Characteristics and in-hospital morbidity trends associated with oral cavity cancer in Brazil

Objectives: To analyze epidemiological aspects, including temporal trends, of in-hospital morbidity from oral cavity cancer in Brazil. Methods: Observational study based on information from the entire Brazilian territory, where data on hospital admissions for oral cancer were analyzed, obtained from the Information System of the Brazilian Cancer Registry. For the temporal series analysis, generalized linear regression model was used with the Prais-Winsten method. Results: Of the 121,971 patients hospitalized with oral cavity cancer, 76.40% were male and 23.60% were female, resulting in a M:F ratio of 3.24:1. Regarding the anatomical region of involvement among hospitalized patients with oral cavity neoplastic lesions, there was a predominance in non- specic places in the mouth, such as the oor of the mouth, soft and hard palate, among others (32.68%), followed by lesions in the region of tongue (28.89%). In this population, the predominant age group was between the fth (31.09%) and sixth decades of life (24.99%); men had oral cavity neoplastic lesions at an earlier age than women. In all regions of the country, the staging of cases diagnosed in the tertiary health network accredited to the José Alencar Gomes da Silva National Cancer Institute (INCA) was late, with an increased tendency for metastasis. The temporal trend of the adjusted in-hospital morbidity rates showed to be increasing in the Northeast, South and Midwest regions for the male gender. For females, they were increasing in the Northeast and South regions. Conclusions: It is concluded that the distribution of in-hospital morbidity rates for oral cavity cancer in the country is irregular, with a predominance of male patients and higher in the Southeast and South regions, and there is a trend for this rate to increase in both genders.


Introduction
Neoplastic lesions primarily located in the lips, oral cavity, salivary glands and oropharynx are classi ed as oral cavity cancer, according to the José de Alencar National Cancer Institute (INCA, Instituto Nacional do Câncer) 1,2 .
In Brazil, 11,180 new cases of oral cavity cancer in men and 4,100 in women are estimated for each year of the 2020-2022 triennium, being, respectively, the 5th and 13th most frequent cancer among all neoplasms 1 . Also, a notable increase in the incidence of this disease has been veri ed among younger people and women 3 . Some risk factors are common for oral cavity cancer, such as the habits of smoking/chewing tobacco and drinking alcohol above the recommended levels, especially when they are associated 4,5 . Other risk factors, such as the presence of human papillomavirus (HPV) infection and excessive exposure to the sun, are determinant for cancers of the oropharynx and mouth, respectively 6 .
In turn, an early diagnosis favors the implementation of usually less invasive and more effective treatments 5 . Behaviors associated to primary prevention, early detection and adequate and timely treatment are associated, among other factors, to the organization and quality of the provided health services 7,8 . In fact, the incorporation of the monitoring of morbidity and mortality from cancer in the health management routine is of utmost importance to implement actions aimed at preventing and controlling cancer and its risk factors 9,10 .
Hence, Brazil has experienced an expansion of the population's access to oral health policies after the creation of the Uni ed Health System (SUS, Sistema Único de Saúde), with the consolidation of Primary Health Care (PHC) and the incorporation of dental care at the three levels of care through the National Oral Health Policy (PNSB, Política Nacional de Saúde Bucal) created in 2004 [10][11][12] .
The National Cancer Prevention and Control Policy establishes that cancer treatment in Brazil is carried out in quali ed health establishments, which generate data from hospital cancer records in Brazil, called the Integrating Module of Hospital Cancer Records (RHC Integrator) 13 .
Considering the scarcity of information, it was decided to study temporal series of in-hospital morbidity from oral cavity cancer, aiming at characterizing disease burden variations by Brazilian macroregion. The trend analysis for this indicator, during the assessed period, is unprecedented in Brazil. Therefore, the aim of this study was to analyze epidemiological aspects, including the temporal trend, of in-hospital morbidity from oral cavity cancer in Brazil, considering hospitalizations in an INCA-accredited health network.

Ethical aspects
This study was exempted from the analysis by the Ethics Committee because it used data obtained from public and unrestricted access databases. This study is reported in accordance with the STROBE (Strengthening the reporting of observational studies in epidemiology) guidelines 14 .

