Validation of disease recordings in Swedish dairy cattle

Jansson Mörk, Marie (2009) Validation of disease recordings in Swedish dairy cattle. Doctoral diss. Dept. of Clinical Sciences, SLU. Acta Universitatis agriculturae Sueciae vol. 2009:66.

Full text available as:

PDF 4618 Kb

Abstract

This thesis examines the completeness of the Swedish dairy disease recording system: it attempts to quantify how much disease the system’s database captures relative to what farmers find and veterinarians treat. Two field studies were conducted. In the first, 177 farmers recorded information about disease events, regardless of whether the disease event had resulted in a veterinary visit. In the second, farm copies of veterinary records (851 records from 112 herds) were collected.

The proportion of disease events receiving veterinary treatment was estimated, and measures of disease incidence based on the farmers’ data were compared with incidences estimated from the Dairy Disease Database (ddd). Further, the completeness of the ddd was estimated based on agreement between information in the ddd and farmer-reporting and herd-copies, respectively. Differential completeness was also evaluated. Finally, the probability of a successfully registered disease event for the whole disease recording process was estimated for five different disease complexes, based on the results of both field studies.

The overall completeness of veterinary treated disease events in the ddd was estimated to be 71% and 75%, based on the farmers’ recordings and on the farm copies, respectively. Differential completeness linked to regions, veterinary employment type and between different groups of animals was found. The probability of a successfully registered disease event (regardless of veterinary treatment) in the ddd varied between 30% for diarrhoea and 72% for puerperal paresis. Whether or not the farmer contacted a veterinarian was found to be the most influential step in the recording process, followed by whether or not the disease record was registered in the raw data file at the Swedish Dairy Association.

Lack of completeness in the ddd will result in conservative disease incidence measures. Underreporting of veterinary treated disease events, as well as undercoverage of farmer-observed events, was found to vary depending on several factors which could introduce bias in estimates based on the ddd, which primarily is a problem if the data are used for epidemiologic research and less so for other areas.