Laboratory Validation and Initial Field Testing of an Unobtrusive Bioaerosol Detector for Health Care Settings

Received: October 7, 2017
Revised: September 24, 2018
Accepted: November 29, 2018

Download Citation: RIS|BibTeX|https://doi.org/10.4209/aaqr.2017.10.0371  

Cite this article:

Kesavan, J., Kilper, G., Williamson, M., Alstadt, V., Dimmock, A. and Bascom, R. (2019). Laboratory Validation and Initial Field Testing of an Unobtrusive Bioaerosol Detector for Health Care Settings. Aerosol Air Qual. Res. 19: 331-344. https://doi.org/10.4209/aaqr.2017.10.0371

HIGHLIGHTS

• The expensive UV-APS was highly accurate across a 100 fold particle concentration.
• The less expensive TACBIO was reasonably accurate in detection efficiency.
• Detection of fluorescent particles >1.5 µm by both systems agreed.
• TACBIO is able to track real-time variations in ambient particles.
• TACBIO can be used as a bioaerosol detector in indoor environments.

Keywords: Aerosol sampling; Hospital clinic; Background aerosols; UV-APS; TACBIO.

INTRODUCTION

Major human infectious diseases may be transmitted by aerosolized microorganisms. The risks of intentional, hostile bioaerosol release have led to military interest in methods for rapid detection and threat reduction. Soldiers and first responders employ bioaerosol detectors as an indication of the need to take precautions such as evacuation or respirator use until detected hazards are eliminated or have passed. While these military bioaerosol detectors are commonly used in outdoor settings, they may potentially be utilized in indoor setting such as hospitals to detect and control airborne infectious organisms.

Broadly speaking, there are two main environmental bioaerosol sampling strategies, with differing strengths and limitations. The first strategy is a two-staged bioaerosol collection and identification system whose strength is the precise and definitive identification of sampled constituents and whose limitation is a long analysis time. This system begins with sample collection, a process lasting 15 minutes to 8 hours depending on the sampling method and goals. Impingers and filters can be used to collect all particles together while an Andersen cascade impactor or similar separator is required for size based separation and determination of particle concentration and size distribution. Collected samples are next transported to a laboratory or analyzed in the field for identification using methods such as polymerase chain reaction (PCR), immunoassay, and culturing.

The strength of the second type of strategy, a bioaerosol detection system, such as the ones studied herein, is the capacity for rapid threat identification with opportunity for real-time threat mitigation. Sampling instruments identify particle characteristics while the particles are still suspended in the air, bypassing the need for sample collection (Hairston, 1997). Laser-induced fluorescent (LIF) detectors are a class of detectors that are used to detect biological particles. Biological particles have components (e.g., tryptophan, flavins, NADH, NADPH) which will provide fluorescence when excited at specific wavelengths (Manninen et al., 2009; Putkiranta et al., 2010). Fluorescent (LIF) aerosol detectors have high time resolution and provide fluorescent aerosol information in seconds, a response time which is suited for use in defense and potentially in health caresettings. These detectors can detect all particles that fluoresce in the observed wavelength but the detector is unable to differentiate between biological and nonbiological source particles. They are also unable to differentiate between species.

A recognized limitation of bioaerosol detection systems is the use of surrogate indicators for agent identification, although methods to overcome this limitation are under development (Saari et al., 2016). Instruments utilize various excitation and emission wavelengths to detect biological particles. In particular, bioaerosol detectors utilize an organism’s fluorescence properties to distinguish biological particles from the background particles. The amount of fluorescence of biological particles depends both on the fluorescent properties of the particles themselves as well as the characteristics of the instrument such as the excitation and emission wavelengths. Manninen et al. (2009) described excitation and emission maps for typical bacteria spores which can be used in engineering fluorescence-based bioaerosol detectors. Furthermore, bacterial and fungal spores may be detected and distinguished from each other through their dissimilar fluorescence spectra (Saari et al., 2013). Individual bacteria may agglomerate into clusters of infectious particles, further increasing particle size.

In addition to fluorescence, bioaerosol detectors measure particle size. Outdoor air particles show a trimodal size distribution, with combustion-sourced acidic, metal-rich particles present in the submicron range, and larger particles derived from crustal surface disruption in the micron sized range. Only particles larger than 0.5 µm and larger than 1.5 µm will be detected by the UV-APS and TACBIO, respectively. Less is known about the distribution of particles in the indoor environment, although infiltration of outdoor sourced particles into the indoor environment is well recognized. Indoor sources of particles include cigarette smoking and home heating using wood as fuel. Cigarette smoke and products of other combustion processes may also show fluorescent properties; however, these particles are small in size (< ~1.5 µm) (Anderson et al., 1989). Taken together, therefore, particle fluorescence and size information can be used to differentiate combustion products from biological particles.

