Colorectal Cancer Deaths in South America: time-series analysis

Colorectal Cancer Deaths in South America: time-series analysis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Colorectal Cancer Deaths in South America: time-series analysis Alvin J. H. Correa, Ernesto D. Freiberg, Silvia J. Birnenbaum, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3001420/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Colorectal cancer (CRC) in South America causes 69,435 deaths annually; mortality rates are heterogeneous across countries. Limited resources in public health systems defy screening and treatment interventions, rendering forecasting and modeling invaluable policy-making tools. We aimed to forecast CRC deaths in individuals aged 55 or older in South America for the period 2020–2030 using time-series analysis and assess heterogeneity in CRC death rates. Methods Time-series analysis were used to forecast CRC mortality in South America in individuals aged 55 + years. ARIMA and Bayesian with Gaussian processes (GP) and Markov-Chain Montecarlo Simulation models were developed. Heterogeneity in CRC mortality rates across countries was evaluated. Data were extracted from the Global Burden of Disease study. Results Between 2020 and 2030, South American countries are expected to record between 680,514 (ARIMA) and 548,372 (GP) CRC deaths in individuals aged 55 or older, both sexes. Brazil and Argentina with the highest burden. The ARIMA model predicts an increase in annual deaths from 52,905 (2020) to 70,811 (2030), while the GP model predicts a slight decrease from 51,343 (2020) to 43,344 (2030). Mortality rates vary significantly between countries, with Uruguay having the highest in 2010 (236.12/100,000) followed by Argentina and Brazil. Conclusion Both the ARIMA and GP models predicted that more than half million people would dye of CRC in South America during the next decade. Mortality rates will be heterogeneous among countries. Accurate forecasting is essential for effective public policies. Continued research and concerted efforts are necessary to address the burden of CRC and to explain causes of varying mortality rates. Colorectal cancer deaths mortality modelling South America ARIMA Gaussian Processes Bayesian analysis Time-series Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Background Colorectal cancer (CRC) ranks as the third leading cause of cancer deaths worldwide [ 1 ] and poses a major health challenge for South America in the next decade [ 2 ]. Cancer is currently the primary or secondary cause of premature death in South American countries and is forecasted to surpass cardiovascular diseases along the 21st century [ 3 ]. The region has the highest incidence rates of all cancers in the Americas, affecting both sexes [ 4 ]. It is estimated that approximately 69,435 people die from CRC each year in Latin America, with CRC being among the top two causes of cancer deaths in South America [ 4 ]. Mortality from CRC is highly variable across the region. Uruguay has the highest standardized mortality rate for CRC in the subregion (21.1/100,000 people/year), and Argentina experiencing the second highest number of deaths after lung cancer, and one of the highest standardized mortality rates for CRC in the subregion as well (17.1/100,000 people/year) [ 5 ]. The cumulative risk of death from CRC among males up to 74 years of age was approximately 2% in Uruguay and Argentina in 2012, twice that of Brazil (1%) [ 6 ]. Neighboring countries have lower mortality rates from CRC and the reasons for this heterogeneity have not been well evaluated and could be attributed to the absence of reliable vital registration systems or other epidemiological and demographic differences. CRC is a potentially preventable disease [ 7 ]. Despite the recent controversy regarding the place of colonoscopy in its early detection [ 8 ], the US Preventive Services Task Force has concluded with high certainty that screening for CRC in adults aged 50 to 75 years has substantial net benefit over morbidity and mortality [ 7 ], with an obvious impact at the community level. The recommended screening options include detection of blood in the stool, detection of DNA markers associated with CRCs, detection of advanced precancerous lesions in cells shed from the colon or rectum into the stool, or direct visualization of the tumor or precancerous lesions with colonoscopy or computed tomography colonography. Structural deficits in the public health systems of South American countries [ 5 ] determine fewer human and technological resources to conduct CRC screening [ 9 ]. There are not enough colonoscopes, biochemical reactants are not widely available, poverty itself imposes additional barriers for screening, and there is less accessibility to high-quality curative treatment [ 6 ]. Having an accurate and well-informed hypothesis about the future allows for better decisions to be made. Anticipating the impact of CRC in the region could be an essential tool for designing public health policies aimed at reducing the impact of the disease on the population and could allow policymakers to set anticipatory scenarios for intervention [ 10 ]. Time-series forecasting involves the collection and analysis of historical observations through modelling. Modeling allows us to understand the long-term behavioral features of the system [ 11 ]. The aim of this study was to use time-series analysis to forecast the expected number of deaths from CRC in individuals aged 55 years and older of both sexes in South American countries over the next decade. The study also assessed the heterogeneity of CRC deaths incidence across countries. Methods Our study aimed to forecast the mortality due to colorectal cancer among people aged 55 years or older in South America, using a dataset of annual deaths by country. To achieve this, we developed two distinct time-series analysis models: an ARIMA model and a Bayesian model based on Gaussian Processes and Markov-chain Montecarlo Simulations. The models were implemented using relevant libraries in Python. We have made our models and databases available on a GitHub repository for easy access [ 12 ]. In addition, we conducted an evaluation of the heterogeneity of CRC deaths across South American countries and discussed our findings in detail. Data Source South American countries were defined according to the Panamerican Health Organization (PAHO)/World Health Organization (WHO): Argentina, Plurinational State of Bolivia, Brazil, Bolivarian Republic of Venezuela, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay [ 13 ]. Population data and projections for 1990–2030 were obtained from the World Population Prospects 2019 published by the United Nations (UN) Population Division [ 14 ]. To develop a model to forecast the expected number of deaths from CRC in South America until 2030, we created a dataset of all deaths due to CRC in the region over 30 consecutive years between 1990 and 2019, for both sexes, 55 years old and older. National tabulated data on the numbers of colorectal cancer deaths were obtained from the University of Washington, Institute for Health Metrics and Evaluation, Global Burden of Disease website [ 15 ]. Heterogeneity analysis of CRC mortality between countries To identify heterogeneity in CRC year mortality between countries in South America, age-specific mortality rates were calculated for each country, for both sexes, and aged 55 years and older. The data were graphed for visual comparison. Statistical Model The observed deaths due to CRC from 1990 to 2019 for every country included were the input to two different inference forecast models for estimating the year mortality from 2020 until 2030: first, an autoregressive integrated moving average (ARIMA) model, and second, a Bayesian model based on Gaussian processes and Markov-Chain Monte-Carlo simulation. There was no patient or public involvement in this study and the protocol was approved by the ethics committee of the JF Kennedy University in Argentina. ARIMA model Our data are the number of deaths distributed over time at intervals of years; the model should fit the data and then be able to forecast the same quantities for the 2020–2030 period. The autoregressive integrated moving average (ARIMA) model is one of the most widely accepted time series models. The parameters p , q , and d in the ARIMA (p, d, q) model were estimated, where p is the number of autoregressive lags, d is the number of differentiations required to make the model stationary, and q is the order of the moving average. Model identification was based on the algorithm introduced by Box and Jenkins [ 16 ]. The stationarity of the series and other assumptions for the ARIMA model were checked. Visually, the series was non-stationary; in addition, it was assessed using the Dicky-Fuller test [ 17 ]. The series was made stationary through differentiation to stabilize the variance. From the empirical autocorrelation function, a provisional model was chosen. The maximum likelihood method was used to calculate the parameter of the autocorrelation estimation. Then, the autocorrelation function was defined and the adequacy of the model, specifically autocorrelation function residuals for white noise, was checked with a second correlogram for residuals. This process was continued until the model adequately fitted the data. To