Prof. Charles P. Gerba

Full text page (long) scroll down from PDF files so some typos could be present)

Appun MICROBIOLOGY, Aug. 1975, p. 229-237 Vol. 30, No. 2 Copyright 0 1975 American Society for Microbiology Printed in U.S.A.

Microbiological Hazards of Household Toilets: Droplet Production and the Fate of Residual Organisms

CHARLES P. GERBA, CRAIG WALLIS, AND JOSEPH L. MELNICK* Department of Virology and Epidemiology, Baylor College of Medicine, Houston, Texas 77025

Received for publication 12 March 1975

Large numbers of bacteria and viruses when seeded into household toilets were shown to remain in the bowl after flushing, and even continual flushing could not remove a persistent fraction. This was found to be due to the adsorption of the organisms to the porcelain surfaces of the bowl, with gradual elution occurring after each flush. Droplets produced by flushing toilets were found to harbor both bacteria and viruses which had been seeded. The detection of bacteria and viruses falling out onto surfaces in bathrooms after flushing indicated that they remain airborne long enough to settle on surfaces throughout the bathroom. Thus, there is a possibility that a person may acquire an infection from an aerosol produced by a toilet.

The transmission of disease by aerosols from MATERIALS AND METHODS toilets has received only limited study. It has been suggested that, aside from coughing and Viruses and virus assays. E. coli bacteriophage

sneezing, this must be the most common pro-MS-2 and a plaque-purified line of type 1 poliovirus (strain LSc) were used in this study. MS-2, like

cess involved in the generation of infectious poliovirus, is a small (25-nm diameter) icosahedral aerosols (6). Darlow and Bale (6) demonstrated ribonucleic acid virus. All bacteriophage assays were the production of bacterial aerosols, with the done by a modification of the agar overlay method as aid of both a liquid impinger and a slit sampler, described by Adams (1). Overlay agar and broth used from flushed bowls containing Serratia for bacteriophage samples were prepared according to marcescens. These aerosols were found to per-Davis and Sinsheimer (8). Stock poliovirus was grown sist for at least 12 min after the flush. The in baboon kidney cells, concentrated 10-fold, and generation of aerosols by toilets seeded with partially purified by membrane chromatography (22). coliform bacteria has been demonstrated by Poliovirus samples were diluted in tris(hydroxy-

Bound and Atkinson (3) and more recently by methyl)aminomethane-buffered saline containing penicillin (100 U/ml) and streptomycin (100 gg/ml).

Newsom (14). The size of particles produced by Poliovirus assays were made with BSC-1 cells by the the flushing toilet was found to be in the range plaque-forming unit method as used in this laboratory

that was capable of reaching the lower respira- (20).

tory tract (6). In addition, pathogenic fecal Bacteria and bacterial assays. A strain of E. coli contaminants, such as Escherichia and Sal-isolated from domestic sewage was used (identifica- monella, have been isolated from the respira-tion based on ImVic test). All coliform assays were tory tract of infected humans (6). performed on Levine eosin methylene blue (EMB) The fallout of droplets containing pathogens agar. Total aerobic bacterial counts were performed

on bathroom surfaces is also of concern, since on Standard Methods agar (BBL, Cockeysville, Md.).

hand contact with contaminated surfaces can Cultures of E. coli used in seeding experiments were resultinself-inoculationbytouchingofthenose grownovernightinTrypticasesoybroth(TSB)(BBL,

Cockeysville, Md.). All bacterial samples were diluted or mouth (11). Hutchinson (12) traced the in tris(hydroxymethyl)aminomethane-buffered sa-

spread of Shigella sonnei in a nursery school to line.

contaminated toilet surfaces. Contact with con-Toilets. Standard household tank or valve toilets taminated surfaces has also been shown to be were used. The tank toilets had a reservoir containing

important in the spread of animal viruses (11). approximately 20 liters,of which 13.7 liters was released during a flush, unless otherwise noted. The

To date no information exists on the genera-bowl of the tank-type toilet contained a volume of 3.5 tion of viral aerosols by household toilets. This liters unless indicated otherwise. The bowl volume of

study was carried out to gather more informa- the valve toilet was approximately 4.2 liters, with a

tion on the fate of both bacteria and viruses in standard but undetermined amount of water re- flushed toilets. leased during a flush. In the valve toilet used, the

229

 230 GERBA, WALLIS, AND MELNICK

amount of water released was dependent on the water line pressure in the building, which was usually constant. Before being seeded with bacteria or viruses the toilets were cleaned with commercially available chlorine-containing cleanser and flushed repeatedly to eliminate any bacteria or viruses naturally present. A solution of 5 g of sodium thiosulfate per liter was then added to inactivate any chlorine present in the water, at a ratio of 1 ml of solution to 1 liter of tapwater. A tank-type toilet was used in all experi- ments unless indicated otherwise.

