Aerosol Science and Technology 33:274���295 (2000) c 2000 American Association for Aerosol Research Published by Taylor and Francis 0278-6826/ 00/ $12.00 + .00 Evaluation of a Methodology for Quantifying the Effect of Room Air Ultraviolet Germicidal Irradiation on Airborne Bacteria Shelly L. Miller and Janet M. Macher DEPARTMENT OF MECHANICAL ENGINEERING, UNIVERSITY OF COLORADO AT BOULDER, BOULDER, CO 80309 (S.L.M.), CALIFORNIA DEPARTMENT OF HEALTH SERVICES, ENVIRONMENTAL HEALTH LABORATORY BRANCH, BERKELEY, CA 94704 (J.M.M.) ABSTRACT. As a result of the recent resurgence in tuberculosis (TB) and the increasing incidence of multidrug-resistant TB, there has been renewed interest in engineering controls to reduce the spread of TB and other airborne infectious diseases in high-risk settings. Techniques such as the use of lamps that produce ultraviolet germicidal radiation may reduce exposure to infectious agents by inactivating or killing microorganisms while they are airborne. We designed and evaluated a test method to quantitatively estimate the ef cacy of germicidal lamps, in conjunction with dilution ventilation, for reducing the concentration of viable airborne bacteria. Bacterial particles were generated in a 36 m3 room and collected with midget impingers at 5���7 locations. The effectiveness of the control technique was determined by comparing concentrations of culturable airborne bacteria with and without the control in operation. Results for a single, 15 W germicidal lamp showed reductions of 50% for Bacillus subtilis (B. subtilis) and Micrococcus luteus (M. luteus) tests with Escherichia coli (E. coli) showed nearly 100% reduction (E. coli were isolated only from the sampler nearest the aerosol source when the lamp was operating). The addition of louvers to a lamp greatly reduced its ef cacy. Decay experiments showed that roughly 4���6 equivalent air changes per hour were achieved for B. subtilis with one or two lamps operating. These preliminary experiments demonstrated that this methodology was well suited for these evaluations and identi ed factors that could be modi ed to re ne the study design for future work. INTRODUCTION Every means of defense is needed in the battle against airborne infectious diseases includ- ing resurgent tuberculosis (TB) and multidrug- Correspondin g author. resistant TB. Engineering controls are impor- tant tools in many settings, for example, lo- cal exhaust and dilution ventilation, air ltra- tion, ultraviolet germicidal irradiation (UVGI), and negative room pressurization (Nardell 1993 Conroy and Franke 1994 Nagin et al. 1994

Aerosol Science and Technology 33:3 September 2000 Evaluation of UV Germicidal Irradiation 275 Segal-Maurer and Kalkut 1994 CDC 1994). The ef cacy, costs, and problems associated with implementation of various control mea- sures have been discussed (Kellerman et al. 1997 Sutton et al. 1998 Stricof et al. 1998 Nazaroff et al. 1998). UVGI has been proposed as a supplemen- tal control in healthcare facilities (CDC 1994 NIOSH 1997 ACOEM 1998) and is also appli- cable in high-risk settings where persons with unrecognized active disease may be present (Riley and Nardell 1989 CDC 1989, 1992, 1994 Macher 1993 Nardell 1995 Barnes et al. 1996 NIOSH 1997 Behrman and Shofer 1998). UVGI has advantages and disadvantages rel- ative to other engineering controls (Macher 1993 CDC 1994 NIOSH 1997) and has re- turned to the arsenal of TB control measures because of its low cost, ease of application, and potential ef cacy, despite past failure in some settings to reduce the concentration of airborne microorganisms substantially (Goldner et al. 1980 Macher et al. 1992, 1994) or to prevent airborne disease transmission (Nardell 1995). UVGI is a form of nonionizing electromag- netic radiation with wavelengths 100���290 nm (CDC 1994). Low-pressure mercury-vapor lamps used for germicidal applications emit line spectra characteristic of mercury with pre- dominant emission at 254 nm, which is the wavelength to which we refer when using the term UVGI in this paper. The optimal wave- length for microbicidal effectiveness ranges from 250���270 nm, depending on the mi- croorganism (Luckiesh 1946 Summer 1962 NIOSH 1972). UVGI has long been used for its bactericidal and viricidal properties (Wells et al. 1942 Luckiesh 1946 Riley et al. 1957 Goldner et al. 1980) and has been shown to kill many microorganisms in water, on sur- faces, and in the air (Hollander 1942 Wells 1955 Riley et al. 1957, 1976 Summer 1962 Collins 1971 David 1973). UVGI that pene- trates to microbial DNA may cause damage suf- cient to interrupt cell replication, leading to death. UVGI can be used for air disinfection in three con gurations: (1) enclosed in mechanical ven- tilation system ducts, (2) enclosed in locally re- circulating units, and (3) in an open con gu- ration irradiating room air (Nagin et al. 1994). Irradiation of room air is achieved by germi- cidal lamps mounted on one or more walls or suspended from the ceiling, above the heads of people occupying a room. The lamp emissions are directed horizontally and toward the ceiling. The focus of this paper is room air UVGI. Determinants of UVGI effectiveness include irradiance level, duration of irradiation, room con guration, lamp placement, lamp age, air movement patterns, and the amount of moisture in the air (CDC 1994). Indoor relative humidity (RH) above 60% may lessen the effectiveness of UVGI, most likely due to changes in microor- ganism susceptibility or reactivation at different humidity conditions (Riley and Kaufman 1972). Suspended microorganisms that are exposed to UVGI may be damaged, the effect depending on the UVGI dose the cells receive and the lethal dose for each microorganism. UVGI dose is calculated from exposure duration and the ir- radiance level delivered to a cell. Riley et al. (1976) estimated the UVGI dose required to kill 90% of aerosolized Mycobacterium tuber- culosis (M. tuberculosis) to be approximately 700 W s cm 2. Animal studies have compellingly demon- strated that germicidal lamps enclosed in air ducts effectively prevent the spread of TB (Riley et al. 1957 Riley and O���Grady 1961). The rst observation of the ef cacy of room air UVGI to control airborne infection was in 1934 (Wells 1955). Riley et al. (1976) achieved rel- atively large rate constants (10���25 h 1 ) for in- activation of Mycobacterium bovis (M. bovis) BCG, a surrogate for M. tuberculosis, depend- ing on the total power of the germicidal lamps. In the latter study, room air mixing was not characterized and the rate constant was esti- mated from the slope of a concentration decay curve which may have led to an overestimation of UVGI ef ciency if the room was not well mixed (Nicas 1996 Nicas and Miller 1999).

