Domínguez A, Muñoz P, Martínez A, Orcau A
Department of Public Health and Health Regulations, Universitat de Barcelona, Spain.
J Epidemiol Community Health. 1996 Jun;50(3):293-8. doi: 10.1136/jech.50.3.293.
This study aimed to investigate the behaviour of two indicators of influenza activity in the area of Barcelona and to evaluate the usefulness of modelling them to improve the detection of influenza epidemics.
Descriptive time series study using the number of deaths due to all causes registered by funeral services and reported cases of influenza-like illness. The study concentrated on five influenza seasons, from week 45 of 1988 to week 44 of 1993. The weekly number of deaths and cases of influenza-like illness registered were processed using identification of a time series ARIMA model.
Six large towns in the Barcelona province which have more than 60,000 inhabitants and funeral services in all of them.
For mortality, the proposed model was an autoregressive one of order 2 (ARIMA (2,0,0)) and for morbidity it was one of order 3 (ARIMA (3,0,0)). Finally, the two time series were analysed together to facilitate the detection of possible implications between them. The joint study of the two series shows that the mortality series can be modelled separately from the reported morbidity series, but the morbidity series is influenced as much by the number of previous cases of influenza reported as by the previous mortality registered.
The model based on general mortality is useful for detecting epidemic activity of influenza. However, because there is not an absolute gold standard that allows definition of the beginning of the epidemic, the final decision of when it is considered an epidemic and control measures recommended should be taken after evaluating all the indicators included in the influenza surveillance programme.
本研究旨在调查巴塞罗那地区流感活动的两个指标的表现，并评估对其进行建模以改进流感疫情检测的有用性。
描述性时间序列研究，使用殡葬服务机构登记的全因死亡人数和报告的流感样病例数。该研究集中于五个流感季节，从1988年第45周至1993年第44周。对登记的每周死亡人数和流感样病例数进行处理，采用时间序列自回归积分滑动平均（ARIMA）模型识别法。
巴塞罗那省的六个大城镇，每个城镇都有超过6万居民且均设有殡葬服务机构。
对于死亡率，所提出的模型是二阶自回归模型（ARIMA(2,0,0)）；对于发病率，是三阶自回归模型（ARIMA(3,0,0)）。最后，对这两个时间序列进行综合分析，以促进对它们之间可能存在的关联的检测。对这两个序列的联合研究表明，死亡率序列可以与报告的发病率序列分开建模，但发病率序列既受之前报告的流感病例数影响，也受之前登记的死亡率影响。
基于总体死亡率的模型对于检测流感的流行活动是有用的。然而，由于没有绝对的金标准来界定疫情的开始，在考虑何时认定为疫情以及推荐何种控制措施时，最终决策应在评估流感监测计划中包含的所有指标之后做出。
