Deposition of Bacteria and Bacterial Spores by Bathroom Hot-Air Hand Dryers

Analyses of colonies on rich, double-strength Schaeffer's glucose (2×SG) medium plates left open for 2 min in men's and women's bathrooms (36 in total) with hand dryers off in three buildings in the basic science research areas of the University of Connecticut (UConn) School of Medicine found an average of 0 to 1 colony/plate (Table 1, environmental samples). Extending exposure of duplicate open plates for 18 h in two bathrooms with hand dryers off gave an average of only 6 colonies/plate (Table 1). However, when plates were exposed to air from hand dryers for 30 s in the 36 bathrooms located on different floors of the three buildings tested (Fig. 1), there was an average of 18, 24, and 60 colonies/plate in each building tested (range, 3 to 254 colonies/plate) in two different experiments separated by 3 to 4 weeks (Table 1). Bathrooms in research areas above the academic building (Fig. 1) had significantly greater average numbers of colonies/plate, i.e., 60, than bathrooms in two other basic science research areas (Fig. 1, buildings 1 and 2), which averaged 24 and 18 colonies/plate, respectively (P < 0.001) (Table 1). Notably, in two of the three buildings in which bacteria dispersed by hand dryers were measured, there was no statistically significant difference between the numbers of bacteria dispersed in men's and women's bathrooms (data not shown), and this subject was not studied further.

These bacterial counts were obtained from plates ∼12 in. from the hand dryer air outlet. Since hands below a hand dryer may be closer than 12 in., we also compared bacterial deposition after exposure of plates to hand dryer air 2 in. or 12 in. from the nozzle (see Table S1 in the supplemental material). While more bacteria were recovered from plates closer to the hand dryer nozzle, this difference did not reach statistical significance (P = 0.175).

If the bacteria found on plates exposed to hand dryer air are coming from room air passing through hand dryers, then HEPA filters in the hand dryers should reduce bacterial surface deposition. Indeed, this is reported to reduce bacteria passing through the hand dryers used in this work by ≥99.9% (30). Therefore, we compared bacterial recovery from five hand dryers before and after retrofitting the dryers with HEPA filters. While bacterial recoveries prior to HEPA filter retrofit were comparable to those seen in previous experiments, bacterial counts at 9 days after HEPA filter retrofitting were reduced ∼4-fold, and this difference was significant (P < 0.001) (Table 2; note that for the hand dryer data in Table 1, air exposure was 30 s whereas it was 1 min for the data in Table 2). Thus, the HEPA filter retrofit reduced, but did not eliminate, microbe deposition by hand dryers.

A second possible source of bacteria deposited on plates by hand dryers is from buildup of bacteria in the hand dryers themselves and then dispersal by the forced air. To test this possibility, the inner surface of hand dryer nozzles from eight different bathrooms, four with and four without HEPA filters, were swabbed to recover bacteria as described in Materials and Methods. However, an average of only ∼4 colonies was recovered per bathroom (data not shown). Thus, it seems extremely unlikely that hand dryers carry within the nozzle a significant reservoir of bacteria that are aerosolized and deposited on plates.

The data above indicated that much, and perhaps all, of the bacteria deposited on plates by hand dryer air is coming from the bathroom air by either passing through hand dryers without HEPA filters or by being pulled into the air coming out of hand dryer nozzles by convection. To test this directly, two Luria broth (LB) (see Materials and Methods) and two blood agar plates in each of five bathrooms in research buildings 1 and 2 were exposed to air blown by small sanitized fans (Fig. S1) for 20 min as described in Materials and Methods. These bathroom air exposures gave averages of 15 and 12 colonies/plate in research buildings 1 and 2, respectively (Table 3) (range, 2 to 33 colonies/plate), demonstrating that the small fan is capable of deposition of aerosolized bacteria from bathroom air. These results did not differ significantly from those for bacterial colonies deposited by hand dryers from the same bathrooms when calculations of the colonies deposited by hand dryers and small fans were corrected for the times for air exposure and rates of airflow from these two sources (Table 3).

