Vol. 43, No. 5APPLIED AND ENVIRONMENTAL MICROBIOLOGY, May 1982, p. 1104-11100099-2240/82/051104-07$02.00/0

Hot Water Systems as Sources of Legionella pneumophila inHospital and Nonhospital Plumbing Fixtures

ROBERT M. WADOWSKY,1 ROBERT B. YEE,'* LORRAINE MEZMAR,1 EDWARD J. WING,2 ANDJOHN N. DOWLING1' 2

Department of Microbiology, Graduate School of Public Health,' and Department of Medicine, School ofMedicine,2 University of Pittsburgh, Pittsburgh, Pennsylvania 15261

Received 16 September 1981/Accepted 1 February 1982

Samples obtained from plumbing systems of hospitals, nonhospital institutions,and homes were cultured for Legionella spp. by plating the samples directly on aselective medium. Swab samples were taken from the inner surfaces of faucetassemblies (aerators, spouts, and valve seats), showerheads, and shower pipes.Water and sediment were collected from the bottom of hot-water tanks. Legion-ella pneumophila serogroups 1, 5, and 6 were recovered from plumbing fixtures ofthe hospitals and nonhospital institutions and one offive homes. The legionellae (7to 13,850 colony-forming units per ml) were also present in water and sedimentfrom hot-water tanks maintained at 30 to 54°C, but not in those maintained at 71and 77°C. Legionella micdadei was isolated from one tank. Thus legionellae arepresent in hot-water tanks which are maintained at warm temperatures or whosedesign results in warm temperatures at the bottom of the tanks. We hypothesizethat hot-water tanks are a breeding site and a major source of L. pneumophila forthe contamination of plumbing systems. The existence of these bacteria in theplumbing systems and tanks was not necessarily associated with disease. Theextent of the hazard of this contamination needs to be delineated.

Legionella pneumophila has been shown to bepresent in aquatic habitats. Initially these bacte-ria were found in cooling towers of air condition-ing systems and a stream in association withoutbreaks of Legionnaires' disease (3, 12). Sub-sequently they were detected in natural waterswithout association with disease (5, 6). Recentreports suggest that the legionellae may alsooccur in potable water systems, particularly inhospitals. The bacteria have been isolated fromshowerheads in hospitals in which there hadbeen cases of the disease (2, 13). We haverecovered the bacteria from a prefilter of adeionization system in a hospital (16). In En-gland, L. pneumophila was isolated from pota-ble water from not only hospitals but also hotels,both with and without known association withdisease (14).These findings indicate that the bacteria may

contaminate potable water systems, and that theoccurrence of Legionella spp. in water systemsand the impact on health need to be delineated.However, surveys for the detection of thesebacteria have been limited because of the lack ofa suitable selective medium. Our developmentof a selective medium, glycine-vancomycin-polymyxin B (GVP) agar (16), has enabled us toconduct a survey of plumbing fixtures in hospitaland nonhospital environments. The contamina-tion of these sites by L. pneumophila and the

role of hot-water tanks as a source of thiscontamination are reported in this paper.

MATERIALS AND METHODS

Samples. Samples of the sediment in plumbing fix-tures were obtained for culturing from three nonhospi-tal institutions, two of which were gymnasiums, fivehomes of faculty and staff members, and three localhospitals. Plumbing fixtures included faucets andshowers. A faucet was sampled by inserting a cottonswab into the spout and vigorously rubbing the swabagainst the internal surface. Subsequently the hot- andcold-water stems were removed. The internal surfaceof each valve seat was then sampled with a cottonswab. Showers were sampled in a similar manner. Ashowerhead was removed, and samples were obtainedfrom the inner surface of the showerhead and theshower pipe.Each cotton swab was used to inoculate an agar

plate. A line of inoculum, approximately 2.5 cm wide,was made by streaking across the diameter of a platewith the swab. With an inoculating needle, the inocu-lum was distributed over the surface of the plate bystreaking perpendicularly to the line of inoculum.Tap water was collected before the removal of

sediment from a plumbing fixture. Both tap water andwater from hot-water tanks were cultured by thespread plate technique by using a bent glass rod todistribute 0.1 or 0.2 ml of water over the surface of anagar plate. When legionellae were not recovered, 1 mlof water was subsequently plated out by the spreadplate technique.

