摘要:Authors Yvonnick Guillois-Bécel , Isabelle Tron , Yann Le Strat , Bertrand Gagnière , Agnès Verrier , Claire Gourier-Fréry , Alain Briand Cellule de l’institut de veille sanitaire en région Bretagne Agence régionale de santé de Bretagne Cire Ouest 6, place des Colombes CS 14253 35042 Rennes cedex France, Observatoire régional de santé de Bretagne 8 D, rue Franz Heller CS 70625 35706 Rennes cedex 7 France, InVS 12, rue du Val d’Osne 94415 Saint-Maurice cedex France Key words: carbon monoxide, carbon monoxide poisoning DOI : 10.1684/ers.2011.0494 Page(s) : 477-84 Published in: 2011 Background: In the Morbihan district, 1,316 council flats were equipped in 2003 with a CO detector with an expected life span of 10 years. The model was a metal oxide semiconductor detector that was claimed to comply with the European standard (EN 50291). Methods: To assess the operation of these detectors, we conducted a descriptive study among tenants, supplemented by laboratory tests of new detectors of the same model and of devices removed after 4 years of operation. During these tests, the detectors were exposed to gases with CO concentrations ranging from 33 to 330 ppm. The time until the alarm went off was compared to the European standard requirements in each test. Results: Residents in 387 households were questioned: 19.5% [16.5-22.9] reported an alarm. The acceptance of detectors was good among tenants; however only 39.8% [35.8-44.0] had read the user manual, and relatively few knew how to respond to alarms appropriately. The incidence of the alarm days – days associated with at least one alarm – rose by a factor greater than 8 between 2004 and 2006. Symptoms consistent with CO poisoning were reported for 7.7% [3.1-18.0] of the alarm days. Alarm circumstances were partially consistent with known sources of CO exposure, and one dangerous situation was detected. None of the new detectors complied with the European standard. The test results appear to suggest that rapid sensor drift is responsible for low-level alarms and that sensor saturation may explain the alarm failures. The latter represented 18.3% of the test results of the removed devices. Conclusions: Assessment of the performance of CO detectors before they are authorised for sale would be beneficial for public health. Providing information to the households equipped with detectors about the use of their safety device would also be useful.