Protocol for Sampling Environmental Sites for Legionellae

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUIY 987, p. 5-58 0099-0/87/075-05$0.00/0 Copyright C) 987, Americn Society for Microbiology Vol. 5, No. 7 Protocol for Smpling Environmentl Sites for Legionelle J. M. BARBAREE,l* G. W. GORMAN, W. T. MARTIN,' B. S. FIELDS,' AND W. E. MORRILL' Respirtory Diseses Brnch, Division of Bcteril Diseses, Center for Infectious Diseses, Centers for Disese Control, Atlnt, Georgi 0, nd 7 Atherton Circle, Dectur, Georgi 005 Received 8 Jnury 987/Accepted 6 April 987 A protocol for smpling environmentl sites ws developed nd used to identify possible sources of Legionell species in support of epidemiologic investigtions t two hospitls. In hospitl A, legionelle were isolted from of 06 (0%) different sites. Three seprte Legionell pneumophil serotypes nd previously unrecognized species were present in different combintions in the positive smples. Two of five cooling towers contined the sme L. pneumophil serogroup monoclonl type (,,,5) s ws isolted from ptients. The sme monoclonl type ws lso isolted from mke-up wter for the two cooling towers, hot wter tnk, wter seprtors in four min ir compressor systems for respirtory therpy, nd cold nd hot wter fucets. In hospitl B, of 7 (8%) smple sites contined legionelle, ll of which were L. pneumophil serogroup. The monoclonl type mtching isoltes from ptients (,,,5) ws found t the highest concentrtion in hot wter tnk, but it ws lso present t four other smple sites. Since legionelle not relted to disese my be found in mny of the sites smpled, n epidemiologic ssocition with the probble source should be estblished before intervention methods, such s disinfection, re undertken. Documenttion of the source for the spred of the etiologic gent cusing legionellosis epidemic cn be problem. The gent, Legionell species, which is ubiquitous in freshwter environments (9, 0), sometimes cuses pneumoni nd hs been implicted s the gent of Pontic fever. The primry mens for trnsmission of pneumoni ppers to be inhltion of erosols contining virulent Legionell species by susceptible individuls (, 5, 7). The potentil sources of wter tht could contin legionelle nd be erosolized in n institution such s hospitl mke the investigtion of the legionellosis outbrek difficult. The selection of smple sites to identify the common sources of legionelle should be bsed on epidemiologic dt proving n ssocition between ptients nd possible exposure to erosols contining the gent. However, the microbiologicl, physicl, nd chemicl conditions of the wter my chnge before epidemiologic dt hve been collected nd nlyzed. Thus, it my be pproprite to obtin smples t the outset of n investigtion to ensure collection of timely nd pproprite specimens. We developed protocol s guide for selecting smple sites t the beginning of legionellosis investigtion. Our selection of sites for the protocol ws bsed on pst lbortory experience in legionellosis investigtions, sites implicted for the spred of legionellosis by other investigtors, nd common chrcteristics of wter distribution nd irconditioning systems. The protocol does not include ll possible smple sites nd contins some sites tht re not pplicble to ll circumstnces. This report describes our results when the protocol ws used to estblish profile of positive sites for Legionell species in two epidemic investigtions. Bckground for protocol. Bsiclly, wter specimens ssocited with buildings re in eight ctegories (Tble ). The first ctegory (A), potble wter outside nd on the boundry of hospitl property, is included to document externl sources of municipl wter tht contin legionelle nd my seed the wter nd plumbing systems of the building. The * Corresponding uthor. reports of other investigtors such s Voss et l. (7, 8) hve indicted tht municipl wter tretment plnts hrbor legionelle nd emphsize the need to identify externl contmintion sources. The second ctegory (B), the generl potble wter system for hospitl or fcility, includes sites tht cn be very importnt for disese trnsmission. The presence of