The most commonly used laser-induced fluorescence (LIF) based bioaerosol detection instrument is the Ultraviolet Aerodynamic Particle Sizer (UV-APS) which is used in laboratory and field testing to quantify nonfluorescent and fluorescent particles in air (Huffman et al., 2010, 2012), and has been utilized to measure biological aerosols in various locations. UV-APS is very expensive (~$150K) and the manufacturer has stopped manufacturing this instrument. To determine a suitable low cost detector to use in public locations, biological aerosol detector candidates are compared to the UV-APS.

There are many commercially available bioaerosol detectors; however, they only provide limited information due to proprietary hardware and software and do not have peer reviewed publications. Laboratory use research instruments such as Wideband Integrated Bioaerosol System (WIBS) (Droplet Measurement Technologies, Longmont, CO) and C-FLAPS (Dycor Technologies Ltd., Edmonton, Canada) provide more information and have peer reviewed publications. Laboratory instruments are more expensive, bulky and heavy and are not suitable for field use.

WIBS is an expensive laboratory instrument that utilizes two fluorescent excitation wavelengths (280 and 370 nm) and two emission wavelengths (310–400 nm and 420–650 nm) to detect the fluorescent properties of particles in addition to determining the light scattering properties of particles (Healy et al., 2012; Robinson et al., 2013; Toprak and Schnaiter, 2013; Healy et al., 2014; Robinson et al., 2017). C-FLAPS is another laboratory instrument that operates by concentrating the particle stream and eliminating particles outside of the 1–10 µm range. Particles are concentrated by a concentrator that eliminates particles larger than 10 µm and smaller than 1 µm. C-FLAPS has a laser diode that provides an excitation wavelength of 405 nm and the emission is measured in two distinct wavelength bands. An advanced alarm algorithm is used to provide exceptional discrimination and interference rejection for biological threat detection. WIBS and FLAPS III instruments are suitable for research but may be too expensive for field use requiring multiple instruments for broader coverage.

A market survey conducted by Emanuel and Caples (2011) provides limited descriptions of biodetectors and their characteristics. Names of a few detectors are listed here and more information about them is provided in the report: AbleSentry (Lockheed Martin, San Diego, CA), ABSS(ATHINA Biological Security System, Chemring Detection Systems, Charlotte, NC), AiroCollect-Detect-288 (Partner Airogistic, LLC, Austin, TX), ARETE TRAP (Threat Reduction Advancement Processor; Arete, Northridge, CA), MAB Biological Alarm Monitor (PROENGIN, Inc., Plantation, FL), kBioTM-FD (Partner Airogistic, LLC, Tucson, AZ), etc. Most of these bioaerosol detectors evaluate the fluorescence properties of particles to determine whether the particle is a bioaerosol.

A low cost, unobtrusive bioaerosol detector, the TACBIO® Gen II (TACBIO), was developed by U.S. Army Edgewood Chemical Biological Center (ECBC), and is a commercially available bioaerosol detector. The LED technology limits the TACBIO unit to detection of particles > 1.5 µm while the UV-APS unit detects particles > 0.5 µm. IBAC™ 2 (FLIR Systems, Inc., www.Flir.com) is a lower cost field instrument that continuously monitors the air for biological aerosol threats. The FLIR website provides some general information about the function of the detector but does not list excitation or emission wavelengths. Instantaneous microbial detection (IMD-A 300 and 350 systems) was developed by BioVigilant (Tucson, AZ) for rapid microbial monitoring methods (RMM). This system detects particle count, size (0.5 to > 10 µm) and biological status (fluorescence) using a 405 nm laser instantaneously and in real time.

The need for effective control strategies for airborne infectious agents, combined with the opportunity presented by this new technology motivated the present research. The first goal of this study was to conduct initial laboratory validation of two bioaerosol detectors (UV-APS and TACBIO) using a well characterized bacterial particle generation system. The second goal was to conduct field validation of the detectors through simultaneous UV-APS and TACBIO sampling of background aerosol concentrations in an outpatient hospital setting. The third goal was to assess the impact of the TACBIO’s size detection limit of > 1.5 µm on the ability to detect variation in biological particles in the health care setting.

METHODS

The laboratory validation determined the comparative ability of the UV-APS and TACBIO unit to detect airborne fluorescent particles utilizing generated biological particles across a range of concentrations with bioaerosol detector counting efficiency as the primary endpoint. The field validation consisted of simultaneous sampling of the two units in a hospital clinic waiting room with concomitant observation of occupant activity. 

UV-APS

The Ultraviolet Aerodynamic Particle Sizer (UV-APS) (Model 3314, TSI Incorporated, Shoreview, MN) is a particle spectrometer that measures each particle’s aerodynamic diameter, light-scattering intensity and fluorescence. A detailed description of the instrument utilizing a continuous laser is provided by Hairston et al. (1997); however, the commercial instrument contains a pulsed laser that provides an excitation wavelength of 355 nm and analyzes emission wavelengths from 420 to 575 nm.

