Document ZJowjom6b3J0L88rO2nBa1GJV

~~ ~~~ Current Microbiology An International Journal @ Springer-VerlagNew York Inc. 1994 Disease Outbreaks and CotAing Towers d Legionella Concentrations &$2. ,tr -.B$ZG. - Shelton,'.' W. Dana Flanders'l and George K. Morris2 .$?$2. lD,,,g,n o- f Epidemiology. School of Public Health, Emory University, Atlanta, Georgia; z&ti;Con Laboratories. Norcross. Georgia, USA stract. This study was designed to investigate the possible association of high colony counts of : .T legionellae from cooling towers and evaporative condensers with Legionnaires' disease outbreaks. We obtained legionellae counts from samples of cooling towers and evaporative condensers that **'- were the likely sources of hvo different Legionnaires' disease outbreaks and compared these counts _ _-,.with those from cooling towers that were not associated with reports of human disease. Among 675 potential control cooling tower water samples from 258 facilities, 136 facilities had one or more cooling towers that met our criteria for inclusion into the study. Samples taken from buildings where an outbreak had occurred had much higher Legionella counts than did samples from other . buildings. Colony counts from the two outbreak-associated facilities were significantly higher than ". '"'colonycounts from other facilities [Wilcoxon Rank Sum Test (Exact), p < 0.011. The results of the study suggest that, among cooling towers that test positive for the presence of Iegionellae, higher colony counts are associated with higher risk of Legionnaires' disease. Legionellosis, including both Legionnaires' disease and Pontiac fever, is caused by organisms of the genus Legionella. An estimated 3,000 to 100,000 cases of Legionnaires' disease occur in the United States annually [l].Most occur as sporadic cases, and the source of exposure often remains undetermined. Previous studies have shown cooling towers and/or evaporative condensers to be responsible for outbreaks [4, 91. Several investigators have suggested an increased risk of Legionnaires'disease from exposure to +-ntaminated building sources [2, 101; however, no relation with Legionella counts has been established. ?e significance of the presence of Legionella bacteria in building water systems in the absence of disease isoftendownplayedowing to the ubiquitous nature of `he Organism [4,6,9, 121. ,-.hblished investigations on outbreaks of k$!Onnaires' disease that have established the expoSUre'Source seldom include colony counts of Le@J?$bahcteria from the identified exposure source. At present, routine environmental culturing is not when sporadic cases occur because no asmiation has been developed between degree of Contaminationand disease [4, 6, 9, 121, and sporadic Communityacquired illnesses are usually not known .. to be associated with further cases. Thus there is a tendency to wait for a contaminated building water system to be associated with illness before any interventions are initiated [9,12]. Because a large percentage of cooling towers have been shown to contain Legionella [13], further study of the association between legionellae concentrations in cooling towers and Legionnaires' disease is warranted. The need for this information was recognized by Fraser in 1980. He called for studies to demonstrate the risk of legionellosis in relation to concentration of L. pneumophifa in cooling tower or evaporative condenser water [SI. This study is an initial step in the investigation of such an association. Materials and Methods In this study, we compared colony counts in samples taken from cooling towers (or evaporative condensers) that were implicated epidemiologically as the source of an outbreak of Legionnaires' disease with the colony counts in samples taken from cooling towers and evaporative condensers not associated with an outbreak. All outbreaks of Legionnaires' disease that occurred in the U.S. between 1988and 1991were eligible for entry into the study if the implicated source was a cooling tower or evaporative con- denser. We identified outbreaks from a listing of outbreaks Ad,d.re.