PPLIED ND ENVIRONMENTL MICROBIOLOGY, Sept. 1984, p. 481-485 99-224/84/9481-5$2./ Copyright C) 1984, merican Society for Microbiology Vol. 48, No. 3 Effects of irflow Rates and Operator ctivity on Containment of Bacterial erosols in a Class II Safety Cabinet JNET M. MCHERt ND MELVIN W. FIRST* Department of Environmental Science and Physiology, Harvard School of Public Health, Boston, Massachusetts 2115 Received 9 December 1983/ccepted 12 June 1984 Biological safety cabinets are frequently relied upon to provide sterile work environments in which hazardous microorganisms can be safely handled. Verification of correct airstream velocities does not, by itself, ensure that adequate protection will be achieved under all users. Instead, the concentration of microorganisms in a cabinet operator's breathing zone must be measured during typical cabinet use conditions to determine whether the exposure is below acceptable limits. In this study, cabinet operator exposures were measured with a personal air sampler. Bacterial spores were released inside a cabinet as a uniform challenge aerosol, and the number of escaping spores was measured for several cabinet arrangements during a number of typical operations. The following were studied to determine their effects on aerosol containment: inflow air velocity, size of access opening, type of operator movements, location of operator's hands, and pace of activity. Other experiments examined differences in aerosol containment for eight typical microbiology operations when performed by six operators who covered a range of body heights and volumes. Class II biological safety cabinets are used in bacteriology, mycology, and virology laboratories, and in pharmacies (i) to protect laboratory personnel from aerosolized material, especially biological hazards, and (ii) to protect work from contamination. In the United States, cabinets are evaluated according to National Sanitation Foundation (NSF) Standard No. 49 (5), which covers requirements for design, construction, and performance. ustralian (6), British (2), German (3), and Japanese (4) standards describe similar requirements for cabinet construction and testing. Laboratory pesonnel working at safety cabinets are protected by an inward flow of air through the work opening. High-efficiency particulate air-filtered exhaust and downflow air prevent contamination of the environment and materials inside a cabinet. Cabinets qualify for NSF certification after demonstrating that an aerosol of Bacillus subtilis spores sprayed toward the window from inside does not penetrate the air barrier (personnel protection). cabinet must control an identical aerosol generated outside the unit so that it does not contaminate the work area (product protection) and show that sideways travel of aerosols over the work surface is kept to a minimum (cross-contamination). Once the velocity profile of the internal downflow air and the face velocity (average velocity of air entering the work opening) have been determined for a cabinet that meets all NSF No. 49 requirements and is certified by the NSF, production models qualify for Standard No. 49 certification if the measured air velocities fall within 1% of those of the NSF-tested cabinet. The biological certification tests measure aerosol containment in unused cabinets operating at optimal airflows. They do not measure containment when an operator is conducting regular procedures that have a potential to disrupt airstreams, nor do they provide information on the effect of poor work practices. Therefore, a method is needed to measure operator exposure to microbiological aerosols originating inside cabinets during the wide variety of procedures * Corresponding author. t Present address: Forsvarets Forskningsanstalt, S-91 82 Ume'a, Sweden. 481 conducted within these enclosures. small, light-weight, and efficient liquid impinger (4a) was used for this purpose. To monitor breathing zone aerosol concentrations, the impinger was worn on the laboratory coat lapels of test subjects working at safety cabinets and conducting a number of frequently used procedures. The accuracy of prevailing recommendations for cabinet use was evaluated with this personal sampler. MTERILS ND METHODS Biological safety cabinet. 