Department of Veterinary Biosciences, University of Helsinki, PO Box 66, FI-00014 Helsinki, Finland.
J Dairy Sci. 2012 Aug;95(8):4309-18. doi: 10.3168/jds.2011-5025.
Information on diseases of dairy cows in Finland is entered into the national disease register. Before the data from such types of secondary database are used, the quality of the data needs to be validated. In this study, 7,324 veterinary records for culled cows were compared against records in the national disease register in the period 2002 to 2008. Evaluation of the national disease register data was done by calculating completeness and correctness values. Completeness is the proportion of events that were physically recorded in the secondary database, and correctness is the proportion of correctly recorded events of all recorded events. The use of both correctness and completeness is important when describing the accuracy of secondary data. The completeness and correctness values for the Finnish national dairy disease register were 83 and 92%, respectively. We found that 39% of all unmatched diagnostic events were unreported by the artificial insemination technician who transferred the data from cow cards to the register. Logistic regression models showed that diagnostic events for cows born into the herd had twice the odds of being transferred compared with events for purchased cows. Diagnostic events for reproductive diseases had higher odds of being transferred to the register compared with all other disease groups, and the odds for transfer of the diagnostic event decreased as the age increased. We also found that if the diagnostic event was the last diagnostic event on the cow card, then its odds of being transferred to the disease register were significantly lowered. Although the Finnish national dairy disease register has good completeness and excellent correctness values, different disease groups, age groups, origin of the cow, and timing of the diagnostic event affect how well diagnostic events are transferred to the Finnish national dairy register.
芬兰的奶牛疾病信息被输入国家疾病登记册。在使用此类二级数据库的数据之前，需要验证数据的质量。在这项研究中，2002 年至 2008 年期间，将 7324 份淘汰奶牛的兽医记录与国家疾病登记册中的记录进行了比较。通过计算完整性和正确性值来评估国家疾病登记数据。完整性是指在二级数据库中实际记录的事件的比例，而正确性是指正确记录的事件占所有记录事件的比例。在描述二级数据的准确性时，同时使用正确性和完整性非常重要。芬兰国家奶牛疾病登记处的完整性和正确性值分别为 83%和 92%。我们发现，所有不匹配的诊断事件中有 39%是由于人工授精技术员在将数据从牛卡转移到登记处时未报告的。逻辑回归模型显示，在群内出生的奶牛的诊断事件转移到登记处的可能性是购买奶牛的两倍。与所有其他疾病组相比，生殖疾病的诊断事件更有可能被转移到登记处，并且随着年龄的增长，转移诊断事件的可能性降低。我们还发现，如果诊断事件是牛卡上的最后一个诊断事件，那么其转移到疾病登记处的可能性就会显著降低。尽管芬兰国家奶牛疾病登记处具有良好的完整性和优秀的正确性值，但不同的疾病组、年龄组、奶牛来源和诊断事件的时间都会影响诊断事件向芬兰国家奶牛登记处的转移情况。