RESULTS

Residual infectious material in toilet bowls. The first group of experiments was conducted to determine the fate of infectious agents in toilet waters after flushing of a typical domestic toilet. Toilet bowls were first cleaned as described in Materials and Methods and then seeded with either an overnight culture of

E. coli or MS-2 phage. In both cases the organisms were suspended in100ml of TSB to simulate the presence of organic matter found in actual fecal material. After the bowl water was mixed, a baseline sample was obtained. The toilet was then flushed, and the toilet water was sampled for residual organisms. This procedure was repeated several times, and the results of a typical experiment for both bacteria and viruses can be seen in Fig. 1. As anticipated, the initial flush eliminated the major proportion of exogenously added organisms. However, after repeated flushes, instead of diminishing, there was often an increase in the number of residual organisms detected in the bowl. In the case of both bacteria and viruses, the number of orga- nisms in the bowl reached a plateau below which their number could not be reduced, even after repeated flushing. From this evidence, it appeared that significant numbers of bacteria and viruses were being adsorbed to the toilet porcelain and then eluted during the flushing action. To test this speculation, the previous experi ment was repeated, and, after the third flush, Tween 80, a nonionic detergent, was added to yield a final concentration of 0.1% in the bowl water. The sides ofthe bowl were then scrubbed with a brush, and a sample was obtained for assay. The results of this experiment for E. coli are shown in Fig. 2.A10,000-foldincreaseinthe bacterial count occurred after the Tween 80 treatment, indicating that bacteria were ad- sorbing to the porcelain surfaces and could be eluted by scrubbing in the presence of an eluent such as Tween 80. When the toilet was flushed after the Tween 80 treatment, almost all of the bacteria were removed from the bowl water

APPL. MICROBIOL. 12 -J 11 I- 10

w 9

I--

z

cr 8

w 7-

-J 0 6 0 5 % 0 0 -J 40 3 0 i 2 3 4 5 6 7

NUMBER OF FLUSHES

FIG. 1. Effect of flushing on removal of exoge- nously added bacteria and viruses to the toilet bowl. The log,, number of organisms indicated above was added to the toilet bowl water, and water samples were removed from the bowl for as say after each flush. Log,1 values of organisms represent the number present in the entire volume contained in the toilet bowl. Symbols: 0, MS-2phage; 0, E. coli.

(Fig. 2). In other experiments it was found that simple mixing ofTween80 into thebowl water or scrubbing the bowl with a brush without addition of Tween 80 were equally efficient in the removal of bacteria from the sides of the toilet bowl.

When the same experiment was performed with poliovirus or MS-2 phage (Fig. 3), an increase in the number of viruses was also noted after Tween 80 treatment, but subsequent flushing resulted in only a gradual loss of virus from the bowl water. Thus, it did not appear that all of the viruses were eluted from the bowl surfaces by the Tween 80 treatment. Perhaps viruses are more difficult to desorb from the porous surface of the porcelain than bacteria.

To determine the number of bacteria usually present in toilets, toilet bowls were monitored in restaurants, service stations, etc. Toilets were treated with Tween 80 and agitated to elute bacteria from the bowl surfaces. A sample of bowl water was then removed and plated on both EMB agar and Trypticase soy agar. The results of these experiments are shown in Table

1. Aerobic bacteria were present at levels from 

VOL. 30, 1975 MICROBIOLOGICAL HAZARDS OF HOUSEHOLD TOILETS 231

_J 1t

0

TWEEN 80 ADDED

I-J

0

04 '

23

234 567

NUMBER OF FLUSHES FIG. 2. Elution of E. coli from toilet bowl surfaces by addition of Tween 80. Same methods as in Fig. 1,

except Tween 80 was added to the bowl water after the flush indicated and the bowl was scrubbed with a brush.

