Centre for Infectious Diseases, National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands.
Epidemiology. 2013 Mar;24(2):244-50. doi: 10.1097/EDE.0b013e31827f50e8.
A proper understanding of the infection dynamics of influenza A viruses hinges on the availability of reliable estimates of key epidemiologic parameters such as the reproduction number, intrinsic growth rate, and generation interval. Often the generation interval is assumed to be similar in different settings although there is little evidence justifying this. Here we estimate the generation interval for stratifications based on age, cluster size, and social setting (camp, school, workplace, household) using data from 16 clusters of Novel Influenza A (H1N1) in the Netherlands. Our analyses are based on a Bayesian inferential framework, enabling flexible handling of both missing infection links and missing times of symptoms onset. The analysis indicates that a stratification that allows the generation interval to differ by social setting fits the data best. Specifically, the estimated generation interval was shorter in households (2.1 days [95% credible interval = 1.6-2.9]) and camps (2.3 days [1.4-3.4]) than in workplaces (2.7 days [1.9-3.7]) and schools (3.4 days [2.5-4.5]). Our findings could be the result of differences in the number of contacts between settings, differences in prophylactic use of antivirals between settings, and differences in underreporting.
正确理解甲型流感病毒的感染动态依赖于可靠估计关键流行病学参数，如繁殖数、固有增长率和代间隔。尽管没有证据证明这一点，但通常假设代间隔在不同环境中是相似的。在这里，我们使用来自荷兰 16 个新型甲型流感（H1N1）集群的数据，根据年龄、集群大小和社会环境（营地、学校、工作场所、家庭）对代间隔进行分层估计。我们的分析基于贝叶斯推断框架，能够灵活处理感染链接和症状发作时间的缺失。分析表明，允许代间隔因社会环境而异的分层最符合数据。具体而言，家庭（2.1 天[95%可信区间= 1.6-2.9]）和营地（2.3 天[1.4-3.4]）中的估计代间隔短于工作场所（2.7 天[1.9-3.7]）和学校（3.4 天[2.5-4.5]）。我们的发现可能是由于环境之间接触次数的差异、环境之间抗病毒药物预防性使用的差异以及漏报的差异所致。
