基于分布滞后非线性模型的归因风险评估方法及应用

Measures and Application for Attributable Risk from Distributed Lag Non-linear Model

目的介绍基于分布滞后非线性模型的归因风险评估方法，并运用该方法评估宁波市气温暴露造成人群死亡的归因风险。方法分布滞后非线性模型通过交叉基函数实现同时描述因变量在自变量维度与滞后维度的分布，使其能够同时评估出暴露因素的滞后效应和非线性效应。收集宁波市2009-2014年人群死亡和气象资料，利用时间序列分析结合分布滞后非线性模型，评估气温造成人群死亡的归因死亡人数和人群归因分值。结果宁波市2009-2014年日均气温与总死亡的累积暴露-反应关系曲线近似呈L型，26℃为最适宜温度。归因于气温暴露造成的死亡人数为29037例（95％CI：19181-38074），占总死亡的13．39％（95％CI：9．19％-17．49％）。低温的归因风险大于高温，归因死亡人数分别为27088例和1977例，归因分值分别为12．49％和0．91％。结论无论高温或低温均与人群死亡增加相关，低温的归因风险更大。

Objective To introduce measures of attributable risk from distributed lag non -linear model（DLNM）, and to apply these methods for estimating the mortality risk attributable to outdoor temperature in Ningbo city. Methods DLNM is based on a cross-basis function that describes simultaneously the shape of the relationship along both the space of the predictor and the lag dimension of its occurrence, and could assess the potentially non-linear and lag effects. The dally data on mortality and meteorological factors were collected from 2009 to 2014 in Ningbo city. A time series study using a DLNM was used to esti-mate the attributable number and fraction to the effect of temperature on mortality. Results The overall cumulative exposure-response curve between temperature and mortality was L-shaped at lag 0 - 14 days, and the minimum-mortality temperature was 26℃. In total,13.39% （95% CI：9. 19% - 17.49% ） of total mortality was attributable to outdoor temperature,while the attribut- able number was 29037 （ 95 % CI： 19181 - 38074）. More attributable deaths were due to cold, with a fraction of 12. 49 % corre- sponding to 27088 deaths,compared with 0.91% and 1977 deaths for heat. Conclusion Both heat and cold were associated with an increased risk of daily mortality, and most mortality burden were caused by cold.