101 to over 109 in the bowls tested.

Droplet production. Dye studies were per- formed to determine if water droplets ejected into the air during flushing reach the seat level of the bowl. Crystal violet dye was added to both the bowl and the tank water, and after covering the bowl with a sheet of white absorb- ent paper the toilet was flushed. By examination of the paper the number of visible droplets produced during flushing was determined. Tank-type toilets produced a random pattern of droplets on the paper, ranging in number from 27 to 104 during a given flush. Valve-type toilets produced fewer visible droplets (between 7 and 10), which were always found towards the rear of the bowl. It appeared that the high pressure of the water coming into the bowl in this type of toilet caused the droplets to be ejected to the rear.

If the bowl was first seeded with E. coli and EMB agar plates were exposed at the seat level (taped to a support and facing the bowl water), coliform colonies appeared on the agar plates in

J

=o 10

0

\ Z

cr:

w

i8 \ TWEEN 80 ADDED

w

0 >6

LA. -J

0

0~ 04

31

0 1234567

NUMBER OF FLUSHES FIG. 3. Elution of MS-2 phage from toilet bowl surfaces by addition of Tween 80. Same methodsas in Fig. 2.

TABLE 1. Recovery of bacteria from toilet bowlsa

Total no./vol in toilet bowl

(log,.)

Sam-Type of Type of Toa

ple establishment toilet Coli-Total no. forms bacteria

1 Hamburger stand Tank 2.50 6.90 2 Service station Tank 5.50 8.05 3 Hospital Valve 5.40 7.50 4 Restaurant Tank 3.10 6.65 5 Motel Valve 3.90 9.25 6 Service station Tank 2.50 7.50 7 Home Tank 3.30 8.25 8 Research institute Valve 3.70 6.80 9 Restaurant Valve 8.10 9.30 10 Research institute Valve 6.10 6.65 11 Research institute Valve 2.28 7.10 12 Research institute Valve 3.10 6.80 13 Service station Valve 4.50 9.40 14 Hospital Valve 2.25 6.00

a Toilets were treated with 220 ml of1%Tween 80 to remove adsorbed bacteria before sampling.

the same pattern as the droplets.

Collection of ejected organisms with gauze. To determine more quantitatively the number of organisms which reach the seat level during

 232 GERBA, WALLIS, AND MELNICK

flushing, cotton gauze was utilized to collect ejected organisms. A series of experiments was first performed to determine the efficiency with which bacteria and virus entrapped by the gauze could be recovered. Suspensions of either bacteria or viruses were added to cut squares of gauze (6 by 6 inches [ca. 15 by 15 cm], two thicknesses, 12 actual layers of gauze), and various solutions were evaluated for their ability to elute the organisms. Gauze prewetted with TSB was found to give the best recoveries of E. coli and MS-2, and the average percentage of recoveries for several experiments is shown in Table 2. It has been shown that enteric viruses adsorbed to a variety of surfaces can be eluted at high pH (17). Thus, glycine buffer adjusted to pH 11.5 was used to elute poliovirus from gauze prewetted with glycine. The average recovery using this method to elute poliovirus from the gauze was 84% (Table 3). With polio- virus it was necessary to concentrate the virus from the gauze eluate before assay. This was accomplished with adsorption onto membranes (Millipore Corp.) as described by Wallis et al. (21).

The following procedure was used to deter- mine the number of organisms reaching the seat level in droplets. The toilet bowl and rim were

TABLE 2.ElutionofE.coliandMS-2phagefrom gauze with TSBa

Pretreatment No. of No.ofor-

of gauze organisms

Organism prior to ad-nlsms eluted % Redition of test placed on fromgauze covery

organism (x 104) (x 10')

E. coli TSB 19.0 19.5 104 19.0 20.0 106 5.0 4.76 95 5.0 5.56 113 5.0 4.90 98 E. coli None, dry 5.0 7.76 15 5.0 5.95 12 19.0 7.00 36 MS-2 TSB 117 106 90 117 89 76 117 92 79 117 80 68 117 91 78

aGauzepads(6by6inches, 2thicknesses, 12actual layers of gauze) were used dry or wetted with 5 ml of TSB. Excess fluids were expressed, and then 0.1 ml of the indicated organisms suspended in tapwater was placed onto the gauze. After 5 min the gauze was soaked in 20 ml of broth, and approximately 7 ml of eluate was expressed from the gauze.

