-lrh /Jen'nniaf !ntemarionul ..... mac llllc. c ccoratne ountams. am \\ 11r poo spas. ... AspJratJOn ol co onJ; ...... IN) for 18 hours at 50°C. A I x5mm pi ug sl icc was.
Preventing Legionnaires' disease Academisch Proefschrift ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam, op gezag van de Rector Magnificus Prof Dr. J.J.M. Franse ten overstaan van een door het college voor promoties ingestelde commissie, in het openbaar te verdedigen in de Aula der Universiteit op woensdag I 0 mei 2000, te 14.00 uur door
Jacobus Leen Kool geboren te Amsterdam.
Promotiecommissie Promotor
Prof. Dr. R.A. Coutinho.
Universiteit van Amsterdam
Co-promotor
Dr. J.C. Butler,
Arctic Investigations Program. CDC. Anchorage, Alaska
Overige !eden
Prof. Dr. 1. Dankert
Universiteit van Amsterdam
Prof. Dr. Jr. D. Kromhout Universiteit van Wageningen: en RIVM. Bilthoven Dr. J.F.P. Schellekens
RIVM. Bilthoven
Prof. Dr. P. Speelman
Universiteit van Amsterdam
Prof. Dr. J.G.P. Tijssen
U niversiteit van Amsterdam
The research presented in this thesis was conducted v.·hile the author was Epidemic Intelligence Service Officer at the Respiratory Diseases Branch. Division of Bacterial and Mycotic Diseases. National Center for Infectious Diseases. Centers for Disease Control and Prevention. Atlanta, Georgia. United States of America.
2
Some say it was radiation, some say there was acid on the microphone, Some say a combination that turned their hearts to stone, But whatever it was, it drove them to their knees. Oh, Legionnaire Is disease.
I wish I had a dollar for everyone that died within that year, Got I em hot by the collar, plenty an old maid shed a tear, Now within my heart, it sure put on a squeeze. Oh, that Legionnaire's disease. Granddad fought in a revolutionary war, father in the War of 1812, Uncle fought in Vietnam and then he fought a war all by himself, But whatever it was, it came out of the trees. Oh, that Legionnaire Is disease.
Bob Dylan: Legionnaire's disease (1978)
3
Table of contents Chapter I :
General introduction ................................................................................ 7
Chapter 2:
Part 1: Communih'-acguired Legionnaires' disease Outbreak of Legionnaires' disease at a bar after basement flooding ...... 31 The Lancet 1998;351: 1030
Chapter 3:
Strengths and limitations of molecular subtyping in a community outbreak of Legionnaires' disease ......................................................... 45
Suhmittedfor publication
Chapter 4:
Part II: Hospital-acquired Legionnaires' disease More than ten years of unrecognized nosocomial transmission of Legionnaires' disease among transplant patients ................................... 65 Infection Control and Ho.spital Epidemiologl' 1998; 19: 898-904 Addendum to Chapter 4 ..........
Chapter 5:
00 . . . . 00 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •
82
Hospital characteristics associated with colonization of water systems by
Legion ella and risk of nosocomial Legionnaires' disease:a cohort study of 15 hospitals ....................................................................................... 83
Infection Control and Hospital Epidemiolog\· 1999; 20: 798-805
Chapter 6:
Part III: Drinking water disinfection and Legionnaires' disease Effect of monochloramine disinfection of municipal drinking water on risk of nosocomial Legionnaires' disease ............................................. 10 I The Lancet 1999; 353: 272-7 Addendum to Chapter 6 ....................................................................... 118
Chapter 7:
General discussion ............................................................................... 119
Legionella in The Netherlands ............................................................. 125 Summary (Eng Iish) .............................................................................. 134 Samenvatting ( nederlands ): Yeteranenziekte voork6men .................... 13 7 Acknowledgments ............................................................................... 141 c·urriculum vitae .................................................................................. 142 Bibliography ...................................
