Fine particulate matter measurements in Swiss restaurants, cafés and bars: What is the effect of spatial separation between smoking and non-smoking areas?


M. Röösli
Institute of Social and Preventive Medicine at Swiss Tropical Institute
Steinengraben 49
CH-4051 Basel
Tel.: +41 61 270 2215
Fax: +41 61 270 2225


Abstract  We performed 124 measurements of particulate matter (PM2.5) in 95 hospitality venues such as restaurants, bars, cafés, and a disco, which had differing smoking regulations. We evaluated the impact of spatial separation between smoking and non-smoking areas on mean PM2.5 concentration, taking relevant characteristics of the venue, such as the type of ventilation or the presence of additional PM2.5 sources, into account. We differentiated five smoking environments: (i) completely smoke-free location, (ii) non-smoking room spatially separated from a smoking room, (iii) non-smoking area with a smoking area located in the same room, (iv) smoking area with a non-smoking area located in the same room, and (v) smoking location which could be either a room where smoking was allowed that was spatially separated from non-smoking room or a hospitality venue without smoking restriction. In these five groups, the geometric mean PM2.5 levels were (i) 20.4, (ii) 43.9, (iii) 71.9, (iv) 110.4, and (v) 110.3 μg/m³, respectively. This study showed that even if non-smoking and smoking areas were spatially separated into two rooms, geometric mean PM2.5 levels in non-smoking rooms were considerably higher than in completely smoke-free hospitality venues.

Practical Implications

PM2.5 levels are considerably increased in the non-smoking area if smoking is allowed anywhere in the same location. Even locating the smoking area in another room resulted in a more than doubling of the PM2.5 levels in the non-smoking room compared with venues where smoking was not allowed at all. In practice, spatial separation of rooms where smoking is allowed does not prevent exposure to environmental tobacco smoke in nearby non-smoking areas.