APPL. MICROBIOL.

first sterilized by igniting alcohol placed on the rim, and the bowl was seeded with the test organism. The bowl was then covered with a piece of gauze (14 by 17 inches [ca. 35.6 by 43.2 cm], 12 thicknesses, double layer) presoaked with 50 ml of TSB or glycine buffer, and the toilet was flushed. The gauze was then soaked for 5 min in 150 ml of eluent with occasional squeezing of the gauze to obtain a maximal amount of eluate. Approximately 100 ml of eluate was usually obtained. The number of bacteria or viruses in the eluate was then quantified. EMB agar placed on top of the gauze held over the bowl indicated that E. coli bacteria did not penetrate the gauze after flushing. Table 4 shows the number of E. coli ejected from two tank-type toilets with different volumes and the amount of variation in the number of organisms recovered from replicate experiments.

The number of bacteria and phage ejected from the toilet during a flush was found to be directly proportional to the amount present at the time of the flush (Fig. 4). Studies with poliovirus (Table 5) indicated that similar quantities of this virus were ejected from the bowl as found for MS-2 phage when the bowl was seeded with plaque-forming units.

106

When the number of seed organisms approached numbers found inhuman stool (about 1012 for bacteria [18] and 101 for virus [16D, as high as 6.6 x 101 coliforms and 2.8 x 103 plaque-forming units of virus were recovered from the gauze.

The number of coliforms and the total number of aerobic bacteria recovered from the gauze

TABiz 3. Elution of polio virus from gauze pads with pH 11.5 glycine buffer

-PFU Gauze pad no.-a PFU added to 1)(x105) detected in eluate(xlO'1) Recovery 1 4.00 3.40 85.0 2 4.00 3.90 97.5 3 4.00 3.00 75.0 4 4.00 3.20 80.0

a Gauze pads (4 by 7 inches, 2 thicknesses, 12 actual layers of gauze) were wetted with 5 ml of pH 11.5 glycine buffer, and 0.1 ml of LSc poliovirus type 1 suspended in tapwater was placed onto the gauze. After 5 min the gauze pads were soaked in 20ml ofpH

11.5 glycine buffer for another 5 min, after which 20 ml ofeluatewassqueezedfromthegauze,andthepH was immediately adjusted to 7.5. PFU, Plaque-form- ing units.

 VOL. 30, 1975 MICROBIOLOGICAL HAZARDS OF HOUSEHOLD TOILETS 233

TABLE 4. Number of E. coli ejected from toilets during flushinga

No.of bacteria added to No.of bacteria toilet bowl (x 10') ejected"

Experimental toilet 1 (bowl volume, 3,500 ml)

9.97 1,260 9.97 1,095 15.70 1,340 105.0 5,500 157.0 3,040 245.0 66,500 Experimental toilet 2 (bowl volume, 5,900 ml)

44.8 1,170 21.8 1,925 37.7 2,440 318.0 1,700 171.0 9,000 283.0 2,630 aBoth toilets had tank volumes of 21,000 ml.

bRepresents number of bacteria collected from gauze pad held at seat level over the bowl during flushing.

5

4z

z 0

C,

a .'

,91

Q CI 0 5 6 7 8 9 lb il l2 LOG 10 SEEDED IN BOWL

FIG. 4. Concentration of bacteria and virus in toilet bowl and numbers ejected during flushing. The log,0 numberofE. coliorMS-2phageindicatedabove was placed in the bowl water, a gauze wetted with TSB was placed over the bowl, and the toilet was flushed. The log,. of the number of organisms re- covered from the gauze is indicated in the ordinate of the above figure. Symbols: 0, MS-2 phage; 0, E.

coli.

when actual fecal material was present are shown in Table 6. These data indicated that numbers of bacteria reaching the seat level of the bowl were not appreciably different when the bowl was seeded with similar numbers of bacteria as cultures, homogenized stool, or as a solid fecal pellet. The apparent increase in the ratio of total bacteria to coliforms may be the result of the use of selective media in the assay of the coliforms; i.e., bacteria may be damaged during aerosolization and fail to grow on the selective media.