Study design and context
This is an observational study of temporal series, in which data on hospital admissions due to oral cavity cancer were analyzed, obtained from the Information System of the Cancer Hospital Registry (SisRHC, Sistema de Informação do Registro Hospitalar do Câncer) in Brazil 13 .
The study covers data from the entire national territory, between the years 2000 and 2015. The database acquisition took place through the IRHC/INCA website (download of databases containing occurrences in all states, per year, from 2000 to 2015) and this was organized and tabulated using the MS Excel software.
All reported cases of hospitalization and classi ed during the rst hospital visit as malignant neoplasms with primary location in the lips, oral cavity, salivary glands and oropharynx (ICD-10 codes C00-C10), according to the INCA 1 classi cation, were included in the study. Therefore, no sample calculation was performed.
Analyzed variables Table 1 shows the variables analyzed in this study, their descriptions and the statistical treatment applied to each one. The codes were grouped by anatomical region for better data presentation. Thus, for this study, the lesions were presented as: "lip" -originally classi ed at the IRH/INCA as C00 (lip); "tongue" -originally C01 (base of the tongue) and C02 (other parts and unspeci ed parts of the tongue); "other parts of the mouth" -originally C03 (gingiva), C04 ( oor of the mouth), C05 (palate) and C06 (other unspeci ed parts of the mouth: mucosa, vestibules, retromolar area); "salivary glands" -originally C07 (parotid gland) and C08 (other major salivary glands and unspeci ed major salivary glands); and "oropharynx"originally C09 (palatine tonsil) and C10 (oropharynx) 9 Age range Variable categorized as 20-29, 30-39, 40-49, 50-59, 60-69, 70-79 years old, 80 years old and over and "no information", considering the sum of these age groups as the total number of cases. This classi cation was chosen for age group presentation, as the assessed pathology shows a relevant incidence from adulthood onwards. There was a disproportionate frequency of cases classi ed as 0 years old, which were considered "no information", indicating problems with registration in the database.

Variables
Description and statistical treatment

Alcohol consumption
Variable categorized into: never consumed, ex-consumer, consumer and not evaluated.

Smoking status
Variable recategorized according to the original database into: never smoked, exsmoker, smoker and not evaluated (the latter also included -not applicable and no information).

Statistical Analysis
With the exception of the in-hospital morbidity rates, data were presented as relative frequency (percentage of cases). To analyze the trend of the in-hospital morbidity rates, generalized linear regression was used with the Prais-Winsten method, which allows rst-order autocorrelation correction to be carried out in the analysis of series of values organized in time. This procedure allowed classifying the rates as increasing (p<0.05 and positive regression coe cient), decreasing (p<0.05 and negative regression coe cient) or stationary (p>0.05) and enabled the quanti cation of annual averages of increase or decrease of the coe cients (annual percent change -APC) and its 95% con dence interval (95%CI) 18 .
This technique was applied on the logarithm of rates. In the analysis of trend and annual variation, the coe cients were unstable in the rst and last two years of the "2000 to 2015" series.   The highest mean in-hospital morbidity rate for men was identi ed in the Southeast region (9.36/100,000 inhab.), whereas, for women, it was identi ed in the Northeast region (2.29/100,000 inhab.). The trend of inhospital morbidity rates was increasing in Brazil, for the entire population and by gender, as well as in the Northeast, South and Midwest regions. Table 3 shows the standardized in-hospital morbidity rates due to oral cancer created to show an equitable distribution of hospital admissions for cases noti ed to INCA between the regions, considering the differences in population distribution.  Table 4 shows the trend of in-hospital morbidity rates for oral cancer, between 2002 and 2013, by macroregion and the total of Brazil, by gender and by overall distribution. The temporal trend of the adjusted in-hospital morbidity rates was increasing in the Northeast, South and Midwest regions for males, whereas, for females, it was increasing in the Northeast and South regions.