obtain the forecast for years ahead of the ARIMA model we created a model equation by replacing the future values of the random shocks with zero and past values with observed residuals. Based on the final selected model, we forecasted the expected annual number of CRC deaths for each country and then conducted a separate analysis for the entire region. The 95% CIs were calculated using the mean squared error of the model. Bayesian model Although ARIMA is widely used in forecasting, it has a fundamental limitation: it is a linear model that may not accurately capture the complex dynamics of biological phenomena. The Bayesian model we propose, on the other hand, recognizes that there exist a family of mathematical functions that can effectively fit the data on CRC mortality over time. To identify the optimal function that fits the data, we assigned a prior probability to each potential function representing yearly CRC deaths by country. Bayesian approach is based on the idea that a priori probabilistic distributions exist for the unknown parameter [ 18 ], that in our case should be one that governs the function that best fit our data about CRC mortality. In this case the prior is a distribution of possible functions able to explain data probabilistic distribution. Functions that exhibited smoother characteristics were deemed more plausible, and thus assigned higher probabilities. Therefore, we performed a Bayesian model [ 19 ] based on Gaussian processes and the Markov-chain Monte Carlo (MCMC) method. Gaussian Processes (GPs) are powerful tools for modeling correlated observations, as is the case with time series. Gaussian processes are a type of probabilistic model commonly used in machine learning and statistics, that can be used with various algorithms to perform regression, classification, and optimization. In essence, a GP defines a distribution over functions, where any finite set of function values has a joint Gaussian distribution. As we were looking for a function that could fit the data and then to be projected into the future to make the forecast. We assumed that uncertainty in the measurements determined no clear parametric form for the function being modeled [ 20 ]. The observed CRC deaths could be explained by many, if not infinite functions; through GP we assign a probability to each of these possible functions, assumed to be normally distributed, and then the mean of this distribution is expected be the best fitting function to the existing data. This function is used to make the forecast. The process is Bayesian, what allows us to incorporate previous believes about the expected behavior of CRC mortality. The GP provides a prior function that captures prior beliefs about the function behavior [ 21 ] in terms of its mean and variance, and given the observations, the prior is updated to form the posterior distribution. This posterior distribution is another GP used to predict the value of the function in the forecast periods. The uncertainty was also quantified, and the hyperparameters of the prior distribution were estimated. These hyperparameters are used to estimate the regular components of the series. These hyperparameters are the mean and variance of the prior and were generated from two gamma distributions by changing the values randomly, until an optimal fitting was achieved. The prior was noninformative, and as deaths are positive values the gamma distributions were preferred. This is the first step in the prior distribution estimation. The second component of the time series, the tendency component, was estimated using a normal distribution with mean µ = 0 and variance s 2 = 4 chosen by convenience. Both components were added to generate the prior distribution for the likelihood calculation, with the data of CRC deaths and a white noise component to maintain the randomness. With these instructions, Markov-Chain Monte Carlo simulations were implemented. Markov Chain Monte Carlo (MCMC) simulations is a class of algorithms used to sample from a probability distribution that is difficult or impossible to sample from directly, as it is the case with the probability distributions derived from the GP. MCMC is a powerful tool for Bayesian inference, which is a statistical framework for updating probabilities based on new data. MCMC algorithms work by generating a sequence of samples from the target distribution, where each sample is dependent on the previous sample. The sequence of samples forms a Markov chain, where the distribution of each sample depends only on the previous sample, not on the entire history of the chain. This allows the algorithm to explore the entire space of possible values and converge to the target distribution, even for complex, high-dimensional distributions. This process generated the posterior of our Gaussian process, with which back-casting or curve fitting to the data and forecasting were estimated. The final output was a projection of the number of CRC deaths per year for the next decade with its 95% Credibility Interval. Results Between 1990 and 2019, South American countries recorded a total of 971,424 deaths due to CRC in individuals aged 55 years and older, of both sexes (Table 1). As expected from the respective populations, most deaths in this region are in Brazil and Argentina. Reported CRC deaths grew in all included countries. The results of the CRC death estimation between 2020 and 2030 for every selected country are provided in two separate tables, one for the ARIMA model (Table 2) and the other for the GP (Table 3). According to the ARIMA model, the number of deaths from colon cancer is expected to increase in South America from 52,905 (95% CI 49,759-56,051) in 2020 to 70,811 (95% CI 68,273-73,349) in 2030, indicating an evident rise over the 2020-2030 period (Table 4). In contrast, the GP model predicts a slight decrease in the number of annual deaths from colorectal cancer from 51,343 (95%CrI 51,306-51,381) in 2020 to 43,344 (95%CrI 43,291-43,397) in 2030 (Table 4). Based on the GP model, the highest burden of CRC deaths is projected to be in Brazil and Argentina, with a total of 139,503 (95%CrI:138,000-141,010) and 73,958 (95%CrI:71,612-76,301) CRC deaths respectively. The estimates for the ARIMA model were slightly higher, at 157,782 (151,366-164,198) and 81,970 (95%CI:79,698-84,242), respectively, for the same period. The figures for both models are presented in separate sets, the ARIMA (Figures 1) and the Gaussian Process model (Figures 2). The totals for the regions are shown in two separate figures for ARIMA (Figure 3) and GP (Figure 4). Both models predict that South America will experience a sizable number of CRC deaths throughout the 2020-2030 period, with the ARIMA model estimating 680,514 (95%CI 652,563-708,465) deaths and the GP model estimating 548,372 (95%CrI 546,924-549,821) deaths. The reported CRC mortality rates are heterogeneous between countries (Table 5). For example, in 2010, the reported CRC mortality rate for Uruguay was 236.12 deaths/100.000, and it was 85.64 deaths/100.000 for Colombia during the same period. It is easier to compare the CRC mortality rates for each age group between countries in the set of figures (Figure 5). CRC mortality rate by age group increased with age (Table 6). It increased in the oldest group during the time series (Figure 6). Discussion This study highlights the significant burden of CRC fatalities in South American countries. Between 1990 and 2019, a total of 971,424 deaths among individuals aged 55 years and older were attributed to CRC. This burden is expected to increase from 2020 to 2030. During the expected period, Brazil and Argentina are expected to experience the most significant burden of CRC fatalities. The data illustrate a significant increase in documented CRC fatalities across all included countries, albeit with notable variation in its mortality rates. We employed two distinct time-series analysis models, namely ARIMA and a Bayesian model using GP and the MCMC simulations, to project annual CRC deaths during the 2020-2030 period. Both models demonstrate that South America will persist in grappling with a substantial number of deaths resulting from colorectal cancer throughout the predicted period. The ARIMA and GP models estimated a total of 680,514 and 548,372 CRC deaths, respectively, for the span of 2020-2030. As noted, the estimates derived from each model displayed discrepancies. The ARIMA model predicts a constant increase in the number of deaths, whereas the GP model predicts a slightly parabolic trajectory. In a study [4], the authors have made a conclusion that there will be an increase in the expected cancer burden in the region by 2040, assuming that rates remain constant. The expected cancer burden in the region is projected to increase by 66% by 2040, with the incidence of new cases rising from 1.5 million to over 2.4 million annually. However, this particular study did not provide projections for colorectal cancer specifically. According to the GLOBOCAN projections [22], the region under current consideration is expected to experience 59,400 annual deaths due to colorectal cancer by 2030. However, our estimations, founded on the ARIMA and GP models, indicate a deviation from this forecast, with an anticipated annual death toll of 70,811 (95% CI 68,273-73,349), and 43,344 (95% CrI 43,291-43,397), respectively, by 