1.2-m (4-ft)-wide, Class II, Type Bi cabinet (Nuire, Inc., Minneapolis, Minn.) was chosen for this study because it has an adjustable work opening, and the arrangement of the supply filters and blowers allowed a test aerosol to be homogeneously mixed in the downflow air (Fig. 1). The cabinet was positioned in a closed room to avoid cross-drafts. The blowers were operated for at least.5 h before testing to assure stabilized airflows. UV lamp inside the cabinet was turned on, with the cabinet operating, for at least 1 h between tests to further reduce the number of residual spores. Study 1 was conducted to investigate the effect of cabinet operating conditions on spore retention. The average inflow air velocity was set at two values, either ca..38 m/s (75 fpm) or ca..5 m/s (1 fpm), and the inflow opening was adjusted to a height of either 2 or 25 cm. For all tests, the downflow velocity was ca..25 m/s. The air velocity and window area opening combinations resulted in airflow rates of ca. 5,, 6,, 7,, and 8,5 liters per min. In study 2, cabinet operating conditions were maintained constant at an inflow air velocity of.52 m/s, downflow velocity of.27 m/s, and window height of 25 cm. Tracer aerosol generation and collection. Bacillus subtilis spores were stored at 4 C and assayed on Trypticase soy agar (BBL Microbiology Systems, Cockeysville, Md.) incubated at 37 C for 18 h. Further incubation at room temperature for 24 h improved the development of color in the colonies without increasing their size significantly and allowed more certain differentiation from other organisms. The spores were diluted to a concentration of 16 to 18 spores per ml in phosphate gelatin diluent (1 g of gelatin and 4 g of Na2HPO4 in 1 liter of distilled water). The same diluent
482 MCHER ND FIRST a) Front Sectional View b) Side Sectional View FIG. 1. Class II, Type B1 biological safety cabinet modified for activity testing. HEP, High-efficiency particulate air filter. was used in the impinger with only.1 g of gelatin to reduce foaming. Two six-jet Collison nebulizers (BGI, Inc., Waltham, Mass.), operated at 2 lb/in2, dispersed ca. 1 ml of the total 4-ml spore suspension in 3 min..18- to 1.3-liter sample of the cabinet aerosol was collected on a membrane filter (Millipore Corp., Bedford, Mass.;.45-ptm pore size) to measure the final concentration in the air of the cabinet. The filter was placed on a Trypticase soy agar plate and incubated. The cabinet spore concentration ranged from 23 to 62 CFU per liter during study 1 and from 24 to 64 CFU per liter in study 2. ir samplers. personal impinger (4a) was clipped to the lapels of a cabinet operator's laboratory coat, over the sternum, 15 to 2 cm from the nose. The impinger was operated at maximum flow for an average sampling rate of 1.3 liters per min, measured with a calibrated rotameter. single sample was collected during the 25 min of activity in each test of study 1. Two impinger samples (14 and 11 min) were collected for each test of study 2. fter sampling, three Trypticase soy agar plates were inoculated with.1 ml of the impinger collecting liquid (1-ml total volume), 8.5 ml of the remaining liquid was filtered, and the filter was placed on a Trypticase soy agar plate. rotating slit-to-agar sampler (New Brunswick Scientific Co., Inc., New Brunswick, N.J.) was placed at the lower edge of the cabinet window sill, 2 cm from each side wall. The slit samplers each collected 25 liters per min. Turntables with 15-cm-diameter agar plates revolved beneath the sampling slots, completing 1 revolution per h. The slit sampler plates were divided into 6 1-min sectors, and the number of spores in the laboratory room air during each activity period was determined from the total number of bacillus colonies growing in the sectors of the two plates onto which air was impacted during that time. Operators. single cabinet operator was used in study 1 to control for differences between people and to assure repeatable operator movements. Three men (Fig. 2, subjects, B, and C) and three women (Fig. 2, subjects D, E, and F) participated in study 2. Three were considered taller than average (, B, and D), and three were shorter than average > 1 ~~~~~~~~~~~~access PPL. ENVIRON. MICROBIOL. (C, E, and F). One subject in each group was heavier, i.e., wider and bulkier, than average (B and F). The women were familiar with the activities conducted in these simulations, whereas the men although familiar with chemistry bench work, were not microbiologists. During the tests, the volunteers wore laboratory coats that were buttoned for two of four tests, one each seated and standing, and worn open for the other two tests. The shorter subjects stood on blocks, as needed, to bring them to suitable working levels (Fig. 2). The height of a laboratory stool was adjusted to accomodate each seated subject. ctivities. In study 1, the operator moved one or both hands either from the center of the cabinet work tray to a side wall or out through the work opening and back inside. The activity was continuous for 2-min periods with a 1-min pause between activities to separate the samples collected on the slit sampler plates. Movements were paced with a metronome to assure