1104

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

HOT WATER SYSTEM AS SOURCE OF LEGIONELLA 1105

TABLE 1. Recovery of L. pneumophila from showerheads and shower pipes in a gymnasium

Site CFU of L. pneumophila per plate from different showers in a gymnasium

Shower pipe >300 >300 >300 300 300 150 115 55 16 16 250 81 43 0Showerhead 69 51 114 23 18 11 3 28 6 2 >300 300 100 0

Water and sediment from hot-water tanks werecollected from the bottom drain valve of each tank.For our initial sampling of the tanks, two samples, 500to 600 ml each, were collected from each tank. Thefirst sample was collected immediately after a drainvalve was opened, and the second one was obtainedafter the water had been permitted to flow for 2 to 3min. Subsequent sampling periods entailed collectionof 500 to 600 ml of water and sediment immediatelyafter the opening of a drain valve.A tap water suspension of the bacteria in a shower-

head was prepared in an effort to isolate Legionellamicdadei. This showerhead had been located in apatient room in which a case of nosocomial pneumoniadue to L. micdadei had occurred. The showerhead wasimmersed in tap water, and the water was vigorouslyagitated on a magnetic stirrer. After 0.2 ml had beencultured by the spread plate technique, this shower-head suspension was incubated at 42°C for 1 week andthen recultured. This procedure was used since a swabsample of the showerhead had not yielded L. micda-dei. In another study, we had found that this treatmentcould be used for the selective enrichment of L.pneumophila (unpublished results).

All cultures were incubated at 37°C and examineddaily for 7 days.

Media. Samples were first obtained from a nonhos-pital institution and homes for culturing. Each samplewas plated on both buffered charcoal-yeast extractagar and GVP agar (16). This procedure was modifiedfor culturing environmental samples from the hospitalsand two additional nonhospital institutions. The use ofbuffered charcoal-yeast extract agar was discontinuedsince we observed that all positive samples weredetected by GVP agar and colony counts of legionellaewere always higher on this medium. Also, differentialGVP agar was used as the selective plating medium.This medium, GVP agar containing 0.001% bromcre-sol purple and bromthymol blue, permitted the differ-entiation of L. pneumophila from L. micdadei andother legionellae (15). (L. micdadei has been previous-ly called the Pittsburgh pneumonia agent and theTATLOCK bacillus [8, 11]. The names, Tatlockiamicdadei and Legionella pittsburgensis have also beenproposed for this bacterium [7, 10].) Differential GVPagar has been successfully used for the isolation of L.pneumophila from environmental samples locally byA. Brown and V. Yu at the Oakland Veterans Admin-istration Hospital and S. Kominos at Mercy Hospitaland in Los Angeles, Calif. by P. Edelstein at theWadsworth Veterans Administration Hospital (per-sonal communications).

Identification of Legionella spp. L. pneumophila andL. micdadei were identified as previously described(16). Serogrouping was performed by the direct immu-nofluorescence test (1) with reagents kindly providedby the Biological Products Division of the Bureau ofLaboratories at the Centers for Disease Control.These reagents included fluorescein isothiocyanate-

conjugated monovalent antisera against L. pneumo-phila serogroups 1 through 6 and L. micdadei.

RESULTSNonhospital environments. The study of non-

hospital environments was initiated originally toevaluate further the use of GVP agar as aselective medium for the isolation of Legionella-ceae from water-related environments. For thispurpose, the uppermost screen of each faucetaerator in a nonhospital institution was sampled.Of 26 aerators, 10 were found to be contaminat-ed with L. pneumophila. Serogroup 1 was isolat-ed from nine of the positive aerators, and sero-group 5 was isolated from one. We then sampledthe inner surface of each of the spouts of 12faucets, the 10 with positive aerators and 2 withnegative aerators. Only two spouts, ones withpositive aerators, harbored L. pneumophila.Tap water, collected from one of the two fau-cets, was cultured and found to contain 20colony-forming units (CFU) of L. pneumophilaper ml.The showerheads, faucet spouts, and aerators

in five homes (two in the city and three in thesuburbs) were sampled. L. pneumophila sero-group 5 was isolated from a faucet spout from asingle home within the city. This faucet did nothave an aerator.We then obtained samples from showerheads

and showerhead pipes in a gymnasium. L. pneu-mophila serogroups 1 and 5 were recoveredfrom 13 of 14 showerheads. The pipes of thesepositive showerheads were also contaminatedwith the legionellae. In 10 out of the 13 positivesamples, the samples from the pipes yieldedhigher colony counts of L. pneumophila thanthose from the showerheads, and this differencein counts was significant (paired Student t test, P< 0.05) (Table 1).