TABLE 5. Number of polioviruses ejected from toilet during flushing

No. ofPFU No. ofPFU Estimated no. of

added to toilet detected in PFU present in

bowl (x 10') eluate eluatea

2.88 1,570 2,802 2.67 675 1,205 3.78 300 536 2.94 893 1,594 aTo determine efficiency of concentration, a portion of the baseline sample was added to a volume of pH 11.5 glycine buffer and concentrated by the same method as the eluate. Estimated number of plaque- forming units (PFU) was calculated as follows: actual number of viruses collected by gauze = (100/% efficiency of elution from gauze) (100/% efficiency of concentration from eluate) (number of viruses de- tected in eluate). Efficiency of elution from gauze, 84%; average efficiency of concentration using this method, 67%.

TABLE 6. Number of bacteria ejected from toilets during flushing using human stool

No.of bacteria present in No.of bacteria toilet water at time of flush e acted' (x 10')

Coliforms Total Coliforms Total

Homogenized stool

0.182 2.8 10 4,500 1.19 1.54 35 2,000 6.47 7.0 35 4,000 7.35 12.9 60 18,000 18.20 28.2 38 8,000 Solid stool 0.0007 0.0035 2 6,000

1.01 9.10 85 137,000 0.168 2.30 145 10,000 0.178 1.68 30 21,000 1.53 3.06 280 7,000 Controlb 0 0 0 2,500 0 0 0 4,500 0 0 0710 0 0 0 1,090

a Represents number of bacteria collected from gauze pad held at seat level over the bowl during flushing.

b Gauze was placed over the bowl, but the toilet was not flushed. Organisms detected in the controls probably represent those naturally present in the bowl water or from contamination during collection of the eluates.

Fallout of airborne E. coli onto bathroom surfaces. To determine if bacteria ejected into air from the bowl during flushing were falling out onto surfaces in the bathroom, EMB agar

 234 GERBA, WALLIS, AND MELNICK

plates were exposed at various times after a toilet seeded with E. coli had been flushed. The design of the experimental bathroom can be seen in Fig. 5. Agar plates were placed at 6-inch (ca. 15-cm) intervals throughout the room for a total number of 50 plates exposed at one time. Four sets of plates were usually exposed 2, 4, and 6 h after a flush, plus a control set exposed 2 h before the toilet was flushed. The results of these experiments are summarized in Table 7. Within the first 2 h, bacteria were usually detected only in a limited area around the toilet, whereas bacterial colonies detected at

later intervals were more randomly distributed throughout the room.

To detect the fallout of airborne particles containing viruses from flushed toilets, 10-by 8.75-inch (ca. 25.4by20.9cm)squaresofgauze mounted on metal screens and wetted with TSB were placed at the locations shown in Fig. 5. At the end of the exposure time (each set of gauze was exposed at 2-h intervals), as much fluid as possible was expressed from the gauze and

assayed for virus. Since it was desirable to keep the amount of elution fluid to as small a volume as possible, a series of experiments was con- ducted to determine the effect of eluate volume on the efficiency of virus recovery. The results of these experiments are shown in Table 8. When the eluate volume was less than 10 ml, the efficiency of virus recovery was appreciably reduced. Thus, enough TSB was added to the

gauze so that the eluate volume did not fall below this amount. The results off all out experiments usingMS-2

64"-

64"

l 1.1~~r

L"I

*_I

, FIG. 5. Floor plan of experimental bathroom with location ofgauzepadsfor viralfallout experiments.

APPL. MICROBIOL.

TABLE 7. Number of bacteria detected after falling out on bathroom surfaces after flushing

Estimated

Avgno. Avg no. no. of

Time of plates of bacteria Venti-

after on which colonies on floor lationc

flush (h) bacteria detected surface

grewa per flush area of

bathrooms

0-2 7.0 9.0 737 Closed

2-4 2.8 3.0 246 Closed

4-6 0 0 0 Closed

0-2 7.8 9.2 639 Open

2-4 2.0 2.0 164 Open

4-6 0 0 0Open

a Represents the average of six experiments for each 2-hperiod. Ineachexperiment 1011E. coliwereadded to the bowl before flushing. Fifty EMB agar plates, mainly on the bathroom floor, were exposed during each 2-h period and replaced by another set of plates.