2030. The issue of cancer mortality projections is a matter of considerable debate [23] given their critical role in intervention planning and evaluation, and the need for careful management of estimates. GLOBOCAN assumes that the national rates will remain constant throughout the prediction period of 2020-2040 and that the national population projections for those years are accurate [24]. This assumption is speculative. Consequently, they base their estimations on age-adjusted incidences and population structure, which differs from the Bayesian model utilized in our estimation. Our study, instead of assuming population structure or current mortality rates, utilizes GP and MCMC simulations to find out a function that fits the past data and which could project the estimations into the future, rather than focusing on the process that generates the observed number of deaths. Both approaches are a matter of debate in statistical science [20]. This research study highlights the extensive fluctuations in mortality rates pertaining to CRC across various countries. The rates ranged from 85.64 deaths per 100,000 individuals in Colombia to 236.12 deaths per 100,000 individuals in Uruguay in the year 2010. It is noteworthy that across all countries, mortality rates demonstrated a positive correlation with age, with the elderly population exhibiting the highest rates. The variable mortality rates associated with CRC in South American countries are not novel discoveries; however, their complex nature requires comprehensive investigation. Numerous factors may contribute to this inconsistency, including variations in the age distribution of the population, divergent exposure to risk factors, and disparities in the quality and precision of reporting. Despite the moderate to high quality of death certificates within the studied populations [25], these certificates may not be satisfactory for accurately recording CRC-related deaths. The validity of utilizing death certificates as a dependable tool for registering CRC-related deaths remains an open question. Although South American countries have made strides in improving the accuracy of reporting [26], registration of specific type of cancer as a cause of death remains sporadic and inadequate in certain cases [27]. The consistent correlation observed between increasing age and rising death rates in all countries aligns with the epidemiology of this particular type of tumor, indicating that any reporting deficiencies, if present, should be systematic. However, it is also plausible that the observed heterogeneity cannot be solely attributed to reporting issues. Nonetheless, this is a speculative interpretation, and further investigation is requisite to examine this matter thoroughly. The current study provides insights into the impact of mortality from colorectal cancer (CRC) in South America during the current decade. It highlights the necessity for efficacious measures for prevention and control to alleviate the impact of CRC on the populace. The use of various models for estimation allows for a thorough comprehension of the anticipated patterns in CRC mortality and establishes a structure for policymakers to conduct successful interventions. Forecast should be considered critically, as robust accurate forecasts are made when future conditions are like past conditions, something that is not always the case. The time span was ten years. How far into the future to forecast is clearly discretional and depends on our tolerance to uncertainty. Conclusion It is estimated that the South American region will witness fatalities ranging from 680,514 (ARIMA) to 548,372 (GP) CRC deaths in individuals aged 55 or older between 2020 and 2030. Brazil and Argentina are expected to encounter the greatest burden. The ARIMA model predicts an increase in fatalities, projecting from 52,905 (2020) to 70,811 (2030), while the GP model forecasts a slight decrease, from 51,343 (2020) to 43,344 (2030), annually. There is significant variation in mortality rates across countries, with Uruguay having the highest mortality rate in 2010 (236.12/100,000), followed by Argentina and Brazil. The variation in mortality rates presents a complex issue that requires further research. Accurate prediction is a vital tool for formulating effective public policies; in consequence, continuous research and concerted efforts are necessary to address the burden of CRC and base policy interventions on models. Abbreviations CRC Colorectal Cancer. GP Gaussian Process. MCMC Markov-Chain Monte Carlo simulations. ARIMA Autoregressive Integrated Moving Average. Declarations Ethics approval and consent to participate: This manuscript include human data that are publicly available. So specific concept from participants was considered not applicable. The ethics committee of the Kennedy University from Argentina approved the research, and an specialist in bioethics (Silvia Birnenbaum, Master's degree in Biomedical Ethics, Department of Ethics, Kennedy University) directly collaborated with the research. Consent for publication Publication was approved by the authorities of the Health Department of the Kennedy University of Buenos Aires, Argentina. University of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019) repository, http:// www.healthdata.org/gbd/2019 Accessed March 25, 2022. Institute for Health Metrics and Evaluation. Used with permission. All rights reserved. Under the FREE-OF-CHARGE NON-COMMERCIAL USER AGREEMENT. Availability of data and materials The datasets analyzed during the current study are available in the University of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019) repository, http:// www.healthdata.org/gbd/2019 We have made our models and databases available on a GitHub repository for easy access (https://github.com/saludglobaluk/colorectal_cancel_mortality_south_america ). Competing interests The authors declare that they have no competing interests. Funding The study was funded by the JF Kennedy University of Buenos Aires, Argentina. Authors' contributions In this manuscript, CJR conceived the study and contributed to the construction of the models and was responsible for manuscript writing; AJCH did the statistical analysis and wrote part of methos; EDF provided informatics assistance and programmed functions with Python; SJB provided ethics assistance in data management, and contributed to organization of research and internal revisions; CAN collaborated with model design and software assistance; NAP managed databases and assisted with software and libraries. Acknowledgements The authors want to acknowledge Mr. Ariel Garofalo from the Kennedy University for his assistance with figures design and manuscript revision. Authors' information (optional) Dr. Carlos Javier Regazzoni, MD, PhD is director of the Institute of Global Health, Faculty of Health Sciences, Kennedy University, Buenos Aires, Argentina. References GBD 2019 Colorectal Cancer Collaborators. Global, regional, and national burden of colorectal cancer and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019 [published correction appears in Lancet Gastroenterol Hepatol. 2022 Aug;7(8):704]. Lancet Gastroenterol Hepatol. 2022;7:627-47. doi:10.1016/S2468-1253(22)00044-9 Pilleron S, Soerjomataram I, Soto-Perez-de-Celis E, Ferlay J, Vega E, Bray F, Piñeros M. Aging and the cancer burden in Latin America and the Caribbean: Time to act. J Geriatr Oncol. 2019;10:799-804. doi: 10.1016/j.jgo.2019.02.014. Bray F, Laversanne M, Weiderpass E, Soerjomataram I. The ever-increasing importance of cancer as a leading cause of premature death worldwide. Cancer. 2021;127:3029–30. https://doi.org/10.1002/cncr.33587 Piñeros M, Laversanne M, Barrios E, Cancela MC, de Vries E, Pardo C, Bray F. An updated profile of the cancer burden, patterns and trends in Latin America and the Caribbean. Lancet Reg Health Am. 2022;13:None. doi: 10.1016/j.lana.2022.100294. Sanguinetti JM, Lotero Polesel JC, Piscoya A, Sáenz Fuenzalida R. Colorectal cancer screening: a South American perspective. Colorectal cancer screening: a South American perspective. Rev Gastroenterol Peru. 2020;40:238-45. Bray F, Piñeros M. Cancer patterns, trends and projections in Latin America and the Caribbean: a global context. Salud Publica Mex. 2016;58:104-17. doi:10.21149/spm.v58i2.7779 US Preventive Services Task Force, Davidson KW, Barry MJ, Mangione CM, Cabana M, Caughey AB, Davis EM, Donahue KE, Doubeni CA, Krist AH, Kubik M, Li L, Ogedegbe G, Owens DK, Pbert L, Silverstein M, Stevermer J, Tseng CW, Wong JB. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2021;325:1965–77. https://doi.org/10.1001/jama.2021.6238 Bretthauer M, Løberg M, Wieszczy P, Kalager M, Emilsson L, Garborg K, Rupinski M, Dekker E, Spaander M, Bugajski M, Holme Ø, Zauber AG, Pilonis ND, Mroz A, Kuipers EJ, Shi J, Hernán MA, Adami HO, Regula J, Hoff G, NordICC Study Group. Effect of colonoscopy screening on risks of colorectal cancer and related deaths. N Engl J Med. 2022;387:1547–56. https://doi.org/10.1056/NEJMoa2208375 Gualdrini U, Iummato LE. Cáncer colorrectal en la Argentina: Organización, cobertura y calidad de las acciones de prevención y control. Ministerio de Salud de la Nación [Internet]. Buenos Aires: Ministerio de Salud; 2018 [Accessed, September 30, 2022]. Disponible en: http://www.msal.gob.ar/ Doubeni CA, Selby K, Gupta S. Framework and Strategies to Eliminate Disparities in Colorectal Cancer Screening Outcomes. Annu Rev Med. 2021;72:383-398. doi:10.1146/annurev-med-051619-035840 Kaur J, Parmar KS, Singh S. Autoregressive models in environmental forecasting time series: a theoretical and application review. Environ Sci Pollut Res Int. 2023;17:1–25. doi: 10.1007/s11356-023-25148-9. – https://github.com/saludglobaluk/colorectal_cancel_mortality_south_america PAHO Countries and Centers: South America. https://www3.paho.org/commoninfo/viewsubregion.php?lang=en&idsubregion=1 Accessed, December 6, 2022. United Nations, Department of Economic and Social Affairs, Population Division (2019). World Population Prospects 2019: Data. UN Population Division Data Portal, Accessed November 2022. https://population.un.org/dataportal/home University of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019). Available: http:// www.healthdata.org/gbd/2019. Accessed March 25, 2022. Institute for Health Metrics and Evaluation. Used with permission. All rights reserved. Under the FREE-OF-CHARGE NON-COMMERCIAL USER AGREEMENT. Helfenstein U. Box-Jenkins modelling in medical research. Stat Methods Med Res. 1996;5:3-22. doi:10.1177/096228029600500102 Earnest A, Evans SM, Sampurno F, Millar J. Forecasting annual incidence and mortality rate for prostate cancer in Australia until 2022 using autoregressive integrated moving average (ARIMA) models. BMJ Open. 2019;9:e031331. doi: 10.1136/bmjopen-2019-031331. D.R. Cox. Foundations of statistical inference: the case for eclecticism. Austral J Statist. 1978;20:43-59 Roberts S, Osborne M, Ebden M, Reece S, Gibson N, Aigrain S. Gaussian processes for time-series modelling. Phil Trans R Soc A. 2013;371:20110550. http://dx.doi.org/10.1098/rsta.2011.0550 Breiman L. Statistical Modeling: The Two Cultures. Statistical Science. 2001;16,3:199–231 Corani G, Benavoli A, Zaffalon M. Time Series Forecasting with Gaussian Processes Needs Priors.” ECML/PKDD (2020). https://arxiv.org/abs/2009.08102v2. https://doi.org/10.48550/arXiv.2009.08102 (Accesed, February 21, 2023) World Health Organization International Agency for Research on Cancer (IARC). GLOBOCAN 2023: estimated cancer incidence, mortality and prevalence worldwide in 2030. Accessed May 22, 2023. Available from: https://gco.iarc.fr/tomorrow/en/dataviz/trends?types=1&sexes=1_2&mode=cancer&group_populations=1&multiple_populations=1&multiple_cancers=1&cancers=8_9&populations= 32_68_76_152_170_600_604_858_862&age_start=11&scale=log&min_zero=0&num_path=4&group_cancers=1 Bray F, Møller B. Predicting the future burden of cancer. Nat Rev Cancer. 2006 Jan;6:63-74. doi: 10.1038/nrc1781. Ferlay J, Laversanne M, Ervik M, Lam F, Colombet M, Mery L, Piñeros M, Znaor A, Soerjomataram I, Bray F (2020). Global Cancer Observatory: Cancer Tomorrow. Lyon, France: International Agency for Research on Cancer. Available from: https://gco.iarc.fr/tomorrow, accessed 22 May 2023. Seitz K, Deliens L, Cohen J, Cardozo EA, Tripodoro VA, Marcucci FCI, Rodrigues LF, Derio L, Sánchez-Cárdenas MA, Salazar V, Samayoa VR, Pozo X, Dykeman-Sabado DA, de la Lanza CC, Algaba NCB, Alvarez GP, Viana L, González T, Pastrana T. Feasibility of using death certificates for studying place of death in Latin America. Rev Panam Salud Publica. 2021;45:e149. doi: 10.26633/RPSP.2021.149. Mahapatra P, Shibuya K, Lopez AD, Coullare F, Notzon FC, Rao C, Szreter S; Monitoring Vital Events. Civil registration systems and vital statistics: successes and missed opportunities. Lancet. 2007;370:1653-63. doi: 10.1016/S0140-6736(07)61308-7. Sierra MS, Forman D. Cancer in Central and South America: Methodology. Cancer Epidemiol. 2016;44 Suppl 1:S11-S22. doi: 10.1016/j.canep.2016.07.020. PMID: 27678312. Tables Tables 1 to 6 are available in the Supplementary Files section Additional Declarations No competing interests reported. Supplementary Files Tables.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3001420","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":207963241,"identity":"3b5811b6-9451-4137-990c-0216edc4fe05","order_by":0,"name":"Alvin J. H. Correa","email":"","orcid":"","institution":"Kennedy University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Alvin","middleName":"J. H.","lastName":"Correa","suffix":""},{"id":207963242,"identity":"af222e2e-2269-49d0-b508-1c296559e07a","order_by":1,"name":"Ernesto D. Freiberg","email":"","orcid":"","institution":"Kennedy University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Ernesto","middleName":"D.","lastName":"Freiberg","suffix":""},{"id":207963243,"identity":"e880c739-f77c-4c08-9dfb-cc610da9c5e2","order_by":2,"name":"Silvia J. Birnenbaum","email":"","orcid":"","institution":"Kennedy University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Silvia","middleName":"J.","lastName":"Birnenbaum","suffix":""},{"id":207963244,"identity":"b2c482b3-3b19-4890-a556-b2677b822b39","order_by":3,"name":"Constanza Avancini","email":"","orcid":"","institution":"Kennedy University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Constanza","middleName":"","lastName":"Avancini","suffix":""},{"id":207963245,"identity":"47efee0d-532d-43a3-83b8-33771c1c240e","order_by":4,"name":"Nicolás A. Popielik","email":"","orcid":"","institution":"Kennedy University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Nicolás","middleName":"A.","lastName":"Popielik","suffix":""},{"id":207963246,"identity":"dd06a6a4-d4c6-4c10-99ff-7d6ad5bf20dc","order_by":5,"name":"Carlos Javier Regazzoni","email":"data:image/png;base64,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","orcid":"","institution":"Kennedy University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Carlos","middleName":"Javier","lastName":"Regazzoni","suffix":""}],"badges":[],"createdAt":"2023-05-30 16:59:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3001420/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3001420/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":38400240,"identity":"59d542cf-be5f-4f26-a78a-a863aab71f6a","added_by":"auto","created_at":"2023-06-12 14:08:06","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":228140,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eARIMA forecasts for South America countries.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eObservations (red line), prediction (dotted line), and 95% confidence Interval (shadow areas).\u003c/p\u003e","description":"","filename":"Binder11.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/5ccbe2f9aba4367017863fa6.png"},{"id":38400242,"identity":"279fc1b1-aa72-48a5-9418-ef12b1ff346a","added_by":"auto","created_at":"2023-06-12 14:08:06","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":356749,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGP forecasts for South America countries.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eReal data (dark line), prediction trained (red dotted line), and prediction (dark dotted line). Shadow are the simulations of Montecarlo.\u003c/p\u003e\n\u003cp\u003e(1)Plurinational State of Bolivia; (2)Bolivarian Republic of Venezuela.\u003c/p\u003e","description":"","filename":"Binder12.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/1018f7d40b2c58f4f9e649c7.png"},{"id":38400239,"identity":"79998bfd-b682-4f91-b39a-24192d1b48a0","added_by":"auto","created_at":"2023-06-12 14:08:06","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":214659,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eARIMA forecasts for South America.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eObservations (red line), prediction (dotted line), and 95% confidence Interval (shadow areas).\u003c/p\u003e","description":"","filename":"Binder13.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/e51ae4c321d96448a6f2c8d4.png"},{"id":38401164,"identity":"05213f97-4e1a-4bf6-afb9-c064021fd470","added_by":"auto","created_at":"2023-06-12 14:16:06","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":155010,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGP forecasts for South America.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eObservations (red line), prediction (dotted line), and 95% credibility Interval (shadow areas).\u003c/p\u003e","description":"","filename":"Binder14.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/ae0691f6ccf8147d0cb5417a.png"},{"id":38400245,"identity":"3a9e4256-b01c-4b7b-a378-a6c52bebb585","added_by":"auto","created_at":"2023-06-12 14:08:06","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":376959,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCRC Mortality rates by age group and country.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCRC deaths/100.000Hab by year.\u003c/p\u003e","description":"","filename":"Binder15.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/2da4190af7b840c777b0fdc4.png"},{"id":38400244,"identity":"274a22b0-53ed-4830-8486-d5a9c1a6fb2d","added_by":"auto","created_at":"2023-06-12 14:08:06","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":13069,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCRC mortality rate by age group.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIndividuals of both sexes aged 55 years and older in South American countries.