Hospital environments. Three hospitals, desig-nated A, B, and C, requested a survey of theirplumbing fixtures. A hospital-related case ofpneumonia due to L. micdadei had occurred inhospital A, and one due to L. pneumophilaserogroup 6 and L. micdadei had occurred inhospital B. No recent known cases of hospital-related legionellosis had occurred at hospital C.However, sporadic cases of hospital-relatedpneumonia due to L. micdadei had occurred inthis hospital, the most recent being three monthsbefore our collection of samples (17).

VOL. 43, 1982

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

1106 WADOWSKY ET AL.

TABLE 2. Isolation of L. pneumophila fromhospital A

Room Plumbing fixture CFU/Type Site cultured plate

1 Faucet Spout >300CWa valve seat 0HWb valve seat 0

2 Faucet Spout 146CW valve seat 21HW valve seat 2

Communal Shower Showerhead 42a CW, Cold water.b HW, Hot water.

In hospital A, we sampled the faucet assembly(faucet spout and valve seats) of the sinks in thetwo rooms in which the patient had been housedand the showerhead in the communal showerroom which had been used by the patient. L.pneumophila serogroup 1 was recovered fromall of these sites (Table 2). The inner surface ofthe faucet spout in room 1 was the most heavilycontaminated. No legionellae could be recov-ered from the valve seats since they containedlarge numbers of non-Legionellaceae bacteriawhich grew on differential GVP agar. However,in room 2 we were able to detect L. pneumophilain the valve seats in addition to the faucet spout.We noted that the faucets in many of the

patient rooms had aerators. Two aerators oneach of nine patient care floors were sampled.Nine of 18 of these aerators were positive for L.pneumophila serogroups 1 and 6.The patient who had been admitted to hospital

B with a pneumonia due to both L. pneumophilaserogroup 6 and L. micdadei was first admittedto a room in the intensive care unit, designated

room 2-ICU (Table 3), and was subsequentlytransferred to a private room, room 3. L. pneu-mophila serogroup 1 was recovered from theaerator, and serogroups 1 and 6 were recoveredfrom the spout of the faucet in the sink in room2-ICU. In the adjacent room (1-ICU), wheremore extensive sampling could be done, sero-group 6 was isolated from all the sites that weresampled, i.e., aerator, spout, and hot- and cold-water valve seats. In the private room, thelegionellae were found in the faucet spout, thefaucet cold-water valve seat, and the mixingvalve and pipe of the shower. Among thesesites, the shower mixing valve was the mostheavily contaminated. Plating of 0.1 ml of watercollected from the faucet yielded one colony ofL. pneumophila serogroup 6. No legionellaewere isolated from the hot-water valve seat ofthe faucet or the showerhead. The showerheadwas dry at the time of sampling, which wouldnot be favorable for the survival of Legionellaspp.

The most extensive sampling was done inhospital C. Aerators in 78 rooms and faucetassemblies and showers in four patient roomswere examined. After the initial sampling, sam-ples were collected from two of the four rooms 2and 4 months later and from one room 4 monthslater. Among the aerators, 73 of 78 from patientrooms were positive for L. pneumophila sero-

group 1 or 5 or both. All aerators have sincebeen removed. Both faucet and shower fixturesin the four patient rooms also initially harboredserogroup 1 or 5 or both (Table 4). The faucetspouts and hot-water valve seats and showerpipes were the most heavily contaminated sites.With the exception of one sample, the cold-water valve seats of the faucets did not yield anylegionellae. The results of the latter samplings

TABLE 3. Isolation of L. pneumophila from hospital B

Room Plumbing fixture L. pneumophilaType Site cultured CFU/plate Serogroup

1-ICU Faucet Spout 22 6CW' valve seat 12 6HWb valve seat 71 6Aerator 56 6

2-ICU Faucet Spout 100 1, 6Aerator 50 1

3 Faucet Spouit 70 6CW valve seat 1 6HW valve seat 0 NAcWater (0.1 ml) 10 6

Shower Showerhead 0 NAPipe 77 6Mixing valve 300 1

a CW, Cold water.b HW, Hot water.c NA, Not applicable.

APPL. ENVIRON. MICROBIOL.

