2014.ISDS.Abstracts.Final.pdf


ISDS Annual Conference Proceedings 2014. This is an Open Access article distributed under the terms of the Creative Commons Attribution-
Noncommercial 3.0 Unported License (http://creativecommons.org/licenses/by-nc/3.0/), permitting all non-commercial use, distribution, 
and reproduction in any medium, provided the original work is properly cited.

ISDS 2014 Conference Abstracts

Transmission Dynamics of Seasonal Influenza in 
Abidjan: Epidemiology and Modeling

Anderson K. N’gattia*1, 2, Daouda Coulibaly1, Djibril Cherif1, Hervé Kadjo3 and Issaka 
Tiembré1, 2

1Epidemiologic surveillance service, Institut National d’Hygiène Publique, Abidjan, Côte d’Ivoire; 2Université Félix Houphouët-Boigny, 
Abidjan, Côte d’Ivoire; 3Institut Pasteur de Cote d’Ivoire, Abidjan, Côte d’Ivoire

Objective
This study aims to determine the epidemiological and clinical 

profiles of influenza infections related to different strains and the 
effect of climatological parameters on the temporal distribution of 
the disease for the prediction.

Introduction
Influenza poses a global health threat. The disease affects all ages, 

often with variable clinical features.
Abidjan, where this study took place, has a long rainy season April-

July with a shorter less intense rainy season October-November. 
Temperatures vary very little during the year. In temperate areas, 
children and adults aged 65 years are risk groups. In these countries 
the seasonality of influenza is clearly defined, with seasonal 
epidemics in cold weather periods (1). But in the tropics, the risk 
groups of influenza are not as well defined. Also, the dynamics of 
influenza transmission and climatological parameters that influence 
it are specific to the tropical region and not as thoroughly studied.

Methods
Cote d’Ivoire’s Institut National d’Hygiène Publique (INHP) 

operates sentinel influenza surveillance sites throughout the country, 
including Abidjan, following WHO guidelines. Staff at each site 
obtain specimens for influenza testing from a sample of the people 
treated each day for influenza-like illness (ILI) or severe acute 
respiratory infection (SARI). Data used in this report include all 
persons with laboratory-confirmed influenza from the sites located 
in Abidjan during 2007 to 2012, and climatological data from the 
SODEXAM (National Weather Service). The 2 test was used to 
compare proportions of cases by influenza type and also by month 
and season. Sequential logistic regression models were used to 
predict the epidemiological and clinical profiles, and multiple linear 
regression was used to assess the association between the number of 
weekly influenza cases and rainfall, relative humidity and ambient 
temperature. Explanatory models of weekly influenza cases based on 
climatological parameters used data from 2007 to 2010. Including 
data from 2011 to 2012 allowed us to test the prediction models. 
Modeling through the Box-Jenkins method was used with ARIMA 
process. Data were analyzed using Stata MP 12.0 software, StataCorp 
LP, College Station, Texas.

Results
A total of 921 influenza cases were identified over the period 

2007-2010. Of these, 663 (72%) were influenza A and 258 (28%) 
were influenza B. Among the influenza A, 60 (9.1%) were H1N1, 
177 (26.7%) seasonal H3N2, 34 (5.1%) pH1N1 (2009 pandemic 
strain), and 392 (59.1%) unsubtyped. The largest number of 
positive specimens was from young children aged 0-4 years (472; 
52.9%). Individuals with pH1N1 were more likely than those with 
other H1N1 to have cough (OR=10.52; 95%CI=1.71 to 64.57) and 
more likely than those with influenza B to be from the age group 
of 5-14 years (OR=3.10; 95%CI=1.32 to 7.3). For patients with 

the H3N2 influenza, cough (OR=6.3; 95%CI=2.466 to 16.347) 
and arthralgia/myalgia (OR =2.54; 95%CI=1.08 to 5.95) were the 
best predictors versus influenza H1N1. The highest monthly and 
seasonally proportions of influenza viruses were observed the months 
corresponding to the long rainy season and the short rainy season 
in which respectively 48.4% and 20.2% of influenza viruses were 
isolated. The multiple linear regression showed association between 
the number of weekly influenza cases and rainfall ( =0,025; p=0.008) 
but not with relative humidity (p=0.108) and ambient temperature 
(p=0.786). ARIMAX models perform best only with the introduction 
of rainfall at lag (epidemiological weeks: EW) EW0(0.187) and EW-5 
(0.175). The ARIMAX(2,0,0)RF, was the best model (performance: 
AIC=1257.7; BIC=1277.6). The prediction of the weekly incidence 
of influenza from EW1(2011) to EW52 (2012) was implemented by 
ARIMAX(2,0,0)RF and fitted best values compared with observed 
values.

Conclusions
The specific epidemiological and clinical profiles of influenza were 

clearly identified in our study. Weekly rainfall was a good predictor 
of the transmission of seasonal influenza in Abidjan, in a setting 
with little change in temperature through the year. This should allow 
physicians to detect influenza cases and strengthen surveillance and 
preventive measures at the approach of rainy seasons.

Keywords
Transmission dynamics; seasonal influenza; epidemiological and 
clinical profiles; climatological parameters; Cote d’Ivoire

References

1. OMS | Grippe (saisonnière) [Internet]. WHO:www.who.int/
mediacentre/factsheets/fs211/fr/index.html

*Anderson K. N’gattia

E-mail: jeanandersonk@yahoo.fr    

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * 7(1):e42, 2015