legionelle in wter heters nd holding tnks is well documented, nd these hve been implicted s sources in outbreks of legionellosis (, 6). The third ctegory (C), phrmcy wter, my be importnt since wter for the preprtion of sterile solutions used for surgicl ptients nd in respirtory therpy equipment is sometimes supplied by the hospitl phrmcy. Legionellosis cses linked to respirtory therpy equipment hve been well documented (). The fourth ctegory (D), ir compressor systems, cn be importnt if ir for respirtors is supplied to ptients from such devices t the fcility. It is not uncommon for excess wter to be observed t times in compressed ir lines. Sites listed in the fifth ctegory (E), potble wter finl distribution outlets, include shower heds, fucets, hemodilysis units, nd ice mkers. The recovery of legionelle from showers hs been documented (6). The sixth ctegory (F), ir-conditioning systems, includes prts of ir-hndling systems such s evportive condensers nd cooling towers. Both hve been reported s sources for legionelle cusing epidemics (, 5, 7). The seventh ctegory (G) is included since evidence of the spred of legionelle from whirlpools (hot tubs) hs been reported (). The eighth ctegory (H) is included for considertion in cse epidemiologic evidence indictes n ssocition between exposure to erosols from source such s decortive fountin or wter from site such s creek. Wter from creek or river my be used s mke-up wter for cooling tower. Since environmentl nd epidemiologic circumstnces my dictte priorities on the selection of smple sites, the protocol ws written s guide for selecting smple sites nd not s mster list of sites tht must ll be smpled. 5 MATERIALS AND METHODS Outbrek sites. The protocol (Tble ) ws used t two hospitls designted s hospitl A nd hospitl B. Hospitl A Downloded from http://em.sm.org/ on October, 08 by guest

VOL. 5, 987 SAMPLING ENVIRONMENTAL SITES FOR LEGIONELLAE 55 TABLE. Protocol for smpling environment sites for legionelle Site nd description Approx no. Vol of smples of smples A. Potble wter outside or on boundry of hospitl property. Tretment plnt (rw nd refined wter). Gurd house or outlying fcility if wter is not fed there from hospitl. Fire hydrnt(s) B. Generl potble wter system for hospitl. Incoming wter pipe(s) 5. Wter softener (pre nd post) 6. Preheter (dischrge side) 7. Primry heter (dischrge side) 8. Circulting pump(s) 9. Holding tnks (cold wter, dischrge side) 0. Expnsion tnk for hot wter (if possible). Bck drin on sprinkler system(s) (trp to prevent bckflushing my be present nd should be smpled). Fireline where it brnches off min system (my be multiple) C. Phrmcy. Wter used for respirtory therpy equipment D. Air compressor system. Vcuum wter source Positive pressure equipment side 5. Condenste from tnk(s) 6. Wter seprtor(s) (directly off compressors) 7. Wter source(s) ner ir intke(s) 8. Air smples where ptients were ill with legionellosis E. Potble wter finl distribution outlets Hemodilysis wter source 9. Before deminerlizer 0. After deminerlizer Intensive cre units. Respirtory therpy (ptient rooms). Crdic. Services with different geogrphicl loctions. Ice mker (entry wter) F. Air-conditioning system 5. Air hndling unit to service where disese occurred (drin pn) Cooling towers 6. Blowdown 7. Wter supply G. Whirlpools 8. Whirlpool (one nerest ir intke system) 9. Whirlpool drin H. Other 0. Decortive fountin(s). Creeks, ponds, nd sites of stgnnt wter NA, Not pplicble. is n cute cre fcility in the New Englnd Are with pproximtely 700 beds nd 8 buildings. A totl of of 5 legionellosis cses were from one of six min buildings, nd ll isoltes from ptients were Legionell pneumophil serogroup with monoclonl ntibody rectivity pttern to ntigens,,,5 (). There were t lest four municipl wter lines leding into the hospitl plumbing system. The wter ws not chlorinted in the hospitl system. Hospitl B is northern midwest peditric hospitl with pproximtely 00 beds. There were three municipl wter intke pipes, nd residul chlorine level of <0.5 ppm (<0.5,ugIml) ws present in some of the wter smples. At the time of the investigtion, there were three confirmed cses of legionellosis. Isoltes from ptients were lso L. pneumophil serogroup tht rected with monoclonl ntibodies to ntigens,,,5. Collection nd tretment of wter smples. The number of smples tken t ech site ws limited owing to resources nd epidemiologic considertions. However, lrge number 7 0 liters 0 liters. 