Identifying high and low risk areas in a region
By Kathryn Lenker and Meghann Pierdon
Porcine reproductive and respiratory disease syndrome virus infects only swine, and is transmitted via all bodily secretions, including urine, feces, colostrum, milk, saliva, semen and nasal secretions.1 PRRSV has been shown to spread via aerosol, so spatial factors play an important part in disease transmission.2
Regional control programs have reported on the location of outbreaks, and in some cases genetic sequences, but the information can be used to further elucidate areas of high risk and low risk as well as when times of high risk are occurring.3
The purpose of this study was to use historical PRRS information from Pennsylvania's PRRS regional control program to investigate whether PRRS cases in sow farms clustered in space, or in time and space, and whether such analysis could illustrate high and low risk areas for PRRS outbreaks.
PRRS regional control program The PRRS regional control has a database of 93 participating sow farms that reported their PRRSV status quarterly for 10 years. From the quarterly report, sow farms that were never PRRSV positive were considered controls, and sow farms that were PRRSV positive farms were considered positive cases.
The Bernoulli case-control method was used for analysis of spatial clustering at the level of the farm. Analysis was performed using SatScan v 9.7. There were 35 case and 58 control farms included in the analysis. Each was included using the latitude and longitude.
Circular scanning windows were used, and likelihood rations and p-values were calculated by comparing the risk within and outside the scanning window. The program generates random datasets (n=999) and compares the real data with the randomly generated data sets. If the maximum likelihood ratio was found to be in the top 5% of the clusters, then it was determined to be significant at P=0.05.
The null hypothesis was that case farms were spatially randomly distributed, and a cluster was identified if the null hypothesis was rejected. Both high and low levels of clustering were investigated.
Relative risk was calculated by comparing the rate of cases inside the circle to those outside the circle. A relative risk greater than 1 is increased risk, and a relative risk less than 1 is decreased risk.
A retrospective spatial temporal cluster analysis at the level of the quarter was also performed including only the case farms and using cylindrical scanning windows.
High risk, low risk areas Figure 1 represents the purely spatial analysis identified two areas of low risk and two areas of high risk. Cluster 1 had a radius of 23.45 km, an increased risk of PRRSV with a relative risk of 1.75, an expected 28.99 cases and an observed 43 cases (P<0.0001).
Cluster 2 had a radius of 126.10 km, a decreased risk of PRRSV with a relative risk of 0.32, an expected 24.27 cases and an observed nine cases (P<0.001).
Cluster 3 had a radius of 53.88 km, decreased risk of PRRSV with a relative risk 0.12, an expected 8.09 cases and an observed one case (P<0.01).
Cluster 4 had a radius of 29.53 km, an increased risk of PRRS with a relative risk of 1.52, an expected 29.66 cases and an observed 40 cases (P<0.05).
Two significant clusters were found in both space and time. One occurred from 2001- 2002 with a radius of 17.44 km, 1.96 expected cases and 12 observed cases (P<0.001). The other significant cluster occurred from 2015- 2016 with a radius of 4.14 km, 1.12 expected cases and seven observed cases (P<0.05).
Understanding the location of these clusters is beneficial to practitioners and farmers when analyzing PRRSV risk at sow farms. When a new sow farm is to be built, Figure 1 may be important to avoid increased risk of PRRSV introduction.
Biosecurity practices in the high-risk areas could be enhanced, such as filtering facilities, to limit risk. Finally, high value animals, such as those serving as genetic replacements, could be routed to low-risk areas.
Acknowledgements Thank you to the Pennsylvania Pork Producer's Council for providing support and funding.
References 1Tousingnant SJP et al.:2014 AJVR 76.1 70-76. 2Fahrion AS et al.:2014 Preventative veterinary medicine 114.3-4 247-258. 3Valdes-Donoso P et al.: 2016 PloS one 11.2.
Lenker is a researcher and Pierdon is an assistant professor, both at the New Bolton Center, University of Pennsylvania School of Veterinary Medicine.