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

HOT WATER SYSTEM AS SOURCE OF LEGIONELLA 1107

TABLE 4. Isolation of L. pneumophila from hospital C

Room Plumbing fixture L. pneumophilaType Site cultured CFU/plate Serogroup

1 Faucet Spout >300, >300, >300a 1CWb valve seat 1, 0, 0 1HWC valve seat >300, >300, 30 1

Shower Showerhead 0, 10, 0 NAdPipe 3, >300, 3 1

2 Faucet Spout 230, ND,e >300 1CW valve seat 0, ND, 0 NAHW valve seat >300, ND, 200 1Aerator 17, None 5

Shower Showerhead 3, ND, 0 1Pipe 5, ND, 0 1

3 Faucet CW valve seat 0, ND, ND NAHW valve seat 14, ND, ND 1, 5Aerator 26, ND, ND 1

Shower Showerhead 4, ND, ND 1, 5Pipe 30, ND, ND 5

4 Faucet Spout >300, >300, 200 1CW valve seat 0, 0, 0 NAHW valve seat >300, >300, 23 1

Shower Showerhead 16, 3, 4 5Pipe 170, 39, 72 1

a Results of samples collected bimonthly.b CW, Cold water.c HW, Hot water.d NA, Not applicable.eND, Not done.

showed that L. pneumophila could persist for upto 4 months in plumbing fixtures.

L. micdadei was isolated from only a singleshowerhead in hospital C. This fixture was lo-cated in a room in which a case of L. micdadeipneumonia had occurred. Initial sampling usinga swab yielded only L. pneumophila. A suspen-sion of bacteria on this showerhead was pre-pared as described above. Culturing of thissuspension once again yielded only L. pneumo-phila. However, 39 colonies of L. micdadei wererecovered from 0.2 ml of the suspension after ithad been incubated at 42°C for 1 week. No L.micdadei was isolated from swab samples whichwere subsequently obtained from the showerpipe. Thus, although L. micdadei was isolated ina hospital where sporadic cases had occurred,the bacteria were apparently present infrequent-ly or in low concentrations (or both) in theplumbing system.

Hot-water tanks. Water and sediment from thebottom of hot-water tanks in the three hospitalswere cultured. We suspected that legionellaemight be present at this site since: (i) L. pneumo-phila can multiply at 32 to 42°C in tap water(R. B. Yee and R. M. Wadowsky, Appl. Envi-ron. Microbiol., in press); (ii) these bacteriawere found more frequently and in larger num-bers in hot-water valve seats than in cold-watervalve seats, particularly in hospital C; and (iii)

the hospitals maintained their hot water at 43 to55°C in conformance with the requirements ofthe Joint Commission on Accreditation of Hos-pitals (9) and for energy conservation. Becausethese tanks were heated by centrally locatedsteam coils, the temperature at the bottom ofthese tanks tended to be lower and within therange for multiplication of the legionellae.Samples were obtained from 12 hospital hot-

water tanks which were maintained at 31 to 54°Cand two which were held at 71 and 77°C. L.pneumophila was recovered from the first sam-ple obtained from each tank which was main-tained at 31 to 54°C (Table 5). The concentrationof legionellae ranged from 7 to 13,850 CFU/ml.The lowest concentration was found in a tank inhospital A. This tank had been completelydrained 4 weeks before our collection of thesample. Appreciably lower numbers of the bac-teria or none were detected in the second samplefrom each tank. In hospital C, two tanks whichserviced the laundry and animal rooms weremaintained at 71 and 77°C. The water and sedi-ment from the bottom of these tanks did notcontain any Legionella.

L. pneumophila was also recovered from thehot-water tanks of two nonhospital institutions(Table 5). The water and sediment from thebottom of the tanks in two gymnasiums har-bored serogroups 1, 5, and 6. The showerheads

VOL. 43, 1982

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

TABLE 5. Isolation of L. pneumophila from hot-water tanksTemp (°C) L. pneumophila

Building No. CFU/mlsampled Tank Sample lb Sample 1 Sample 2b Serogroup

Hospital A 2 54 31, 38 7, 1,340 0 1, 6Hospital B 4 31-54 28-33 80-600 0-30 1, 6Hospital C 6 43-45 NDc 570-13,850 60-8,750 1, 5