00 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •
143
5
General introduction
Chapter 1: General introduction Legionellosis Infection with bacteria of the genus Legionella can cause two distinct clinical syndromes, grouped together under the name legionellosis. The syndrome that will be the subject of this thesis is Legionnaires' disease, a severe and potentially fatal form of pneumonia. The second is Pontiac fever. which will be discussed in short hereunder. Pontiac fever is a self-limited influenza-like illness. It is characterized by sudden onset of fever (39°-40.5° C I 102°-l 05° F). malaise. myalgia. headache, and a non-productive cough. but not pneumonia. It usually occurs in explosive outbreaks in which >90% of exposed individuals develop the symptoms. The incubation time is usually 24 to 48 hours and patients recover spontaneously within 2-5 days without treatment. Mortality due to Pontiac fever is extremely rare. Legionella is believed to be the causative organism because Pontiac fever patients show a rise in antibodies to L
pneumophila. The pathogenesis of Pontiac fever is not well understood. Viable Legionella organisms have not been recovered from Pontiac fever patients. Some researchers believe that it is not an infection but rather a febrile hypersensitivity reaction to inhaled antigens from amoebae and/or legionellae.1.2
Others have
suggested that it may be an infection with legionellae or Legionella-like organisms that are unable to multiply in human host cells.
3
Legionnaires' disease Legionnaires' disease was first recognized during the 1976 annual convention of the American Legion (an organization of veteran soldiers) in Philadelphia. In the weeks following the convention, 221 persons became ill and 34 died of a previously unknown disease. All victims had either stayed at the downtown Bellevue-Stratford Hotel, had spent time in the lobby, or had been within a distance of one block from the 4
hotel. Media around the world paid attention to this mysterious disease. After five months of intensive investigation researchers recovered the bacteria from guinea pigs inoculated with lung tissue from patients who had died during the outbreak. 5 Legionella does not grow on standard growth media and is not visible on Gram stain; therefore it had not been identified as a pathogen before. Special growth media were later developed (among others, BCYE-a: buffered charcoal-yeast extract agar enriched with a-ketoglutarate).
Examination of stored clinical specimens showed that
7
Chopter I
Legionellu infec ti on has occ urred at least ~ in ce 19-+3. in sporad ic case~ a~ ''ell a~ in o utbrea ks of L eg i o nn a ire~· disea~e. 6 ·-·' Exam in at ion of ~tored :-,era a l ~o re\'ea led that wa~ a~~ocia t ed
with an outbreak of influenza-like illn e\s in a health 9 depa rtm ent in Ponti ac. Michigan. in 1968. wh ich had been named Pon ti ac fe\ ·er Legion ello
Reservoir and route of infection Legionellu i ~ a ubiquit o u ~ orga ni sm th at li\'e'> as an Intrace llul ar parasite of amoe bae in aquati c e1wironments (Figures I and 2). 111 ' 11 It is found in low co nce ntrati om in ri vers. lakes. and ~o il arou nd the ,,_,·orld and as ~ u c h probab ly ca u se~
ve ry lilli e disease. Man-made techn ology. howe\·er. can turn it int o a dangerou'> pathogen. It thri \·es in water of hi gh temperature (77' - 108 F I 25 °--+2° C) and it can co lon ize the biofilm that co, · e r ~ th e in side of tanks and pipe~. Ses~ile ("fixed") biofilm bacteria suc h as Legionellu and Pseudomonas filter nutrients out of large amou nt'> of pass in g water and thereby ca n survive in low -nutri ent environm ents lik e potable wa ter. Drinking water disinfectants such as free ch lorine penetrate poorl y into
Figure 1: Protozoa filled with Legionella. Courtesy B. Fields. CDC.
