WA) and its bundled software. .... HVAC ducts service the space, the relative locations of each being indicated by ... facility from the hallway) and has a set point of between 8 to 10 ACH as maintained by a Metasys® Building Management.
Articles
U.S. Environmental Protection Agency (EPA). Proceedings: EPA nanotechnology and the environment: applications and implications STAR progress review workshop. EPA Document Number: EPA/600/R-02/080. Washington, DC: 2003. Warheit DB, Sayes CM, Reed KL, Swai KA. Health effects related to nanoparticle exposures: environmental, health and safety considerations for assessing hazards and risks. Pharmacol Therapeut. 2008;120:35-42. Wittmaack K. In search of the most relevant parameter for
quantifying lung inflammatory response to nanoparticle exposure: particle number, surface area, or what? Environ Health Perspect. 2007;115:187-94. Woodrow Wilson International Center for Scholars (WWICS). The nanotechnology consumer products inventory. Washington, DC: 2006. Available from: www. nanotechproject.org/. Yokel R, MacPhail RC. Engineered nanomaterials: exposures, hazards, and risk prevention. J Occup Med Toxicol. 2011;6:7.
Characterization of Respirable Aerosols Generated During Routine Laboratory Procedures Halli E. Miller1, Shane Patzlsberger1, Diane J. Rodi2, and Richard T. Robinson1* 1Medical
College of Wisconsin, Milwaukee, Wisconsin and 2Argonne National Laboratory, Argonne, Illinois
Abstract Biosafety Level 2 (BSL-2) and Biosafety Level 3 (BSL-3) facility lab personnel routinely perform procedures that are capable of producing respirable aerosols. While these procedures are considered safe when performed in the closed environment of a biosafety cabinet (BSC), there are, nevertheless, limited data regarding the nature of the aerosols these procedures produce. This lack of aerosol data poses a significant challenge to biosafety professionals, who are charged with assessing the risks associated with handling infectious materials and communicating the importance of proper engineering controls to BSL-2/BSL-3 facility staff. This article characterizes the extent and nature of respirable aerosols produced during routine laboratory procedures when these procedures are performed in an open environment (i.e., outside a BSC). As demonstrated, homogenization, vortexing, and pipetting each produce aerosols of differing characteristics and aerogenic potentials. These characteristics have led to the development of an Exposure Risk Model for routine lab methods to assist biosafety professionals in their efforts to prevent inhalation exposures.
Keywords Aerosols, Aerogenic Potential, Homogenizer, Vortex, and Pipetting
Introduction The degree and size distribution of generated infectious particulates are a major determinant of occupational exposure risk in biomedical laboratories. A general rule of
www.absa.org
thumb for Biosafety Level 2 (BSL-2) facility protocol is to carry out aerosol-producing procedures within a biological safety cabinet (BSC). However, data-driven guidance for biological safety officers is limited to only a few studies (Bennett & Parks, 2005; Kenny & Sabel, 1968; Kupskay, 2002), as funding for such basic procedural studies has diminished since the 1960s. Kupskay (2002) assembled a summary of the data available for aerosol production during laboratory operations, primarily dating from the 1950s. Data sets from Kenny and Sabel (1968) and Bennett and Parks (2005) primarily focus on accident scenarios rather than standard operating procedures. To quantitate the risk of accidental exposure to an aerosolized pathogen, Dimmick (1973) also introduced the concept of “the spray factor,” which is a ratio of the aerosol produced over the starting volume; the higher the spray factor for a procedure, the more likely it would produce infectious particulates at a high enough number to make lab personnel sick (Dimmick, 1973). Analysis of the patterns of laboratory acquired infections (LAIs), however, indicates that anywhere from 60% to 80% of LAIs are not the result of an identifiable inadvertent release (Harding & Byers, 2006). This infers that standard agent handling protocols might be accountable for the exposure and resulting illness, not inadvertent release. The contribution of contact transmission (secondary to transfer of locally settled droplets followed by handto-face transfer) to these LAI occurrences is difficult to measure in any quantitative fashion. This article reports the results of investigation into the degree and size/spatial distribution of particulates produced by three methods that are routinely used in BSL-2 and BSL-3 facilities: homogenization of infectious samples
Applied Biosafety
Vol. 19, No. 1, 2014
19
Articles
with a hand-held homogenizer; mixing samples via vortex; and pipetting the homogenate onto selective agar to enumerate pathogen burden. This article demonstrates that these methods are point sources of detectable aerosols that would likely come into contact with lab staff ( 12" from the point source) if carried out on a lab bench. The extent to which particulates capable of lower airway deposition (i.e., those 5 μm) are produced varied with a number of parameters. It is anticipated that these data can be used by biosafety professionals when assessing the risks associated with homogenizing infectious materials and communicating the importance of proper BSL-2 engineering and administrative controls to facility staff.