reproducibility. Each activity was conducted slowly or quickly in the front or rear of the cabinet. The entire routine of motions was repeated at least three times for each combination of inflow air velocity and access opening height. The activities of study 2 reproduced a number of routine culture manipulations. Three methods of mixing a culture in a test tube were compared, i.e., placing a tube on a Vortex mixer, inverting a screw-capped tube, and rolling a tube between two hands. The test tubes were filled with water by pipette before the mixing procedures and were emptied into a flask afterward. The rack of test tubes and the Vortex mixer were placed behind the front-rear center line of the work tray, 39 cm from the window and 25 cm from the rear wall. The operations of the second half of study 2 simulated processing of plate cultures. set of petri dishes was repeatedly (i) inoculated by pipette with ca..1 ml of water from a flask, (ii) touched with a bent glass rod to spread the inoculum over the agar surface, and (iii) streaked with a wire loop dipped into a flask. The procedures were first performed at a slow pace, while standing and while sitting, to familiarize the subjects with the carefully timed routines and were then performed at a pace twice as fast. Protection factor. The British standard (2) requires that cabinets demonstrate a protection factor of 15, i.e., an operator using a cabinet may experience no more than 1-5 times the exposure received when the work is conducted on an open bench. The protection factor (PF) is calculated as 193cm 189cm 165cm 18cm 16cm 155cm height - 2 86kg 1 kg 55kg 68kg 51kg 75kg weight 15 - o ~~~~~~~~~~~~cm e B C D E F TEST SUBJECT FIG. 2. Morphometric data for test subjects of study 2. Impinger location relative to open window is indicated by triangles:, standing;, seated. Shaded areas indicate the heights of blocks on which shorter subjects stood.
VOL. 48, 1984 the ratio between the transfer index for a room with turbulent ventilation (V-1) and the exposure experienced at a given point (nlns), or: PF = NsIVn where N is the number of particles liberated (N 2 3 x 18 spores), s is the sampling rate (liters per minute) (s 5 liters per min), V is the effective ventilation rate for open bench work (V is assumed to be 1 m3/min), and n is the number of particles recovered at sampling rate s (for a PF of 15, n s CFU). The NSF standard no. 49 requirements for personnel protection convert to PFs very close to the British limit. The PF for the all-glass impinger sampler location (center of cabinet) is 7.5 x 14, (n is 1 CFU, N is 18 spores, s is 6 samplers x 12.5 liters per min). t the slit sampler location (sides of cabinet), the NSF protection factor is 1.1 x 15 (n is 5 CFU, N is 18 spores, s is 2 samplers x 28 liters per min). Statistical analyses. Multiple regression analysis was used to relate the concentration of tracer spores measured by the air samplers (the dependent, or response, variable) with a number of independent, or explanatory, variables describing the airflows in the cabinet, the activities of the operator, and conditions in the laboratory. Simple linear regression describes, by the slope and intercept of a line, the relationship between a dependent and a single independent variable over a range of values of each. Multiple regression accommodates any number of predictor variables for a given response variable, generating a regression coefficient for each that indicates whether or not the factor is significant in explaining the variation seen in the dependent variable (1). The sign of the regression coefficient expresses the direction of the correlation between the two factors. With this method, although cabinet aerosol concentration and room air temperature and relative humidity were not constant, their effects on aerosol escape could be accounted for over the range of test levels of the other variables, i.e., airflow velocity, height of access opening, type of operator movement, and operator factors. Because of the continuous nature of the impinger's operation, one measurement of aerosol concentration in the cabinet user's breathing zone was taken for the entire procedure of each of the 17 tests of study 1 for which breathing zone data were available. Each of the 21 tests for which slit sampler data were collected was subdivided by type, location, and speed of movement into 168 individual measurements (21 tests x 2 movements x 2 locations x 2 speeds). These data were also categorized and analyzed by the individual, two-factor, descriptor variables: type of motion, location of hands, and speed of movements. Separate impinger samples were collected for the mixing and plating operations of study 2 (N = 6 operators x 4 tests per