%' reprint requests to: Brian G. Shelton, Pathcon Labs, 270 Scientific Drive, Suite 3, Norcross, GA 30092. USA. I I i 360 CURRENTMICROBI investigated by the Centers for Disease Control (CDC)between 1983 to 1991 [l]. We obtained legionellae counts from one of the outbreak-associated cooling towers from a publication describing -the investigation performed by the CDC in a peer-reviewed journal [4]. For the other outbreak, the legionellae counts from the implicated cooling towers were determined by our laboratory. Only those outbreaks with two or more cases of Legionnaires' disease were eligible, and only those outbreaks with patients that showed symptoms of pneumonia, radiographic evidence, and culturepositive sputum were included in the study. We excluded outbreaks that occurred from sources other than cooling t<lwersor evaporative condensers. We identified hvo outbreak that met these criteria. Outbreak description. Outbreak 1, investigated y our laboratory, occurred in a hospital, and the cooling towers :re implicated as the source. The outbreak strain was Legionella neumophilu sero- group 1, monoclonal subtype 1, 2,5,6. The OUi reak consisted of three cases that occurred from June, 1989, to hgust, 1989. All cases were hospitalized patients who were 1 :ated in a single building at the hospital. The cooling towers, wh rh were connected side by side, were a common exposure source ft all cases since all cases were on the same wing, on the same floor. nd the air intakes of the building were situated downwind from a r faced the cooling tower drift. The potable water system was also i. m m o n exposure source to all cases. However, no viable Legi~ *(lapnewnophila were detected from the potable water system af r three thorough investigations. The samples from two coolir towers yielded Legionellapneumophila serogroup 1. monoclona ubtype 1.2.5.6 and Legiodla pneumophilu serogroup 1, monoc nal subtype 1, 6 at Concentrations ranging from 120colony formir mits (CFU)/ml to 3600 C R l / m l (mean = 1917 CFLJ/rnl, n = 6 The high variation in colony counts resulted because one of thc wen had much higher counts than the other tower. Outbreak 2 was associated with an evapora: 2 condenser at a retirement hotel between June 10 and July . 1988 [JJ. The evaporative condexer was epidemiologically plicated as the source of the outbreak. which included six case 41. The association was also supported by microbiological sampling. The outbreak strain was Legionella pneumophilu SL group 1. monoclonal subtype 1, 2, 5, 6. The outbreak strain w: demonstrated in the evaporative condenser water at concentra ns greater than 9000 CFU/ml [J]. We ccmpared the colony counts in sa. des from these outhrcak-associated towers with counts in con )I samples from other cooling towers. All samples sent to ou laboratories for analysis from cooling towen or evaporative con nsen from 19SS to 1991 were eligible if the building was nc reported to be associated with an outbreak. Cooling tower water mples(n = 675) were analyed for Legionella concentrations fi II 258 buildings with no previously known association with Leg inaires' disease. Because of the similarity between cooling towe an d evaporative condensers with respect to design, operation, .id potential for exposure, evaporative condensers were con? x e d as cooling towers in this study. Cooling tower samples that sted negative for Legiunellu were excluded from the comparisor ;roup. If we had several samples from cooling towers or evap) ative condensers from the same building. we used the mean conci.:itration. Control water samples were collccted and Xcnt to the lahoratory, next day (AM) deliven. All samples were procrsscd on thc clay of arrival by the laboratory for viablc Lcyiottcllu. All samples. including the onc outhreak sample scnt to our I:ilwrator), were :\n;ilyzed in duplicntc. Snmplcs were analyzed with Buffered Cli:ircoal \'enst Extract agar (BCYE) with sclectivc mtibiotics and glycine as previously d e m i d IS]. In tained by filter concentration of 100 Nuclepore polycarbonate membrane fol ml of filtered sterile water for furth treated by the method of Bopp et al. only if a significant number of competing non-legio were detected. Typical colonies on BCYE agar wer tested