Home > Research > Publications & Outputs > Compatibility between livestock databases used ...

Associated organisational unit


Text available via DOI:

View graph of relations

Compatibility between livestock databases used for quantitative biosecurity response in New Zealand

Research output: Contribution to Journal/MagazineJournal articlepeer-review

<mark>Journal publication date</mark>06/2016
<mark>Journal</mark>New Zealand Veterinary Journal
Issue number3
Number of pages7
Pages (from-to)158-164
Publication StatusPublished
Early online date10/11/15
<mark>Original language</mark>English


AIM: To characterise New Zealand's livestock biosecurity databases, and investigate their compatibility and capacity to provide a single integrated data source for quantitative outbreak analysis. METHODS: Contemporary snapshots of the data in three national livestock biosecurity databases, AgriBase, FarmsOnLine (FOL) and the National Animal Identification and Tracing Scheme (NAIT), were obtained on 16 September, 1 September and 30 April 2014, respectively, and loaded into a relational database. A frequency table of animal numbers per farm was calculated for the AgriBase and FOL datasets. A two dimensional kernel density estimate was calculated for farms reporting the presence of cattle, pigs, deer, and small ruminants in each database and the ratio of farm densities for AgriBase versus FOL calculated. The extent to which records in the three databases could be matched and linked was quantified, and the level of agreement amongst them for the presence of different species on properties assessed using Cohen's kappa statistic. RESULTS: AgriBase contained fewer records than FOL, but recorded animal numbers present on each farm, whereas FOL contained more records, but captured only presence/absence of animals. The ratio of farm densities in AgriBase relative to FOL for pigs and deer was reasonably homogeneous across New Zealand, with AgriBase having a farm density approximately 80% of FOL. For cattle and small ruminants, there was considerable heterogeneity, with AgriBase showing a density of cattle farms in the Central Otago region that was 20% of FOL, and a density of small ruminant farms in the central West Coast area that was twice that of FOL. Only 37% of records in FOL could be linked to AgriBase, but the level of agreement for the presence of different species between these databases was substantial (kappa >0.6). Both NAIT and FOL shared common farm identifiers which could be used to georeference animal movements, and there was a fair to substantial agreement (kappa 0.32–0.69) between these databases for the presence of cattle and deer on properties. CONCLUSIONS: The three databases broadly agreed with each other, but important differences existed in both species composition and spatial coverage which raises concern over their accuracy. Importantly, they cannot be reliably linked together to provide a single picture of New Zealand's livestock industry, limiting the ability to use advanced quantitative techniques to provide effective decision support during disease outbreaks. We recommend that a single integrated database be developed, with alignment of resources and legislation for its upkeep.