00 ml 00 ml.00 ml.00 ml NA". l00 ml. Wet swb > (06) of different sites were smpled in ssocition with hospitl A, nd 7 sites were smpled t hospitl B. Ech of the five cooling towers t hospitl A ws smpled two to three times. All wter smples of. were collected in sterile -liter or 50-ml polypropylene continers. Lrger smples, such s wter from the city tretment plnt, were collected in 5-gllon (9-liter) plstic continers. Potble wter from fucets ws collected fter llowing the fucet to run stedily for pproximtely min. All wter smples were shipped in lrge insulted continers nd were received unrefrigerted in the lbortory for processing within 8 h of collection. Potble wter smples concentrted by filtrtion nd in some cses (when mny nonlegionelle were present) were cid treted s described by Gormn et l. (). When swbs were used, they were sterile cotton swbs dipped in sterile wter before use. Medi nd isoltion procedures. The following medi were inoculted (0.-ml inoculum per plte) with ech concen- Downloded from http://em.sm.org/ on October, 08 by guest

56 BARBAREE ET AL. APPL. ENVIRON. MICROBIOL. TABLE. Results with the environmentl smple site protocol Hospitl A Hospitl B Smple site ctegory Sites positive/ Sites smpled Sites positive/ sites smpled sites smpled Sites smpled Potble wter outside 0/8 5 outside fire hydrnts 0/ fire hydrnts or on the boundry Rw, erted, nd refined wter t Rw nd refined wter from the of hospitl property the wter tretment plnt municipl wter tretment plnt Generl potble wter /9 6 incoming pipes /6 sites where wter enters plumbsystem for hospitl 6 hot wter heters ing system for buildings ( site/ high zone pumps building) hot wter tnk drins hot wter heters hot wter heter inlet fireline connector wter line before cooling tower Potble wter finl / cold or hot wter fucets /8 5 showers distribution outlets 5 ertors fucets Air compressor system 5/9 5 compressed ir lines 0/ ir compressor receiver tnk wter seprtors 5 compressor wter supply lines excess wter lines from ir compressors Air compressor condenstes Air-conditioning system 9/0 5 cooling tower sumps ( tower/ /7 cooling tower sumps building) Mke-up wter for of the Mke-up wter cooling towers Whirlpools nd other /7 ertors 0/ Wter supply for dilysis unit whirlpool drins whirlpool operting room spry humidifier All legionelle isolted from smple sites t hospitl B were L. pneumophil serogroup. trted or nonconcentrted wter smple: two pltes of buffered chrcol-yest extrct gr contining lphketoglutrte (BCYE); two pltes of glycine-polymyxin B-nisomycin-vncomycin medium (GPAV), which is BCYE supplemented with glycine (0.%), polymyxin B (00 units/ml), nisomycin, (80,ug/ml), nd vncomycin (5,ug/ml); one plte of BCYE without L-cysteine; nd one plte of GPAV without L-cysteine. The lst two pltes were used s negtive controls. An djustment ws mde for dilution or concentrtion when the CFU of legionell-like orgnisms were clculted; 00% recovery ws ssumed fter filtrtion. Severl or ll legionell-like orgnism colonies from t lest one plte of GPAV per smple were confirmed s legionelle bsed on colony morphology, L- cysteine requirement, nd direct fluorescent-ntibody tests (). Rectivity to L. pneumophil serogroup monoclonl ntibodies. L. pneumophil serogroup cultures were tested by Roger McKinney (Immunology Lbortory, Meningitis nd Specil Pthogens Brnch, Division of Bcteril Diseses, Centers for Disese Control) for rections to monoclonl ntibodies (5). Rectivity ptterns to the ntibodies were used to subgroup types. Immunoutordiogrphic procedure. The immunoutordiogrphic procedure reported by Mrtin et l. () ws used to detect colonies of L. pneumophil serogroup contining monoclonl ntigen. Ascites fluid contining monoclonl ntibody to ntigen