operator x 2 impinger samples per test = 48). On the slit TBLE 1. Comparison of tracer concentration outside a cabinet at two inflow velocities and two window heightsa vg concn [CFUfiter x 14 + SEM (N)] Regression variable Personal impinger Slit-to-agar sampler Inflow velocity.38 m/s 7.91 2.37 (7) 1.8 4.8 (56)b.5 m/s 5.7 2.36 (1) 4.68 1.8 (112)b Height of access opening 2 cm 1.2 3.13 (9)C 8.12 3.1 (88)C 25 cm 1.82.919 (8)C 5.16 1.67 (8)C b p =.2, as indicated by regression analysis. c Significant difference, P c.5. CONTINMENT OF BCTERIL EROSOLS 483 TBLE 2. Comparison of tracer concentration outside a cabinet for two types of operator movements, two locations of hands, and two speeds of workinga Regression variable vg concn (CFUfiter x 14 ± SEM) Hand motion Side to side.671 ± O.O911b Through window 12.7 ± 1.97b Location of work Front half 4.16 ±.737c Rear half 9.26 ± 2.2c Speed of motion Slow pace 5.71 ±.976 Fast pace 7.71 ± 1.95 Results are from the slit-to-air sampler (N = 84). b Significant difference, P c.1, as indicated by regression analysis. c Marginally significant difference, P =.6. sampler plates, these activities were subdivided into five mixing and three plating operations (N = 6 operators x 4 tests per operator x 8 samples per test = 192). RESULTS Work opening height, examined in study 1, was a statistically significant predictor of the spore concentration at both sampling locations (Table 1). Work opening height was negatively correlated with aerosol penetration of the air barrier, i.e., tracer concentration was higher outside the cabinet at either inflow velocity for a 2-cm-high window than it was for a 25-cm-high window. Inflow air velocity was significantly, and also negatively, correlated with room air concentration at the sides of the cabinet (Table 1). erosol concentration within the cabinet was a highly significant factor at the slit sampler location (P =.1). s would be expected, more spores were collected outside the cabinet when the concentration was higher inside. Room air temperature and relative humidity were not correlated with aerosol concentration. Separate regressions were run on the slit sampler data for each type of motion, then for each location of hands, and finally for each speed (Table 2). larger number of spores was collected when hands were moved through the window than when they were moved sideways. fter accounting for the effects of air velocity, window height, and type of movement, work location, i.e., close or far from the access opening, was marginally significant. More aerosol escaped when movements began in the rear of the cabinet. The difference between the two working speeds was not statistically significant, but a higher average concentration was seen during faster movements. In these tests, the limit of detection, a single colony, was.8 CFU per liter of air for the slit sampler, i.e., 1 CFU/ 1,2 liters of air; and.33 CFU per liter for the impinger, i.e., 1 CFU/3 liters. n acceptable level of leakage under the NSF test is an aerosol concentration of.17 CFU per liter at the sides of a cabinet, the slit sampler site; and.27 CFU per liter in the center of a cabinet, all-glass impinger site. The concentrations detected during these tests ranged from s.8 to 1.2 CFU per liter (about 7 times the acceptable level) at the sides of the cabinet. The spore concentration in an operator's breathing zone ranged from s.33 to.66 CFU per liter (about 24 times the acceptable level). Calculated as protection factors, the slit sampler results ranged from 5 x 14 to infinity (no spores collected) during side-to-side movements of the hands within the
484 MCHER ND FIRST cabinet and ranged from 4 x 12 to 2 x 15 when hands came through the window. Regression analysis revealed no significant differences between the eight activities of the second study (Table 3). Whether the laboratory coat was unbuttoned and free to flap around or closed did not affect aerosol containment, nor did the pace of the activities. Operator sex was an important predictor of the number of spores collected at either sampler location. More spores escaped from the cabinet into an operator's breathing zone, but not out the sides of the cabinet, when the operators stood than when they were seated. The position of the impinger was higher relative to the access opening when the operators stood (Fig. 2). The samples collected in the breathing zone were positively correlated with operator height and weight (Fig. 3), except for subject F, the shortest, but third heaviest, person. Shorter and lighter, i.e., narrower, persons caused greater aerosol leakage at the sides of the cabinet, the slit sampler location (Fig. 3B). DISCUSSION Our test of a safety cabinet's containment capabilities differed