Description of the UV-APS is provided in the manual and in journal articles (Hairston et al., 1997); however, a brief description is provided below. Similar to the operation of the Aerodynamic Particle Sizer (APS), particles are accelerated through the UV-APS and travel past two red laser beams. The scattered light produces one double-crested beam profile which allows a particle to create a single continuous signal that has two crests. The time measured for the particle to travel between the two laser beams is compared to a calibration curve to determine the aerodynamic particle size. Particles with only one crest or more than two crests are not used in size distribution calculations but logged for concentration. These data are utilized to obtain a particle’s aerodynamic size and size based on scattered light. The fluorescence of particles is measured by a second pulsed UV light. The UV-APS is heavy (35 kg = 76 lbs), bulky (2.63 ft³ = 0.072 m³) and has an air flow rate of 5 L min^–1. The particle size range (0.5–20 µm) is subdivided into 52 channels and the fluorescence signal is divided into 64 channels. The sample duration and the number of samples are adjustable and can be selected based on the sampling criteria. The UV-APS manual indicates that there is < 10% coincidence for 0.5 µm particles when the number concentration is 1500 particles cm^–3. Similarly, there is < 10% coincidence for 10 µm particles when the concentration is 600 particles cm^–3. Characteristics of the UV-APS are provided in Table 1.

 
TACBIO Gen II Detector

The TACBIO Gen II is a localized point detector that works on the basis of induced fluorescence. Fig. 1 illustrates the outside and inside view of the TACBIO. It illuminates aerosols with a deep-UV light source, and the scattered and fluorescent light is measured by photomultiplier tubes. The excitation wavelength used by the system is 270 nm and the emission is measured in the wavelengths from 315 to 700 nm. Unlike other fluorescence detectors, it uses a high power light emitting diode (LED) instead of a laser. The UV-LED light is focused down to a beam spot size of3 mm × 3 mm square. The detection zone is where that precisely-aligned beam spot intersects the 1/8′′ (3 mm) diameter airflow. Aerosol particles are drawn into the system at a rate of approximately 1.0–1.3 L min^–1 with a small internal pump. The TACBIO is small (0.28 ft³ = 7.93 × 10^–3 m³), light weight (3.2 lbs), and can be battery powered. In quantities of 10,000 or more the cost per unit is currently estimated at $2,000, which is significantly cheaper than all other options. These properties make it useful for tactical purposes and for forming a relatively dense network of detection points. To that end, the TACBIO has a Global Positioning System (GPS) antenna and can communicate using its Ethernet and USB ports. It is compliant with communication software architectures Common CBRNE Sensor Interface (CCSI) and Integrated Sensor Architecture (ISA), and it can be controlled remotely via its webserver user interface. Characteristics of the TACBIO are provided in Table 1.

Fig. 1. Unobtrusive Environmental Aerosol Sampler (TACBIO aerosol detector). Outside view and the internal workings are shown in this figure. Specifications are shown in Table 1.

The TACBIO II system operates in 2 modes, either internet-connected with time set remotely to GMT or without internet connection with a start time, defined manually as the time of TACBIO activation. Sampling interval is 5 seconds with the number of fluorescent particles reported as output for each 5 second interval. The TACBIO has a wide particle count range (0–10,000 particles per liter) which allows it to be used in environments ranging from clean rooms to high dust areas. For the TACBIO, the coincidence occurs when the concentration is > 10 particles cm^–3.

 
Laboratory Validation

Ink Jet Aerosol Generator

Monodispersed bacteria aerosols were generated using an Ink Jet Aerosol Generator (IJAG) to determine the detection efficiency of the UV-APS and the TACBIO. The IJAG was developed to produce lower concentrations of particles to represent clean rooms or environments with low bioaerosol concentrations. The concept of using an ink jet cartridge for near-monodisperse aerosol generation was first developed at the U.S. Army Edgewood Chemical Biological Center (ECBC) to enable testing of bioaerosol detection instruments in aerosol static chambers, flow-through chambers, and clean rooms. The ink jet approach was of particular interest because the nozzles are of the order of 50 µm in diameter, which accommodates transmission of liquid suspensions of bacterial spores, where the spores are typically on the order of 1 µm in diameter.

A detailed description of the IJAG is provided in Kesavan et al. (2014) and a brief description is provided below. The IJAG dispenser contains a liquid solution of bacteria. Computer-generated pulses create primary and satellite particles. HEPA-filtered airstreams aerodynamically separate and remove undesired satellite particles. The primary particles are counted by a light scattering system located right below the aerosol generation location. Variously sized monodisperse particles are generated by adjusting the concentration of spores in water. Generated primary particles travel through an oven which evaporates the liquid resulting in the final dried primary particle.

The IJAG has been utilized in many studies, especially to calibrate the Aerodynamic Particle Sizer (APS) using 0.95 to 13.3 µm solid and liquid particles (Kesavan et al., 2014). Kesavan et al. (2013) employed the IJAG system to generate near-monodisperse clusters (geometric standard deviation ≈ 1.15) of Bacillus atrophaeus spores in the size range of 2.8 to 4.4 µm for Ultraviolet C inactivation studies.