Methods Particle Count Measurements Particle concentration data were collected using a Fluke 985 Particle Counter (Fluke Corporation, Everett, WA) and its bundled software. As recommended in the manufacturer’s instructions, the particle counter was “blanked” prior to any particle measurements using the provided zero count filter. Shortly after blanking the Fluke 985, measurements of 0.3μm, 0.5 μm, 1 μm, 2 μm, 5 μm, and 10 μm particle concentrations were made for homogenization, vortexing, and pipetting using the setups described in the Results section. Homogenization, Vortexing, and Pipetting For aerosol recordings during homogenization, 25 mL of Dulbecco’s Modified Eagle Medium (DMEM; Life Technologies, Carlsbad, CA) or complete DMEM (i.e., DMEM containing 10% fetal bovine serum (FBS) was transferred into a 50 mL polypropylene conical (Corning, Corning NY) and affixed—as described in Results—under a Tissue-Tearor 985370-XL homogenizer (Biospec Products, Bartlesville, OK). After immersing the homogenizer probe portion under the media meniscus, the homogenizer was turned on at either low, medium, or high speeds. Following a delay period (to allow the homogenizer to reach a continuous speed), particle count measurements were then collected at 1" intervals from the conical lip; a ~22 second collection period was used at each interval (i.e., the amount of time necessary to collect data from 1 L room air). To ensure aerosols measured came from the fluid and not the motor, the authors attempted to measure small particle aerosols generated by the motor, which in this study is largely surrounded by casing. They attempted to take a background reading of any particles produced by the homogenizer by letting it run without the probe portion being in any liquid (i.e., dry); note, however, that this runs counter to the manufacturer’s instructions to not let the homogenizer probe run dry. After the standard warm-up/delay period, the homogenizer quickly began to overheat and release small metal shards. None of these phenomena occurs when the homogenizer is used per the manufacturer’s instructions. Measuring the number of particles produced by the homog20
www.absa.org
enizer motor was not feasible; furthermore, this number is also likely to be minimal given the extent to which the motor is encased. In terms of impact on data, whatever particles were produced by the homogenizer in the 22-second warm-up period would, presumably, be uniform throughout all time points and distances. Therefore, the data would be “normalized” in this regard. For aerosol recordings during vortexing, a 50 mL and 15 mL conical were each filled with 1/5 their total volume and held on the cup head of a Vortex Genie 2 (Scientific Industries, Bohemia, NY); the Vortex Genie dial was set to a speed of 5-6. Following a delay period, particle count measurements were then collected at 1" intervals starting at 4" from the conical lip; a ~22 second collection period was used at each interval. Aerosols produced by homogenization and vortexing were measured as the number of particles per liter of room air (Log10 transformed). For homogenization and vortexing, particle count measurements were collected for a minimum of 3 separate days (i.e., each method was performed once each day for 3 days), with the each day’s background aerosol levels subtracted for data analysis. For aerosol recordings during pipetting, a plastic Petri plate containing 18 mL of solidified Luria Broth (LB) agar was placed on a lab bench, with the particle counter probe directly above the agar. For an individual “trial,” a 200 uL volume was pipetted onto the agar surface a total of four times; the cumulative particle numbers detected at the Petri plate lip over the four iterations was then recorded. A total of two trials were performed in this manner, with each trial on a separate day.