One of the organisms recovered from multiple bathrooms was B. subtilis (see below), the species of strain PS533 used to prepare large amounts of spores in a research laboratory on the 2nd floor of research building 1 (Fig. 1). B. subtilis is nonpathogenic, and its spores are found throughout the environment, including the human gut, and are also present in many probiotic preparations ingested by both animals and humans (31). Since B. subtilis PS533 is kanamycin resistant (Km^r) because it contains the plasmid pUB110 (32), we determined numbers of Km^r and spore-forming colonies deposited on plates by hand dryers in different bathrooms (Table 1). Of note, we recovered Km^r sporeformers from all floors of the three research buildings tested, regardless of distance from the spore laboratory on the 2nd floor of building 1 (Fig. 1). While there was no statistical relationship by floor between the distance of the bathroom from the spore laboratory and the number of Km^r spore-forming bacteria recovered (Table S2 and data not shown), there was a statistically significant increased percentage of Km^r spores recovered from building 1 bathrooms overall compared with that from bathrooms above the academic building and building 2 (Table 1). We next extracted plasmid DNA from 25 Km^r spore formers, and all had a plasmid that comigrated with plasmid pUB110 (Fig. 2 and data not shown). In addition, 12 of these plasmids that were randomly chosen had a BamH1/EcoR1 restriction enzyme digestion pattern consistent with pUB110 (Fig. 2 and data not shown), strongly indicating that these isolates are strain PS533 bacteria. In total, these latter isolates comprised ∼2.5 to 5% of all bacteria deposited on plates by hand dryer air. Km^r spore-forming B. subtilis bacteria, most likely PS533, also comprised 3 to 4% of strains isolated from bathroom air blown on plates by small fans (see below).

We next wanted to determine whether B. subtilis PS533 and perhaps other sporeformers were deposited on surfaces as spores or as growing or sporulating cells. To answer this question, microbes from hand dryer air in a bathroom in building 2 were collected in liquid and in control liquid open to bathroom air as described in Materials and Methods. Aliquots of the liquid were heated to kill growing mesophilic bacteria but not spores, and aliquots of heat-treated control and hand dryer air-exposed liquid were plated for colony recovery, again as described in Materials and Methods. No colonies were recovered from the control liquid, while 16 colonies were recovered from the two plates spread with 200 μl of hand dryer air-exposed samples (40 colonies/ml). All but two of these colonies were Km^r, and all but one formed significant amounts of spores on LB plates. Analysis of the bacteria in these 16 colonies by a matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) Biotyper (see below) found that one colony could not be identified, one of the non-Km^r organisms was Bacillus pumilus, and the remainder were B. subtilis colonies. These data confirm that at least some of the B. subtilis bacteria recovered from agar plates are deposited on surfaces as spores by hand dryers. However, since the total number of bacteria in the liquid exposed to hand dryer air was not determined, it was not possible to directly determine the percentage of total bacteria dispersed by hand dryer that were present as spores.

To identify spores present in bathroom air itself, small fans were used to blow air in two bathrooms for 1 h at two LB plates in each bathroom; after incubation for 48 h at 37°C, these four plates had an average of 57 colonies. The air blown in each bathroom was also directed at one sterile nitrocellulose filter on purified agarose plates lacking germinants such that spores that adhered to the filters would not germinate. The filters were heated at 70°C for 45 min to kill cells, but not spores, of mesophilic species and then transferred to LB plates that were incubated at 37°C for 48 h. The 11 colonies that grew on the two heated filters were then picked on LB and LB plus Km (10 μg/ml) plates, as well as 2×SG sporulation medium plates. After incubation for 24 to 48 h at 37°C, 4 colonies were Km^r, and microscopic examination showed that all 11 colonies had formed spores and that the spore and cell morphologies of 10 of the 11, including the 4 Km^r colonies, were indistinguishable from those of B. subtilis. Identification of 7 of these colonies (3 Km^r colonies and 4 colonies sensitive to kanamycin) showed that all were B. subtilis bacteria (data not shown; see below). Km^r sporeformers were ∼3.5% of total colonies deposited from environmental air as, for the two plates examined, there were 4 Km^r spore formers from 114 (2 plates × 57 colonies each) total colonies.

