【摘要】 目的探讨某三甲医院死亡病例的时间分布情况,从大体上了解该院死亡病例周期性变化趋势,为后期工作开展及应对提供参考。方法本次研究收集2011年1月至2015年12月该三甲医院的所有死亡患者的数据,包括患者的死亡时间,使用SPSS 21.0对死亡时间的分布进行分析,并加以预测,以2015年实际观察值验证模型的预测效果。结果 2011-2014年的数据变化中存在明显的季节因子,周期性变化明显,其中加法模型R2最高,因此选择加法模型进行拟合,模型拟合效果判定指标,RMSE值=8.776;MAPE值=8.368;MAE值=6.552;标化后的BIC值=4.586。本次研究结果中预测误差值在20%以内,且实测值均落在95%CI的范围内,经配对t检验发现实测值与预测值之间差异无统计学意义(t=0.448,P=0.663),实测值与预测值之间相关系数r=0.780(P=0.003)。结论新疆某三甲医院的死亡病例的时间分布符合指数平滑相加模型,存在明显的季节变化周期,可能与选择性就医有关。本结果显示该院在1、5、9月份的死亡数上升,应在工作中做好应对,做好死者家属的安抚及处理工作。更多还原

【Abstract】 Objective To explore time distribution of deaths in tertiary hospital for a better understanding of cyclical factors and trends of hospital deaths occurred, and to provide reference for the further development of targeted policies. Methods The data of the death of patients in a tertiary hospital from January 2011 to December 2015 were collected. SPSS21.0 was used to analyze the distribution of the time of death. The effect of forecast model was verified by actual observed value in 2015. Results There were obvious seasonal factors in the change of data from 2011to2014, and there was significant periodic change. Among them, the additive model R2 was the highest, which was chosen for fitting, with the effect determination indexes, of RMSE=8.776, MAPE=8.368, MAE=6.552, BIC after standardization=4.586. The predicted error values were within 20%, and the measured values were within the range of 95% CI.There was no significant difference between the measured and the predicted values by the paired t test(r=0.448, P=0.663). Correlation coefficient between measured value and predicted value was r=0.780(P=0.003). Conclusion Time distribution of deaths of tertiary hospital in Xinjiang is in line with the exponential smoothing additive model, with obvious seasonal change, which may be related to selective treatment. The results show that the number of deaths risen in the January, May, September, and it is suggested that corresponding measures should be taken to deal with the deceased and to appease their families.更多还原