The aerosol output from the IJAG can be calculated directly from the particle count rate provided by the light scattering system and the air flow rate exiting the IJAG. In addition, the output of the IJAG can be delivered directly into a sampler or detector inlet to measure the particle size and number concentration as shown in Kesavan et al. (2014).

Test Organism

Clusters of biological spores, generated at various rates, were used in this test to determine the detection efficiency of the UV-APS and TACBIO. Three organisms were used in this test: Bacillus atrophaeus var. globigii (Bg), Bacillus thuringiensis (Bt), and Bacillus anthracis Sterne (BaS). All three species are used as simulants for Bacillus anthracis Ames and are similar in size (~1 µm).

Laboratory Testing

The detection efficiency of the UV-APS and the TACBIO were determined using the IJAG in a clean air chamber. HEPA filters were installed on top of the clean air chamber from which filtered air was blown downwards. Three species of organisms (Bt, BaS, and Bg) with various aerosol generation rates (1, 2, 3, 5, 10, 50, and 100 particles sec^–1) were utilized in this test. Generated monodisperse Bt, BaS, and Bg cluster sizes were measured with a UV-APS that showed mean number diameters of 2.1 ± 1.1, 2.3 ± 1.1, and 2.6 ± 1.2 µm, respectively.

IJAG output was delivered at the inlet of the two detectors to ensure that all particles entered the detector. Connecting pieces were machined to smoothly transition the aerosol from the IJAG to the detectors without any loss. More information about the transition piece is provided in a publication by Kesavan et al. (2014).

For this laboratory test, the UV-APS was turned on to begin to sample the air and then the IJAG particle generator was turned on. The UV-APS sampled for a total of 40 seconds, while the IJAG generated particles for 30 seconds; a longer sampling time allowed all the particles generated by the IJAG to be sampled by the UV-APS. The UV-APS reported the number of particles counted. Air sampling of the clean chamber was also periodically conducted to confirm that the chamber air was free of particles. The counting efficiency of the UV-APS was determined using the equation provided below. Eight repeats were conducted for each condition. Additional information on testing is provided by Kesavan et al. (2014).

The sample time of the TACBIO was 5 seconds and was not adjustable; therefore, the TACBIO test was conducted with a continuous sampling protocol. The IJAG generated particles for 2 minutes followed by 1 minute of no particle generation. The particle generation rate was then increased to the next level during the minute of no particle generation. The detection efficiency of the TACBIO was determined by comparing the counts obtained by the TACBIO and the generation rate of IJAG for each 5 second interval, during the 2 minutes of aerosol generation. Data were also analyzed for the minute of no particle generation to confirm that the chamber was free of particles.

Field Testing: Environmental Aerosol Detection and Analysis

Field testing occurred in the waiting/reception room of a subspecialty adult outpatient clinic shared by pulmonary and gastroenterology specialty physicians (Fig. 2). The reception room included 20 chairs and 3 check-in desks, with access to 10 exam rooms and clinical staff workspaces. The detectors were placed in the corner of the waiting area on an end table off the ground, a placement that had the air intake at a height similar to the breathing zone of a seated patient. Both the UV-APS and TACBIO sampled the air from 0927 to 1600 with adult patients and their families entering, waiting and exiting the reception area. An observer noted time-activity patterns during the sampling period.

Fig. 2. Site of Field Validation for Unobtrusive Environmental Aerosol Detector. The UV-APS and TACBIO were positioned adjacent to each other in an Outpatient Sub-specialty Clinic. Data were collected during clinic operations.

The air intake for the building HVAC system is drawn from the roof, behind the elevator and center of the building. Air exchange operates with an economizer with a minimum of 25% total air returned. Air filtration is MERV 8 with 30–40% pre-filters, capturing particles 3–10 µm in diameter. During the test period, the HVAC system operated with constant ventilation to the building.

Analysis

Continuous UV-APS data were first analyzed for the entire particle readout (particles > 0.5 µm) and expressed by count and mass, and also by total particles, fluorescent and nonfluorescent particles. UV-APS data were also analyzed for the particle fraction sized > 1.5 µm. Correlation analysis for the two monitors compared the UV-APS > 1.5 µm size fraction with the total TACBIO particle counts (which are only able to include particles sized > 1.5 µm). For the correlation analysis, the simultaneous samples were collected at a 5 second sampling frequency/window resulting in over 4000 readings for this sampling period.

