Page 236 - Contributed Paper Session (CPS) - Volume 3

CPS2003 Bruno de S. et al.

Structured additive Regression Modeling of
pulmonary tuberculosis infection
2
2
1
1
Bruno de Sousa , Carlos Pires , Dulce Gomes , Patrícia Filipe , Ana Costa-
Veiga , Carla Nunes
3
3,4
1 Faculty of Psychology and Education Sciences, University of Coimbra, Portugal
2 Centro de Investigação em Matemática e Aplicações, Instituto de Investigação e Formação
Avançada, Universidade de Évora, Departamento de Matemática, Escola de Ciências e
Tecnologia, Portugal
3 CISP - Centro de Investigação em Saúde Pública, National School of Public Health,
Universidade Nova de Lisboa, Portugal
4 H&TRC - Health & Technology Research Center, ESTeSL, Lisbon School of Health Technology,
Instituto Politécnico de Lisboa, Av. D. João II, Portugal

Abstract
Tuberculosis (TB) is one of the top 10 causes of death and the leading cause
from a single infectious agent (above HIV/AIDS). In 2017, the World Health
Organization (WHO) estimated 10.0 million people developed TB and 1.3
million deaths (range, 1.2–1.4 million) among HIV-negative people with an
additional 300 000 deaths from TB (range, 266 000–335 000) among HIV-
positive people. Studies that understand the socio-demographic
characteristics, time and spatial distribution of the disease are vital to
allocating resources in order to improve National TB Programs. The database
includes information from all confirmed Pulmonary TB (PTB) cases notified in
Continental Portugal between 2000 and 2010. Following a descriptive analysis
of the main risk factors of the disease, a Structured Additive Regression (STAR)
model is presented exploring possible spatial and temporal correlations in PTB
incidence rates in order to identify the regions of increased incidence rates.
Three main regions are identified as statistically significant areas of increased
PTB incidence rates in Continental Portugal. STAR models proved to be a
valuable and effective approach in identifying PTB incidence rates and will be
used in future research to identify the associated risk factors in Continental
Portugal, yielding high-level information for decision-making in TB control.

Keywords
Structured Additive Regression Models; Pulmonary Tuberculosis; Spatial-
Temporal Epidemiology; Full Bayesian; Empirical Bayesian

1. Introduction
Pulmonary Tuberculosis (PTB) is an infectious disease which affects
millions of people every year, being the second most deadly infectious disease
worldwide after the human immunodeficiency virus (HIV) [1]. The disease is
caused by the bacillus Mycobacterium tuberculosis that affects mainly the

225 | I S I W S C 2 0 1 9