Calculated as follows: (total floor surface area of bathroom/total surface area of all agar plates exposed) (total number of colonies on agar plates).

cThe ventilation of the room was considered closed when all air vents in the room were sealed and the bathroom door remained closed.

TABLE 8. EffectofeluatevolumeonrecoveryofMS-2 phage from gauze padsa

Vol of wetting Avg vol of Avg % recovery fluid (ml) eluate (ml) of virus

35 11.3 72

30 10.1 78

25 5.0 59 22 4.2 46 a Gauze pads (10 by 8.75 inch) were wetted with the amount of TSB indicated above. Virus was added as two drops of 0.05 ml each. After 15 min as much fluid as possible was expressed from each ofthe gauze pads. This eluate was then assayed for virus. 1.2 x 103 to 6.0 x 10" viral plaque-forming units were added to each gauze pad.

phage are shown in Table 9. As was observed for the bacteria, most of the virus appeared to fall-

out within 2 h after the flush. However, small numbers of virus were often detected on control sets of gauze exposed before the toilet flush, indicating that virus from experiments per- formed several days previously was still present in the room. These background counts were subtracted from those obtained after each toilet flush.

Natural contamination of bathroom sur- faces by coliforms. Using 2.5-inch (6.4-cm) diameter Rodac plates of Levine EMB agar,

various surfaces in a number of household and

 VOL. 30, 1975 MICROBIOLOGICAL HAZARDS OF HOUSEHOLD TOILETS 235

TABLE 9.NumberofMS-2phage falling out on flushing, and even continual flushing could not bathroom surfaces after toilet flushing remove a persistent fraction. This was attrib- Total uted to the adsorption of the organisms to the

Total no. of PFU Estimated PFU after no. ofPFU

Time eluted correction on floor Venti-

after from all for ef-surface lationc

gauze ficiency area of padsa of bathroom

elution'

0-2 4,107 5,175 35,468 Closed 2-4 244 307 2,532 Closed 4-6 0 0 0 Closed 0-2 209 263 1,540 Open 2-4 0 0 0Open 4-6 0 0 0Open

'Represents the average of five to six experiments for each 2-h period. In each, approximately 4 x 10$ plaque-forming units (PFU) of MS-2 phage were

added to the bowl before flushing. Eluate from eight gauze pads was pooled before assay. The location of the gauze pads in the bathroom can be seen in Fig. 5.

The average efficiency of elution was approxi- mately 79%.

e

cThe ventilation of the room was considered closed when all air vents in the room were sealed and the bathroom door remained closed.

TABLE 10. Detection of naturally occurring coliforms on bathroom surfaces"

No. of

agar Maxi- plates mum Nof on %Sam-no.of

Surface sam-which plea coliforms

coli-detected

pleapom positive o

forms on a were single de-plate tected

Walls 54 11 20 5 Floor 120 31 26 >100 Seat, toilet 70 27 38 15 Rim, toilet 35 10 28 >60 Flush handle 9 1 11 2 Bathtubs, sinks, 125 6 5 >100

cabinets, etc.

aRodac plates ofEMB agarwere used.

public bathrooms were sampled for the presence of coliforms. The results are summarized in Table 10. Over 20 bathrooms were tested and coliforms were detected in all, indicating that surface contamination in the bathroom is com-

mon.

DISCUSSION

Considerable numbers of bacteria and viruses were shown to remain in the bowl water after

porcelain surfaces of the bowl, with gradual elution occurring after each flush. This ad- sorbed fraction could explain the heavy bacte- rial aerosols detected by Darlow and Bale (6) after the second flush of toilets seeded with bacteria. In toilet bowl waters tested by us, chlorine was usually absent or was present in very small amounts (<1 mg/liter), probably due to its rapid loss to the atmosphere. In addition, organic matter in the stool would combine with any free chlorine that might be present. Thus, a gradual build-up ofboth bacteria and viruses in toilets could occur during regular use.