ws provided by R. McKinney. Air-smpling methods. A six-stge Andersen ir smpler contining GPAV s the medium for impct of ir ws used for ir smpling t rte of 0.08 m/min. Under norml circumstnces, selective medium should not be used with the Andersen ir smpler, but it ws used in this cse to void overgrowth by nonlegionelle. Air ws smpled for nd 0 mins, respectively, from ir compressor lines nd mbient ir in five legionellosis ptient rooms in hospitl A. Also, n ll-glss impinger contining 0.5% queous yest extrct ws used to collect ir t rte of 0.008 m/min for 0 min in ech of the five rooms. The broth ws stined by the direct fluorescent-ntibody procedure nd plted (six times) onto olbcye gr. Pltes were incubted for > dys t 5 C in humid tmosphere of ir plus.5% CO nd exmined periodiclly for legionell-like orgnisms. RESULTS Tble summrizes results from using the smple site protocol (Tble ) s guide t both hospitls. More smples were tken nd were positive t hospitl A ( of 06) thn t hospitl B ( of 7). A vriety of legionelle were recovered t hospitl A (Tbles nd ). In the first ctegory (A), smples of municipl wter before it entered hospitl building were negtive for legionelle. In the generl potble wter system (ctegory B), L. pneumophil serogroup monoclonl group,,,5 ws isolted from wter tnk in building E. In building B, three tnks contined L. pneumophil. Serogroups nd were present in two tnks, nd serogroup ws present in the third one. Phrmcy wter (ctegory C) ws not used for respirtory therpy nd ws not smpled. In ctegory D, legionelle were not recovered in ir smples of the compressed ir lines used for respirtory therpy in five ptient rooms, but the wter seprtors in the ir compressors did contin L. pneumophil. Extensive smpling of the potble wter finl distribution outlets (ctegory E) showed combintions of L. pneumophil serogroups nd nd n unrecognized species in some of the hot nd cold wter plumbing systems in five buildings (Tble ). In the ir-conditioning system (ctegory F), two of five cooling towers contined L. pneumophil serogroup monoclonl Downloded from http://em.sm.org/ on October, 08 by guest

VOL. 5, 987 SAMPLING ENVIRONMENTAL SITES FOR LEGIONELLAE 57 Ctegory of smple site TABLE. Legionelle isolted from smple sites t hospitl A Sites with L. pneumophil SG- SG- SG- nd SG- Other legionelle Air-conditioning Cooling towers b, c, Cooling towers b nd c L. pneumophil SG-5 in system nd 5c cooling tower Mke-up wter for cool- Mke up wter for cool- L. pneumophil nd preing towers b nd c ing tower viously unrecognized species in mke-up wter for cooling tower Generl potble wter Building E, hot wter Building B, low-zone Building B,c low-zone hot system tnkb hot wter tnk no. wter tnks no. nd Cold wter-finl Sink fucet in room in sink fucets,b one ech Unrecognized species from distribution outlets building Ab sink in room in buildings A sink fucet in room fucet in ech of nd B ech in buildings B nd C rooms in building Bb,c Hot wter-finl Wter dischrge line Sink fucet in ech of Sink fucetd in ech of 6 distribution outlets from hot wter heter rooms in building rooms in building B nd in building Db B nd room in room in building C building C Air compressor system Air compressor wter seprtors in buildings Bb c nd Cd Whirlpools, phrmcy, Whirlpoolc Unrecognized species lso other Hemodilizer wter in whirlpool nd hemodilizer wter SG, Serogroup. b L. pneumophil serogroup monoclonl group,,,5. cl. pneumophil serogroup monoclonl group not contining ntigen. d L. pneumophil serogroup undetermined monoclonl group. type,,,5. L. pneumophil serogroup 5 ws found in only one cooling tower. In ctegory G, whirlpool nd wter being used with hemodilizer contined combintion of L. pneumophil serogroups nd nd previously unrecognized Legionell species. No smples were tken from sites covered in ctegory H. L. pneumophil serogroup ws isolted from ll positive sites t hospitl B. The highest numbers of legionelle found in the potble wter system were in the hot wter tnk (.5 x 07 CFU/liter). Of the finl distribution outlets smpled, three ertors nd wter from one shower hed nd