INTRODUCTION

The World Health Organization classifies Non-Communicable Diseases (NCDs) as the leading cause of morbidity and mortality globally¹. Malignant neoplasms account for 41 million deaths annually, representing 71% of all global deaths². The incidence and mortality rates of cancer are rising rapidly worldwide, reflecting demographic, epidemiological, and nutritional transitions, as well as an increase in cancer risk factors^3,4.

The latest GLOBOCAN estimates for 2020 reported approximately 19.3 million new cancer cases (18.1 million excluding non-melanoma skin cancer) and 10 million cancer-related deaths (9.9 million excluding non-melanoma skin cancer)¹. In the Americas, 20.9% of global cancer incidence and 14.2% of cancer-related mortality were reported, compared to other continents¹. Among the highest expected incidence rates are found in Australia, New Zealand, Northern Europe, and Western Europe¹.

In Brazil, the National Cancer Institute (INCA) estimates 704,000 new cancer cases for the 2023-2025 period⁵. Excluding non-melanoma skin cancer, breast and prostate cancers each account for 15% of new cases⁵. In 2020, the global mortality rate for breast cancer, age-adjusted to the world population, was 11.84 deaths per 100,000 women. In the Southeast and South regions of Brazil, the highest rates were 12.64 and 12.79 deaths per 100,000 women, respectively⁶. Age over 50 years is a primary risk factor for breast cancer development; however, genetic factors (mutations in the BRCA1 and BRCA2 genes), family history of cancer, late menopause, obesity, and frequent exposure to ionizing radiation also represent significant risk factors^4,7.

Breast cancer is the most prevalent cancer among women (excluding non-melanoma skin cancer) across all regions in Brazil, with an incidence of 61.61 new cases per 100,000 women⁶. The Southeast region has the highest estimated risk, with 81.06 cases per 100,000 inhabitants⁵. In the state of Espírito Santo, 790 new cases of female breast cancer were estimated for 2022⁵.

The Brazilian Hospital Cancer Registries (HCR) are a key data collection tool for cancer, helping to qualify healthcare services for planning purposes. These registries enable the evaluation of healthcare quality, the development of clinical research, and the creation of public health policies^8-10. The National Cancer Institute (INCA) uses the HCR to monitor the epidemiological evolution of cancer every three years, offering nationwide training to improve hospital management, clinical treatments, health promotion, and disease prevention efforts. Additionally, these data are used for clinical and epidemiological research^8-10.

Advanced-stage breast cancer represents a barrier to early diagnosis and the initiation of treatment, making public health planning, especially at the primary care level, essential. Furthermore, public programs that provide epidemiological data on the social determinants of health, which shape cancer risk factors, are crucial¹¹.

A recent analysis of a time series up to 2020, encompassing the entire Oncology Care Network of Espírito Santo-comprising a CACON and seven High Complexity Oncology Assistance Units (UNACON) aimed at guiding cancer surveillance actions in the region, including the monitoring and evaluation of the HCRs of hospitals in the state oncology network, particularly regarding malignant breast neoplasm cases, remains unclear.

Thus, the objective of this study is to evaluate the trend of incompleteness in the variables of the HCR for malignant breast neoplasms in the hospitals of the Oncology Care Network of Espírito Santo, Brazil.

METHODS

Study design, location and period

This is a retrospective time series study using secondary data from the Oncology Care Network of Espírito Santo (ES), which consists of one High Complexity Oncology Center (CACON) and seven High Complexity Oncology Assistance Units (UNACON). The study covers the period from 2000 to 2020.

The secondary data used were retrieved from the SIS-HCR, part of the INCA Integrated System, as well as from the Hospital Cancer Registries of the state of Espírito Santo, collected in collaboration with the State Health Department of Espírito Santo (SESA/ES). The state of Espírito Santo has an Oncology Care Network distributed across three health regions: Central-North, Metropolitan, and South (Figure 1)^9,10.

Source: Adapted from (Espírito Santo, 2023).

Figure 1 : Spatial Distribution of the Oncology Care Network in Espírito Santo 

RESULTS

During the historical series from 2000 to 2020, a total of 16,587 cases of female breast cancer were recorded from the HCR across the entire Oncology Care Network of the state of Espírito Santo. Table 1 presents the percentage of incompleteness of the variables in the breast cancer tumor registration form for the evaluated historical series (Table1).