for cysteine requirement. Cysteine analyzed for immunofluorescent reactions valent Legonella antisera to determine Monoclonal subtyping was perf0 and clinical isolates by the CDC by methods pr Ill]. We used an exact version of the Wilcoxon R test the cooling null hypothesis of towers associated no difference bemeen colony with an outbreak and t h m of othoe,r towers. We restricted this analysis to the wanner cooling tmer operational months of May. lune, July. August, September, and October. As well, frequency tables of the mean numben of Legionella in control cooling tower samples per faciliq were created. The mean numbers ofkgzbd/u fromthe contfol towen were compared with numbers in "outbreak" towen to study differences bemeen the two groups. For additional descriptive analysiswe constructed bistogrm to summarize the relative frequency of the mean CFU/ml from control facilities. We classified counts as "high" if the mean CFU/ml was greater than or equal to 1600. the upper 5 percentile of mean CFUlml from control building cooling tower samples. All counts less than 1600 uere considered "low." We assessed the association between outbreaks and "high" counts, using Fisher exact confidence limits. The upper 5 percentile was chosen arbi- trarily for purposes of anal!.su. not to suggest this as a guideline level. Results .- We processed 675 cooling tower samples from 258 facilities that met our criteria. We processed 12 control cooling tower samples from 8 facilities in 1988, 65 control samples from 21 facilities in 198% 190control samples from 71 facilitiesin 1990,and 408 samples from 15s facilities in 1991. We excluded 122 of these comparison facilities because the d 0 n Y counts were zero. Of the 132excluded facilities,8,119 30, and 73 were excluded in the years 1988, 1 9 8 9 7 1990, and 1991 respectively. None of these excluded facilities was known to be associated with an Out- break. The final comparison population was 2s6 cooling tower samples from 136facilities. AS shownin Fig. 1, mean colony counts from the comparison cooling towers were substantially lower than those from the outbreak-associated towers. By restriction of the analysis to the r cooling tower operational months, outbreak samples ranked firstand eighth out of 101total saxnpleS.uslng the Wilcoxon R3nk Sum Test (eat , q h niaires' Disease and Cooling Towers 361 ~ Control Facilites,ne138 0 Outbreak Fadlites. n=2 L- +L :y<* ' I:$%. .I >;&!$; ai lo 10 11 to 30 31 to 100 101 to 500 sol to lo00 1001 to 2Ooo G;P?. e W l O M l L (mUnCFWml) fig, 1. Distributionof mtony counts in positive facilities. 2001 to 4000 >a1 Table 1. Wilcoxon b.i 7:St (Exact), p-value = 0.0063 ~omllingfor cooling zk-tr b\Fntional months May rtirou& October m / m t per t a s ? Number of outbreak facilities Number of control facilities rOand <IO0 100-499 500-999 1000-1499 1500-2999 t 3000 0 68 0 18 05 00 15 13 samples had si--mf? higher colony counts than the control s m p j s p d u e = 0.0063. Table 1 shows wth,,ae.-`t*edrissatrmibpulteiso.nrofaLy~i -nella counts in cooling tower y toScooling tower operational months. oi ~kunrestricted data was also s1t!a:tistically si_gniiicuii I Legionella maxxzxions from each of the differ- e d - o u t b r e a k - e t k cooling towers were above 19W/d.tht q ~5 perrcentile of the compari- Fn group. T~LL cL?- Y of 136 facilities with no ?$break (10 of X psitive samples) had mean CaO?*l,O- ny counts eqaal m or above 1600 CFU/ml com- Pj?pd.with 2 of 1 -ties associated with an out- break. The odds &pas infinite with a 95% Fisher `Exact lower h i t of 2.6, again indicating that high colom- xuxs are very strongly associated with having had s outbreak (Fisher Exact P- "+e = 0.00s). r, I ?. I4*- Discussion The results of our study suggest a strong reiatiun between high Legionella counts in cooling towers with Legionnaires' disease. An extensive epidemiologc investigation to identify a common source was conducted for one of the outbreaks, as previously described (outbreak 2). Although an extensive epidemiologic investigation was not performed in outbreak 1, all three cases were located on the same side of the building that faced the cooling towers, the