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

Modeling Spatial Heterogeneity with Excess Zeroes 
from School Absenteeism dSata

Xiaoxiao Song*1, 2, Qi Zhao1, Changming Zhou1, Tao Tao1, Lars Palm3, Vinod Diwan4, Hui 
Yuan5 and Biao Xu1

1Dept. of Epidemiology, School of Public Health, Fudan University, Shanghai, China; 2School of Public Health, Kunming Medical 
University, Kunming, China; 3Future Position X, Gavle, Sweden; 4ICHAR, Karolinska Instituet, Stockholm, Sweden; 5Jiangxi Provincial 
Center for Disease Prevention and Control, Nanchang, China

Objective
To describe and explore the spatial heterogeneity via Random 

effects zero-inflated Poisson model (RE-ZIP) for absenteeism 
surveillance in primary school for early detection of infectious disease 
outbreak in rural China.

Introduction
Absenteeism has been considered as a potential indicator for 

the early detection of infectious disease outbreaks in population, 
especially in primary schools. However, in practice this data are 
often characterized by an excess of zeros and spatial heterogeneity. 
In a project on integrated syndromic surveillance system (ISSC) in 
rural China(1), Random effect zero-inflated Poisson (RE-ZIP) model 
was applied to simultaneously quantify the spatial heterogeneity for 
“occurrence” and “intensity” on school absenteeism data.

Methods
Daily health-related school absenteeism at 62 primary schools 

in two counties of Jiangxi Province were reported to the web-based 
platform of ISSC during 24 continuous months (from April 1th, 2012 
to June 31st, 2014). The RE-ZIP has two components that correspond 
to two zero generating processes with two random intercept ( 1 
and 2). The first process is governed by a binary distribution that 
generates structural zeros. The second process is governed by a 
Poisson distribution that generates counts, some of which may be zero 
(named sampling zero) (2), see Eqation.1-2. Firstly, artificial data was 
generated to illustrate that the two random effects could appropriately 
describe the spatial heterogeneity for “occurrence” and “intensity” of 
school absenteeism (Figure_1). Then, the spatial heterogeneity for 
“occurrence” and “intensity” in real absenteeism data from the 62 
primary schools were assessed and visualized through quantifying 
the two random effects in each individual school. All analyses were 
processed using SAS NLMIXED(3).

Results
In total, 32283 health-related absenteeism records were reported 

from the 62 primary schools in the study period. Figure_2 displayed 
clearly the spatial heterogeneity for both “occurrence” and “intensity” 
in these primary schools indicated by 1 and 2 of the two random 
effects.

The spatial variation in these two counties were also visualized 
graphically these on map in Figure_3 and Figure_4.

Conclusions
In infectious disease surveillance, the key questions are “whether 

an outbreak will happen” or/and “how large the epidemic will be”, 
which could be answered through observing and monitoring the 
spatial pattern of health related school absenteeism. Findings from 
this study suggest that RE-ZIP model is an effective tool in the 
surveillance of health-related school absenteeism data with high 
spatial heterogeneity.

Keywords
School absenteeism; spatial heterogeneity; Random effects zero-
inflated Poisson model; occurrence; intensity

Acknowledgments

This study was funded by [European Union’s] [European Atomic Energy 
Community’s] Seventh Framework Programmer ([FP7/2007-2013] 
[FP7/2007-2011]) under grant agreement no. [241900].

References

1. Yan W-r, Nie S-f, Xu B, Dong H-j, Palm L, Diwan VK. Establishing
a web-based integrated surveillance system for early detection of 
infectious disease epidemic in rural China: a field experimental study. 
BMC medical informatics and decision making2012;12(1):4.

2. Hu M-C, Pavlicova M, Nunes EV. Zero-inflated and hurdle models
of count data with extra zeros: examples from an HIV-risk reduction 
intervention trial. The American journal of drug and alcohol 
abuse2011;37(5):367-75.

3. Wang J, Xie H, Fisher JF. Multilevel models: applications using SAS®: 
Walter de Gruyter; 2011.

*Xiaoxiao Song

E-mail: chinasxx@gmail.com    

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * 7(1):e159, 2015