\u003c/p\u003e","description":"","filename":"Binder16.png","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/bb8e3ac38aa7e9fab93b552b.png"},{"id":40863831,"identity":"307a16b7-33ab-476f-b393-b88696e9e686","added_by":"auto","created_at":"2023-08-01 06:52:27","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1454174,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/d92f4ecd-be5a-4371-8f97-3bb4bceb1525.pdf"},{"id":38401163,"identity":"1c6a49ff-f572-426f-9020-2681a0a6f3c8","added_by":"auto","created_at":"2023-06-12 14:16:06","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":64199,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-3001420/v1/7a5a878d1e31ab67ff21f24d.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Colorectal Cancer Deaths in South America: time-series analysis","fulltext":[{"header":"Background","content":"\u003cp\u003eColorectal cancer (CRC) ranks as the third leading cause of cancer deaths worldwide [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e] and poses a major health challenge for South America in the next decade [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Cancer is currently the primary or secondary cause of premature death in South American countries and is forecasted to surpass cardiovascular diseases along the 21st century [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. The region has the highest incidence rates of all cancers in the Americas, affecting both sexes [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. It is estimated that approximately 69,435 people die from CRC each year in Latin America, with CRC being among the top two causes of cancer deaths in South America [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Mortality from CRC is highly variable across the region. Uruguay has the highest standardized mortality rate for CRC in the subregion (21.1/100,000 people/year), and Argentina experiencing the second highest number of deaths after lung cancer, and one of the highest standardized mortality rates for CRC in the subregion as well (17.1/100,000 people/year) [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The cumulative risk of death from CRC among males up to 74 years of age was approximately 2% in Uruguay and Argentina in 2012, twice that of Brazil (1%) [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Neighboring countries have lower mortality rates from CRC and the reasons for this heterogeneity have not been well evaluated and could be attributed to the absence of reliable vital registration systems or other epidemiological and demographic differences.\u003c/p\u003e \u003cp\u003eCRC is a potentially preventable disease [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Despite the recent controversy regarding the place of colonoscopy in its early detection [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], the US Preventive Services Task Force has concluded with high certainty that screening for CRC in adults aged 50 to 75 years has substantial net benefit over morbidity and mortality [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e], with an obvious impact at the community level. The recommended screening options include detection of blood in the stool, detection of DNA markers associated with CRCs, detection of advanced precancerous lesions in cells shed from the colon or rectum into the stool, or direct visualization of the tumor or precancerous lesions with colonoscopy or computed tomography colonography.\u003c/p\u003e \u003cp\u003eStructural deficits in the public health systems of South American countries [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] determine fewer human and technological resources to conduct CRC screening [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. There are not enough colonoscopes, biochemical reactants are not widely available, poverty itself imposes additional barriers for screening, and there is less accessibility to high-quality curative treatment [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Having an accurate and well-informed hypothesis about the future allows for better decisions to be made. Anticipating the impact of CRC in the region could be an essential tool for designing public health policies aimed at reducing the impact of the disease on the population and could allow policymakers to set anticipatory scenarios for intervention [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTime-series forecasting involves the collection and analysis of historical observations through modelling. Modeling allows us to understand the long-term behavioral features of the system [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. The aim of this study was to use time-series analysis to forecast the expected number of deaths from CRC in individuals aged 55 years and older of both sexes in South American countries over the next decade. The study also assessed the heterogeneity of CRC deaths incidence across countries.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eOur study aimed to forecast the mortality due to colorectal cancer among people aged 55 years or older in South America, using a dataset of annual deaths by country. To achieve this, we developed two distinct time-series analysis models: an ARIMA model and a Bayesian model based on Gaussian Processes and Markov-chain Montecarlo Simulations. The models were implemented using relevant libraries in Python. We have made our models and databases available on a GitHub repository for easy access [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. In addition, we conducted an evaluation of the heterogeneity of CRC deaths across South American countries and discussed our findings in detail.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData Source\u003c/h2\u003e \u003cp\u003eSouth American countries were defined according to the Panamerican Health Organization (PAHO)/World Health Organization (WHO): Argentina, Plurinational State of Bolivia, Brazil, Bolivarian Republic of Venezuela, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Population data and projections for 1990\u0026ndash;2030 were obtained from the World Population Prospects 2019 published by the United Nations (UN) Population Division [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. To develop a model to forecast the expected number of deaths from CRC in South America until 2030, we created a dataset of all deaths due to CRC in the region over 30 consecutive years between 1990 and 2019, for both sexes, 55 years old and older. National tabulated data on the numbers of colorectal cancer deaths were obtained from the University of Washington, Institute for Health Metrics and Evaluation, Global Burden of Disease website [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eHeterogeneity analysis of CRC mortality between countries\u003c/h2\u003e \u003cp\u003eTo identify heterogeneity in CRC year mortality between countries in South America, age-specific mortality rates were calculated for each country, for both sexes, and aged 55 years and older. The data were graphed for visual comparison.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Model\u003c/h2\u003e \u003cp\u003eThe observed deaths due to CRC from 1990 to 2019 for every country included were the input to two different inference forecast models for estimating the year mortality from 2020 until 2030: first, an autoregressive integrated moving average (ARIMA) model, and second, a Bayesian model based on Gaussian processes and Markov-Chain Monte-Carlo simulation. There was no patient or public involvement in this study and the protocol was approved by the ethics committee of the JF Kennedy University in Argentina.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eARIMA model\u003c/h2\u003e \u003cp\u003eOur data are the number of deaths distributed over time at intervals of years; the model should fit the data and then be able to forecast the same quantities for the 2020\u0026ndash;2030 period. The autoregressive integrated moving average (ARIMA) model is one of the most widely accepted time series models. The parameters \u003cem\u003ep\u003c/em\u003e, \u003cem\u003eq\u003c/em\u003e, and \u003cem\u003ed\u003c/em\u003e in the ARIMA (p, d, q) model were estimated, where \u003cem\u003ep\u003c/em\u003e is the number of autoregressive lags, \u003cem\u003ed\u003c/em\u003e is the number of differentiations required to make the model stationary, and \u003cem\u003eq\u003c/em\u003e is the order of the moving average. Model identification was based on the algorithm introduced by Box and Jenkins [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. The stationarity of the series and other assumptions for the ARIMA model were checked. Visually, the series was non-stationary; in addition, it was assessed using the Dicky-Fuller test [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. The series was made stationary through differentiation to stabilize the variance. From the empirical autocorrelation function, a provisional model was chosen. The maximum likelihood method was used to calculate the parameter of the autocorrelation estimation. Then, the autocorrelation function was defined and the adequacy of the model, specifically autocorrelation function residuals for white noise, was checked with a second correlogram for residuals. This process was continued until the model adequately fitted the data. To obtain the forecast for years ahead of the ARIMA model we created a model equation by replacing the future values of the random shocks with zero and past values with observed residuals. Based on the final selected model, we forecasted the expected annual number of CRC deaths for each country and then conducted a separate analysis for the entire region. The 95% CIs were calculated using the mean squared error of the model.