seven fucets contined L. pneumophil. Wter outlets in 0 of the ptient rooms (7%) were positive for L. pneumophil. L. pneumophil serogroup monoclonl type,,,5 ws isolted from one shower hed, three fucets, nd hot wter tnk (Tble ). One of four cooling towers used in the ir-conditioning system contined L. pneumophil serogroup. Legionell species were not isolted from the wter supply for one dilysis unit. DISCUSSION Although legionelle were not isolted from wter before it entered either hospitl property, periodic smpling would be needed to rule out intermittent entry of legionelle from outside sources such s municipl wter reservoirs. Legionelle hve been isolted from wter tretment plnts (7, 8). It is interesting tht three serotypes of L. pneumophil nd nother species were coexisting t different sites t hospitl A, yet the cses were cused only by one monoclonl type of L. pneumophil serogroup. Only one of five cooling towers contined serogroup 5 s the predominnt type, wheres serogroup ws predominnt in the other four. Perhps the microbil flor nd environmentl conditions were more optiml for serogroup 5 in tower, or the contminting source of legionelle ws different for tht tower. The predominnce of L. pneumophil mong other legionelle in the two towers implicted s the source for spred of legionellosis my be evidence of its superior bility to survive, compete with other microbil flor, nd multiply to concentrtion necessry for the trnsmission of legionellosis. Fliermns (8) reported tht L. pneumophil serogroup (Knoxville strin) multiplied nd survived longer thn representtive strins of serogroups,, nd in diffusion chmbers t 55 C. Thus, some legionelle hve equivlent virulence chrcteristics, but respond differently to environmentl conditions. The bility of Legionell strin to compete with other bcteri nd multiply in plces such s cooling towers nd hot wter tnks cn be crucil fctor in the trnsmission of legionellosis. Use of the protocol reveled tht L. pneumophil serogroup monoclonl type,,,5 ws in the mke-up TABLE. Monoclonl rectivity Monoclonl rectivity ptterns of L. pneumophil serogroup isoltes from hospitl B Source of smple,,,5 showerhed fucetsb hot wter tnk, shower fucetsc,,6 fucet,,5,6,7,8,9 shower" No rectivity shower fucets Ech source is different smple site unless noted otherwise. b One isolte ws from n ertor. c One isolte ws from n ertor wsher. d This shower is the sme one lso contining L. pneumophil serogroup with, rectivity pttern. Downloded from http://em.sm.org/ on October, 08 by guest

58 BARBAREE ET AL. wter for towers nd nd presumbly ws responsible for seeding the towers. These two towers were reported s the source for the outbrek t hospitl A (). Although no legionelle were isolted from compressed ir lines tht were supplying ir for respirtory therpy equipment, the presence of legionelle in the wter seprtor in the compressed ir system poses potentil hzrd if mlfunction llows the entry of erosols from the wter seprtors. Limited use of the protocol t hospitl B did show tht L. pneumophil serogroup monoclonl type,,,5 ws probbly being distributed from the hot wter tnk to outlets, such s the shower heds. Epidemiologic dt hd implicted the showers s sources of exposure to the gent. The isoltion of vriety of legionelle t hospitl A demonstrted tht rndom smpling without n epidemiologic evlution nd compring isoltes from the environment nd from ptients could led to flse conclusions bout sources of epidemic strins. The protocol should be stremlined to hve only five ctegories, with wter from the phrmcy, whirlpools, nd other sites s one ctegory. Also, ctegory A should include well, nd humidifiers should be dded to ctegory F. High priority for smpling should be given to the hot wter tnks, outlets in the rooms of legionellosis ptients, cooling tower wter (if it is possible tht erosols of this wter could hve been brethed by ptients), respirtory therpy equipment, nd holding tnks for wter. The lck of legionellosis cses cused by legionelle other thn the cse isolte types (L. pneumophil serogroup monoclonl type,,,5) present in the wter reinforces the need for documenting n epidemiologic ssocition before disinfection mesures re undertken. Prudent use of the protocol to gther smples in timely mnner should support dt from n epidemiologic study. ACKNOWLEDGMENTS We thnk Roger McKinney for his work on the monoclonl ntibody group determintions for L. pneumophil serogroup nd epidemic investigtors Brry Dvis, Pul Grbe, Luri Mrkowitz, nd Mike Linnn for their contributions to this study. LITERATURE CITED. Arnow, P. M., T. Chow, D. Weil, E. N. Shpiro, nd C. Kretzschmr. 