Data Analysis Particle concentration data were downloaded from the Fluke 985 Particle Counter and analyzed using Microsoft Excel 2010. Analyzed data were plotted using Graph Pad Prism.
Results Background Particulate Sizes and Concentrations Vary in a Standard BSL-2 Laboratory Setting In an indoor setting, variability in airborne particulate concentration varies greatly depending on where measurements are taken with respect to supply/exhaust vents and traffic routes (Allander & Strindehag, 1973). Prior to collection of experimental data from mock-infectious materials, baseline concentrations of particles within each size range in different sectors of the facility were measured to determine which areas would provide the lowest particle noise levels. A graphical representation of the facility and the relative locations of its supply and exhaust HVAC vents are depicted in Figure 1A. To determine the baseline particle concentrations, the usable area of the lab was divided into 132 different sectors; for each particle size, concentration measurements were collected three times over as many days. As shown in the heat maps in Figures 1B-1F, the mean concentrations of (Figure 1B) 0.3 μm, (Figure 1C) 0.5 μm,
Applied Biosafety
Vol. 19, No. 1, 2014
Articles
(Figure 1D) 1 μm, (Figure 1E) 2 μm, and (Figure 1F) 5 μm particulates varied in a statistically significant manner among different areas of the facility. The concentration of 0.3 μm particles was the highest in the room (103.9-104.1 particles/L room air), while aerogenic particles 5 μm in size were much lower (100.4-101.5 particles/L room air). As a negative control, it was confirmed that all particle concentrations dropped to zero when the particle counter was placed within a Class II Type A1 BSC (data not shown). Since the encircled bench spaces in Figure 1A demonstrated the lowest background concentrations for noise particulates for each size, all subsequent measurements of aerosol particle production were carried out at this location within the facility.
Aerosol Production as a Function of Distance During Homogenization To characterize the aerosols produced during homogenization, the setup shown in Figure 2 was established. Specifically, a 50 mL conical polypropylene culture tube containing a defined volume of media (25 mL) was fixed in place with the probe end of a clamped homogenizer immersed 1" below the media meniscus. To estimate the three-dimensional profile of the aerosol spray, initial particle concentration measurements within the five size ranges (0.3 μm, 0.5 μm, 1 μm, 2 μm, and 5 μm particulates) were determined at defined horizontal distances from the point source (0"-12" from the lip of the conical), as well as defined heights and times posthomogenization (Figure 2). Data were controlled for speed
Figure 1 The concentrations of 0.3 m, 0.5 m, 1 m, 2 m, and 5 m size range particulates in each area of the facility were measured to establish the areas with minimal background particulate noise levels. Shown in (A) is a schematic of the facility which contains 5 work benches, 2 BSCs and a single exit to the hallway. Three supply and three exhaust HVAC ducts service the space, the relative locations of each being indicated by encircled “a’s” and “e’s”, respectively. (B-F) Air flow is single pass maintained as a net negative with respect to the adjacent hallway (i.e., air flows into the facility from the hallway) and has a set point of between 8 to 10 ACH as maintained by a Metasys® Building Management System. Lab space was segregated into 132 different sectors with each sector depicted by a square on the map. Within each sector, the concentrations of (B) 0.3 m, (C) 0.5 m, (D) 1 m, (E) 2 m, and (F) 5 m size range particulates were determined in the absence of any aerosol-producing activities. Data are represented in a heat-map format as overlaid onto the facility schematic. The adjacent legend matches each color to a corresponding particulate range concentration (Log10/L room air). Based on these initial baseline particulate counts, the area encircled by a dashed-line in (A) was chosen for experimental measurements as this location represented the greatest signal to noise region.