from NSF certification protocol by (i) the addition of an operator working at a cabinet, (ii) sampling in the operator's breathing zone with a personal sampler rather than with fixed area samplers, and (iii) introducing a test aerosol in such a manner that a uniform concentration of spores existed throughout the downflow cabinet air, not a heavy concentration delivered at high velocity in the center of the work area. The results demonstrate that an inflow air velocity of.5 TBLE 3. Comparison of tracer concentrations outside a cabinet for two or eight activities, two working speeds and laboratory coat fits, male and female operators in seated and standing positions' vg concn (CFU/liter x 14 ± SEM) Regression variable Personal Slit-to-agar impingerb sampler' ctivity Test tubes 261 ± 51.4 Fill test tubes 85.5 ± 12.3 Vortex test tubes 73.5 ± 12.6 Invert test tubes 9.2 ± 12.5 Roll test tubes 92.8 ± 11.4 Empty test tubes 69.4 ± 6.3 Plating 282 ± 57.6 Inoculate plates 67.2 ± 8.96 Spread inoculum 67. ± 7.39 Streak plates 65.8 ± 7.86 Pace Slow 267 ± 46.7 81.3 ± 5.2 Fast 276 ± 64.8 71.6 ± 5.2 Coat Closed 263 ± 42.2 77.4 ± 5.6 Open 28 ± 67.8 75. ± 4.62 Sex Women 167 ± 24.2d 69.4 7.7e Men 376 + 69.7d 83.4 9.7e Posture Seated 19 ± 34.1e 75.8 ± 4.37 Standing 353 + 68.2e 77.1 + 5.8 bn = 24. ' N = 24 for activity subsambles; N = 96 for other averages. d Marginally significant difference, P c.7. e Significant difference, P ±.5, as indicated by regression analysis. -J w (3 4X -J CD Cl) 4 6 5 F 4 F 3-2 1 1 1 1 9 8 7 6 5 WEIGHT (kg) 5 6 7 8 9 1 -. ID 155 165 175 185 195 HEIGHT (cm) WEIGHT (kg) 5 6 7 8 9 1 E F C E C N N ' PPL. ENVIRON. MICROBIOL. ~~~D B ~~~~BB 155 165 175 185 195 HEIGHT (cm) FIG. 3. Effects of height ( ) and weight ---) on tracer concentration at (operator's breathing zone, impinger sampler) and B (corners of the work opening, slit sampler). Letters indicate operator identification. Separate lines drawn for men (, B, and C) and women (D, E, and F). Standard error of the mean bars are on concentration means at heights only. m/s contained the aerosol better than did an airflow of.38 m/s. 25-cm-high work opening was more protective than a 2-cm-high opening. It is generally thought that a larger opening provides more opportunity for aerosol particles to escape, but in these tests, inflow air velocity was maintained constant when the window was raised. Given identical air velocities, a larger volume of air was drawn in through a wider opening, and the greater air flux provided better protection against loss of aerosol particles from the interior. The operator's movements in study 1 were intentionally artificial to maintain a reproducible test protocol by which to study the effects on cabinet containment of various combinations of work opening height and inflow air velocity. In most cases, the spore concentration outside was within the acceptable limits for the unmanned United States and British tests. That withdrawing hands completely through a window
VOL. 48, 1984 caused more disruption of the air barrier and greater aerosol leakage was expected. Particles are extracted from a cabinet as contaminated air is withdrawn along with a person's arms and hands, and contamination can be shed from arms and equipment. It is therefore essential, for maximum personnel protection, that supplies be arranged to avoid reaching through a window during aerosol generating operations. That working in the rear of a cabinet was less safe than working in the front was surprising and contrary to recommended practice. When cultures are handled behind the downflow air split of a cabinet, aerosol sprays are contained in the air exhausted from the rear (Fig. 1). Contamination is not expected in the portion of air moving toward the front grille, in the direction of an operator. In this study, however, tracer spores were mixed with all cabinet air, even that in the front half. The presence of an operator's body closer to a window when reaching to the rear of a cabinet obstructed the inflow of air and created sufficient turbulence to cause cabinet air to spill out. Significantly different levels of exposure were measured for the men and women, although whether this was due to a difference in level of skill and dexterity or to the positive correlation of maleness with height and weight cannot be clearly determined. Men, having more weight in the upper body half than women, are wider at the shoulders and present a broader obstacle at a work opening. The shorter subjects did not stand or sit considerably lower at the cabinet than did the taller ones, because their absolute heights were augmented to allow them to work comfortably at the access opening (Fig. 2). To reach equipment in the rear half of the