Implementation of a Data System to Better Characterize Brazilian Colorectal Cancer Screening Population
Implementation of a Data System to Better Characterize Brazilian Colorectal Cancer Screening Population
Colorectal cancer (CRC) is the third most common type of cancer among men and the second among women in Brazil. Despite the high incidence and significance of CRC in Brazil, very little is known about its prevalence among the asymptomatic population. Recently, a CRC screening program was implemented at the Cancer Hospital of Barretos. Characterization of the clinical findings detected in the screening population and the prevalence of basal CRC might contribute to better organization of the program and define the best strategy for a future national screening program. We hypothesize that recruitment and the early outcomes of our screening program based on the fecal immunochemical test (FIT) will differ from the outcomes corresponding to other populations due to sociodemographic differences. Aims: i. To implement a data collection and storage system for follow-up of the screening program participants and to measure early outcomes (adenoma, advanced adenoma and cancer) and associate them with sociodemographic risk factors; ii. to quantify the risk of CRC in the Brazilian population and to develop algorithms for risk stratification of CRC screening; and iii. to compare the risk stratification to other countries with low, medium and high incomes. Methods: Individuals aged 50 to 65 years will be included in the HCB screening program from November 2017 to December 2018. The following data will be collected from all participants: sociodemographic and ethnic (skin color) characteristics; risk factors for CRC, such as smoking and drinking; comorbidities, including diabetes mellitus and arterial hypertension; and FIT, colonoscopy and histopathology examination results. Data collection will be performed using the REDCap data collection/database system. The risk score will be formulated using the Chi-square test (or Fisher's exact test) and simple logistic regression, and the regression coefficients will be calculated. Then, the model identified for the training sample will be replicated with a validation sample. The resulting score will be used to calculate the sensitivity, specificity, positive predictive value, negative predictive value, accuracy, area under the receiver operating characteristic (ROC) curve and Kolmogorov D statistic.
Variables and data collection Upon entry into the program, the following data will be collected from all participants: sociodemographic and ethnic (skin color) characteristics; risk factors for CRC, such as smoking and drinking; comorbidities, including diabetes mellitus and arterial hypertension; and the FIT, colonoscopy and histopathology examination results.
Data collection will be performed using the REDCap electronic data capture tools hosted at Barretos Cancer Hospital (Paul et al., 2009). For this purpose, clinical forms were developed containing questions relative to the data to be analyzed in the present project.
Colonoscopy procedure Colonoscopies will be performed according to the routine procedures of the Department of Prevention Endoscopy, HCB. Bowel preparation will begin on the eve of the test and will consist of a residue-free liquid diet and oral intake of 10 mg of bisacodyl (5 mg/tablet, Boehringer Ingelheim do Brasil Química e Farmacêutica Ltda, São Paulo, Brazil) at 2:00 and 6:00 pm. On the test day, preparation will begin 5 hours before the procedure and will consist of the oral intake of 500 ml of 20% mannitol.
Prior to the colonoscopy, the patients will be sedated with fentanyl (0.05 mg/ml - Cristália Produtos Químicos Farmacêuticos Ltda, São Paulo, Brazil), midazolam (5 mg/5 ml - Produtos Roche Químicos e Farmacêuticos S.A., Rio de Janeiro, Brazil) and propofol (10 mg/ml - Cristália Produtos Químicos Farmacêuticos Ltda, São Paulo, Brazil) via the intravenous route. Then, a flexible endoscope will be introduced up to the cecum and terminal ileum.
Oxygen will be supplied at a flow rate of 2 l/min throughout the procedure. The heart rate and oxygen saturation will be monitored. All tests will be performed using high-definition colonoscopes (Olympus 180), and CO2 insufflation will be performed by senior endoscopists with extensive experience in diagnostic colonoscopy.
The following data will be recorded throughout the procedure: colonoscope introduction and withdrawal times; quality of bowel preparation according to the Boston scale; and location and characteristics of the detected lesions. Accurate reports of the study outcomes (adenoma, advanced adenoma and carcinoma), including the number, size and precise location, will have paramount importance. The detected lesions will be classified following the Paris classification (Paris Classification, 2002) as follows: type 0-I, polypoid (0-Is: sessile; 0-Isp: sub-pedunculated; and 0-Ip: pedunculated) and type 0-II, non-polypoid (0-IIa: slightly elevated surface; 0-IIb: flat; 0-IIc: slightly depressed; and 0-III: excavated).
Statistical analysis All statistical analyses will be performed using the SPSS software for Windows, version 21.0. In all cases, the significance level will be set to 0.05.
The sample will be described in terms of the mean, standard deviation, minimum, maximum and quartiles for quantitative variables and through frequency tables in the case of qualitative variables.
The risk score will be formulated as follows:
The sample will be composed of individuals diagnosed or not with cancer. Individuals with adenoma or advanced adenoma and individuals without cancer will be excluded from the analysis. Cases with a negative FIT will be considered No Cancer. Cases with a positive FIT will be subjected to colonoscopy, the results of which will allow the classification of each participant as Cancer or No Cancer. The sample will be randomly divided into two groups. The first group will comprise 75% of the data and be named the training sample. The second group will comprise the remaining 25% of the data and be named the validation sample.
The training sample will be used to develop the risk score. First, the relationship of all of the participants' characteristics with the outcomes (Cancer/No Cancer) will be investigated using the Chi-square test (or Fisher's exact test) and simple logistic regression. All characteristics with a p-value less than 0.2 in the previous assessment and other characteristics that the investigators judge relevant will be included in a multiple logistic regression model; then, the regression coefficients will be calculated.
Next, the model identified for the training sample will be replicated with the validation sample; the resulting score will be used to calculate the sensitivity, specificity, positive predictive value, negative predictive value, accuracy, area under the receiver operating characteristic (ROC) curve and Kolmogorov D statistic.
If the model is considered discriminant, nomograms will be plotted, and risk groups will be defined based on the risk score.
Inclusion Criteria:
Exclusion Criteria:
guimaraes.dp@gmail.com551733216600 ext. 7267