www.absa.org
Applied Biosafety
Vol. 19, No. 1, 2014
21
Articles
Figure 2 Experimental setup for particulate measurement produced via homogenization. A tissue homogenizer was clamped onto a ring stand, with the probe immersed into an immobilized 50 mL conical polypropylene culture tube containing 25 mL of either DMEM or cDMEM. With the probe tip positioned 1” under the media meniscus, the homogenizer was turned on (at low, medium or high speed) with particle counts collected at positions along the indicated dashed lines. Specifically, while the homogenizer was running, particle counter (PC) measurements were made at 1” intervals along a 12” parallel path beginning at the lip of the conical (dashed line); at a point 6” from the lip of the conical (indicated by an ~), additional measurements were made at intervals -2.3” below to 11.8” above (vertical dotted line).
of homogenization and protein concentration by collecting data at low, medium, or high speed and with both DMEM and complete DMEM (cDMEM, which is supplemented with 10% fetal bovine serum). Figure 3 shows the concentration profile for size ranges of (Figure 3A) 0.3 μm, (Figure 3B) 0.5 μm, (Figure 3C) 1 μm, (Figure 3D) 2 μm, and (Figure 3E) 5 μm particulates at 1" intervals from the conical lip, taken at low (left panel), medium (middle panel), and high (right panel) homogenizer speed. With the exception of 5 μm particulates (Figure 3E), homogenization produced a net detectable aerosol that peaked in concentration at roughly 6" away from the lip of the conical (Figures 3A-3D). The presence of protein in the media did not significantly affect measured aerosol characteristics, as both DMEM (open squares) and cDMEM (closed squares) displayed similar patterns of particulate concentration. As evidenced by the 0.3 μm data (Figure 3A), concentrations measured up against the tube lip during homogenization were reproducibly below background concentration levels. These values fall below the dashed lines of Figure 3 and are assumed to reflect a vortex effect near the conical lip. The generation of (Figure 3A) 0.3 μm, (Figure 3B) 0.5 μm, (Figure 3C) 1 μm, and (Figure 3D) 2 μm size aerosols were all modulated by homogenizer speed, with progressively higher speeds resulting in lower aerosol production (compare the highest values observed in the left panel of Figure 3A to those of the middle and right panels of Figure 3A). Collectively, these data demonstrate that homogenization of aqueous samples results in an aerosol that is mostly concentrated 6" away from the point source, the concentration of measured particulates is in22
www.absa.org
versely related to the homogenizer speed, and the presence of protein in the aqueous sample does not significantly affect the aerosol patterns as measured by this technique.
Aerosol Properties as a Function of Height During Homogenization Characterization of aerosols as a function of height above and below the tube lip was carried out at the measured optimized distance of 6" along the vertical using the same setup as shown in Figure 2. Concentration data at heights ranging from 2.3" below the conical lip to 11.8" above the conical lip are plotted in Figure 4; data are expressed as the percent of total aerosol produced as a function of height. As shown, among the total aerosol produced for each particle size, the greatest concentrations were found between 2.3" and 7.1" above the lip of the conical tube (Figures 4A, 4C, 4E, 4G, and their associated regression line maxima). Notably, upon removing the aerosol point source (i.e., turning off the homogenizer), aerosols of each particulate size dropped to baseline values within 20 seconds (Figures 4B, 4D, 4F, 4H, 4J). These data point to the possibility that aerosols generated during homogenization either settle or are removed from the data collection area by HVAC within 20 seconds of generation. Aerosol Production During Vortexing Vortexing is another commonly used technique in BSL-2 facilities that may serve as a source of potentially infectious aerosol particulates. The experimental setup shown in Figure 5A was used in conjunction with a 15 mL or 50 mL conical polypropylene culture tube atop a vortex-
Applied Biosafety
Vol. 19, No. 1, 2014
Articles
Figure 3 Characterization of aerosols generated by sample homogenization (distance). Measurements of (A) 0.3 m, (B) 0.5 m, (C) 1 m, (D) 2 m and (E) 5 m size range particulates at each interval were taken using DMEM (closed squares, ) or cDMEM (open squares,
) while the homogenizer was set on low (left panel), medium (middle panel) or high speed (right panel). After correcting for background particulate levels—measured just prior to turning on the homogenizer— net concentration values were plotted as a function of the distance from the point source. Data points represent the mean (±S.D.) particulate concentration at each distance, over three separate trials on as many days. The dashed lines in plots A, D, and E are positioned at experimental-background = 0 to emphasize negative data points (i.e., the background particulate concentrations were greater than the net particulates produced by homogenization).