cabinet, however, the shorter subjects moved closer to the window and stretched further than did the taller ones. Nevertheless, shorter operators did not appear to be exposed to higher breathing zone aerosol concentrations because of this. Rather, the taller and, to some extent, the bulkier subjects caused escaping spores to remain at the center of the cabinet. Higher aerosol concentrations were collected at the sides of the cabinet for the thinner operators, and to some degree for the shorter operators, suggesting that leakage occurred peripherally when a narrow person worked close to the open window. Identifying cabinet downflow air separations with smoke tubes and observing researchers working at biological safety cabinets revealed that few work comfortably behind the air split, because shorter persons cannot easily reach far into a cabinet, whereas taller ones hunch over to adjust eye level and arm position. Subject F presented an interesting combination of predictor factors. ll aerosol samples were very low (Fig. 3), which was consistent with her sex or degree of skill, but confounded the trend for the correlation between breathing zone aerosol concentration and height and that between peripheral aerosol escape and weight. It is reasonable to assume that the effectiveness of a cabinet is compromised when workers' movements are rapid, as opposed to slow and deliberate. This tendency was seen in our tests, but it was not statistically significant. lthough major advantages of using bacterial aerosols are the ability to detect very low concentrations of escaping CONTINMENT OF BCTERIL EROSOLS 485 organisms and the moderate cost compared with other tracer and detector systems, we realize that in-place biological tests cannot be undertaken in laboratories where the release of a bacterial aerosol would compromise clean work. Various other materials, e.g., salt and dye aerosols, and gases, have been used to measure containment. potassium iodide aerosol is widely used in Great Britain, and a dioctyl phthalate aerosol is the challenge in the ustralian cabinet test. Workers can also be monitored for exposure to agents handled in research laboratories or those encountered in clinical specimens, etc., when a personal sampler is used with culture media suitable for the organisms under investigation. Conclusions. This study demonstrated that testing biological safety cabinets with persons working at them, conducting their usual culturing operations, provides different information on cabinet performance than does static testing. Further containment tests of this nature should be undertaken to evaluate other cabinet designs and laboratory procedures on a comparative basis and to measure the effects on aerosol containment of various combinations of activities, inflow and downflow air velocities, access opening heights, and cabinet widths. Finally, it is important to keep in mind that even well-operated biological safety cabinets lose a very small fraction of aerosolized particles. When absolute containment is required, a glove box must be used. CKNOWLEDGMENTS We thank the six volunteers who served as the test subjects and who remain anonymous; L. Harding; R. Fink, D. Liberman, and F. Schaefer of the Harvard University and Massachusetts Institute of Technology, Cambridge, Mass. Biological Safety Offices for reviewing early versions of the manuscript; and H. Feldman for guidance with the statistical analyses. J.M.M. was supported by a Public Health Service Educational Resources Center training grant from the National Institute for Occupational Safety and Health while conducting this research. LITERTURE CITED 1. rmitage, P. 198. Statistical methods in medical research, p. 32-34. Blackwell Scientific Publications, Boston, Mass. 2. British Standards Institute. 1979. BS 5726, specification for microbiological safety cabinets. British Standards Institute, London. 3. Deutsche Forschungsgemeinschaft. 198. Emphelungen zum Einsatz von Mikrobiologischen Sicherheitskabinen. Deutsche Industrie Normen, Bonn. 4. Japanese ir Cleaning ssociation. 1983. Proposed standard for class II biological safety cabinets. Jpn. ir Clean. ssoc. 1:17-34. (In Japanese.) 4a.Macher, J. M., and M. W. First. 1984. Personal air samplers for measuring occupational exposures to biological hazards. m. Ind. Hyg. ssoc. J. 45:76-83. 5. National Sanitation Foundation. 1983. Standard no. 49 for class II (laminar flow) biohazard cabinetry. National Sanitation Foundation, nn rbor, Mich. 6. Standards ssociation of ustralia. 1982. ustralian standard 2252-1982. Biological safety cabinets, part 2-laminar flow biological safety cabinets (class II) for personnel and product protection. Standards ssociation of ustralia, North Sydney, New South Wales.