2 71, 77 ND 0 ND NAdGymnasium 1 2 43 34, 48 220, 310 0, 10 1, 5Gymnasium 2 1 49 43 80e 0 1, 6

a Temperature at which hot water was maintained in tank for use.b Sample 1 was collected immediately after drain valve at bottom of tank was opened; sample 2 was collected 2

to 3 min later.c ND, Not done.d NA, Not applicable.e Sample also contained 20 CFU of L. micdadei per ml.

and showerhead pipes in gymnasium 1 werecontaminated with legionellae as describedabove. One of the hot-water tanks in gymnasium2 was also found to contain 20 CFU of L.micdadei per ml. The presence of legionellae inthese tanks was not unexpected since the waterand sediment at the bottom was 34 to 48°C.Samples were then collected from the hot-

water tanks in hospital C and the two gymnasi-ums 2 months and 2.5 weeks, respectively, afterthe initial sampling period. With the exception ofthe samples from tanks 4 and 5 in hospital C, allof the samples were again positive for Legion-ella (Table 6). The sediment in tanks 4 and 5 inhospital C had been removed 2 days earlier byletting the water flow from the drain valves for15 min. Samples from three of the other fourtanks in hospital C yielded lower concentrationsof legionellae than had been obtained previous-ly. The concentration of legionellae in the sec-ond samples from the gymnasiums were essen-tially the same or higher than those found in theinitial samples.

Water and sediment was also collected fromhot-water tanks in four homes. Three of thesetanks were the modern type with gas burnerslocated at the bottom of the tank, whereas thefourth was an old type having a side burner. L.pneumophila (470 CFU/ml) was isolated onlyfrom the water and sediment from the oldertank.

DISCUSSIONThe isolation of L. pneumophila from shower-

heads (2, 13) was the first evidence that thisbacterium could inhabit the plumbing systems ofhospitals. Subsequently Tobin et al. (14) showedthat L. pneumophila was present in the potablewater of hospitals and hotels in Great Britain.These workers had to concentrate the bacteria inthe water samples and to use guinea pig inocula-tion for the detection of the legionellae.Our development of a selective medium (16)

has greatly facilitated the procedure for theisolation of these bacteria from environmentalsources. We have been able to conduct an

TABLE 6. Isolation of L. pneumophila from hot-water tanks in hospital C and gymnasiums'

Hot-water tank Sample 1 Sample 2

Temp (OC) CFU/mI Temp (°C) CFU/ml

Hospital C1 43 750 35 4402 49 1,210 27 1,4603 43 570 38 804 43 1,100 43 05 43 3,360 38 06 44 13,850 34 3,460

Gymnasiums1 48 220 30 8502 34 310 30 3003 43 80 38 270

aThe interval between collection of samples 1 and 2 was 2 months for hospital C and 2.5 weeks forgymnasiums.

APPL. ENVIRON. MICROBIOL.1108 WADOWSKY ET AL.

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

HOT WATER SYSTEM AS SOURCE OF LEGIONELLA

extensive survey of the plumbing of hospitals,nonhospital buildings, and homes. Our findingssupport the above reports that L. pneumophilais present in the plumbing systems of hospitalsand nonhospital institutions. L. micdadei ap-pears to be an infrequent contaminant. On theother hand, L. pneumophila is often present onthe inner surfaces of plumbing fixtures in thesebuildings where it may occur in relatively largenumbers. The most common sites of contamina-tion are aerators, faucet spouts, hot-water valveseats, and shower pipes. Tap water from con-taminated fixtures usually contains small or un-detectable numbers of legionellae.

This contamination of plumbing systems byLegionella spp. raises a number of questionsconcerning the source of these bacteria, theirvirulence, and the relationship of this contami-nation to the occurrence of disease, includingthe significance of the quantity of legionellaepresent. In addition, our findings may be uniqueto the buildings that we have sampled. Theavailability of our selective medium for theisolation of Legionella should facilitate the ex-amination of plumbing systems and hot-watertanks in other localities.The sources of the legionellae which contami-

nate the plumbing fixtures remain to be deter-mined. All of the institutions that we havesurveyed receive their water from the samemunicipal water supply system. To date, therehave been no published reports which implicatepotable reservoir water. We can only speculatethat the bacteria may be present infrequently orin very low concentrations or both. These smallnumbers of legionellae may be enriched in thewarm environment at the bottom of hot-watertanks.