www.absa.org
Applied Biosafety
Vol. 19, No. 1, 2014
23
Articles
Figure 4 Characterization of aerosols generated by sample homogenization (height and time post-generation). DMEM was homogenized at low speed while PC measurements were collected 6” away at seven different heights (Heights 1-7) above and/or below the conical lip. The number of particles at a given height (NHeight 1) was divided by the total number of particles counted at all heights (NHeight 1-7) to determine the percent of total aerosol present at each height. Represented in (A, C, E, G, I) are the percent of total aerosol values for (A) 0.3 m, (C) 0.5 m, (E) 1 m, (G) 2 m and (I) 5 m particulates as a function of height above the aerosol source, as well as a non-linear regression of the same data (dashed lines with associated R2 values). Immediately upon removing the aerosol point source, (B, D, F, H, J) the decay in number of (B) 0.3 m, (D) 0.5 m, (F) 1 m, (H) 2 m, and (J) 5 m particulate aerosols was monitored via PC measurement. Shown are the total number of particles detected (over a 10 sec collection period) as a function of the time after removing the aerosol source. Data shown are representative of three separate trials on as many days.
24
www.absa.org
Applied Biosafety
Vol. 19, No. 1, 2014
Articles
er cup head (each filled with media at 1/5 of the total volume of the tube). The vortexer was activated on medium speed while particle measurements were made at 1" intervals beginning 4" from the conical lip. The results of this analysis are plotted in Figures 5B-5F for particles 0.3 mμ, 0.5 μm, 1 μm, 2 μm, and 5 μm in size. Only 0.5 μm-sized particles using a 15 mL conical culture tube were generated in quantities above baseline through vortexing. These concentrations, however, were roughly 1 order of magnitude
or more lower than those produced by homogenization (compare concentrations of 0.5 μm particles in Figure 5C to that produced by the homogenizer in Figure 3B). Collectively, these data demonstrate that for particles 0.5 μm or larger, vortexing can create a measureable aerosol that— albeit modest relative to that produced by homogenization—is within a size and distance range that may contact lab personnel either directly or indirectly by surface deposition on their hands or directly within their breathing zone.
Figure 5 Characterization of particles produced during vortexing. Shown in (A) is the setup for determining the extent to which aerosols were generated during sample vortexing. Specifically, while vortexing using either a 50 mL or 15 mL conical (each filled with 1/5 of their total volume), particle count measurements were made at one inch intervals beginning 4” from the conical lip at the same vertical height. Shown are the mean concentrations (±SD) of (B) 0.3 m, (C) 0.5 m, (D) 1 m, (E) 2 m, and (F) 5 m particulates as a function of distance from the point source, with particle counts from 15 mL conicals represented by closed squares (), and particle counts from 50 mL conicals represented by open squares (
). Data represent the combined results of three separate experiments.
