Diseño de un modelo de simulación para evaluar intervenciones en salud preventiva del cáncer.

Abstract

In Colombia, cancer represents a growing public health problem, it is one of the leading causes of death in the world, although activities aimed at prevention are carried out, there are multiple challenges in the implementation. Particularly, cervical cancer, in the National Demographic and Health Survey it was found that only 66.5% of the women surveyed have had a cytology exam every year. Consequently, it is necessary to promote improvements in prevention actions. According to WHO data, between 30% and 50% of cancer deaths could be prevented through measures aimed at modifying or avoiding the main risk factors. However, despite the development of these prevention strategies, the population has low levels of adherence.

This behavior can be understood from the context of the Health Belief Model (HBM) proposed by the U.S. Public Health Service which seeks to explain the obstacles that arise in the population to adopt protective behavior in disease prevention. The results of each intervention are related to multiple social, economic, political, cultural and organizational variables. Therefore, for the design of interventions, it is necessary to study the reasons why individuals do not apply existing health care knowledge or to identify the reasons why the impact of prevention measures is still limited.

The purpose is to design a computational simulation model to evaluate and support interventions targeting cancer prevention. Agent-based simulation models (MBA) have become a common tool for modeling and understanding complex, nonlinear systems of this type. MBA is about understanding how different interventions may impact individual behaviors and the interaction patterns of these behaviors, as well as identifying the most effective interventions to improve population health. These models can be adjusted and recalibrated according to the needs of a certain target population and allow the evaluation of long-term results of interventions carried out for screening, prevention and other purposes.

The construction of the simulation model began with the elaboration of a conceptual model in order to structure the problem, reduce its level of complexity and graphically represent the dynamics of the variables of interest to be used. According to Perera & Liyanage (2000), the most common limitation in the development of simulation models is the availability of the data to be used; therefore, the definition of the purpose, the identification of the variables and the definition of the actors were based on what was reported in scientific articles; as for the design of the model, the interrelations between actor and context were considered based on a project whose impact was directly related to the present work.

For the construction of the model, the steps recommended in the ODD protocol (Overview, Design concepts, Details) of Grimm et al. (2006) and the steps proposed by Macal and North (2010) were used as a guide, contemplating 7 steps: (i) identify the agents and obtain a theory of their behavior, (ii) identify the relationships of the agents and establish a theory of the interaction between them, (iii) define a modeling platform and a model development strategy, (iv) collect the necessary data to implement the simulation model, through literature and reliable sources that provide true figures and data, (v) establish the initial system conditions, (vi) test the agent behavior models by randomly inspecting the agent attributes, comparing the results with other models and evaluating the model performance under extreme conditions, (vii) run the model and analyze the results by linking the micro-level behaviors of the agents with the macro-level behaviors of the system.

The determinant rule in the simulation model is the decision made by an agent to attend or not to attend a cytology, this probability of attendance was calculated through a logistic regression, the results allowed to identify which aspects significantly impact on the decision. Once the computational model was built, public health interventions extracted from scientific articles and framed in the HBM constructs were included in the simulation. This process was divided into three stages: (i) sensitivity analysis using the NetLogo application tools, (ii) analysis of the logistic regression results (coefficients and odds ratios) and (iii) simulating changes in the agents to impact certain variables.

The construction of the computational model allowed understand the dynamics associated with the early detection of HPV in the context of women aged 18 to 69 years in the city of Bogota with Sisben categories I, II and III. By replicating the real behavior of the agents (based on beliefs and attitudes represented in the HBM questionnaire), the probability of attendance to the cytology was calculated, which took an average value of 55.6%. The simulation evidence that the adoption of the protective behavior of attending a cytology depends on external factors (lead time, treatment received by health workers, availability of epidemiological information) which, at the same time, have repercussions on internal factors: perceived barriers and susceptibility. On the other hand, five variables related to personal and cultural beliefs about cervical cancer (age, Sisben category, severity, perceived benefits and motivation) impacted the decision made by the agent. Regarding the interventions evaluated, the best scenario is presented by the education-oriented intervention, particularly, providing information about the disease, how to prevent it and the importance of regular check-ups. This intervention increased by 19.42% the average probability of attending a cytology, when compared to the scenario that does not include interventions. Finally, it was found that the most effective interventions are focused on reducing misinformation and promoting healthy lifestyle habits.