中文摘要：

为了探讨沙尘天气，特别是沙尘暴对人群呼吸系统健康的影响。对过去9年（1995-2003）沙尘暴高发地区甘肃省武威市的6所大中型医院沙尘天气高发季节（3月1日-5月31日）每日呼吸系统疾病入院人数进行了调查．采用半参数广义相加模型（Semi-parametric generalized additive model，GAM），在控制了长期趋势、日历效应（calendar effect）以及气象因子等混杂因素后，考虑到具体年份的差异。分别对每年建立GAM回归模型（滞后7d）并进行分析．然后应用Meta-分析方法的固定效应模型对各年GAM分析结果（即相对危险度RR）进行效应量合并，计算合并效应量（RR＋），以此表示沙尘天气对呼吸系统疾病日入院人数的综合影响．结果表明：（1）对于沙尘暴的影响，各年呼吸系统疾病日入院相对危险度（RR）合并效应量（RR＋）随沙尘暴滞后天数的不同而不同。且在滞后3、4、5d时每天的RR＋均有统计学意义，其RR＋分别为1．40（95％CI为1．06-1．86）、1．34（95％CI为1．01-1．77）、1．73（95％CI为1．35～2．23）；（2）扬沙天气在其滞后第6天RR＋有统计学意义（RR＋为1．14（95％CI为1．01-1．30）；（3）浮尘天气与扬沙天气类似，也在其滞后第6天RR＋有统计学意义，其RR＋为1．12（95％CI为1．00-1．25）．（4）沙尘天气对人群健康的影响，可根据滞后不同天的RR＋值变化，分为滞后早期效应、滞后中期效应和滞后后期效应三类；（5）沙尘天气引发和加重的呼吸系统疾病主要为各种肺炎、上呼吸道感染和感冒以及急、慢性支气管炎．这些结果表明。呼吸系统日入院人数沙尘天气。特别是沙尘暴与有关联系。且均表现为滞后效应。而且沙尘暴的影响强度大于扬沙天气或浮尘天气的影响．

英文摘要：

To investigate the association between dust events, especially dust storms, and daily respiratory hospitalization and how the association （ RR ＋ ） was distributed across various lags, six good-sized hospitals in Wuwei city of Gansu province, China. were selected. During the period from March 1 ^at to May 31^at in every year （from 1995 to 2003 ） when dust events occurred frequently annually, an investigation about daily respiratory hospitalization was conducted. After controlling long time trend, calendar effect and other meteorological confounding factors, we fit a semi-parametric generalized additive model （GAM） in each year （from 1995 to 2003 ） to allow for year-specific differences, then combined the calculated year-specific relative risks （RRs） by using fixed effect model of meta-analysis and the combination effect size of year-specific results （ RR ＋ ） were obtained. Using these models that considered simultaneously the effect of dust events up to 7 days, The distributions of RR ＋ across lags of dust events were found. Our results show that （ 1 ） for dust storms, there were 3,4, 5-day delayed effects of dust storms, their RR ＋ were 1.40 （95 % CI 1.06 - 1.86）, 1.34 （95 % CI 1.01 - 1.77 ） and 1.73 （95%CI 1.35 -2.23）, respectively; （2） There was 6-day delayed effect of blowing dust （RR ＋ = 1. 14（95% CI 1.01 - 1.30） ; （3） Similarly, the 6-day delayed effect of floating dust （ RR ＋ = 1.12 （95% CI 1.00 - 1.25 ） also was found ; （4） According to the distribution of RR ＋ across various lags, the damaging effects of dust events on population health were divided into as initial-delayed effect, medium term- delayed effect and last-delayed effect. It might be due to different sensitives and responses of populations with the different damaging effects on dust events and different toxicological roles of various harmful factors in dust events; （5） The dominant respiratory diseases induced and aggravated by dust events included pneumonia, infl