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

Statistical Monitoring of Condemnations from 
Slaughterhouses

Flavie Vial*1, Sarah Thommen2 and Leonhard Held2

1Vetsuisse Faculty, Veterinary Public Health Institute, Bern, Switzerland; 2Institute of Social and Preventive Medicine, Zürich, 
Switzerland

Objective
We evaluate the performance of the improved Farrington algorithm 

for the detection of simulated outbreaks in meat inspection data.

Introduction
Meat inspection data are routinely collected over several years 

providing the possibility to use historical data for constructing 
a baseline model defining the expected normal behaviour of the 
indicator monitored. In countries in which the reporting of data 
is compulsory (e.g. in the EU), coverage of the majority of the 
slaughtered population is ensured.

Methods
This study focuses on the whole carcass condemnations (WCC) 

of cattle slaughtered in Switzerland between 2007 and 2012 (1). 
The monthly time series was retrospectively modelled using the 
function {hhh4} in the R {surveillance} package (2). The mean 
monthly incidence was decomposed additively into an auto-regressive 
component; and an endemic component multiplied by the offset to 
adjust for variation in the number of total animals slaughtered per 
month. The retrospective model was used to simulate a baseline series 
to which an outbreak was added at a random starting point ti. Outbreak 
sizes were randomly generated as Poisson with mean equal to k times 
(with k from 2 to 10) the standard deviation of the baseline count at 
ti (1000 simulations for each k). The improved Farrington algorithm 
(3) was applied for outbreak detection using the {faringtonFlexible} 
function:

Results
Algorithm A includes parameters chosen according to the results 

of the retrospective analysis (no trend and no seasonality). Algorithm 
B includes more historic data (b=3 instead of b=2) and considers 
monthly seasonal effects (w=0). Both algorithms produced less than 
one false alarm per year. In terms of probability of detection (POD), 
algorithm A outperforms algorithm B at the detection of larger 
outbreaks (k=5 to 10), but the latter is more appropriate when looking 
to detect smaller outbreaks (Figure 1).

Conclusions
The improved Farrington algorithm led to low false positive rate 

(FPR) but the POD was low for small outbreaks. Furthermore, more 
than 50% of the outbreak cases had already occurred by the time the 
outbreak was detected. The low temporal resolution of this dataset 
currently limits its use for early detection.

Table 1: Outbreaks attributes in WCC data (averaged over 1000 simulations 
for each k)

POD and FPR of algorithms A &B for different outbreak sizes (k)

Keywords
Meat inspection; veterinary public health; syndromic surveillance

Acknowledgments

Marion Zumbrunnen extracted the data & Sebastian Meyer helped with 
the R programming.

References

1.Vial F, Reist M. Evaluation of Swiss slaughterhouse data for integration 
in a syndromic surveillance system. BMC Vet Res. 2014 Jan;10:33.

2.Höhle M. { surveillance} : An R package for the monitoring of infectious 
diseases. Comput Stat. 2007 Aug;22(4):571–82.

3.Noufaily A, Enki DG, Farrington CP, Garthwaite PH, Andrews NJ,
Charlett A. An improved algorithm for outbreak detection in multiple 
surveillance systems. Stat Med. 2013 Mar 30;32(7):1206–22.

*Flavie Vial

E-mail: flavie.vial@vetsuisse.unibe.ch    

Online Journal of Public Health Informatics * ISSN 1947-2579 * http://ojphi.org * (1):e169, 201