Monitoring of nosocomial invasive aspergillosis and early evidence of an outbreak using cumulative sum tests (CUSUM)

Abstract

In order to provide a statistically based evaluation of the incidence of invasive aspergillosis (IA) over time, we applied the cumulative sums (CUSUM) methodology, which was developed for quality control and has already been applied for the surveillance of hospital-acquired infections. Cases of IA were recorded during a 5-year period. Incidence rates of cases assumed to be hospital-acquired, i.e. nosocomial IA (NIA), were analysed using CUSUM tests. Relationships between NIA, fungal contamination and construction or renovation work were tested using time-series methods. Between January 2002 and December 2006, 81 cases of NIA were recorded. CUSUM analysis of NIA incidence showed no significant deviation from the expected monthly number of cases until August 2005, and then the CUSUM crossed the decision limit, i.e. identified a significant increase in NIA as compared with the reference period (January 2002 to December 2004). Up to April 2006, the learning-curve CUSUM stayed over its limit, supporting an ongoing outbreak involving 24 patients, and then it significantly decreased in May 2006. Follow-up after May 2006 indicated no out-of-control situation, supporting a return to the baseline situation. In haematology wards, significant links were found between NIA incidence and fungal contamination of several sites at each ward (mainly unprotected common sites). An environmental source of contamination could be suspected, but no significant relationship was found between NIA incidence and ongoing construction or renovation. In conclusion, the CUSUM test proved to be well suited for real-time monitoring of NIA and for early identification and follow-up of an outbreak.