The Friedman test indicated no significant difference in the scores across the years (p=0.228), suggesting stability and consistency in the categories assigned to each variable. Table 2 presents the significant trends of either an increase or decrease in the incompleteness of the tumor registration form variables, as assessed using the Mann-Kendall test. The analysis results revealed a variety of incomplete patterns in the epidemiological variables examined.

Upon examining the sociodemographic variables of the patients, no significant trends of incompleteness were found for age (p=0.6077), race/ethnicity (p=0.9278), or educational level (p=0.8044). However, variables such as place of birth (p=0.0145), number of clinics attended (p=0.0218), clinical treatments (p<0.001), family history of cancer (p<0.001), alcohol consumption (p<0.001), and smoking (p<0.001) exhibited significant trends of incompleteness.

Regarding the variables related to diagnosis and treatment, patterns of increased incompleteness were observed. The variable “Federal Unit of residence” demonstrated significance (p=0.0112), suggesting an increase in data incompleteness. Similarly, the patients’ place of origin also exhibited a significant trend of increased incompleteness (p=0.0112). Other breast cancer-related variables, such as previous diagnoses and treatments (p=0.0085), key diagnostic criteria (p<0.001), tumor laterality (p=0.0086), presence of multiple primary tumors (p=0.0136), other tumor clinical staging (p=0.0124), and the pTNM classification system (p=0.0037), also showed significance with an increase in incompleteness over time (Table 2).

In figure 2, the graphs display the percentage of incompleteness for the sociodemographic, health, and lifestyle variables in the present historical series. An upward trend in incompleteness is observed for variables such as place of birth, marital status, and patient occupation, while the other variables showed fluctuations over time.

Note: The dashed line represents the temporal trend, while the solid black line indicates the temporal evolution of incompleteness.

Figure 2 : Trends in the incompleteness of sociodemographic variables in Hospital Cancer Records of women with breast cancer in the Oncology Care Network of Espírito Santo from 2000 to 2020 (N = 16,587) 

The trend curves for the incompleteness of clinical information in the tumor registration form variables of the HCR are shown in figure 3. All variables exhibited significant changes throughout the historical series. Variables such as key diagnostic criteria and disease status at the end of treatment had a high percentage of incompleteness in the year 2000. However, over the years of the temporal series, this percentage decreased, indicating an improvement in cancer records across all hospitals.

Note: The dashed line represents the temporal trend, while the solid black line indicates the temporal evolution of incompleteness.

Figure 3 : Trends in the incompleteness of clinical variables in Hospital Cancer Records of women with breast cancer in the Oncology Care Network of Espírito Santo from 2000 to 2020 (N = 16,587) 

On the other hand, variables such as treatment clinics and most relevant exams showed a low percentage of incompleteness in the early years of the series but experienced an increase in incompleteness over time. Other variables followed a different trend pattern. For instance, the occurrence of multiple primary tumors, screening date, and treatment start date showed a low incompleteness early in the series but experienced a spike in incompleteness in the middle years, with a subsequent decrease in the following years.

DISCUSSION

Studies that evaluate data incompleteness provide a comprehensive view of the completeness of epidemiological variables in cancer registries, offering insights into data quality improvement within the Oncology Care Network of Espírito Santo. Our findings highlighted that certain variables, such as family history of cancer, smoking, alcoholism, pTNM, and disease status at the end of treatment, showed significant trends of decreased data completeness.

In Brazil, a study using a prostate cancer database found that sociodemographic variables, such as race/color, were classified as having regular incompleteness, while the place of birth showed a 14.25% incompleteness in 2000 (regular) but achieved excellent ratings in 13 years and good ratings in 3 subsequent years of the time series¹⁰. The education level variable was classified as poor in 2009 (24.14% of missing data) and 2010 (38.31% of missing data) but received excellent ratings in 8 years and good ratings in 7 years. In contrast, in our study, the education level variable consistently received an excellent score throughout the entire time series¹⁰.

