Fernández-Pérez C, Tejada J, Carrasco M
Servicio de Medicina Preventiva, Hospital Universitario San Carlos, Madrid, Spain.
Int J Epidemiol. 1998 Apr;27(2):282-8. doi: 10.1093/ije/27.2.282.
The present study describes the use of time series analysis in the evaluation of the incidence of nosocomial infection. The main hypothesis analysed was that monthly occurrence of nosocomial infection in a hospital may be related to work-related factors such as the control and training of personnel imposed by a surveillance system, strikes supported by medical personnel and movement of personnel. Time series analysis was used to quantify, model and statistically evaluate these interventions.
The data employed (March 1982-December 1990) were supplied by the nosocomial infection surveillance system of a primary-care general hospital. The monthly time series incidence of nosocomial infections (measured as percentage cumulative incidence) was analysed by curve fitting, autoregressive, integrated and moving average (ARIMA) modelling (Box-Jenkins) and intervention and dynamic regression analysis.
The imposed control and training of personnel by the surveillance system was associated with a 3.63% decrease in the accumulated monthly incidence of nosocomial infection from 7.82% to a baseline level of 4.19%. There was a strong indication that an increase of infection incidence of 4.34% corresponded to a medical strike. This increase was maintained over the following months raising the baseline level to 4.84%. An increase of 0.18% was associated with each new nursing contract. Evidence was obtained for the possible relationship between incidence of nosocomial infection and vacation periods.
The results suggest the need for strict control of the activities of hospital personnel and for the adoption of certain preventative measures during vacation periods to avoid an undesirable increase in the incidence of nosocomial infections.
本研究描述了时间序列分析在医院感染发生率评估中的应用。分析的主要假设是，医院中每月医院感染的发生情况可能与工作相关因素有关，如监测系统对人员的控制和培训、医务人员支持的罢工以及人员流动。时间序列分析用于量化、建模和统计评估这些干预措施。
所使用的数据（1982年3月至1990年12月）由一家基层综合医院的医院感染监测系统提供。通过曲线拟合、自回归、积分和移动平均（ARIMA）建模（Box-Jenkins）以及干预和动态回归分析，对医院感染的每月时间序列发生率（以累积发生率百分比衡量）进行了分析。
监测系统对人员实施的控制和培训使医院感染的累积月发生率从7.82%下降了3.63%，降至4.19%的基线水平。有强有力的迹象表明，感染发生率增加4.34%与一次医疗罢工相对应。这一增加在随后几个月持续存在，将基线水平提高到4.84%。每签订一份新的护理合同，感染发生率增加0.18%。获得了医院感染发生率与假期之间可能存在关系的证据。
结果表明需要严格控制医院工作人员的活动，并在假期采取某些预防措施，以避免医院感染发生率出现不良增加。