www.absa.org
Applied Biosafety
Vol. 19, No. 1, 2014
25
Articles
Aerosol Production During Micro-pipetting To enumerate pathogen burden in experimental samples, BSL-2 facilities commonly pipette organ or tissue culture homogenates onto selective agar. After 1 or more days, colonies become apparent and can be counted and extrapolated back to the source. While pipetting occurs in close proximity to the agar surface, the elastic nature of agar raises the possibility that microscopic amounts of pipetted materials can actually bounce off the agar surface, landing on lab personnel directly or on the work surface. To detect mock infectious material redirection from agar surfaces, the particle counter was set up above the lip of an agar-containing Petri plate. 200 uL of media were pipetted onto the agar surface (Figure 6A) while simultaneously measuring airborne particulate concentrations. The results of 44 independent experiments are shown in Figure 6B; these demonstrated that out of 44 pipetting repetitions, 32 produced measureable above-baseline particulates in the 1 μm to 10 μm size range (73%). The largest number of detected particles fell within the 0.5 μm size range followed by 1 μm, 2 μm, 5 μm, and 10 μm in size. Collectively, these data demonstrate that routine pipetting onto agar surfaces results in the deposit of microscopic particles onto surrounding surfaces, with the frequency and quantity of production inversely correlated with average particulate diameter.
Discussion Research into host-pathogen interactions remains a national research priority, and infection remains a leading cause of death worldwide. Experimental models of hostpathogen interactions are predominantly studied in BSL-2 facilities and, to a lesser extent, in containment labs (BSL-3 or BSL-4 facilities). Studies typically comprise controlled interactions between a human pathogen (either of viral, bacterial, fungal, or protozoan origin) and an in vitro or in vivo host. Regardless of the model system or facility used, certain lab methods are routine for assessing pathogen bur-
den in these hosts. These methods include homogenization of infected cells or tissues, vortexing samples to equalize distribution of an agent, and pipetting onto agar media. Each of these manipulations is a potential aerosol point source, capable of exposing lab personnel to an infectious agent. However, data regarding the nature of the aerosols produced by these methods are limited, making it difficult for biosafety professionals to accurately assess the risks associated with handling pathogens outside of a BSC. This study demonstrates that homogenization, vortexing, and pipetting produce a measureable aerosol with characteristic properties and concentrations. Among the three methods, homogenization produced aerosols with the greatest concentration of particles capable of infecting the lower airways ( 0.5μm based on the review of Green [Green, 1968]). Furthermore, the aerosols produced by homogenization traveled further than those produced by vortexing and remained detectable 20 seconds post-homogenization. By way of comparison, the aerosols produced by vortexing were more limited in scope and tube-size dependent (i.e., 15 mL tubes produced more 0.5 μm and 1 μm particles than 50 mL tubes, standardizing for volume percent). Curiously, the particle count data suggest that net particle counts were reduced below baseline noise by vortexing, suggesting a local vacuum effect. Bernoulli’s Principle predicts that when vortices are formed in stirred fluids, a dynamic pressure is created that is lowest in the core region closest to the axis. Similar effects have been observed with bathtub whirlpools that draw a column of air down the core or with aerodynamics of airplane engine forward vortices which suck water and small stones into the core (and then into the engine). This may account for why higher homogenization speeds were less effective at generating aerosols than lower homogenization speeds. On a final note, although the majority of aerosol particulates produced by pipetting were 1 μm and would not allow for lower airway infection, they are nevertheless capable of transferring an infectious agent to lab personnel or the work surface. The
Figure 6 Characterization of particles produced during pipetting. Shown in (A) is the setup for detecting whether and how many particles are bounced off an agar surface while pipetting. While pipetting 200 uL of media onto the agar surface, particle count measurements were collect at the lip and vertical height of the petri plate. A single “pipetting trial” consisted of four pipetting repetitions of 200 uL media. Shown in (B) are the data from these pipetting trials, with each dot representing the number of particles reflected (as a function of particle size) for a single pipetting trial. Data represent the combined results of two separate experiments. 26