air in:skes faced the cooling towers, the cooling towers were the only source at the facility to contain the same Legionella subtype as the outbreak strain, and no further cases occurred once the cooling towers were decontaminated. This further suggests that the cooling towers determined to be the cause for outbreak 1 were the actual exposure source of the outbreak. ' As with other infectious organisms, our results suggest that the higher the concentration of Legionella in an exposure SGLIrce, the higher the risk of disease. Previous epidemiolcgic studies have demonstrated an association of Legionnaires' disease with both duration and proximity of expnsure to a legionelhe-contaminated source with Legionnaires' disease [4], yet none have attempted to associate the concentration of Legionella in the source of exposure. Our hypothesis-that higher colony counts are associated with higher risk-is biologically plausible. It is plausible to conclude that higher legionellae counts in cooling tower water would result in higher i 362 legionellae counts in air. The concentration of Le- counts may have preferentially Submi gionelfa in cooling towers should be a crude measure This would have biased the results away of potential for exposure, even though the distance and the true association would be less t from cooling towers to susceptible individuals may reported in this study. Alternatively, a vary. Although the infective dose of Legionella is not in the opposite direction, as there may known, if the infective dose required to cause tendency for facilities with indoo Legionnaires' disease is greater than one CFU, then complaints to preferentially sub concentration is important. Because infective dose would have biased the results tow presumably varies among individuals because of dif- the true association would be eve ferences in susceptibility, the number receiving an one described in this study. A pote infective dose shouirl increase as the concentration in fication bias also exists, since we an exposure source iricreases. Therefore, it is plau- results from two different laborat sible that the risk shou!d be higher among those if the laboratory that analyzed the comp&n samples exposed to sources with higher colony counts. and outbreak 1 tended to record lower counts than In reviewing the literature, we identified addi- the laboratory that analyzed outbreak 2, then the true tional outbreaks that did not meet our eligibility odds ratio would be lower than the odds ratio criteria. These outbreaks were excluded because they reported in this study. On the other hand, if the occurred outside the United States and/or were laboratory that analyzed the comparison samplesand Pontiac fever rather than Legionnaires' disease out- outbreak 1 tended to record higher counts than the breaks and/or were unpublished in the scientific laboratory that analyzed outbreak 2, then the true literature and/or occurred in a year when control odds ratio would be higher than the odds ratio samples were not collected. In general, colony counts reported in this study. from these outbreaks were high and consistent with Some potential limitations of our study include our results. For example, colony counts from highly (i) the small numbers of outbreaks included, (ii) suspected cooling tower samples from three different failure to study other factors that may predict disease outbreaks in Australia were high. The mean levels risk such as the virulence of the organism, the from the cooling towers associated wi.h the Daw distance from the target person to the source, and the Park, Wollongong, and Bumie Tasman 3 outbreaks susceptibility and immune status of the target person, were 5500 CFU/ml (Dr. Trever Steele, 1'92, personal (iii) concentrations from the anal>zed sample may communication) (n = 2; 1000 CFU/m! and 10,000 not represent concentrations in the source at the time CFU/ml), 2000 CFU/ml [7], and 280.300 CFU/ml of the outbreak, (iv) lack of randomly sampled (Dr. Trever Steele, 1992, personal communication) comparison facilities, (v) not all outbreaks were respectibely. An outbreak of Pontiac fever in the U.S. investigated, and the total number