RESULTS

Laboratory Validation

Fig. 3 shows the detection efficiency results of UV-APS and TACBIO for challenges with three organisms. The detection efficiency of UV-APS was close to 100% and ranged from 99 to 110% for all particle generation rates and species. On the other hand, detection efficiency was slightly lower for the TACBIO compared to UV-APS and ranged from 88.6 to 100.8% for Bt and BaS and was even lower for Bg (62.7–81.7%). TACBIO detected Bt and BaS particles at a significantly higher rate compared to Bg particles due to the higher fluorescence properties of Bt and BaS. Similar results were observed by Pan et al. (2014) in their study that looked at the fluorescence cross section and peak fluorescence cross section using an excitation wavelength of 266 nm, which is similar to the wavelength of TACBIO. The counting efficiency results of the UV-APS and TACBIO tested with a Student’s t-test were significantly different with a p-value of < 0.001 for all organisms tested.

Fig. 3. Laboratory Validation: Detection Efficiency of UV-APS (left) and TACBIO (right) for three different spore forming bacteria (Bt, BaS, and Bg) at aerosol generation rates of 1, 2, 3, 5, 10, 50, and 100 particles per second. Bt is shown by the slanted bars, BaS the open bars and Bg the speckled bars.

UV-APS Measurements in the Hospital Clinic

Fig. 4 shows results of more than 6 hours of continuous in-clinic sampling using the UV-APS detector, comparing size characteristics and fluorescence. The top panel shows the effect of particle size on the quantity and variability of detected particles. The left panels show particles > 0.5 µm, providing a comprehensive sampling of ambient air. The right panels show particles > 1.5 µm, the size fraction chosen because it is the subset of the airborne particles also detected by the TACBIO. The top panel UV-APS data show the 47 fold greater total particle counts for the > 0.5 µm fraction (range: 2–8 cm^–3, average: 4.65 cm^–3; top left), when compared with 1.5 µm fraction (range: 0.02–0.8 cm^–3, average: 0.13 cm^–3; top right) measured by the same UV-APS. The > 0.5 µm fraction showed a mid-morning decline in total particle count, then a gradual afternoon rise. The > 1.5 µm fraction showed a very different temporal pattern with marked short term variation, and no cross day trends. The lowest values were measured at the noon hour and the end of the clinic session, when fewer patients were in the waiting room. The middle panel (Figs. 4(c)–4(d)) shows the impact of characterizing particle fluorescence measured by the UV-APS on detection patterns. Only a very small proportion (2%) of the > 0.5 µm particles fluoresced (middle left), while a much larger proportion (58%) of the > 1.5 µm particles fluoresced (middle right). The bottom panel correlates UV-APS measured fluorescent and nonfluorescent particles. The bottom left panel shows the noncorrelation of > 0.5 µm fluorescent and non-fluorescent particle counts, indicating they arise from different sources. The bottom right panel shows the close correlation of > 1.5 µm fluorescent and nonfluorescent particle counts (correlation coefficient = 0.782, p-value < 0.001) indicating that they have a common origin.

Fig. 4. Field Sampling: Results of continuous in-clinic sampling from 927 to 1600 using the UV-APS and displayed by size characteristics and fluorescence. Top panel: Total particles greater than 0.5 µm, measured by UV-APS (left panel) and greater than 1.5 µm, measured by UV-APS (right panel). Middle panel: fluorescent and non-fluorescent particles greater than 0.5 µm, measured by UV-APS (left panel) and greater than 1.5 µm, measured by UV-APS (right panel). Bottom panel: Correlation of fluorescent and non-fluorescent particles greater than 0.5 µm, measured by UV-APS (left) and greater than 1.5 µm, measured by UV-APS (right).

Table 2 shows particle concentrations in an occupied outpatient hospital clinic on a summer day as reported by the UV-APS bioaerosol detector. The calculated mass concentration is proportional to the diameter cubed. The comparison of the two size fractions shows the contribution of the large particles to the mass concentration. The maximum mass concentration for all particles > 0.5 µm occurred at 15:41 with 67 µg cm^–3 for nonfluorescent and 52 µg cm^–3 for fluorescent particles (Fig. 5). The maximum number concentration for all particles > 0.5 µm occurred at 15:55 with 8.2 particles cm^–3 for nonfluorescent and 0.6 particles cm^–3 for fluorescent particles (Fig. 5).

 

Fig. 5. Illustrates two examples (at 13:49:14 and 12:14:44) of all and fluorescent particle distribution measured by UV-APS. The top 3D graphs show an example of relatively low fluorescent distribution for number (left) and mass (right). Bottom 3D graphs show an example of high fluorescent measurement for number (left) and mass (right). The higher fluorescent signature was recorded when a mother and her infant were seated near the instruments. Note that the number and mass concentration scales are not equal for the graphs.

Side by Side Comparison of UV-APS and TACBIO II

Fig. 6 shows over 6 hours of continuous in-clinic normalized particle counts for particles > 1.5 µm simultaneously measured by the UV-APS (blue) and TACBIO (orange) detectors. The significant correlation of UV-APS and TACBIO measurements is shown in Fig. 7 (Pearson’s r = 0.458; linear regression: R² = 0.0.209, p < 0.00001; time matched: n = 1222).