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eBayesian model\u003c/h2\u003e \u003cp\u003eAlthough ARIMA is widely used in forecasting, it has a fundamental limitation: it is a linear model that may not accurately capture the complex dynamics of biological phenomena. The Bayesian model we propose, on the other hand, recognizes that there exist a family of mathematical functions that can effectively fit the data on CRC mortality over time. To identify the optimal function that fits the data, we assigned a prior probability to each potential function representing yearly CRC deaths by country. Bayesian approach is based on the idea that a priori probabilistic distributions exist for the unknown parameter [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], that in our case should be one that governs the function that best fit our data about CRC mortality. In this case the prior is a distribution of possible functions able to explain data probabilistic distribution. Functions that exhibited smoother characteristics were deemed more plausible, and thus assigned higher probabilities. Therefore, we performed a Bayesian model [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] based on Gaussian processes and the Markov-chain Monte Carlo (MCMC) method. Gaussian Processes (GPs) are powerful tools for modeling correlated observations, as is the case with time series. Gaussian processes are a type of probabilistic model commonly used in machine learning and statistics, that can be used with various algorithms to perform regression, classification, and optimization. In essence, a GP defines a distribution over functions, where any finite set of function values has a joint Gaussian distribution. As we were looking for a function that could fit the data and then to be projected into the future to make the forecast. We assumed that uncertainty in the measurements determined no clear parametric form for the function being modeled [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe observed CRC deaths could be explained by many, if not infinite functions; through GP we assign a probability to each of these possible functions, assumed to be normally distributed, and then the mean of this distribution is expected be the best fitting function to the existing data. This function is used to make the forecast. The process is Bayesian, what allows us to incorporate previous believes about the expected behavior of CRC mortality. The GP provides a prior function that captures prior beliefs about the function behavior [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] in terms of its mean and variance, and given the observations, the prior is updated to form the posterior distribution. This posterior distribution is another GP used to predict the value of the function in the forecast periods. The uncertainty was also quantified, and the hyperparameters of the prior distribution were estimated. These hyperparameters are used to estimate the regular components of the series. These hyperparameters are the mean and variance of the prior and were generated from two gamma distributions by changing the values randomly, until an optimal fitting was achieved. The prior was noninformative, and as deaths are positive values the gamma distributions were preferred. This is the first step in the prior distribution estimation. The second component of the time series, the tendency component, was estimated using a normal distribution with mean \u0026micro;\u0026thinsp;=\u0026thinsp;0 and variance s\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;4 chosen by convenience. Both components were added to generate the prior distribution for the likelihood calculation, with the data of CRC deaths and a white noise component to maintain the randomness. With these instructions, Markov-Chain Monte Carlo simulations were implemented.\u003c/p\u003e \u003cp\u003eMarkov Chain Monte Carlo (MCMC) simulations is a class of algorithms used to sample from a probability distribution that is difficult or impossible to sample from directly, as it is the case with the probability distributions derived from the GP. MCMC is a powerful tool for Bayesian inference, which is a statistical framework for updating probabilities based on new data. MCMC algorithms work by generating a sequence of samples from the target distribution, where each sample is dependent on the previous sample. The sequence of samples forms a Markov chain, where the distribution of each sample depends only on the previous sample, not on the entire history of the chain. This allows the algorithm to explore the entire space of possible values and converge to the target distribution, even for complex, high-dimensional distributions. This process generated the posterior of our Gaussian process, with which back-casting or curve fitting to the data and forecasting were estimated. The final output was a projection of the number of CRC deaths per year for the next decade with its 95% Credibility Interval.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eBetween 1990 and 2019, South American countries recorded a total of 971,424 deaths due to CRC in individuals aged 55 years and older, of both sexes (Table 1). As expected from the respective populations, most deaths in this region are in Brazil and Argentina. Reported CRC deaths grew in all included countries. The results of the CRC death estimation between 2020 and 2030 for every selected country are provided in two separate tables, one for the ARIMA model (Table 2) and the other for the GP (Table 3).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAccording to the ARIMA model, the number of deaths from colon cancer is expected to increase in South America from 52,905 (95% CI 49,759-56,051) in 2020 to 70,811 (95% CI 68,273-73,349) in 2030, indicating an evident rise over the 2020-2030 period (Table 4). In contrast, the GP model predicts a slight decrease in the number of annual deaths from colorectal cancer from 51,343 (95%CrI 51,306-51,381) in 2020 to 43,344 (95%CrI 43,291-43,397) in 2030 (Table 4). Based on the GP model, the highest burden of CRC deaths is projected to be in Brazil and Argentina, with a total of 139,503 (95%CrI:138,000-141,010) and 73,958 (95%CrI:71,612-76,301) CRC deaths respectively. The estimates for the ARIMA model were slightly higher, at 157,782 (151,366-164,198) and 81,970 (95%CI:79,698-84,242), respectively, for the same period. The figures for both models are presented in separate sets, the ARIMA (Figures 1) and the Gaussian Process model (Figures 2). The totals for the regions are shown in two separate figures for ARIMA (Figure 3) and GP (Figure 4). Both models predict that South America will experience a sizable number of CRC deaths throughout the 2020-2030 period, with the ARIMA model estimating 680,514 (95%CI 652,563-708,465) deaths and the GP model estimating 548,372 (95%CrI 546,924-549,821) deaths.\u003c/p\u003e\n\u003cp\u003eThe reported CRC mortality rates are heterogeneous between countries (Table 5). For example, in 2010, the reported CRC mortality rate for Uruguay was 236.12 deaths/100.000, and it was 85.64 deaths/100.000 for Colombia during the same period. It is easier to compare the CRC mortality rates for each age group between countries in the set of figures (Figure 5). CRC mortality rate by age group increased with age (Table 6). It increased in the oldest group during the time series (Figure 6).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study highlights the significant burden of CRC fatalities in South American countries. Between 1990 and 2019, a total of 971,424 deaths among individuals aged 55 years and older were attributed to CRC. This burden is expected to increase from 2020 to 2030. During the expected period, Brazil and Argentina are expected to experience the most significant burden of CRC fatalities. The data illustrate a significant increase in documented CRC fatalities across all included countries, albeit with notable variation in its mortality rates.\u003c/p\u003e\n\u003cp\u003eWe employed two distinct time-series analysis models, namely ARIMA and a Bayesian model using GP and the MCMC simulations, to project annual CRC deaths during the 2020-2030 period. Both models demonstrate that South America will persist in grappling with a substantial number of deaths resulting from colorectal cancer throughout the predicted period. The ARIMA and GP models estimated a total of 680,514 and 548,372 CRC deaths, respectively, for the span of 2020-2030. As noted, the estimates derived from each model displayed discrepancies. The ARIMA model predicts a constant increase in the number of deaths, whereas the GP model predicts a slightly parabolic trajectory.