The performance of culture-based methods and microscopy for quantification of noninfectious airborne microorganisms in epidemiological studies of highly contaminated work environments.

Airborne levels of microorganisms traditionally have been measured by culture-based methods. Culture-based methods are suitable for the detection of infectious agents, but their suitability for the detection of microorganisms with toxic and allergic effects is less clear, because these effects do not depend on viability of the organisms. During the last 15 years several noncultural methods have been developed for the quantification of airborne microorganisms, including microscopic methods. Microscopy may be expected to provide more valid exposure estimates of microorganisms than culture-based methods, because live and dead microorganisms can be detected. However, their validity may also depend on the ability to differentiate between species. The literature was searched for epidemiological studies in which exposure-response analyses were carried out using culture-based methods and/or microscopy. The influence of several factors on exposure-response associations were considered: design; population size; analytical method; sampling method; exposure levels; outcome; and confounder adjustment. Thirteen studies were found, including a total of 49 exposure-response analyses, and 45% of the analyses showed associations. It was found that the potential of microscopic methods to uncover exposure-response associations was only marginally better than that of culture-based methods (47 and 44%, respectively). Exposure-response associations were more often found with fungi (70%) than with gram-negative bacteria (50%) or total bacteria (22%), perhaps because fungal exposure is more strongly associated to respiratory outcomes than exposure to bacteria. But the shortcomings of the measurement methods may also be important. Further development of measurement methods for bacteria is therefore needed. The complex composition of bioaerosols in many work environments necessitate the assessment of exposure to multiple agents and multivariate statistical analysis of exposure-response associations.