Tsutsui T, Minami N, Koiwai M, Hamaoka T, Yamane I, Shimura K
Applied Epidemiology Section, National Institute of Animal Health, 3-1-5, Kannondai, Tsukuba, Ibaraki 305-0856, Japan.
Prev Vet Med. 2003 Sep 30;61(1):45-58. doi: 10.1016/s0167-5877(03)00160-0.
When foot-and-mouth-disease (FMD) was identified in Miyazaki prefecture in March 2000, Japan conducted an intensive serological and clinical survey in the areas surrounding the index herd. As a result of the survey during the 21 days of the movement-restriction period, two infected herds were detected and destroyed; there were no other cases in the months that followed. To evaluate the survey used for screening the disease-control area and surveillance area, we estimated the herd-level sensitivity of the survey (HSe) through a spreadsheet model using Monte-Carlo methods. The Reed-Frost model was incorporated to simulate the spread of FMD within an infected herd. In the simulations, 4, 8 and 12 effective-contact scenarios during the 5-day period were examined. The estimated HSes of serological tests (HSeE) were 71.0, 75.3 and 76.3% under the 4, 8 and 12 contact scenarios, respectively. The sensitivity analysis showed that increasing the number of contacts beyond 12 did not improve HSeE, but increasing the number of sampled animals and delaying the dates of sampling did raise HSeEs. Small herd size in the outbreak area (>80% of herds have <20 animals) seems to have helped in maintaining HSeE relatively high, although the serological inspection was carried out before sero-positive animals had a chance to increase in infected herds. The estimated herd-level specificity of serological tests (HSpE) was 98.6%. This HSpE predicted 224 false-positive herds (5th percentile estimate was 200 and 95th percentile was 249), which proved close to the 232 false-positive herds actually observed. The combined-test herd-level sensitivity (serological and clinical inspections combined; CTHSe), averaged 85.5, 87.6 and 88.1% for the 4, 8 and 12 contact scenarios, respectively. Using these CTHSes, the calculated probability that no infected herd was overlooked by the survey was > or =62.5% under the most-conservative, four-contact scenario. The probability that no more than one infected herd was overlooked was > or =89.7%.
2000年3月在宫崎县确认口蹄疫疫情后，日本在疫群周边地区开展了密集的血清学和临床调查。在行动限制期的21天调查期间，检测到两个感染畜群并予以扑杀；在随后的几个月里未出现其他病例。为评估用于筛查疾病控制区和监测区的调查，我们通过使用蒙特卡洛方法的电子表格模型估算了调查的畜群水平敏感性（HSe）。纳入了里德-弗罗斯特模型以模拟口蹄疫在感染畜群中的传播。在模拟中，研究了5天期间4种、8种和12种有效接触情况。在4种、8种和12种接触情况下，血清学检测的估计畜群水平敏感性（HSeE）分别为71.0%、75.3%和76.3%。敏感性分析表明，将接触数量增加到12种以上并不会提高HSeE，但增加采样动物数量和推迟采样日期会提高HSeE。疫情爆发地区畜群规模较小（超过80%的畜群动物数量少于20头）似乎有助于使HSeE保持相对较高水平，尽管血清学检测是在感染畜群中血清阳性动物数量有机会增加之前进行的。血清学检测的估计畜群水平特异性（HSpE）为98.6%。该HSpE预测有224个假阳性畜群（第5百分位数估计为200，第95百分位数为249），这与实际观察到的232个假阳性畜群相近。联合检测（血清学和临床检查相结合）的畜群水平敏感性（CTHSe）在4种、8种和12种接触情况下分别平均为85.5%、87.6%和88.1%。使用这些CTHSe，在最保守的4种接触情况下，计算得出调查未遗漏任何感染畜群的概率≥62.5%。未遗漏超过一个感染畜群的概率≥89.7%。