276 S. L. Miller and J. M. Macher Aerosol Science and Technology 33:3 September 2000 Epidemiological studies have shown that UVGI can reduce transmission of airborne infectious diseases in hospitals (McLean 1961), military housing (Willmon et al. 1948), and classrooms (Wells et al. 1942 Perkins et al. 1947 Wells and Holla 1950). However, irradiation of class- rooms was found not to be suf cient to inter- rupt measles transmission if children also were exposed elsewhere in the community (Nardell 1995). There are also anecdotal experiences with UVGI that support its use. Stead et al. (1996) found no documented tuberculin skin test con- versions in 159 hospital employees who were exposed for 3 months to a patient with cavi- tary TB. The isolation rooms at this hospital used UVGI and enhanced ventilation. Burk et al. (1978) reported the lack of active TB or posi- tive skin tests among 514 infants who were ex- posed over a 3 month period to a nursery super- visor with pulmonary TB. The nursery also used UVGI and enhanced ventilation. Other success- ful experiences with UVGI in Denver, Boston, and New York have been reported (Healthcare Hazardous Materials Management 1994). Measurements of UVGI effects in real-life settings are limited. Goldner et al. (1980) found that the ef ciency of UVGI in operating rooms with ceiling-mounted lamps was 74 and 40%, respectively, for species of Bacillus and Micro- coccus. Lidwell (1994) reported a twelve-fold reduction in viable, ambient airborne bacteria in an operating room with high levels of UVGI (290 W cm 2). Macher et al. (1992, 1994) saw a 14���19% reduction in the concentration of viable, ambient bacteria in a well-ventilated hos- pital waiting room equipped with wall-mounted lamps. Recommended limits for occupational expo- sures to UVGI, based on a combination of an- imal data and voluntary human exposures with eye or skin irritation as the endpoint, have been published (NIOSH 1972 AIHA 1991 ACGIH 2000). The guideline for 8 h exposure to 254 nm UVGI is 6 mJ cm 2, with an average permis- sible exposure irradiance for an 8 h period of 0.2 W cm 2 [(6 mJ cm 2 103 W s mJ 1 )/ (8 h 3600 s h 1)]. Permissible irradiance lev- els for shorter or longer exposures are calculated similarly (CDC 1994). Broad-spectrum ultravi- olet radiation has been associated with an in- creased risk for squamous and basal cell carci- nomas of the skin, and the International Agency for Research on Cancer has classi ed UV-C as ���probably carcinogenic to humans (Group 2A)��� (IARC 1992). Short-term overexposure to UVGI can cause erythema and photokeratoconjunctivitis (NIOSH 1972). These skin and eye reactions have been seen in workers occupying an irradi- ated area for extended periods of time (Jones and Gillen 1992 Noll 1995 Brubacher and Hoffman 1996). Overexposure in a laboratory has also been documented (Murray 1990). Other health hazard evaluations have noted worker overexposure as well as inadequate lamp main- tenance, labeling of lamps, worker training, and worker use of personal protective equipment (NIOSH 1992a, 1992b, 1992c). Advice on se- lecting and installing germicidal lamps is sum- marized elsewhere (CDHS 1990 Macher 1993 CDC 1994 Nardell 1995), as are recommenda- tions on preventing worker overexposure (CDC 1994 NIOSH 1992a, 1992b, 1992c). To expand the limited experimental data on the effect of irradiating room air, we developed and tested a methodology to quantify the dam- age UVGI caused to airborne bacteria. We in- vestigated two experimental designs, a steady- state and a decay method, similar to approaches others have used to study the effect of UVGI (Wells and Wells 1938 Riley et al. 1976). We designed, equipped, and characterized a cham- ber with two wall-mounted germicidal lamps, a bacterial aerosol generation system, a ventila- tion system, and multiple air sampling locations. We tested our methodology on three aerosolized bacteria chosen to span a range of UVGI sensi- tivity. UVGI ef cacy was determined by com- paring measurements of the steady-state con- centration of culturable airborne bacteria with and without UVGI. Decay experiments were interpreted to estimate UVGI equivalent air ex- change rates. In addition, the spatial variability