Tweet
BMC Public Health. 2016 Sep 13;16:972. doi: 10.1186/s12889-016-3503-1.
Effects of climatological parameters in modeling and forecasting seasonal influenza transmission in Abidjan, Cote d'Ivoire.
N'gattia AK1,2, Coulibaly D3, Nzussouo NT4, Kadjo HA5, Ch?rif D3, Traor? Y3,6, Kouakou BK5, Kouassi PD3,7, Ekra KD3,6, Dagnan NS3,6, Williams T4, Tiembr? I3,6.
Author information
Abstract
BACKGROUND:
In temperate regions, influenza epidemics occur in the winter and correlate with certain climatological parameters. In African tropical regions, the effects of climatological parameters on influenza epidemics are not well defined. This study aims to identify and model the effects of climatological parameters on seasonal influenza activity in Abidjan, Cote d'Ivoire.
METHODS:
We studied the effects of weekly rainfall, humidity, and temperature on laboratory-confirmed influenza cases in Abidjan from 2007 to 2010. We used the Box-Jenkins method with the autoregressive integrated moving average (ARIMA) process to create models using data from 2007-2010 and to assess the predictive value of best model on data from 2011 to 2012.
RESULTS:
The weekly number of influenza cases showed significant cross-correlation with certain prior weeks for both rainfall, and relative humidity. The best fitting multivariate model (ARIMAX (2,0,0) _RF) included the number of influenza cases during 1-week and 2-weeks prior, and the rainfall during the current week and 5-weeks prior. The performance of this model showed an increase of >3 % for Akaike Information Criterion (AIC) and 2.5 % for Bayesian Information Criterion (BIC) compared to the reference univariate ARIMA (2,0,0). The prediction of the weekly number of influenza cases during 2011-2012 with the best fitting multivariate model (ARIMAX (2,0,0) _RF), showed that the observed values were within the 95 % confidence interval of the predicted values during 97 of 104 weeks.
CONCLUSION:
Including rainfall increases the performances of fitted and predicted models. The timing of influenza in Abidjan can be partially explained by rainfall influence, in a setting with little change in temperature throughout the year. These findings can help clinicians to anticipate influenza cases during the rainy season by implementing preventive measures.
KEYWORDS:
Abidjan; Climatological parameters; Influenza; Modeling
PMID: 27624302 DOI: 10.1186/s12889-016-3503-1
[PubMed - in process] Free full text
Effects of climatological parameters in modeling and forecasting seasonal influenza transmission in Abidjan, Cote d'Ivoire.
N'gattia AK1,2, Coulibaly D3, Nzussouo NT4, Kadjo HA5, Ch?rif D3, Traor? Y3,6, Kouakou BK5, Kouassi PD3,7, Ekra KD3,6, Dagnan NS3,6, Williams T4, Tiembr? I3,6.
Author information
Abstract
BACKGROUND:
In temperate regions, influenza epidemics occur in the winter and correlate with certain climatological parameters. In African tropical regions, the effects of climatological parameters on influenza epidemics are not well defined. This study aims to identify and model the effects of climatological parameters on seasonal influenza activity in Abidjan, Cote d'Ivoire.
METHODS:
We studied the effects of weekly rainfall, humidity, and temperature on laboratory-confirmed influenza cases in Abidjan from 2007 to 2010. We used the Box-Jenkins method with the autoregressive integrated moving average (ARIMA) process to create models using data from 2007-2010 and to assess the predictive value of best model on data from 2011 to 2012.
RESULTS:
The weekly number of influenza cases showed significant cross-correlation with certain prior weeks for both rainfall, and relative humidity. The best fitting multivariate model (ARIMAX (2,0,0) _RF) included the number of influenza cases during 1-week and 2-weeks prior, and the rainfall during the current week and 5-weeks prior. The performance of this model showed an increase of >3 % for Akaike Information Criterion (AIC) and 2.5 % for Bayesian Information Criterion (BIC) compared to the reference univariate ARIMA (2,0,0). The prediction of the weekly number of influenza cases during 2011-2012 with the best fitting multivariate model (ARIMAX (2,0,0) _RF), showed that the observed values were within the 95 % confidence interval of the predicted values during 97 of 104 weeks.
CONCLUSION:
Including rainfall increases the performances of fitted and predicted models. The timing of influenza in Abidjan can be partially explained by rainfall influence, in a setting with little change in temperature throughout the year. These findings can help clinicians to anticipate influenza cases during the rainy season by implementing preventive measures.
KEYWORDS:
Abidjan; Climatological parameters; Influenza; Modeling
PMID: 27624302 DOI: 10.1186/s12889-016-3503-1
[PubMed - in process] Free full text