of outbreaks that was associated with a cooling tower with colony occurred in the U.S.is unknown, and (vi) samples counts greater than 10,000 CFU/ml (Dr. James from one of the outbreaks [4]could have been Barbaree, 1992, personal communication). An out- handled differently from other samples. These limita- break associated with a cooling tower thzt occurred in tions could not be addressed in this study because of a U.S. prison, however, yielded only 10 CFU/ml (Dr. its retrospective nature, but would be important to Barry Fields, 1992. personal communic,ition), and a cooling tower associated with an outbr( ak currently under investigation yielded a Legionella count of 1500 CFU/ml in the laboratories at the ZDC (CDC Memorandum Epi-93-64-1, March 22, 1992), and in consider in subsequent studies. Despite these h i t a tions, the results of our study suggest that Cooling towers with very high colony counts may be Substantially more likely to be the source of an outbreak than cooling towers with lower counts. our laboratory it yielded a Legionella count of 2800 CFU/ml (mean = 2150 CFUIml). A cooling tower ACKNOWLEDGMENTS outbreak that occurred in 1983 had colony counts of 6000 Legioneffa CFUIml [9]. These outbreaks were not included in the analysis, but appear to support We thank Drs. Philip Brachman, Eugene GangaroSa, and John Boring, Division of Epidemiolog. School of Public Health, Emory University for their valuable comments on this manuscript.Wea'@ o u r findings. Although only two outbreaks met our thank Georse W. Gorman for his help in processing samples. inclusion criteria, the result is quite stable. Therc is a potential for selection bias in the Literature Cited control group. as these were essentially volunteer samplcs. For example. facilities with low legionellae 1 B.irbarce Jhl (1991) Controlling L q o n e h in cooling towers` ASIIRAE Journal 33.35-1` et 31.: Legionnaires' Disease and Cooling Towers 363 u VL, Stout J, Goetz A, Muder RR, Taylor F (1953) aceae in the hospital water supply. Epidemiologic disease and evaluation of a method for control of ial Legionnaires' disease and Pittsburgh pneumonia. .i.r&,p.a q-CA, Summer JW,Morris GK, Wells JG (1951) Isolation + fof,r&mella spp. from environmental water samples by low-pH .*`.treatment and use of a selective medium. J Clin Microbiol -tf' 13:714-719 4 Brehan RF, Cozen W,Fields BS, hfastro TD, Carr SJ, Spika JS,Mascola L (1990) Role of air sampling in investigation of an outbreak of Legionnaires' disease associated with exposure to aerosols from an evaporative condenser. J Infect Dis 161:12571261 5. e n t e n for Disease Control (19%) Procedures for the recove y of Legionellu from water by Gonnan GW, Barbaree JM, Feeley JC. Dev Manual, CDC Publication, Atlanta, GA 6. e n t e r s for Disease Control (1955) Lcgionellosis-Staffordshin,England. and Wayne Count)., htichigan. MMWR 34:34350 7. Christopher PJ, Noonan LM, Chiew R (1987) Epidemic of Legionnaires' disease in Wollon_eong.bled J Aust 147:117-1'75 3. Fraser DW (1980) Legionellosk: evidence of airborne transmission. Ann NY Acad Sci 353:61-66 9. Garbe PL, Davis BJ, Weisfeld JS, Markowitz L, Miner P, Garrity F, Barbaree JM, Reingold AL (1985) Nosocomial Legionnaires' disease: epidemiologic demonstration of cooling towers as a source. J Am Med Assoc 254521-524 10. Helms CM, Massanari RM, Zeitler R, Streed S, Gilchrist MJR, Hall N, Hausler WJ, Sywassink J, Johnson W, Wintermeyer L, Hierholzer WJ (1953) Legionnaires' disease associated with a hospital water system: a cluster of 24 nosocomial cases. Ann Intern Med 99:172-178 11. Joly JR, McKinney RM, Tobin JO, Bibb WF, Watkins ID, Ramsay D (1986) Development of a standardized subgrouping scheme for Legionella pnelonophila serogroup 1 using monoclonal antibodies. J Clin Microbiol23:765-771 12. Redd SC,Cohen ML (1987) Legionellu in water: what should be done? J Am Med Assoc 257:1221-1222 13. Shelton BG, Morns GK,Gorman GW (1993) Reducing risks associated with Legionella bacteria in building water systems. In Barbaree JM, Breiman RF, Dufour AP (eds) Legiorlella: current status and emerging perspectives. Washington: American Society for Microbiology, pp 279-251