\u003c/p\u003e\n\u003cp\u003eIn a study [4], the authors have made a conclusion that there will be an increase in the expected cancer burden in the region by 2040, assuming that rates remain constant. The expected cancer burden in the region is projected to increase by 66% by 2040, with the incidence of new cases rising from 1.5 million to over 2.4 million annually. However, this particular study did not provide projections for colorectal cancer specifically. According to the GLOBOCAN projections [22], the region under current consideration is expected to experience 59,400 annual deaths due to colorectal cancer by 2030. However, our estimations, founded on the ARIMA and GP models, indicate a deviation from this forecast, with an anticipated annual death toll of 70,811 (95% CI 68,273-73,349), and 43,344 (95% CrI 43,291-43,397), respectively, by 2030. The issue of cancer mortality projections is a matter of considerable debate [23] given their critical role in intervention planning and evaluation, and the need for careful management of estimates. GLOBOCAN assumes that the national rates will remain constant throughout the prediction period of 2020-2040 and that the national population projections for those years are accurate [24]. This assumption is speculative. Consequently, they base their estimations on age-adjusted incidences and population structure, which differs from the Bayesian model utilized in our estimation. Our study, instead of assuming population structure or current mortality rates, utilizes GP and MCMC simulations to find out a function that fits the past data and which could project the estimations into the future, rather than focusing on the process that generates the observed number of deaths. Both approaches are a matter of debate in statistical science [20].\u003c/p\u003e\n\u003cp\u003eThis research study highlights the extensive fluctuations in mortality rates pertaining to CRC across various countries. The rates ranged from 85.64 deaths per 100,000 individuals in Colombia to 236.12 deaths per 100,000 individuals in Uruguay in the year 2010. It is noteworthy that across all countries, mortality rates demonstrated a positive correlation with age, with the elderly population exhibiting the highest rates. The variable mortality rates associated with CRC in South American countries are not novel discoveries; however, their complex nature requires comprehensive investigation. Numerous factors may contribute to this inconsistency, including variations in the age distribution of the population, divergent exposure to risk factors, and disparities in the quality and precision of reporting. Despite the moderate to high quality of death certificates within the studied populations [25], these certificates may not be satisfactory for accurately recording CRC-related deaths. The validity of utilizing death certificates as a dependable tool for registering CRC-related deaths remains an open question. Although South American countries have made strides in improving the accuracy of reporting [26], registration of specific type of cancer as a cause of death remains sporadic and inadequate in certain cases [27]. The consistent correlation observed between increasing age and rising death rates in all countries aligns with the epidemiology of this particular type of tumor, indicating that any reporting deficiencies, if present, should be systematic. However, it is also plausible that the observed heterogeneity cannot be solely attributed to reporting issues. Nonetheless, this is a speculative interpretation, and further investigation is requisite to examine this matter thoroughly.\u003c/p\u003e\n\u003cp\u003eThe current study provides insights into the impact of mortality from colorectal cancer (CRC) in South America during the current decade. It highlights the necessity for efficacious measures for prevention and control to alleviate the impact of CRC on the populace. The use of various models for estimation allows for a thorough comprehension of the anticipated patterns in CRC mortality and establishes a structure for policymakers to conduct successful interventions. Forecast should be considered critically, as robust accurate forecasts are made when future conditions are like past conditions, something that is not always the case. The time span was ten years. How far into the future to forecast is clearly discretional and depends on our tolerance to uncertainty.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIt is estimated that the South American region will witness fatalities ranging from 680,514 (ARIMA) to 548,372 (GP) CRC deaths in individuals aged 55 or older between 2020 and 2030. Brazil and Argentina are expected to encounter the greatest burden. The ARIMA model predicts an increase in fatalities, projecting from 52,905 (2020) to 70,811 (2030), while the GP model forecasts a slight decrease, from 51,343 (2020) to 43,344 (2030), annually. There is significant variation in mortality rates across countries, with Uruguay having the highest mortality rate in 2010 (236.12/100,000), followed by Argentina and Brazil. The variation in mortality rates presents a complex issue that requires further research. Accurate prediction is a vital tool for formulating effective public policies; in consequence, continuous research and concerted efforts are necessary to address the burden of CRC and base policy interventions on models.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eCRC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eColorectal Cancer.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eGP\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eGaussian Process.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMCMC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eMarkov-Chain Monte Carlo simulations.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eARIMA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eAutoregressive Integrated Moving Average.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eEthics approval and consent to participate:\u003c/h2\u003e\n\u003cp\u003eThis manuscript include human data that are publicly available. So specific concept from participants was considered not applicable. The ethics committee of the Kennedy University from Argentina approved the research, and an specialist in bioethics (Silvia Birnenbaum, Master\u0026apos;s degree in Biomedical Ethics, Department of Ethics, Kennedy University) directly collaborated with the research.\u003c/p\u003e\n\u003ch2\u003eConsent for publication\u003c/h2\u003e\n\u003cp\u003ePublication was approved by the authorities of the Health Department of the Kennedy University of Buenos Aires, Argentina.\u003c/p\u003e\n\u003cp\u003eUniversity of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019) repository, http:// www.healthdata.org/gbd/2019 Accessed March 25, 2022. Institute for Health Metrics and Evaluation. Used with permission. All rights reserved. Under the FREE-OF-CHARGE NON-COMMERCIAL USER AGREEMENT.\u003c/p\u003e\n\u003ch2\u003eAvailability of data and materials\u003c/h2\u003e\n\u003cp\u003eThe datasets analyzed during the current study are available in the University of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019) repository, http:// www.healthdata.org/gbd/2019 \u003c/p\u003e\n\u003cp\u003eWe have made our models and databases available on a GitHub repository for easy access (https://github.com/saludglobaluk/colorectal_cancel_mortality_south_america ).\u003c/p\u003e\n\u003ch2\u003eCompeting interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThe study was funded by the JF Kennedy University of Buenos Aires, Argentina.\u003c/p\u003e\n\u003ch2\u003eAuthors\u0026apos; contributions\u003c/h2\u003e\n\u003cp\u003eIn this manuscript, CJR conceived the study and contributed to the construction of the models and was responsible for manuscript writing; AJCH did the statistical analysis and wrote part of methos; EDF provided informatics assistance and programmed functions with Python; SJB provided ethics assistance in data management, and contributed to organization of research and internal revisions; CAN collaborated with model design and software assistance; NAP managed databases and assisted with software and libraries.\u003c/p\u003e\n\u003ch2\u003eAcknowledgements\u003c/h2\u003e\n\u003cp\u003eThe authors want to acknowledge Mr. Ariel Garofalo from the Kennedy University for his assistance with figures design and manuscript revision.\u003c/p\u003e\n\u003ch2\u003eAuthors\u0026apos; information (optional)\u003c/h2\u003e\n\u003cp\u003eDr. Carlos Javier Regazzoni, MD, PhD is director of the Institute of Global Health, Faculty of Health Sciences, Kennedy University, Buenos Aires, Argentina.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eGBD 2019 Colorectal Cancer Collaborators. Global, regional, and national burden of colorectal cancer and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019 [published correction appears in Lancet Gastroenterol Hepatol. 2022 Aug;7(8):704]. Lancet Gastroenterol Hepatol. 2022;7:627-47. doi:10.1016/S2468-1253(22)00044-9\u003c/li\u003e\n \u003cli\u003ePilleron S, Soerjomataram I, Soto-Perez-de-Celis E, Ferlay J, Vega E, Bray F, Pi\u0026ntilde;eros M. Aging and the cancer burden in Latin America and the Caribbean: Time to act. J Geriatr Oncol. 2019;10:799-804. doi: 10.1016/j.jgo.2019.02.014.\u003c/li\u003e\n \u003cli\u003eBray F, Laversanne M, Weiderpass E, Soerjomataram I. The ever-increasing importance of cancer as a leading cause of premature death worldwide. Cancer. 2021;127:3029\u0026ndash;30. https://doi.org/10.1002/cncr.33587\u003c/li\u003e\n \u003cli\u003ePi\u0026ntilde;eros M, Laversanne M, Barrios E, Cancela MC, de Vries E, Pardo C, Bray F. An updated profile of the cancer burden, patterns and trends in Latin America and the Caribbean. Lancet Reg Health Am. 2022;13:None. doi: 10.1016/j.lana.2022.100294.