Predictors for oral cancer in Brazil

Introduction: The incidence of lip, oral cavity and oropharynx cancer in Brazil is one of the highest worldwide. Objective: This study aimed to identify predictors for oral cancer in Brazil between 2010 and 2013. Method: Through a time series study in which 14,959 primary head and neck cancer diagnoses were evaluated. The variables of interest were gender, age, race, education level, family history of cancer, alcohol consumption, smoking, and previous cancer diagnosis. The outcome variable was divided into “oral cancer” and “cancer of other head and neck regions.” The data were analysed by multiple binary logistic regression; α=5%. Result: The protective factor was: approximately 12 years of education (OR = 0.85). The risk factors were: being an ex-consumer (OR=1.19) or consumer (OR=1.11) of alcohol, tobacco use (OR=1.35) and a prior diagnosis of cancer that went untreated (OR=1.21). Conclusion: Was concluded that the oral cancer had the following predictors compared to other types of head and neck cancer during the same period: approximately 12 years of education (protective factor) and ex-consumer or consumer of alcohol, smoking and previous diagnosis of cancer that went untreated (risk factors). Descriptors: Cancer; public health; social determinants; epidemiology.


INTRODUCTION
Cancer is a growing worldwide public health problem 1,2 .Recently, there have been approximately 300,000 new cases of cancer in the lip, oral cavity and oropharynx and approximately 140,000 deaths 3 .The incidence of these types of cancer in Brazil is one of the highest worldwide 4,5 .
Cancers located in the lip, oral cavity and oropharynx regions are generally grouped together in epidemiological studies because they have similar histopathological characteristics and risk factors 6 .The increase in cancers located in the head and neck regions [7][8][9][10] has attracted academic interest, especially in the search for preventable risk factors related to habits and life styles 11,12 .
Population-based cancer registries (PBCR's) provide permanent information on the number of new cases of cancer and allow the collection of information on the incidence and conditions that may affect disease occurrence, thereby strengthening epidemiological evaluations 13 .
The aim of this study was to identify predictors of cancer located primarily in the lip, oral cavity and oropharynx regions in Brazilian patients during the period between 2010 and 2013 using population-based cancer registries.

Study Population
This study was a time series study in which all primary head and neck cancer diagnoses (codes C00-C13, C30.0, C31.0 and C31.1 according to the international classification of diseases (ICD 10-2011) 6 between 2010 and 2013 were selected.These data were obtained from the PBCRs of the National Cancer Institute (Instituto Nacional do Cancer -INCA) and comprised the hospital cancer records (HCRs) provided by the HCR integrator, which brings together information from 260 Brazilian hospitals on approximately 25 platforms.These data are provided free of charge on INCA's website 14 .

Data Collection
A total of 30,342 cancer records related to the study period (2010-2013) were collected from around the country.Of these, only those that contained complete information on the independent variables of gender, age, race, education level, family history of cancer, alcohol consumption, smoking, and previous diagnosis of cancer were considered for the purposes of the sample.In total, 14,959 records with complete information were collected.
The dependent variable was the location of the cancer.This variable was divided into "oral cancer" (located in lip, oral cavity and oropharynx) and "cancer of other head and neck regions" to create a dichotomous outcome.The reference categories for each of the independent variables were gender (male), age, skin colour (white), literacy (no education), family history (absence of family history), alcoholism (non-consumers), smoking (non-consumers) and previous diagnosis of cancer (none).
In total, 10,464, 10,284, 5,746 and 3,848 cases of cancer of the head and neck were recorded in the Brazilian HCRs in 2010, 2011, 2012 and 2013, respectively, corresponding to 30,342 cases.However, some of the records were excluded from the analysis due to a lack of information regarding any of the variables of interest, resulting in the inclusion of 14,959 records in the general database.

Statistical Analysis
The data were analysed using R software (3.1.3)(Bell Laboratories; Auckland-New Zealand).A normality test (Kolmogorov-Smirnov) was performed for the "age" variable, which revealed a non-normal distribution type (p = 0.000).The Mann-Whitney test was performed to establish the difference in the ages of patients with primary cancer of the lip and oral cavity and those with cancer in other regions of the head and neck.
A univariate binary logistic regression was conducted in which each variable was evaluated as an outcome predictor with a 5% significance level.Then, all variables with a significance below 5% were included together in a multivariate model.This model was adjusted using the backward method.A 5% significance level was adopted for the permanence of the variables in the final model.