All of the hot-water tanks in our institutionswere constructed of copper-lined steel. Untilother types of tanks are sampled our findingsmay apply only to tanks of this type of construc-tion.Our present working hypothesis is that the

water and sediment at the bottom of hot-watertanks may be a major source of the legionellaethat contaminate the plumbing fi:.tures andpipes. This high degree and frequency of con-tamination in the institutions may have resultedfrom the maintenance of hot-water tanks attemperatures optimal for the growth of L. pneu-mophila. The bacteria probably concentrate andmultiply at the bottom of the tank, where thetemperature is somewhat lower, by adhering tosediment which accumulates there. The agita-tion of the bacteria-laden sediment may thenlead to introduction of large numbers of bacteriainto the plumbing system. Dispersed sedimentmay settle on and thereby contaminate plumbingfixtures and pipes. The continued presence of

these bacteria at these sites may be the result oftheir ability to survive and multiply on thesesurfaces and periodic reseeding from hot-watertanks. Our hypothesis is supported by two find-ings. First, sequential sampling from the bottomdrain valves showed that the legionellae weremainly in the water and sediment at the bottomof the tanks (Table 5). Second, we have foundthat L. pneumophila is able to grow in tap waterat 32 to 42°C (Yee et al., submitted for publica-tion). The temperatures of the samples wereusually within this range.Another hypothesis to be considered is the

liklihood of the legionellae multiplying in thecirculating hot water per se or on the surfaces ofthe pipes and valves of the circulating hot-watersystems in the institutions (or both). We plan toobtain weekly samples of hot water just beforeits reentry into a highly positive hot-water tankand to sample valves and pipes of a circulatinghot water system.Our findings on a limited number (five) of

single-family houses suggest that Legionellamay occur infrequently in the plumbing fixturesof these buildings. We speculate that this infre-quent occurrence may be related to the type ofhot-water tank that is present. Houses in whichlegionellae were not detected utilized modernhot-water tanks. In these tanks, the heat isapplied at the bottom of tanks, thereby maintain-ing temperatures which kill the bacteria accumu-lating there. In contrast, the legionellae werepresent in the water and sediment at the bottomof an older domestic tank which had a side-burner to heat the water. The temperature at thebottom of this tank was within the range for thegrowth of L. pneumophila. Interestingly, thishome was the only one in which L. pneumophilawas recovered from a plumbing fixture. Thesampling of hot-water tanks in additional homesis needed to test this hypothesis.Our study provides further evidence that L.

pneumophila is frequently found in high num-bers in plumbing systems without any associa-tion with disease. Thus caution must always beapplied in implicating a contaminated plumbingsystem as a source of the bacteria in an outbreakof Legionnaires disease.Decontamination of plumbing systems should

be considered when there is an association withdisease. In one hospital, hyperchlorination ofthe cold water and maintenance of the hot waterat 60°C was followed by cessation of an outbreakof Legionnaires disease (4). If our hypothesis istrue that hot-water tanks are a source of thecontamination of plumbing systems, then drain-age of the sediment from the tanks may beeffective in appreciably decreasing or eliminat-ing this contamination. This simple, inexpensiveprocedure may be employed even when no

VOL. 43, 1982 1109

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from

1110 WADOWSKY ET AL.

associated disease is found. It may supplementraising the temperature of the water in the tanksovernight. We are testing these two measures inhospital A and the two gymnasiums.

ACKNOWLEDGMENTS

We thank Ronald Speranza for excellent technical assist-ance, Patricia Carothers for secretarial assistance, the physi-cal plant staffs of the institutions and the infection controlpersonnel of the hospitals for advice and assistance in thecollection of samples, and Monto Ho and A. William Pascullefor advice and critical review of the manuscript.

This study was supported in part by Public Health Serviceresearch grant AI 17047 from the National Institute of Allergyand Infectious Diseases and sponsored by the EnvironmentalEpidemiology Center of the Graduate School of Public Healthof the University of Pittsburgh under the support of coopera-tive agreement CR80681-01-2 with the U.S. EnvironmentalProtection Agency.

LITERATURE CITED

1. Cherry, W. B., B. Pittman, P. P. Harris, G. A. Hebert,B. M. Thomason, L. Thacker, and R. E. Weaver. 1978.Detection of Legionnaires' disease bacteria by directimmunofluorescent staining. J. Clin. Microbiol. 8:329-338.