In another temporal series study conducted in Brazil, the authors evaluated the consistency of epidemiological variables in hospital records, such as origin and marital status, and demonstrated high consistency and stability in their records over the years, indicating a well-maintained data process⁹. However, for clinical variables like cancer staging by TNM, the classification was “very poor”. Similarly, in our study, the TNM variable consistently received a “very poor” score throughout the entire time series⁹.

In the Midwest region of Brazil, a study conducted in the state of Mato Grosso explored the quality of cancer hospital records by accessing data completeness and consistency. The study demonstrated that some variables, such as “marital status” and “sex,” showed high consistency over time, indicating stable data quality¹⁶. The authors also emphasized that in that state, variables such as “race/skin color,” “education,” and “occupation” faced challenges in terms of completeness, with variations in the filling rates over the years¹⁶.

In the state of São Paulo, a study on the completeness trends of medical records of elderly women with breast cancer showed that only the variable “family history of breast cancer” followed a decreasing trend of incompleteness, leading to an improvement in records over the years¹⁷. However, variables such as race/color, years of education, use of oral contraceptives, duration of oral contraceptive use, hormone replacement therapy, and breastfeeding showed worsening trends in records, with increasing incompleteness¹⁷.

In the United States, a study¹⁸ investigated racial and ethnic disparities in breast cancer survival by analyzing how tumor characteristics, sociodemographic factors, and treatment could mediate these disparities. The research found an increased risk of death from breast cancer for Black women compared to White women, especially when considering tumor subtype. In age-adjusted models, among women with estrogen receptor-positive tumors, Black women were three times more likely to die from breast cancer than White women in the first two years after diagnosis¹⁸.

In a systematic review and meta-analysis¹⁹ of 20 studies involving 14,103 Black women and 76,111 White women, it was found that Black women had a 19% higher likelihood of dying from breast cancer compared to White women, after adjusting for age, stage, and socioeconomic status¹⁹. The improvement in the quality of demographic data records contributes to the identification of risk factors for the development of breast cancer. A study conducted in São Paulo, Brazil, highlighted the importance of identifying risk factors and genetic polymorphisms for population screening and early breast cancer diagnosis, emphasizing that carcinogenesis is multifactorial and depends on specific tumor characteristics²⁰.

Regarding the TNM variable, our study observed a “poor” score throughout most of the observed period. Similarly, other studies also reported “poor” quality in this variable^16,21. Na In research conducted in Cuiabá, Mato Grosso, the completeness quality was classified as “very poor” for the TNM variable and “poor” for staging¹⁶. In another study conducted nationwide in Brazil²², a low percentage of inconsistency was found in staging for five cancers, including breast cancer. However, for the tumor laterality variable, the same nationwide study found a very high error rate in data filling, which contrasts with our study, where the completeness quality for this variable was “excellent” throughout almost the entire evaluated period.

Detecting significant trends in sociodemographic and clinical variables is essential for creating strategies for the prevention, diagnosis, treatment, and follow-up of cancer patients^23,24. Identifying increases or decreases in variables related to risk factors, alcoholism, and smoking indicates the need to enhance the quality of HCRs. Another notable variable is family history of cancer, where individuals with a family history of cancer, particularly in first-degree relatives, are at higher risk of developing certain types of cancer due to genetic and environmental factors that may be shared within families^25-27. The quality of cancer hospital records is crucial for reliable epidemiological analysis and for creating effective health policies^28,29. Identifying variables with missing data or inconsistencies highlights the ongoing importance of improving data collection and monitoring.

The study has some limitations, such as being restricted to data from the HCR of a single Brazilian state, which limits the generalization of the findings to other states. Therefore, caution should be exercised when interpreting the external validity of the results. While HCRs provide valuable information about the quality of healthcare services, they do not fully represent the national epidemiology of cancer.

CONCLUSION

Based on the analysis of variables from the Hospital Cancer Registry (HCR) related to malignant breast cancer cases in the Oncological Care Network of Espírito Santo, a significant trend of incompleteness was observed in essential variables for monitoring and planning oncological care. This finding highlights the need to improve the quality of records and enhance the training of the teams responsible for data collection, aiming to obtain more robust and complete data. Improving the records can support more effective public policies and provide better targeted and more appropriate care for breast cancer patients.