Fig. 6. Continuous sampling in the waiting room of an active outpatient health clinic from 0927 to 1600. Shown are counts of UV-APS Measured Fluorescent Particles larger than 1.5 µm in size and Fluorescent Particles Measured by the TACBIO over the Measurement Time.

Fig. 7. Correlation of larger than 1.5 µm in size fluorescent particle concentration measured by TACBIO and UV-APS concurrently in an outpatient health care setting (Pearson’s Correlation r = 0.458; linear regression R² = 0.209, P < 0.00001). Only time matched points were selected for this correlation (n = 1222).

The aerosol concentration in the clinic varied during the measurement time as shown in Fig. 6. The lower counts corresponded to lower occupancy in the waiting room as the noon hour and end of the day approached. The number concentration increased when there was increased activity next to the samplers. For example, increased particle counts were observed when two small children were playing next to the samplers from 12:02 to 13:16. More detailed correlation of activity with the number concentration will be provided in a separate paper.

DISCUSSION

This study used a two-pronged laboratory and field validation approach to assess the unobtrusive TACBIO environmental aerosol detector for possible use in a health care setting.

The first goal of this study was to conduct initial laboratory validation of two bioaerosol detectors (UV-APS and TACBIO). Study results showed that the expensive UV-APS was highly accurate across a 100 fold generation rate for all 3 test organisms, and has been considered as the gold standard. The less expensive TACBIO was reasonably accurate, but showed a reduction in detection efficiency for 1/3 organisms tested. Expected accuracy was defined using the laboratory validation test results. The laboratory validation demonstrates a capture rate as low as 63% and as high as 100.8% for the TACBIO. The anticipated use of this detector is to identify perturbations, rather than to measure precise quantities. With this intended use, acceptable accuracy is a detection level that allows detection of perturbations that will occur in field conditions. Our field testing showed a range of particle levels that exceeded the level of particle concentration when there was no activity. Additional field testing is needed to characterize the background aerosols in various health care settings using the UV-APS and TACBIO.

The second study goal was to conduct field validation of the TACBIO with the UV-APS, since it is a more accurate reference standard. This was accomplished through simultaneous measurement of indoor aerosol concentrations in an outpatient hospital clinic waiting/reception area.

Study results showed a wide variation in ambient airborne particle counts over time, with close correlation between the two detectors. These findings provide reassurance that the inexpensive and unobtrusive TACBIO is able to track real-time variations in ambient particles despite its reduced retrieval efficiency in the laboratory validation setting. Low cost, small size, and low weight of the TACBIO make it suitable for use as a field instrument.

The third goal was to assess the impact of the TACBIO’s size detection limit of > 1.5 µm on sampling results in this health care setting, where human sourced infectious particles are the primary concern. We found concordance in the numbers of fluorescent particles > 1.5 µm detected with simultaneous UV-APS and TACBIO sampling. While the TACBIO will not detect smaller particles, these data indicate a positive signal can be obtained despite its inability to detect smaller particles.

Our air sampling did detect a large number of nonfluorescent particles in the 0.5–1.5 µm range. An important finding was that these smaller particles were nonfluorescent, and therefore are most probably nonbiological in nature.

The UV-APS detector showed a large number of particles in the 0.5 to 1.5 µm range as it was present in the clinic area throughout the day. These small particles were uniformly nonfluorescent and thus are not human sourced particles but could be produced by vehicle exhaust or cooking in the cafeteria. The failure of the TACBIO to detect these small nonfluorescent particles is an acceptable instrument characteristic. The observed size limit of > 1.5 µm markedly enriches the air sample for fluorescent material. Taken together, these results indicate the TACBIO offers potential as bioaerosol detector in indoor environments.

Strengths of the study include the use of a novel, highly standardized bioaerosol generation system for the laboratory validation with inclusion of 3 separate bacterial strains across a 10² concentration range. The TACBIO’s 1.5 µm cut-off size is a consequence of its LED technology, so understanding its impact on the interpretation of sampling results is important. Another strength of the technology is the real-time sampling of both size and fluorescence and the use in this study of the UV-APS detector to evaluate the characteristics of particles in 2 size fractions. A final strength is the performance of the field validation with simultaneous sampling in an indoor health care setting. We are unaware of previous studies reporting particle concentration measurements from a hospital environment using these detectors.

Limitations of the study include the inability of the UV-APS to sample particles smaller than 0.5 µm due to detector performance characteristics. Detector sampling results were not coupled with a sampling and identification system, limiting the ability to infer the composition of the detected particles. Future studies should collect and identify the particles in air so that correlations could be conducted with direct reading instrument results. Nonbiological materials such as nutrient broth and peptone water can also provide fluorescent signals which can affect the measurements of needed organisms. Even though measuring fluorescent properties of particles is an efficient method to identify biological particles, only a fraction of single biological particles emit enough fluorescence to be detected by instruments. For example, the larger sized particles could be clusters of bacteria, or bacteria attached to other substances such as mucus. These limitations indicate a need for caution in interpreting detector results, and for additional field testing and validation studies for this promising technology.