Droplets produced by flushing toilets were found to harbor both bacteria and viruses placed in the toilet before flushing. The average human stool weighs approximately 100 g and contains a total of about 10l2 bacteria (18), of which 10i0 or more are coliforms (19). In in- fected persons, up to 10"1 Salmonella (19) and 108 to 10"1 Shigella (14) have been detected in the stool. Concentrations as high as 10i0 Salmo- nella paratyphi B per g of feces have been detected in carriers (19). The number of polio- viruses present in the stool of infected persons canbeashighas 106pergoffeces,givingatotal of 10. in the stool (16). For these values, the number of infectious organisms ejected from the bowl would range from 1,000 to over 10,000 based on data obtained in this study.

Organisms collected on the gauze probably represent those contained in the larger-sized droplets that quickly settle out on surfaces in the bathroom and do not take into account the organisms present in smaller droplets which may be airborne for considerable lengths of time. The detection of coliform bacteria and viruses falling out onto surfaces in the bathroom after flushing indicated that these organisms remain airborne long enough to settle on surfaces throughout the bathroom. The number of

E. coli detected on the agar plates is probably a minimal value since airborne bacteria are dam- aged during aerosolization and by environmen- tal stresses, making growth on selective media more difficult (13). These data also do not take into account the accumulation of organisms on the walls and other surfaces of the bathroom. The number of viruses calculated to be falling out onto the floor surface of the bathroom was found to be over a log greaterthan that detected when the gauze was held over the bowl. This may be due to large numbers of virus being present in smaller-sized drops which do not 

236 GERBA, WALLIS, AND MELNICK

impact onto the gauze during flushing. Whereas the number of both bacteria and viruses deter- mined to be impacting on surfaces was less when the bathroom door and vents were left open, it should be pointed out that the orga- nisms not settling out in the room under these conditions are probably being carried to other locations by air currents.

The presence of fecal organisms on bathroom surfaces is undoubtedly widespread, as evi- denced by the isolation of coliform bacteria on surfaces in all of the bathrooms sampled. Hutchinson (12) detected the presence of Shi-

gella sonnei on bathroom floors and toilet seats. He also found that this organism could persist for as long as 17 days on wooden water closet seats. Newsom (14) reported that Salmonella survived for 11 days after desiccation when suspended in either tapwater or feces. In this present study, large numbers of coliform bacte- ria were found on several occasions under sham- poo bottles in bathrooms, which might indicate that regrowth or prolonged persistence of these organisms may occur where organic material

has a chance to accumulate. In addition, en- tero viruses, as well as members of the adenovi- rus and reovirus groups, have been found to survive desiccation on surfaces (4).

Doses of less than 102 Shigella flexneri have been found to be capable of initiating infection in man (9). Also, Darlow et al. (7) have shown that the lethal dose of Salmonella typhimurium in mice was lower when they were infected by inhalation rather than ingestion. With virus, however, the minimal infectious dose for humans may be as low as one tissue culture infectious dose (15). Thus, it would appear that the numbers of bacteria and viruses ejected from the toilet are sufficiently large to initiate infection, especially in the case of viruses.

In a study of airborne transmission of coxsack- ievirus type 21, Couch et al. (5) detected only 1 50% tissue culture infectious dose per 3.5 ft3

(0.10 mi) in a barracks in which aerosol trans- mission of the virus between humans occurred. In an army barracks in which transmission of adenovirus was occurring, air sampling revealed the presence of only 1 50% tissue culture infectious dose in 920 ft3 (25.76 min) of air (2). The results of our study would seem to indicate that greater concentrations of virus would exist as aerosols in bathrooms in which infectious material has been flushed. The spread of viral disease by aerosols from toilets may take on added importance in those viral diseases in which low numbers of viruses are excreted in nasopharyngeal secretions as compared to the amount excreted in the feces. In addition, the APPL. MICROBIOL.

overall importance of enteric virus transmission by respiratory secretions is still in dispute. For

example, the spread of poliovirus within families when pharyngeal excretion occurred differed little from that following purely fecal excretion (10). Also, recent studies with adeno- virus type 4 indicate that fecal sources may be more important than respiratory sources in the spread of this virus (10).

The significance of enteric virus disease transmission by contact with surfaces harboring

infectious material should not be overlooked, as evidenced by the recent findings of Hendley et

al. (11) on the transmission of rhinovirus from fomites to the hands and self-inoculation of the eyes and nose. Additional evidence would be necessary to substantiate the role of aerosols in the epidemiology of disease transmission by toilets, but the results of this study indicate that it appears to be a distinct possibility.