(A)
www.absa.org
Applied Biosafety
Vol. 19, No. 1, 2014
Articles
Figure 7 Exposure Risk Model for Routine Lab Methods
duration of pathogens’ viability on work surfaces varies greatly with the species under investigation (Kramer et al., 2006). It is important to note that these data, in addition to supporting the conclusions above, also demonstrate the importance of effective HVAC systems in diminishing the risk of aerogenic infection. By filtering air coming into a laboratory, HVAC systems help reduce the introduction of pathogens known to exist in outside air (Brodie et al., 2007); however, HVACs also eliminate from room air those pathogens introduced into the air by infectious individuals (Allander & Strindehag, 1973) or routine lab methods. Specifically, in this laboratory, the decay of particles was observed within 20 seconds of eliminating the aerosol point source (Figures 4B, 4D, 4F, 4H). This highlights the importance of assessing the sufficiency of a given facility’s HVAC system for minimizing exposure risk at each individual BSL-2 and BSL-3 research facility. Collectively, from these data the authors have developed an exposure risk model of these routine lab methods and the aerosols they respectively generate (Figure 7). As represented by the purple triangle (Figure 7, tapering off from left-to-right), the methods of pipetting, vortexing, and homogenizing result in the production of particles of decreasing size. Likewise, and as represented by the pink triangle (Figure 7, tapering off from right-to-left), the methods of pipetting, vortexing, and homogenizing result in the production of aerosols of increasing concentration and spread. Due to the combination of these two properties (i.e., particle size and aerosol spread), pipetting is more likely to spread infectious agents to the immediate vicinity of activities (i.e., hand and lab bench deposit), while homogenization is the most likely to result in exposure at a greater distance from production via direct inhalation of the agent. Building on the work of Dimmick (1973), it is assumed that the greater the concentration of micro-organism in the solution being homogenized, the greater the number of microorganisms per particle and thus likelihood of personnel exposure. However, it is important to note that aerosols generated by coughing do not necessarily demonstrate a linear correlation between aerosol droplet size and number of colony forming units (CFU) contained within (Fennelly et al., 2012). In conjunction with related studies (Bennett & www.absa.org
Parks, 2005), these data form a basis for risk determination of infection potential and the need for the implementation of engineering controls in BSL-2 facilities.
Acknowledgments This work was supported by the American Biological Safety Association (ABSA) through an Elizabeth R. Griffin Research Foundation grant (to R.T.R). *Correspondence should be addressed to Richard T. Robinson at rrobinson@ mcw.edu.
References Allander C, Strindehag O. Particle concentrations in patient rooms with various types of ventilation. J Hyg (Lond). 1973;71:633-40. Bennett A, Parks S. Microbial aerosol generation during laboratory accidents and subsequent risk assessment. J Appl Microbiol. 2005;100:658-63. Brodie EL, DeSantis TZ, Parker JPM, Zubietta IX, Piceno YM, Anderson GL. Urban aerosols harbor diverse and dynamic bacterial populations. PNAS. 2007;104:299-304. Dimmick RL. Laboratory hazards from accidentally produced airborne microbes. Dev Ind Microbiol. 1973;15:44-7. Fennelly KP, et al. Variability of infectious aerosols produced during coughing by patients with pulmonary tuberculosis. Am J Respir Crit Care Med. 2012;186:450-57. Green GM. Pulmonary clearance of infectious agents. Ann Rev Med. 1968;19:315-36. Harding AL, Byers KB. Chapter 4: Laboratory-acquired infections. In Fleming DO, Hung DL, editors. Biological safety: principles and practices, 4th ed. Washington, DC: ASM Press; 2006. p.63 Kenny MT, Sabel FL. Particle size distribution of Serratia maccescens aerosols created during common laboratory procedures and simulated laboratory accidents. Appl Microbiol. 1968;16:1146-50. Kramer A, Schwebke I, Kampf G. How long do nosocomial pathogens persist on inanimate surfaces? A systematic review. BMC Infect Dis. 2006;6:130. Kupskay B. Risk assessment for enteric pathogens in the biosafety level 2 laboratory. Appl Biosafety. 2002;7:120-32.
Applied Biosafety
Vol. 19, No. 1, 2014
27