This study determined the sampling efficiency of the UV-APS and TACBIO using the IJAG aerosol generation rate of 1 to 100 particles per second, which only covers 2 orders of magnitude of aerosol concentration. However, indoor and outdoor aerosol measurements indicate a wider range of bioaerosol concentrations. The aerosol concentrations used in this test were limited by the aerosol generator. Future tests could use other aerosol generators to generate higher concentrations of aerosols.

The laboratory experiments used 3 organism types. The samples were washed and larger particles and clusters were removed to prevent clogging of the aerosol generator. The final solution that was aerosolized was plated to confirm that the sample was free of contamination. During the laboratory validation studies, the detection efficiency of both the UV-APS and the TACBIO were determined by dividing the particles counted by the instruments by the particle generation rate of the IJAG. Particles generated by the IJAG may be slightly different from the set generation rate. Generated particles ranged ± 7% of the set aerosol generation value. Future studies should use the actual number of particles generated instead of the generation rate.

Previous investigators have evaluated the BioScout, another low cost bioaerosol detector (Saari et al., 2014, 2015). The BioScout employs a 405 nm continuous wave laser diode to excite autofluorescence from individual particles while the TACBIO uses a 270 nm laser diode. Thus BioScout is similar to TACBIO as both instruments collect scattered light and fluorescence to analyze each particle; however, the two instruments use very different excitation and emission wavelengths. Saari et al. (2015) have compared the particle detection efficiency of the low cost BioScout to the UV-APS. Their measurements indicate higher detection efficiency of BioScout compared to UV-APS and higher correlation efficiency was seen for fluorescent bioaerosol particles (FBAP) for both instruments. We are unaware of studies directly comparing the BioScout with the TACBIO, but both seem to share the ability to detect short term variation in fluorescent bioaerosol particles under field conditions. The demonstrated ability of the BioScout to track and recognize source intrusion (e.g., an air pollution episode) (Saari et al., 2014) is likely to be a capacity of the TACBIO as well, one that will need to be demonstrated in field conditions.

Extensive sampling of outdoor particles (EPA Integrated Science Assessment) shows a trimodal distribution of outdoor ambient particles. The finer mode particles are derived from combustion sources, while the coarse mode particles are derived from disruption of crustal materials. There may also be seasonal and diurnal variation in outdoor sourced fluorescent bioaerosol particles (FBAP) which were measured by bioaerosol detectors. For example, Saari et al. (2015) conducted measurements in an outdoor environment and showed two modes during summer, fine-mode (0.5–1.5 µm) and coarse-mode (1.5–5 µm); however, the fine mode dominated during winter months as a result of the ground being covered with snow, preventing the resuspension of crustal-derived particles (Saari et al., 2015). Huffman et al. (2012) used a UV-APS to measure fluorescent and nonfluorescent particles in the rainforest of central Amazonia during the wet season. FBAP concentration had a consistent diurnal cycle where the concentration was lowest during the day and rising after sunset and peaking before sunrise. Measurements observed in outdoor environments by Saari et al. (2015) observed increased biological particles at nighttime due to natural release of fungal spores at night. It is possible that detector algorithms may need to account for site-specific background patterns.

While the primary purpose of our field validation was to correlate the UV-APS and TACBIO detectors in a health care setting, this limited sampling period demonstrated intriguing variations in particle counts with differing temporal patterns for the total (dominated by smaller particles) and larger particle size fractions. The aerosol concentration of particles > 0.5 µm, measured with the UV-APS detector, decreased from 0927 to approximately 1000, then gradually rose until 1600, which was the end of sample time. Change in the number of small particles over time may be due to factors such as cooking in the cafeteria, exhaust from the traffic around the building parking lots, or differential infiltration of outdoor-sourced particles related to ventilation or changes in wind direction. These changes in small particles amounts would not be detected by the TACBIO system. Additional investigation would be required to understand whether this intraday variation is a consistent pattern, and if so, the source of the variation. Small particle concentration did not appear to correlate with the amount of human activity in the waiting room. On the other hand, the aerosol concentration of particles > 1.5 µm in size fluctuated from minute to minute, but did not show the more gradual, intraday changes in concentration. There were dips at times when waiting room occupancy decreased. Further studies are needed to associate occupancy and variability of large particle concentrations. Our data show that in the health clinic, fluorescent particles were unequally distributed across the size spectrum: Only a small fraction of the smaller sized particles (0.5–1.5 µm) demonstrated fluorescence and over half the particles > 1.5 µm exhibited fluorescence. Anecdotally, higher fluorescent particle counts were observed for larger particles (> 1.5 µm) when significant amount of activity was going on such as children playing next to the detectors and when people were coughing next to it. The fluorescence characteristics of a particle may indicate that the particle is biological in origin which includes infectious organisms in air. An additional study is needed to evaluate the hypothesis that in this indoor space, smaller particles (< 1.5 µm) are produced as a result of combustion processes such as cooking while larger particles (> 1.5 µm) are produced as result of human activity, and include infectious organisms.

