Time series modeling of pneumonia admissions and its association with air pollution and climate variables in Chiang Mai Province, Thailand

Apaporn Ruchiraset,Kraichat Tantrakarnapa

doi:10.1007/s11356-018-3284-4

Abstract

This study aimed to predict the number of pneumonia cases in Chiang Mai Province. An autoregressive integrated moving average (ARIMA) was used in data fitting and to predict future pneumonia cases monthly. Total pneumonia cases of 67,583 were recorded in Chiang Mai during 2003–2014 that the monthly pattern of case was similar every year. Monthly pneumonia cases were increased during February and September, which are the periods of winter and rainy season in Thailand and decreased during April to July (the period of summer season to early rainy season). Using available data on 12 years of pneumonia cases, air pollution, and climate in Chiang Mai, the optimum ARIMA model was investigated based on several conditions. Seasonal change was included in the models due to statistically strong season conditions. Twelve ARIMA model (ARMODEL1–ARMODEL12) scenarios were investigated. Results showed that the most appropriate model was ARIMA (1,0,2)(2,0,0)[12] with PM10 (ARMODEL5) exhibiting the lowest AIC of − 38.29. The predicted number of monthly pneumonia cases by using ARMODEL5 during January to March 2013 was 727, 707, and 658 cases, while the real number was 804, 868, and 783 cases, respectively. This finding indicated that PM10 held the most important role to predict monthly pneumonia cases in Chiang Mai, and the model was able to predict future pneumonia cases in Chiang Mai accurately.

Highlights

The autoregressive integrated moving average (ARIMA) model was first proposed in 1976 and widely used for predicting and early warning analyzing of infectious diseases (Luz et al 2008; Reichert et al 2004; Yi et al 2007) and for predicting future air quality status from various aspects of development in several countries (Konovalov et al 2009; Pedro Muñoz Miguel et al 2017)
When comparing between air pollution (ARMODEL3) and climate variables (ARMODEL4), the results in this study indicated that climate variables played a more important role concerning pneumonia than air pollution variables because ARMODEL4 exhibited lower Akaike Information Criterion (AIC) values than ARMODEL3
The study aimed to predict the number of pneumonia cases in Chiang Mai Province

Summary

Introduction

The autoregressive integrated moving average (ARIMA) model was first proposed in 1976 and widely used for predicting and early warning analyzing of infectious diseases (Luz et al 2008; Reichert et al 2004; Yi et al 2007) and for predicting future air quality status from various aspects of development in several countries (Konovalov et al 2009; Pedro Muñoz Miguel et al 2017). Examining associations between environmental factors and adverse health outcomes is more advantages using the ARIMA model (Imai et al 2015; Sharafi et al 2017; Unkel et al 2012). Of the estimated nine million child deaths in 2007, around 20% or 1.8 million were due to pneumonia (WHO and UNICEF 2009). To control this risk, the World Health Organization (WHO) and United Nations International Children’s Emergency Fund (UNICEF) launched a Global Action Plan for Pneumonia Prevention and Control (WHO and UNICEF 2009) in 2009.

Objectives

Methods

Results

Discussion

Conclusion