal, stomach, female breast, prostate, and all types of cancer in 2020, 2021, and 2022, for Brazil and Brazilian macro-regions
Full size table

COVID-19 mortality in cancer patients was inversely associated with the rate of hospital beds maintained by the public health system, regardless of the HDI of the RGI. When these analyses were carried out for each region of Brazil, only the Northeast and South regions showed a statistically significant association, also with a negative direction. In the analysis between the outcome and the general availability of hospital beds – i.e., the rate of SUS plus private inpatient hospitalization beds per 100 thousand inhabitants – only the South showed a statistically significant association, which was negative (Table 3).

Table 3 Rate of hospital beds and SUS hospital beds (beds per 100,000 inhabitants) in the Intermediate Geographic regions of Brazil, by macro-region and for Brazil, between 2020 and 2022, and its association with mortality from COVID-19 in individuals with cancer
Full size table

Discussion

This study found that mortality from all types of cancer and the five most common types in Brazil was lower than expected in the COVID-19 pandemic years (2020–2022), and stomach cancer was the type that showed the greatest difference between observed and predicted rates. During the pandemic, mortality from COVID-19 in cancer patients (i.e., cancer as a contributing cause), a direct impact of the pandemic on this population, was higher, respectively, in 2021 and 2020, and the highest rates followed the waves of the pandemic in the country. The findings showed that this mortality was negatively associated with the number of hospital beds in the public health system (SUS) per inhabitant, considering the 133 RGI in Brazil, suggesting a protective role of the availability of hospital care concerning deaths due to COVID-19 in cancer patients.

The present study’s findings seem to align with the competing cause hypothesis, as the lower death rates from all cancers appear to have been partly compensated by the death rate from COVID-19 in cancer patients. This explanation supports that many individuals with cancer, at a higher risk of death from this disease, may have died from COVID-19, as this population had a higher risk of death from COVID-19 than the general population [24,25,26]. However, this rationale needs additional reflection: the contributing cause of death is likely underreported in the mortality information system, especially considering the first pandemic year, in which the unprepared health services – responsible for completing the causes of death – were under massive pressure and overload in the fight against COVID-19.

A study in Brazil identified that mortality from ill-defined causes increased in 2020 compared to the period between 2015 and 2019 [20]. In the present study, death rates from cancer as the underlying cause were adjusted for ill-defined causes, according to redistribution techniques established in the literature [22], but there are no statistical procedures to correct the underreporting of deaths due to COVID-19 that had cancer as a contributing cause. Thus, the present study suggests that mortality from COVID-19 in cancer patients (1) is probably much higher than recorded, and (2) was not just a competing cause but an additional cause of death in this population since the first year of the pandemic. In other words, mortality from COVID-19 not only took the lives of patients who would die from cancer but added preventable deaths to the mortality burden of this population, especially at the peaks of the pandemic. The results from 2021 contribute to these rationales, as both mortality from cancer as the underlying cause of death and mortality from COVID-19 in cancer patients increased in 2021 compared to 2020. Also, the findings suggested that this dynamic in mortality persisted for at least three years after the pandemic’s start.

According to Maruthappu et al. [27], who use the survival analysis of Quaresma, Coleman, and Rachet [28], colorectal, female breast, and prostate cancer are considered treatable, as their five-year survival rate is greater than 50%. It is reasonable to expect that if cancer mortality increases as a result of delays in diagnosis and treatment, this impact would be more evident in treatable cancers – i.e., those that would benefit more from timely diagnosis and treatment. This study did not confirm this hypothesis for the three years after the pandemic started, as colorectal, female breast, and prostate cancer also showed observed rates lower or very similar than expected for the period. However, the pandemic indeed affected the care of treatable cancers in Brazil. The number of mammograms performed by the SUS in 2020 decreased by more than 40%, reaching a 67% reduction in Rondônia, a state in the North of the country [29]. Also, the situation remained unfavorable in 2021: 15% fewer exams than in 2019 [30]. Analysis that compared new cases of colorectal and anal cancer in a reference center for cancer treatment in São Paulo reported a decrease in newly diagnosed patients referred and an increase in locally advanced disease from March to July 2020 compared to the same period in 2019 [31]. Concerning prostate cancer, the number of therapeutic procedures (radical prostatectomy plus radiotherapy) carried out by the SUS reduced by 22% between August 2020 and March 2021, compared to the same period in previous years [32]. So, it is hasty to conclude that cancer mortality did not increase as a consequence of the COVID-19 pandemic. As cancer mortality is not an outcome that occurs immediately, analyses with longer follow-ups are essential [12,13,14,15].

The largest difference between observed and predicted rates in this study was found for stomach cancer, a type of cancer that has a low five-year survival rate (18.8%) [28] and is comparable to untreatable cancers. Some factors may be influencing this result: (1) the methodology used for prediction, which considered the last three pre-pandemic years to reflect the more recent pattern of the outcome, may have underestimated the decline in mortality that has been occurring in Brazil in recent decades [33], and (2) the possibility that patients with stomach cancer are more susceptible to die from COVID-19 due to social [34] and/or clinical vulnerabilities – which needs further investigation.

