Environmental Drivers of Bacillus-Positive Blood Cultures in a Cancer Hospital, Sapporo, Japan

Abstract

The Bacillus species is a well-documented causative pathogen of nosocomial bloodstream infection. The present study aimed to identify climatological variables that are associated with Bacillus-positive blood culture in Sapporo, Japan. All cases with Bacillus-positive blood cultures from January 2011 to December 2016 were retrospectively analysed. Climatological data from 2011 to 2016, including daily mean temperature and absolute humidity, were retrieved from the Japan Meteorological Agency. Employing a hazard-based statistical model to describe the non-homogeneous counting process in which temperature and absolute humidity act as explanatory variables, we computed all possible models with variable lengths of time lag. Akaike Information Criterion was computed to identify the best fitted model. High wavelet power at 12 months was identified for the period from 2013 onwards, which coincided with the time period in which sampling multiple sets of blood culture has been recommended. The temperature-only model with a lag of six days yielded a high sensitivity value (72.1%) and appeared to be the optimal model to predict Bacillus-positive blood culture with the highest area under the receiver operating characteristic curve value. Temperature was identified as a climatological driver of Bacillus-positive blood culture. Our statistical modelling exercise offers an important message for infection control practices to improve awareness among healthcare workers of the identified association and mechanically controlled in-room temperature.