98. Nosocomil legionnires' disese cused by erosolized tpwter from respirtory devices. J. Infect. Dis. 6:60-67.. Bollin, G. E., J. F. PLouffe, M. F. Pr, nd B. Hckmn. 985. Aerosols contining Legionell pneumophil generted by shower heds nd hot-wter fucets. Appl. Environ. Microbiol. 50:8-.. Cherry, W. B., nd R. M. McKinney. 98. Direct immunofluorescence tests for dignosis of legionellosis. Am. J. Pthol. 0:8-5.. Ciesielski, C. A., M. J. Blser, nd W. L. Wong. 98. Role of APPL. ENVIRON. MICROBIOL. stgntion nd obstruction of wter flow in isoltion of Legionell pneumophil from hospitl plumbing. Appl. Environ. Microbiol. 8:98-987. 5. Cordes, L. G., D. W. Frser, P. Skliy, C. A. Perlin, W. R. Else, G. F. Mllison, nd P. S. Hyes. 980. Legionnires' disese outbrek t n Atlnt, Georgi country club: evidence for spred from n evportive condenser. Am. J. Epidemiol. :5-. 6. Cordes, L. G., A. M. Wiesenthl, G. M. Gormn, J. P. Phir, H. M. Sommers, A. Brown, V. L. Yu, M. H. Mgnussen, R. D. Meyer, J. S. Wolf, K. N. Shnds, nd D. W. Frser. 98. Isoltion of Legionell pneumophil from shower heds. Ann. Intern. Med. 9:95-97. 7. Dondero, T. J., R. C. Rendtorff, G. F. Mllison, R. M. Weeks, J. S. Levy, E. W. Wong, nd W. Schffner. 980. An outbrek of legionnires' disese ssocited with contminted cooling tower. N. Engl. J. Med. 0:65-70. 8. Fliermns, C. B. 985. Ecologicl niche of Legionell pneumophil, p. 75-6. In S. M. Ktz (ed.), Legionellosis. CRC Press, Inc., Boc Rton, Fl. 9. Fliermns, C. B., W. B. Cherry, L. H. Orrison, S. J. Smith, D. L. Tison, nd S. H. Pope. 98. Ecologicl distribution of Legionell pneumophil. Appl. Environ. Microbiol. :9-6. 0. Fliermns, C. B., W. B. Cherry, L. H. Orrison, nd L. Thcker. 979. Isoltion of Legionell pneumophil from nonepidemicrelted qutic hbitts. Appl. Environ. Microbiol. 7: 9-.. Grbe, P. L., B. J. Dvis, J. S. Weisfeld, L. Mrkowitz, P. Miner, F. Grrity, J. M. Brbree, nd A. L. Reingod. 985. Nosocomil legionnires' disese: epidemiologic demonstrtion of cooling towers s source. J. Am. Med. Assoc. 5:5-5.. Gormn, G. W., J. M. Brbree, nd J. C. Feeley, 98. Procedures for the recovery of Legionell from wter. Developmentl mnul. Centers for Disese Control, Atlnt, G.. Mngione, E. J., R. S. Remis, K. A. Tit, H. B. McGee, G. W. Gormn, B. B. Wentworth, P. A. Bron, A. W. Hightower, J. M. Brbree, nd C. V. Broome. 985. An outbrek of Pontic fever relted to whirlpool use, Michign, 98. J. Am. Med. Assoc. 5:55-59.. Mrtin, W. T., J. M. Brbree, nd J. C. Feeley. 98. Detection nd quntittion of Legionell pneumophil by immune utordiogrphy, p. 99-00. In C. Thornsberry, A. Blows, J. C. Feeley, nd W. Jkubowski (ed.), Legionell: Proceedings of the nd Interntionl Symposium, Atlnt, Georgi. Americn Society for Microbiology, Wshington, D.C. 5. McKinney, R. M., W. L. Thcker, D. E. Wells, M. C. Wong, W. J. Jones, nd W. F. Bibb. 98. Monoclonl ntibodies to Legionell pneumophil serogroup : possible pplictions in dignostic tests nd epidemiologic studies. Zentrlbl. Bkteriol. Hyg. Abt. Orig. A 55:9-95. 6. Neill, M. A., G. W. Gormn, C. Gibert, A. Roussel, A. W. Hightower, R. M. McKinney, nd C. V. Broome. 985. Nosocomil legionellosis, Pris, Frnce: evidence for trnsmission by potble wter. Am. J. Med. 78:58-588. 7. Voss, L., K. S. Button, R. C. Lorenz, nd 0. H. Tuovinen. 986. Legionell contmintion in preopertionl tretment plnt. J. Am. Wter Works Assoc. 78:70-75. 8. Voss, L., K. S. Button, nd 0. H. Tuovinen. 985. Legionell in metropolitn wter distribution system. Environ. Tech. Lett. 6:9-8. Downloded from http://em.sm.org/ on October, 08 by guest