The North and Northeast had the lowest mortality from COVID-19 in cancer patients in the three pandemic years. There is a substantial chance of underreporting in these regions, as these regions were severely affected by the pandemic, in which the first spikes in mortality due to COVID-19 occurred in the country [35, 36]; and have the most room for improvements concerning the quality of records in SIM [37]. For Brazil and each region separately (results not shown), these rates were higher in the periods corresponding to the peaks of deaths from COVID-19 in the country. These are expected results, as these deaths are a direct consequence of the pandemic and followed its dynamics of resurgence and slowdown. Studies with the American population have identified similar results [38, 39]. The rate of COVID-19 in cancer patients, which is not irrelevant, makes it clear that relying only on the underlying cause of death may lead to underestimating the impact of the pandemic on patients with cancer.

The density of SUS hospital beds was negatively associated with mortality from COVID-19 in cancer patients in the Northeast, South, and the whole country, in an analysis adjusted for the HDI of the analysis units – the 133 RGI in Brazil. These regions correspond to an intermediate geographical stratification between states and cities, which always includes large urban centers. The RGI likely covers the entire path through the health system that a resident of these goes through, with larger cities acting as a reference for smaller nearby cities, which do not have hospitals or health services of greater technological complexity [40]. If a cancer patient with COVID-19 lives in a small town without a hospital and needs hospital attention, they are likely to be referred for hospital treatment in a more structured city close to their local residence. This characteristic makes RGI an opportune territorial division to study the determinants of mortality from COVID-19 in cancer patients, identifying which factors are associated with more favorable regional outcomes. The availability of beds may be crucial in protecting this vulnerable population from preventable deaths, as cancer patients were more likely to develop severe cases of COVID-19 [24,25,26] and, consequently, to depend more on the hospital network. The present results align with this rationale; however, additional investigations that analyze the flow of bed occupancy and/or ICU beds in different regions can contribute to this understanding and support decisions on allocating health resources in possible future crises. Also, the significant results for SUS beds, even with adjustment for HDI, point to the crucial role of the public health system for the cancer population during this outbreak.

The decrease in death rates from cancer as the underlying cause that the present study found in the first year of the pandemic is compatible with the findings of other studies in Brazil [19] and other countries [16,17,18]. However, this study observed lower decreases than those reported for Brazil [19]. Differences in the definition of the reference rate for comparison and, mainly, the characteristics of the present study could explain these discrepancies. As far as we know, this study is the first to analyze cancer mortality using the final version of the SIM from 2020 to 2021, as well as extending the analysis to the year 2022, and to use data from the 2022 Census of Brazil to calculate the mortality rates. The final versions of the annual mortality databases contain more recorded deaths than preliminary versions, and the 2022 Census reported a smaller population than estimates based on previous censuses. These differences could justify the discrepancies in the results.

The main limitation of this study concerns the registration of deaths. As already mentioned, the quality of recording the underlying cause of death decreased during the pandemic, with an increase in the percentage of deaths with an ill-defined underlying cause compared to previous years [20], which may be an expected consequence of a collapsing health system. To minimize this limitation, the methodology of this study included the redistribution of ill-defined deaths, considering variations by sex, age group, month, and RGI for the analysis of the underlying cause of death; however, no statistical correction replaces qualified registration.

The mortality from cancer was lower than expected during the three years of the COVID-19 pandemic in Brazil. However, this outcome must continue to be monitored, as an increase in mortality due to the burden that the new disease imposed on the health system, especially at the hospital level, cannot yet be ruled out. As it is a chronic disease whose cure may require years of treatment, which causes sequelae and has the potential for recurrence – factors in part related to access to timely diagnosis and treatment – the deleterious impact of COVID-19 on cancer care may have an impact on mortality from this disease years after the pandemic’s control. The literature reports that cancer care disruptions during the COVID-19 pandemic could lead to significant life loss but also argues that this damage could be mitigated by increasing diagnostic and treatment capacity in the short term to address the service backlog [14]. The results of the present study highlight the need to understand what consequences the pandemic may have brought on health services, which can be worked on to prevent the increase in mortality in the coming years. Surveillance in health in the coming years is critical, as well as monitoring the production of cancer diagnosis and treatment services and the delay in the onset of treatment.

Also, concerning the cancer patients, COVID-19 not only took the lives of patients who would die of cancer – it was not one cause of death that replaced another; it likely added preventable deaths to the mortality burden of this population. The availability of SUS hospital beds may have acted as a protective factor against this additional mortality source for the cancer population in Brazil. This evidence can join the accumulated knowledge about the COVID-19 pandemic to assist in equitable strategies to mitigate the COVID-19 impact on the cancer patients and to contribute to decision-making in health crisis.