A variety of floras were recovered from blood agar plates exposed to hand dryer air, with and without HEPA filters, while only 2 colonies were recovered from MacConkey plates that select for enteric Gram-negative organisms (Table 4) (note that only one of these colonies could be identified). A sample of 70 colonies from hand dryer air in different bathrooms with or without HEPA filters, selected based on differing colony morphologies, was subjected to identification by the MALDI-TOF Biotyper as described in Materials and Methods. Eight colonies were not identified and might be environmental organisms not present in the database used for bacterial identification. The 62 identified bacteria represented 21 species, with a wide variety of environmental bacteria, including Staphylococcus aureus (Table 4). While several other potential opportunistic human pathogens were recovered from multiple bathrooms, we did not identify all bacteria from all bathrooms, so these data almost certainly underrepresent bacterial diversity across bathrooms (see Discussion).

A sample of 34 colonies deposited on plates by small fans directly from bathroom air, including 7 colonies deposited as spores as described above, was also subjected to identification. The 7 colonies deposited as spores were all B. subtilis bacteria, and 25 of the remaining 27 colonies were identified, including potential human pathogens such as Acinetobacter baumannii (Table 4).

We surveyed 36 hot-air Xlerator hand dryers (model XL-SB; Excel Dryer, Inc., East Longmeadow, MA, USA) without HEPA filters (but see below) in 18 men's and 18 women's bathrooms in or adjacent to two basic science research areas in the UConn School of Medicine (building 2, with 14 bathrooms on floors 1 to 7 and building 1 with 14 bathrooms on floors 1 to 7) and in areas above the academic building (8 bathrooms; floors 4 to 7). These bathrooms are interconnected through hallways (Fig. 1) but are not within clinical areas. Access to these bathrooms was unrestricted during sampling of hand dryer air or bathroom air itself (see below). Unless otherwise noted, hand dryer airflow was directed onto rich 2×SG medium agar plates (50) for 30 s with covers off and held in gloved hands ∼12 in. from the air outlet of the dryers. However, in a few experiments, hand dryer air exposure was for 1 min, or the plates exposed were 5% tryptic soy broth-sheep's blood agar plates (Hardy Diagnostics, Santa Maria, CA) or were MacConkey's agar (Remel, Lenexa, Kansas) to select for enteric Gram-negative organisms. Plates were then covered and incubated for ∼40 h at 37°C, and colonies were counted. Two 2×SG plates were also incubated open for 2 min in each bathroom, routinely with the hand dryers off. Duplicate 2×SG plates were also left open for 2 min in the laboratory where PS533 spores were prepared but not adjacent to areas where spores were being isolated and purified, and again ≤1 colony/plate was recovered (data not shown).

HEPA filters were retrofitted in hand dryers in five bathrooms according to the manufacturers' specifications. These hand dryers were tested 2 days before and 9 days after this retrofit as described above, but exposure of plates was for 1 min in order to increase the numbers of colonies obtained from the hand dryers that contained HEPA filters.

To test for bacteria in bathroom air, we determined the numbers of bacteria deposited on plates by small fans (Haptime portable mini fan, model YGH-539; Haptimetech, China). The fans were sanitized by swabbing all exposed surfaces with 50% ethanol and were placed ∼3.5 in. above open 2×SG plates on bathroom counters (see Fig. S1 in the supplemental material); airflow at the edge of these fans (diameter, ∼2.5 in) was measured at 400 linear ft/min with a TSI VelociCalc thermo-anemometer (TSI, Inc., Shoreview, MN) but was slightly lower at the center. After exposure to air moved by these fans, for 20 min unless noted otherwise, plates were closed, incubated for ∼40 h at 37°C, and colonies were counted.

To assess the presence of bacteria on the inner surfaces of hand dryer nozzles, cotton swabs were moistened in sterile phosphate-buffered saline ([PBS] 10 mM KPO₄ buffer, pH 7.8, 150 mM NaCl) and used to swab the entire inner surface of the hand dryer nozzles in four bathrooms for ∼20 s. The swabs were placed in 300 μl of PBS for 15 min and vortexed for ∼30 s, and then 200 μl was spread on 2×SG medium plates, the plates were incubated for ∼36 h, and the colonies were counted.

