摘要
目的 比较不同统计方法在手足口病季节性研究中的应用价值,为手足口病季节性防控提供科学依据。方法 以南宁市手足口病为例,收集2006-2009年南宁市各乡镇、城区手足口病的新发病例数,利用流行曲线、季节指数、变异系数、圆形分布法以及扫描统计量等方法分析手足口病的季节性趋势。结果 流行曲线直观描述南宁市手足口病高发月份集中在4-7月,12月和1月呈现小高峰。季节指数判断1月、4-7月为南宁市手足口病的高发季节,以5-6月份发生强度最大。圆形分布显示南宁市手足口病的高发日期为5月10日,高发时段跨度大。变异系数显示5-6月和9月的变异程度较低,为潜在的季节性月份。扫描统计量显示南宁市手足口病高发时段集中在4月下旬到7月上旬之间,以5-6月份最为高发;高发时段的相对风险为非高发时段的1.33-1.85倍。结论 不同统计方法在检测手足口病季节性规律时的侧重面、精确性和实际意义有所不同,探测到高发时段的价值和意义亦有所不同。变异系数描述疾病在某时段内流行强度的变异度,变异度高说明疾病流行强度的强弱交替,但变异度低可能意味着普遍高发或普遍低发。流行曲线、季节指数虽可直观描述手足口病的季节趋势,但缺乏统计推断,可能受随机误差的影响。圆形分布可定量探测出具有统计学意义的手足口病高发日期和高发时段,但可能因高发时段跨度较大,而失去实际意义。扫描统计量兼顾手足口病的时空信息,在校正单纯性时间变化趋势和周内效应后,可探测到更精确、稳定和更具实际意义的高发时段,在探讨手足口病的季节趋势中具有优势。
Objective To compare the utility of statistical methods for detecting seasonality of HFMD. Methods Daily data on notified cases of HFMD over the period of 2006 - 2009 was collected from Nanning prefecture. Epidemic curve, seasonality index, coefficient of variance, circular distribution and scan statistics were employed to detect the seasonality of HFMD. Resuits Period of April to July was found to be the months with higher HFMD incidence by epidemic curve, as well as December ＆ January with secondary incidence peak. Seasonality index found that January and April to July were the months of higher HFMD incidence ,May and June were the incidence peak of HFMD as well. Circular distribution showed that 10th May was the peak onset of HFMD ,and the duration of higher HFMD incidence was over llth January to 5th September. May and June were the potential months with higher HFMD incidence, as well as lowest coefficient of variance. And period of late April to early July was detected as higher HFMD incidence cluster by scan statistics,with relative risk of 1.33 - 1.85. Conclusion As far as various statistical methods were concerned, the capacity of detecting seasonality was inconsistent. Coefficient of variance just showed the magnitude of variance of HFMD incidence, rather than identifying months with higher incidence of HFMD. Considering the shortage of statistic inference, the seasonality of HFMD detected by seasonality index would be disturbed by random error, although it Could statistically describe the seasonal trend of HFMD incidence. To some extent, the capacity of circular distribution was determined by parameter, and the outcome of detecting seasonality might be impractical. In contrast, scan statistics might be more powerful,practical and robust to detect seasonality of HFMD,by integrating spatiotemporal information of HFMD and adjusting for effect of day-of-week, as well as effect of purely temporal trend.
出处
《中国卫生统计》
CSCD
北大核心
2015年第3期410-413,共4页
Chinese Journal of Health Statistics
基金
广西科学研究与技术开发计划项目(桂科攻10124001A-65)
广西自然科学基金(2013GXNSFBA019125)
广西研究生教育创新计划资助项目(YCSZ2013029)