The aerosol concentration in the hospital clinic environment was 4.65 particles cm^–3 for particles > 0.5 µm (2% of which fluoresced) and 0.13 particles cm^–3 for particles > 1.5 µm (57.7% of which fluoresced). The UV-APS has been used by other investigators to measure total and fluorescent particle size distribution and concentration in various ambient environments. Table 3 shows examples of particle concentrations detected by the UV-APS in other settings, in order to provide a perspective on this study’s UV-APS sampling results. In our study, fluorescent particles were defined as particles in fluorescent channels 3–64 as it was defined by Huffman et al. (2010). However, minor differences in cut-off sizes used with the UV-APS by other investigators limits direct comparison of sampling results. Coarse particles were defined as particles larger than 1 µm by Huffman et al. (2010); however, our study evaluated particles larger than 0.5 and 1.5 µm as the TACBIO measures particles larger than 1.5 µm very efficiently; therefore, the 1.5 µm was the size selected for analysis in our study.

Our indoor sampling was limited to a single summer day, in an occupied space in a health care clinic. We evaluated coarse particles > 1.5 µm in size over a 6 hour period in the summertime and showed that fluorescent particles accounted for 57% of the total particle counts and 55% of the total particle mass. Outdoor air contains significant amount of pollen and fungal spores; on the other hand, a clinic will have lower number of total and biological particles due to filtered air and a significant amount of cleaning.

Additional validation of the TACBIO will be needed to understand its use in the health care setting. The amount of fluorescence of biological particles depends on the fluorescent properties of the particles as well as the characteristics of the instrument such as the excitation and emission wavelengths. Instruments utilize various excitation and emission wavelengths to detect biological particles. Mold and high bioaerosol concentrations in buildings can

be detected to prevent adverse health effects. Agranovski et al. (2003a, b) reported that fluorescent particle fraction varied between 2 and 52% for bacteria and 48 and 99% for fungal spores. Detecting fluorescence is an efficient method to detect molecules such as NADH, NADPH, tryptophan, and flavins that are present in microbial cells. These sources of variation will need to be understood when developing a sampling and control strategy.

Aerosol transmissible diseases/pathogens have been identified as needing attention in the health care setting, both for patients and those with health related occupations. Major human infectious diseases may be transmitted by aerosolized microorganisms. These include bacterial diseases such as tuberculosis, legionellosis, meningitis caused by Neisseria meningitidis, and inhalation anthrax; viral diseases such as colds, flu, chickenpox, smallpox, measles, and hantavirus pulmonary syndrome; and fungal diseases such as aspergillosis, psittacosis, and histoplasmosis (Hinds, 1999; Katz and Salem, 2016). Contagious diseases have the potential to become outbreaks when microorganisms are released from infected people in public settings, or through intentional aerosol dispersion as a hostile act. In the health care setting, exposure of humans to aerosolized microorganisms can occur with transmission from infected people (Li et al., 2004) or from disruption of reservoirs such as during renovation of occupied spaces (Kanamori et al., 2015).

Biological particles in air vary from submicron to many millimeters in size. Cough droplets containing virus and bacteria may be in the millimeter range while individual virion particles and clusters of virus particles may be in submicron range. Review of various bioaerosol particle size distributions in health care settings indicates that particle sizes range in size from 0.3 to 50 µm and bioaerosol particles arise from activities including surgical dental procedures, mechanical ventilators, bed making and resuspension (Thomas, 2013). The smallest-particle information reported in these studies may be limited by the lower size detection limit of the particle measurement instrument used in these studies. In health care settings, infection control programs focus on control of both aerosol and contact modes of transmission. The motivation for detector development and characterization is to provide a tool for use as part of a control program.

Our interdisciplinary investigative team recognizes the long-term need to develop strategies to address infectious aerosol exposures in health care settings. We have assembled a team with unique and complementary expertise, and taken the first steps to address this complex issue. The presentation of field data is intended to provide the reader with a perspective on the range of particle concentrations detected in a real-world health care setting. We think this is a very useful (and difficult to obtain) dataset.

CONCLUSION

Overall, these data indicate that a low cost TACBIO bioaerosol detector can be utilized in a health care setting to detect fluorescent particles larger than 1.5 µm in diameter. More extensive sampling as well as the identification of associated patterns will be needed to understand variations in the particle concentration with respect to clinical occupancy and the time of day. Future validation studies should also include pairing a bioaerosol detector with a bioaerosol sampling/identification system to further characterize the fluorescent particles in relation to clinical activity.

 
ACKNOWLEDGEMENT

This research was supported in part by an internship appointment (VJA) at ECBC administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and ECBC. Manuscript writing was performed while VJA held a Postdoctoral Associateship Award at ECBC which was administered by the National Research Council.