2. Cordes, L. G., A. M. Wisenthal, G. W. Gorman, J. P.Phair, H. M. Sommers, A. Brown, V. L. Yu, M. H. Mag-nussen, R. D. Meyer, J. S. Wolf, K. N. Shands, and D. W.Fraser. 1981. Isolation of Legionella pneumophila fromshower heads. Ann. Intern. Med. 94:195-197.

3. Dondero, T. J., R. C. Rentorff, G. F. Mailison, R. M.Weeks, J. S. Levy, E. W. Wong, and W. Schaffner. 1980.An outbreak of Legionnaires' disease associated with acontaminated air-conditioning cooling tower. N. Engi. J.Med. 302:365-370.

4. Fischer-Hoch, S. P., J. O.'H. Tobin, A. M. Nelson, M. G.Smith, J. M. Talbot, C. L. R. Bartlett, M. B. Gillett, J. E.Pritchard, R. A. Swann, and J. A. Thomas. 1981. Investi-gation and control of an outbreak of Legionnaires' diseasein a district general hospital. Lancet i:932-936.

5. Fliermans, C. B., W. B. Cherry, L. H. Orrison, S. J.Smith, D. L. Tison, and D. H. Pope. 1981. Ecologicaldistribution of Legionella pneumophila. Appl. Environ.Microbiol. 41:9-16.

6. Fliermans, C. B., W. B. Cherry, L. H. Orrison, and L.

Thacker. 1979. Isolation of Legionella pneumophila fromnon-epidemic related aquatic habitats. Appl. Environ.Microbiol. 37:1239-1242.

7. Garrity, G. M., A. Brown, and R. M. Vickers. 1981.Tatlockia and Fluoribacter: two new genera of organismsresembling Legionella pneumophila. Int. J. Syst. Bacter-iol. 30:609-614.

8. Hebert, G. A., B. M. Thomason, P. P. Harris, M. D.Hicklin, and R. M. McKinney. 1980 "Pittsburgh pneumo-nia agent": a bacterium phenotypically similar to Legion-ella pneumophila and identical to the TATLOCK bacteri-um. Ann. Intern. Med. 92:53-54.

9. Joint Commission of Accreditation on Hospitals. 1981.Accreditation, manual for hospitals, p. 43. Joint Commis-sion on Accreditation of Hospitals, Chicago.

10. Pasculle, A. W., J. C. Feeley, R. J. Gibson, L. G. Cordes,R. L. Myerowitz, C. M. Patton, G. W. Gorman, C. L.Carmack, J. W. Ezzell, and J. N. Dowling. 1980. Pitts-burgh pneumonia agent: direct isolation from human lungtissue. J. Infect. Dis. 141:727-732.

11. Pasculle, A. W., R. L. Myerowitz, and C. R. Rinaldo, Jr.1979. New bacterial agent of pneumonia isolated fromrenal-transplant recipients. Lancet 11:58-61.

12. Politi, B. D., D. W. Fraser, G. F. Mallison, J. V. Mohatt,G. K. Morris, C. M. Patton, J. C. Feeley, R. D. Telle, andJ. V. Bennett. 1979. A major focus of Legionnaires'disease in Bloomington, Indiana. Ann. Intern. Med.90:587-591.

13. Tobin J. O'H., J. Beare, M. S. Dunnili, S. Fisher-Hoch, M.French, R. G. Mitchell, P. J. Morris, and M. F. Muers.1980. Legionnaires' disease in a transplant unit: isolationof the causative agent from shower baths. Lancet 1i:118-121.

14. Tobin, J. O'H., R. A. Swann, and C. L. R. Bartlett. 1981.Isolation of Legionella pneumophila from water systems:methods and preliminary results. Br. Med. J. 282:515-517.

15. Vickers, R. M., A. Brown, and G. M. Garrity. 1981. Dye-containing buffered charcoal-yeast extract medium fordifferentiation of the family Legionellaceae. J. Clin. Mi-crobiol. 13:380-382.

16. Wadowsky, R. M., and R. B. Yee. 1981. A glycine-con-taining selective medium for isolation of Legionellaceaefrom environmental specimens. Appl. Environ. Micro-biol. 42:768-772.

17. Wing, E. J., F. J. Schafer, and A. W. Pasculie. 1981.Successful treatment of Legionella micdadei pneumoniawith erythromycin. Am. J. Med. 71:836-840.

APPL. ENVIRON. MICROBIOL.

on May 7, 2018 by guest

http://aem.asm

.org/D

ownloaded from