RESULT
A total of 8,147 (54.4%) of the total evaluated records (14,959) described oral cancers.The mean age of the subjects was 59.11 years (±12.75),and the median age was 58.00 years.The mean age of patients with cancer of the lip, oral cavity and oropharynx was 59.39 years (±12.65 years) and of patients with cancer of other head and neck region cases was 58.90 years (±12.82),with no significant difference between the groups (p=0.647).
Table 1 shows that both the oral cancer and cancers of other head and neck regions were primarily concentrated in males with white skin with an incomplete primary education.There was a lower incidence in individuals with higher education levels.The cancer in the two regions considered for the outcome was primarily found in individuals with no family history of cancer, alcohol and tobacco consumers and those who had a previous diagnosis of cancer and had not undergone treatment.
Of all the cancers found in the head and neck, the most common histological type was squamous cell carcinoma (ICD 8070/3), both of the oral region (89.0%) and of other head and neck regions (89.0%).This cancer type affected 83.6% of all individuals.
The univariate analysis for each of the variables of interest to the outcome is shown in Table 2.
The data in Table 3 was obtained for the "literacy" variable, where having reached the secondary education level reduced the chance by 1.16-fold.
Alcoholism was a risk factor for both ex-consumers (OR = 1.19) and consumers (OR=1.11),as was smoking (OR=1.35)and a previous diagnosis of another cancer that was not treated (OR=1.21).
The ages of individuals affected by oral cancer and other head and neck regions corresponded to the ages reported by Casati et al. 19 , who found that head and neck cancers usually developed between 50 and 70 years of age.However, Chor et al. 17 did not establish the age variable as a risk factor for cancer in lip, oral cavity and oropharynx; this result differed from the study by Ribeiro et al. 11 , where increasing age was associated with cancers of the lip and oral cavity.
There was a predominance of squamous cell carcinoma in both the lip, oral cavity and oropharynx regions and in the other head and neck regions, which also occurred in other parts of the world 2,17,[20][21][22][23][24][25] .
The male gender is more commonly associated with cancers of the head and neck, especially those that involve the lip, oral cavity and oropharynx 11,17 .Radoï et al. 23 identified an association between the consumption of alcohol and tobacco and the occurrence of oral cancer in men in France.
Ribeiro et al. 11 and Weatherspoon et al. 2 found that higher education levels behaved as protective factors for the occurrence of cancer of the lip and oral cavity.The identification of factors associated with the development of the oral cancer contribute to the prevention of cancer in these areas.The finding that the main risk factors are preventable is important information for health managers and should encourage campaigns for the prevention of a cancer that has a considerable number of comorbidities and an average survival of only 5 years over the last 40 years (30-40%) [22][23][24] .
In addition to the issues of comorbidities and survival, it is also important to note that the highest number of oral cavity cancer deaths was found in individuals between the fifth and sixth decades of life and in patients who had a low level of education 23 .
The attention of oral health managers and professionals should be drawn to preventive measures, including patient education in which potential risk factors are explained, and to establishing the patient profile associated with cancer of the lip, oral cavity and oropharynx 2 .
The study of HCR data may have limitations because the data are not always complete 11,18 ; this limitation is inherent in studies using secondary data.In light of this, it was necessary to exclude 50.69% of the collected records to perform a multiple analysis.Although incomplete data were excluded, the number of cases evaluated allowed a robust multivariate analysis with the generation of predictors for the occurrence of oral cancer during the evaluated period.
The characteristics of this study did not allow us to examine the association of cases with human papilloma virus (HPV) infection, which has been associated with the development of squamous cell carcinoma of the head and neck 17,21,25 , especially in the oropharyngeal region 17 .However, the characteristics did permit the association of factors frequently reported to be risk factors in studies with different designs, such as alcoholism 17,22 and smoking 22 .The latter factor has also been associated with squamous cell carcinoma due to its promotion of mutations in the TP53 gene 17 .The present study demonstrates that in Brazil, the model adjusted with the variables from the national databases offers evidence of a greater risk of cancer of the lip, oral cavity and oropharynx than that of other head and neck regions over a four-year period when associated with the risk factors alcoholism (ex-consumer and consumer) and smoking as a previous diagnosis of cancer that went untreated.
The predicted protective factor of oral cancer were the completion of approximately 12 years of education during the evaluated period compared to other types of cancer in the head and neck.Being an ex-consumer or consumer of alcohol, smoking and having a previous diagnosis of cancer that went untreated were risk factors.

Table 1 .
Absolute and relative frequency values for the independent variables included in the study according to the cancer location, in Brazil (2010-2013) (n=14,959)