LITERATURE CITED

1. Adams, M. H. 1959. Bacteriophages. Interscience Pub- lishe's, New York. 2. Artenstein, M. S., and W. S. Miller. 1966. Air sampling for respiratory disease agents in army recruits. Bacte- riol. Rev. 30:571-572. 3. Bound, W. H., and R. I. Atkinson. 1966. Bacterial aerosol from water closets. A comparison of two types of pan and two types of cover. Lancet i:1369-1370. 4. Buckland, F. E., and D. A. J. Tyrrell. 1962. Loss of infectivity on drying various viruses. Nature (London) 195:1063-1064. 5. Couch, R. B., R. G. Douglas, Jr., K. M. Lindgren, P. J. Gerone, and V. Knight. 1970. Airborne transmission of respiratory infection with coxsackievirus A type 21. Am. J. Epidemiol. 91:78-86. 6. Darlow,H.M.,andW.R.Bale.1959.Infectivehazardsof water-closets. Lancet i:1196-1200. 7. Darlow, H. M., W. R. Bale, and G. B. Carter. 1961. Infection of mice by the respiratory route with Salmo- nella typhimurium. J. Hyg. 59:303-308. 8. Davis, J. E., and R. L. Sinsheimer. 1963. The replication of bacteriophage MS-2. I. Transfer of parental nucleic acid to progeny phage. J. Mol. Biol. 6:203-207.

9. DuPont, H. L., R. B. Hornick, M. J. Snyder, J. P. Libonati, S. B. Formal, and E. J. Gangarosa. 1972. Immunity in shigellosis. II. Protection induced by oral live vaccine or primary infection. J. Infect. Dis. 125:12-16. 10. Fox, J. P. 1973. Introduction and spread of respiratory viral pathogens in families, p. 487-493. In J. F. Hers and K. C. Winkler (ed.), Airborne transmission and airborne infection. J. Wiley and Sons, New York.

11. Hendley, J. O., R. P. Wenzel, and J. M. Gwaltney, Jr. 1973. Transmission of rhinovirus colds by self-inocula- tion. New Engl. J. Med. 288:1361-1364. 12. Hutchinson, R. I. 1956. Some observations on the method of the spread of sonne dysentery. Mon. Bull. Minist. Health Public Health Lab. Serv. 15:110-118.

13. Kingson, D. 1971. Selective media in air sampling: a review. J. Appl. Bacteriol. 34:221-232. 14. Newsom, S. W. B. 1972. Microbiology of hospital toilets. Lancet ii:700-703. 15. Plotkin, S. A., and M. Katz. 1967. Minimal infective dosesofvirusesformanbytheoralroute, p. 151-156.In G. Berg (ed.), Transmission of viruses by the water 

VOL. 30, 1975 MICROBIOLOGICAL HAZARDS OF HOUSEHOLD TOILETS 237

route. J. Wiley and Sons, New York. enteric carriers. J. Hyg. 52:67-70.

16. Sabin, A. B. 1955. Behavior of chimpanzee-avirulent 20. Wallis, C., S. Grinstein, J. L Melnick, and J. E. Fields. poliomyelitisvirusesinexperimentallyinfectedhuman 1969.Concentrationofvirusesfromsewageandexcreta volunteers. Am. J. Med. Sci. 230:1-8. on insoluble polyelectrolytes. Appl. Microbiol. 17. Sobsey, M. D., C. Wallis, M. Henderson, and J. L. 18:1007-1014. Melnick. 1973. Concentration of enteroviruses from 21. Wallis, C., M. Henderson, and J. L. Melnick. 1972. large volumes of water. Appl. Microbiol. 26:529-534. Enterovirus concentration on cellulose membranes.

18. Stanier, R. Y., M. Doudoroff, and E. A. Adelberg. 1970. Appl. Microbiol. 23:476-480. The microbial world. Prentice-Hall, Englewood Cliffs, 22. Wallis, C., and J. L. Melnick. 1967. Concentration of N.J. enteroviruses on membrane filters. J. Virol. 1:472-477. 19. Thomson, S. 1954. The number of bacilli harboured by