To look for B. subtilis PS533 in bacteria dispersed by hand dryer air, colonies from overnight incubations of plates exposed to hand dryer air were picked onto LB plates (51) (per liter, 5 g of yeast extract, 10 g of tryptone, 15 g of agar, and 150 mmol of NaCl) with Km (10 mg/liter) and incubated overnight at 37°C. The morphology of cells in Km^r colonies was then determined by phase-contrast microscopy. Cells with a bacillus-like morphology were further tested for spore formation by picking cells onto 2×SG plates, incubating the plates for 48 h at 37°C to allow sporulation, and then examining colonies for spores by phase-contrast microscopy. Some sporulating Km^r colonies were also tested for plasmid pUB110 by plasmid isolation and purification from cells grown overnight in LB at 37°C using a QIAprep Spin Miniprep kit for plasmid DNA purification (Qiagen, Valencia, CA). Purified plasmid and plasmid digested with the restriction enzymes BamH1 and EcoR1, as well as purified plasmid pUB110 from growing cells of B. subtilis strain PS533, were subjected to 1% agarose gel electrophoresis with DNA markers. Gels were stained with ethidium bromide and photographed.

To determine if spore-forming bacteria deposited on plates by hand dryers were deposited as spores or as growing or stationary-phase cells, we took advantage of the fact that spores are much more resistant to wet heat than are germinated spores or growing or stationary-phase cells of the same organism (52–54). Twenty milliliters of sterile PBS was placed in each of three 1-liter sterile lyophilization flasks, and the open flasks were held 10 to 12 in. below the dryer and given 10 min of hand dryer air exposure. As a control, another three flasks with liquid were left open to room air on the counter during the course of the experiment. The total volume per flask of PBS left after hand dryer air exposure was ∼14 ml; 500 μl was removed, heated at 65°C for 30 min to kill germinated spores or growing or stationary-phase cells (53, 54), and cooled to 23°C. Two hundred microliters of each heat-treated PBS sample was then plated in duplicate on blood agar plates that were incubated for ∼20 h at 37°C, and colonies were counted.

To directly identify spores present in bathroom air itself, the small fans described above were used to blow air in two bathrooms for 1 h at 2 LB medium plates in each bathroom to determine total bacteria deposited. The bathroom air in the same two bathrooms was also blown by small fans for 1 h at sterile nitrocellulose filters on purified agarose plates lacking all germinants, such that spores that adhered to the filters would not germinate. The filters were then heated at 70°C for 45 min to kill cells of mesophilic species, but not spores, and the filters were transferred to LB medium plates that were incubated at 37°C for 48 h. Heat treatment of the damp filters was more severe than that of the liquid samples because dry-growing cells are more resistant to heat than cells in liquid (52, 53). The colonies that grew on the heated filters were then picked on LB and LB plus Km (10 μg/ml) plates, as well as 2×SG sporulation medium plates.

Colonies were prepared for definitive identification of various bacteria as follows: (i) air from hand dryers with and without HEPA filters was directed for 1 min at blood agar plates or MacConkey's agar plates; (ii) colonies from heat-resistant organisms deposited in liquid by hand dryer air were picked to blood agar plates; (iii) bathroom air moved by small fans was directed for 1 h at blood agar plates; or (iv) colonies from spores deposited on filters on purified agarose plates by bathroom air moved by small fans were picked onto blood agar plates. Plates were incubated at 37°C for 24 to 40 h, selected individual colonies of various representative morphologies were picked to fresh blood agar plates, and plates were incubated at 37°C overnight. Bacteria in the resultant colonies were identified by matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry using a MALDI Biotyper instrument (Bruker GmbH, Bremen, Germany) in the Hartford Hospital clinical microbiology laboratory (55).

To compare the bacterial colony levels between buildings, floors, distance from hand dryer, and the use of the small fan, a generalized linear model with Poisson distribution, a log link function, and a scaled deviance for overdispersion was used. The comparison of pre-HEPA to post-HEPA filter bacterial levels was performed using generalized estimating equations with Poisson distribution, log link, and an unstructured correlation matrix with robust estimator. Descriptive statistics include means and standard deviations. All analyses were conducted in SPSS, version 24, and statistical significance was set at 0.05.