Discussion
This study identi ed that cases of hospital admissions due to oral and oropharyngeal cancer in Brazil, from 2000 to 2015, comprised mostly men, with a predominance of individuals aged 50-59 years, with low level of schooling, white and brown ethnicity, and with more advanced stage tumors, with metastasis already identi ed. In addition, the temporal trend of the in-hospital morbidity rates for oral cavity cancer in the period 2002-2013 showed to be increasing, both for males and females. The analysis of this information is unprecedented, considering the in-hospital morbidity of this disease in Brazil.
In all regions, the smoking and alcohol consumption factors were relevant to the development of the disease among hospitalized patients, as the consumption or ex-consumption of alcohol and tobacco was shown to be at higher amounts than those who declared they had never drunk alcoholic beverages or used tobacco products. Moreover, the observed increase in the consumption of alcohol-tobacco 19,20 has been associated with the increase in the disease prevalence, especially in female patients.
In all regions, the cases diagnosed at advanced staging (with metastasis) at the time of the rst hospital admission showed higher percentages than cases classi ed as initial staging/with localized involvement.
Data from the INCA Hospital Cancer Records warn that, as most patients arrive at hospitals at an advanced stage of the disease, the treatment is no longer curative, being in most cases mutilating 21 .
It was found that, for the entire Brazilian territory, the trend of in-hospital morbidity rates for oral cancer, between the years 2002 and 2013, was increasing, both for males and females. An increasing trend was also observed when analyzing both genders together. According to this nding, a study analyzing the granting of social security bene ts in Brazil for cancer of the oral cavity (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) showed this trend was increasing. Social security bene ts related to cancer involvement in other parts of the mouth, nasopharynx, oropharynx, oor of the mouth and palate showed the same increasing trend 22 .
The growing trend in the evidenced overall picture of hospitalizations may be linked to the initial hypothesis of better case registration after the consolidation of the nationwide hospital integrated information system.
As the analysis of these data is unprecedented in the literature, it was not possible to compare them with other studies on hospitalization trends due to oral cavity cancer. However, considering the results of studies that indicate a stability or decrease in mortality rates related to neoplastic conditions of the oral cavity 19,23 in contrast to the growing trend of hospitalization observed in the present study, this investigation suggests better access to health services throughout the historical series.
Users of the public health system who previously did not have access to treatment may be able to have access to the hospital network and attempt to be cured 24 . This reasoning would also be compatible with the results indicated with advanced staging, since, as this is a chronic disease, people would come to the health service network with a previously established case severity 25 .
It is worth mentioning that the inclusion and expansion of the family health network close to Brazilian communities experienced during the analyzed historical period has shown the potential to reach the population at risk for the development of oral cavity cancer 26, 27 . In this sense, a recent study shows that the inclusion of oral health teams is associated with early and timely detection of these diseases 8 . Which, associated with the possibility of access to treatments in the hospital network, would result in lower mortality rates, while in-hospital morbidity is increasing 28 . A recent study indicated the association between the number of oral health teams and the timely diagnosis of oral cavity cancer 8 .
Therefore, it is reiterated that studies of epidemiological surveys for oral cancer diagnosis and surveillance are of utmost importance, allowing better management of public policies in the health area and data that support planning by health professionals, so that users have greater chances of treatment and cure 29,30 .
As Brazil is a country with a large territorial extension and has many regional variations, both geographical and populational, the importance of developing research by state to assess the speci c characteristics of its population is highlighted. Thus, this study met its purpose of identifying trends and characterizing epidemiological aspects related to morbidity from oral and oropharyngeal cancer in Brazil, showing data that had not been published before in the literature.
We point out some limitations of the present study, since by using public information systems, it was observed that some data were under-recorded (classi ed as not applicable or missing), which seems to correspond to the failure to complete the medical records at the time of hospitalization and also the migration of some databases from regional platforms to a national platform, particularly in the state of São Paulo, which has a single state database 15 .
Nevertheless, it is considered that the data shown here constitute the best and most comprehensive information on oral cancer morbidity available in the Brazilian territory. In addition, these data are already consolidated and have a policy of information monitoring and checking at state and national levels to avoid duplication of information 13 , giving them reliability and robustness.
Our results demonstrate that greater attention should be paid to the early detection and treatment of oral cancer, reinforcing that dedicated professional action is still required in the ght against the disease and educating the population about risk factors and signs and symptoms of oral cavity cancer. Also, it is necessary to expand cancer care, by involving the Primary Health Care level and high complexity centers.
We conclude that the distribution of in-hospital morbidity rates for oral cavity cancer in the country is irregular, with a predominance of the male gender and in the Southeast and South regions, and that there is a trend for this rate to increase in both genders.

Declarations
Ethics approval and consent to participate All information is of public domain, with no identi cation of patients and for that reason, there was no need for project approval by the Ethics Committee for Research with human beings. The exemption was taken by the Ethics Committee in Human Beings of the Federal University of Mato Grosso do Sul