\u003c/li\u003e\n \u003cli\u003eSanguinetti JM, Lotero Polesel JC, Piscoya A, S\u0026aacute;enz Fuenzalida R. Colorectal cancer screening: a South American perspective. Colorectal cancer screening: a South American perspective. Rev Gastroenterol Peru. 2020;40:238-45.\u003c/li\u003e\n \u003cli\u003eBray F, Pi\u0026ntilde;eros M. Cancer patterns, trends and projections in Latin America and the Caribbean: a global context. Salud Publica Mex. 2016;58:104-17. doi:10.21149/spm.v58i2.7779\u003c/li\u003e\n \u003cli\u003eUS Preventive Services Task Force, Davidson KW, Barry MJ, Mangione CM, Cabana M, Caughey AB, Davis EM, Donahue KE, Doubeni CA, Krist AH, Kubik M, Li L, Ogedegbe G, Owens DK, Pbert L, Silverstein M, Stevermer J, Tseng CW, Wong JB. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2021;325:1965\u0026ndash;77. https://doi.org/10.1001/jama.2021.6238\u003c/li\u003e\n \u003cli\u003eBretthauer M, L\u0026oslash;berg M, Wieszczy P, Kalager M, Emilsson L, Garborg K, Rupinski M, Dekker E, Spaander M, Bugajski M, Holme \u0026Oslash;, Zauber AG, Pilonis ND, Mroz A, Kuipers EJ, Shi J, Hern\u0026aacute;n MA, Adami HO, Regula J, Hoff G, NordICC Study Group. Effect of colonoscopy screening on risks of colorectal cancer and related deaths. N Engl J Med. 2022;387:1547\u0026ndash;56. https://doi.org/10.1056/NEJMoa2208375\u003c/li\u003e\n \u003cli\u003eGualdrini U, Iummato LE. C\u0026aacute;ncer colorrectal en la Argentina: Organizaci\u0026oacute;n, cobertura y calidad de las acciones de prevenci\u0026oacute;n y control. Ministerio de Salud de la Naci\u0026oacute;n [Internet]. Buenos Aires: Ministerio de Salud; 2018 [Accessed, September 30, 2022]. Disponible en: http://www.msal.gob.ar/\u003c/li\u003e\n \u003cli\u003eDoubeni CA, Selby K, Gupta S. Framework and Strategies to Eliminate Disparities in Colorectal Cancer Screening Outcomes. Annu Rev Med. 2021;72:383-398. doi:10.1146/annurev-med-051619-035840\u003c/li\u003e\n \u003cli\u003eKaur J, Parmar KS, Singh S. Autoregressive models in environmental forecasting time series: a theoretical and application review. Environ Sci Pollut Res Int. 2023;17:1\u0026ndash;25. doi: 10.1007/s11356-023-25148-9.\u003c/li\u003e\n \u003cli\u003e\u0026ndash; https://github.com/saludglobaluk/colorectal_cancel_mortality_south_america\u003c/li\u003e\n \u003cli\u003ePAHO Countries and Centers: South America. https://www3.paho.org/commoninfo/viewsubregion.php?lang=en\u0026amp;idsubregion=1 Accessed, December 6, 2022.\u003c/li\u003e\n \u003cli\u003eUnited Nations, Department of Economic and Social Affairs, Population Division (2019). World Population Prospects 2019: Data. UN Population Division Data Portal, Accessed November 2022. https://population.un.org/dataportal/home\u003c/li\u003e\n \u003cli\u003eUniversity of Washington, Institute for Health Metrics and Evaluation. Global Burden of Disease (GBD 2019). Available: http:// www.healthdata.org/gbd/2019. Accessed March 25, 2022. Institute for Health Metrics and Evaluation. Used with permission. All rights reserved. Under the FREE-OF-CHARGE NON-COMMERCIAL USER AGREEMENT.\u003c/li\u003e\n \u003cli\u003eHelfenstein U. Box-Jenkins modelling in medical research. Stat Methods Med Res. 1996;5:3-22. doi:10.1177/096228029600500102\u003c/li\u003e\n \u003cli\u003eEarnest A, Evans SM, Sampurno F, Millar J. Forecasting annual incidence and mortality rate for prostate cancer in Australia until 2022 using autoregressive integrated moving average (ARIMA) models. BMJ Open. 2019;9:e031331. doi: 10.1136/bmjopen-2019-031331.\u003c/li\u003e\n \u003cli\u003eD.R. Cox. Foundations of statistical inference: the case for eclecticism. Austral J Statist. 1978;20:43-59\u003c/li\u003e\n \u003cli\u003eRoberts S, Osborne M, Ebden M, Reece S, Gibson N, Aigrain S. Gaussian processes for time-series modelling. Phil Trans R Soc A. 2013;371:20110550. http://dx.doi.org/10.1098/rsta.2011.0550\u003c/li\u003e\n \u003cli\u003eBreiman L. Statistical Modeling: The Two Cultures. Statistical Science. 2001;16,3:199\u0026ndash;231\u003c/li\u003e\n \u003cli\u003eCorani G, Benavoli A, Zaffalon M. Time Series Forecasting with Gaussian Processes Needs Priors.\u0026rdquo; ECML/PKDD (2020). https://arxiv.org/abs/2009.08102v2. https://doi.org/10.48550/arXiv.2009.08102 (Accesed, February 21, 2023)\u003c/li\u003e\n \u003cli\u003eWorld Health Organization International Agency for Research on Cancer (IARC). GLOBOCAN 2023: estimated cancer incidence,\u0026nbsp;\u003cbr\u003emortality and prevalence worldwide in 2030. Accessed May 22, 2023. Available from: https://gco.iarc.fr/tomorrow/en/dataviz/trends?types=1\u0026amp;sexes=1_2\u0026amp;mode=cancer\u0026amp;group_populations=1\u0026amp;multiple_populations=1\u0026amp;multiple_cancers=1\u0026amp;cancers=8_9\u0026amp;populations=\u003cbr\u003e32_68_76_152_170_600_604_858_862\u0026amp;age_start=11\u0026amp;scale=log\u0026amp;min_zero=0\u0026amp;num_path=4\u0026amp;group_cancers=1\u003c/li\u003e\n \u003cli\u003eBray F, M\u0026oslash;ller B. Predicting the future burden of cancer. Nat Rev Cancer. 2006 Jan;6:63-74. doi: 10.1038/nrc1781.\u003c/li\u003e\n \u003cli\u003eFerlay J, Laversanne M, Ervik M, Lam F, Colombet M, Mery L, Pi\u0026ntilde;eros M, Znaor A, Soerjomataram I, Bray F (2020). Global Cancer Observatory: Cancer Tomorrow. Lyon, France: International Agency for Research on Cancer. Available from: https://gco.iarc.fr/tomorrow, accessed 22 May 2023.\u003c/li\u003e\n \u003cli\u003eSeitz K, Deliens L, Cohen J, Cardozo EA, Tripodoro VA, Marcucci FCI, Rodrigues LF, Derio L, S\u0026aacute;nchez-C\u0026aacute;rdenas MA, Salazar V, Samayoa VR, Pozo X, Dykeman-Sabado DA, de la Lanza CC, Algaba NCB, Alvarez GP, Viana L, Gonz\u0026aacute;lez T, Pastrana T. Feasibility of using death certificates for studying place of death in Latin America. Rev Panam Salud Publica. 2021;45:e149. doi: 10.26633/RPSP.2021.149.\u003c/li\u003e\n \u003cli\u003eMahapatra P, Shibuya K, Lopez AD, Coullare F, Notzon FC, Rao C, Szreter S; Monitoring Vital Events. Civil registration systems and vital statistics: successes and missed opportunities. Lancet. 2007;370:1653-63. doi: 10.1016/S0140-6736(07)61308-7.\u003c/li\u003e\n \u003cli\u003eSierra MS, Forman D. Cancer in Central and South America: Methodology. Cancer Epidemiol. 2016;44 Suppl 1:S11-S22. doi: 10.1016/j.canep.2016.07.020. PMID: 27678312.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 6 are available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Colorectal cancer, deaths, mortality, modelling, South America, ARIMA, Gaussian Processes, Bayesian analysis, Time-series","lastPublishedDoi":"10.21203/rs.3.rs-3001420/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3001420/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eColorectal cancer (CRC) in South America causes 69,435 deaths annually; mortality rates are heterogeneous across countries. Limited resources in public health systems defy screening and treatment interventions, rendering forecasting and modeling invaluable policy-making tools. We aimed to forecast CRC deaths in individuals aged 55 or older in South America for the period 2020–2030 using time-series analysis and assess heterogeneity in CRC death rates.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTime-series analysis were used to forecast CRC mortality in South America in individuals aged 55 + years. ARIMA and Bayesian with Gaussian processes (GP) and Markov-Chain Montecarlo Simulation models were developed. Heterogeneity in CRC mortality rates across countries was evaluated. Data were extracted from the Global Burden of Disease study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBetween 2020 and 2030, South American countries are expected to record between 680,514 (ARIMA) and 548,372 (GP) CRC deaths in individuals aged 55 or older, both sexes. Brazil and Argentina with the highest burden. The ARIMA model predicts an increase in annual deaths from 52,905 (2020) to 70,811 (2030), while the GP model predicts a slight decrease from 51,343 (2020) to 43,344 (2030). Mortality rates vary significantly between countries, with Uruguay having the highest in 2010 (236.12/100,000) followed by Argentina and Brazil.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBoth the ARIMA and GP models predicted that more than half million people would dye of CRC in South America during the next decade. Mortality rates will be heterogeneous among countries. Accurate forecasting is essential for effective public policies. Continued research and concerted efforts are necessary to address the burden of CRC and to explain causes of varying mortality rates.\u003c/p\u003e","manuscriptTitle":"Colorectal Cancer Deaths in South America: time-series analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-06-12 14:08:01","doi":"10.21203/rs.3.rs-3001420/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"bdd332a8-63cc-414a-a61f-75df84f77728","owner":[],"postedDate":"June 12th, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2023-08-01T06:44:19+00:00","versionOfRecord":[],"versionCreatedAt":"2023-06-12 14:08:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3001420","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3001420","identity":"rs-3001420","version":["v1"]},"buildId":"rHA-KDH7Qsr4HCuvH75dn","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}