8
Cr.'nl!rul inrroduct ion
bi ofilm 12 and Legionello i-, furth er ~hi e l ded off b) th e amoebae th at it pa ra~ i t i '>e, Free chl orine leveb in muni cipa l drinl-;ing water are ge nerall y ;, uffic ient to neutraliLe plankt oni c (free-fl oa ting) co lifo rm bac teri a. but are oft en too lo"" to kill Legionl!!!u li \·in g in bi ofilm. In additi on. many drinkin g wa ter di , infec tant ;, ;, uch as free chl orin e do not reac h di ;,tal ;, ites in a water di ;,tributi on ,y;, tem. ca n di ;,;,ipa te qui ckl y in hea ted water. and are often re mo\·ed durin g \\'ater filterin g ;, uch a;, occ ur;, in whirl poo l bath ,. Wh en a su;,ce ptibl e pe r,o n inh ales a co nt aminated ae ro;,o l consiq in g of 11
dropl et;, of the ri ght size to reac h th e al\'eo li ( I to 5 mi cron) he or ' he ca n de\'e lop th e di ;,ea;,e. 14 Legionellu infec t>. hum an al\·eo lar mac roph age;, in th e ;,a me way it infech amoe bae. 15 Ae roso l dropl eh can co ntain amoeba! \'e;, ic le;, fi li ed with Legionello and thi s may co ntribute to more di;, tant di ;,;,e min ati on and to hi gher ill\·as i\'e ness of th e bac teri a. 16 ' 17 Th e co mbin ati on o f hi gh te mperature and potenti al fo r aeroso l form ati on ca n be 14 found in. for exa mple. air co nditi onin g coo ling towe rs (Fi gure 3 ). showe rs. fauce ts. air humidi fier;,. respiratory therapy equipment. and \\·hirlpoo l bath s: all these de\·ices
Fig ure 2: Electron microscope picture of amoeba opsonizing L pneumophi/a. Courtesy B. Fields, CDC .
9
Chapter I
Aerosol
!
' ' ' ' '
Fan
A1r Handling
Dnft - - Ehmmator
!
....
!
,,
!
AT
J
!
Unit
.....
' - Air 4... F¥ w ~
'
' ! ~
~
,...' ' ' ' ' Water
Water (80 F)
A1r
- - Ftll
.....
!
~
Wa ter (95 F)
~
~ ~
-
Aefngerant
Nozzles
''
to buildtng
Heated
-.......... spray
!
Cooled Atr
!
-....
I
Chtlled Refngerant
Condenser
(hea t source)
I
! !
' '
.....
Figure 3: Anatomy of a cooling towe r. A cooling tower is the part of an air co nditioning system that is usually located outside a building . It exchanges heat by evaporating sprayed water; aerosol is created in the process . The water in the sump often has a temperature that is favorable for amplification of Legionella. Courtesy J. Butler, CDC .
have frequentl y been desc ribed as the so urce of Leg ionn aires' di sease o utbreaks. Additionally , anecdotal reports suggest th at other aeroso l producin g dev ices, suc h as medicati on nebulizers. an ultraso ni c mi stin g machine. an e lec trical s ump pump. and a . d'1sease. IR-OJ carpet c Ieaner ca n transmtt (Mi cro-) aspiration of water contaminated w ith Legionella ha s also been described. albeit less frequ e ntly , as a route of tran smi ss ion.
Thi s may occur
predominantly in perso ns with swa ll ow in g disorders or in conjun cti o n w ith nasogastri c feed ing. n
Person-to-perso n transmission of Leg ionn aires' di sease has never been
documented and co lon ization of th e nasophary nx probabl y does not occ ur e ither. At the tim e of writing of thi s thes is, 43 Legionella spec ies a nd 65 serogro ups were de sc ribed.
'·"-l
2
In add iti o n many Legionella- like amoeba! pathogen s (LLAPs)
have been identifi ed . so me of whi ch are capab le of infecting human s. LLAPs do not grow on stand ard Legionella medi a such as BCYE-a; they can o nl y be cultured in g rowth media co nta inin g li ve amoebae.
25 26 '
Legionel/a pneumoph ila ca used the
maj ority (9 1o/c ) of reported Leg io nnaires' di sease cases in the United States; L pneumophila serogro up l was found in 7 1o/c of fully identifi ed c linica l isolates* of
In realitY the proportion oflegionellosis cases that are caused br L pneumophilo serogroup I mar be someH·hatloH·er because gro tt·th media and diagnostic techniques are optimi::edfor this strain. resulting in a higher rate of detection than for other species and serogroups. 111 '
10
General introduction
cases reported from 1980 to 1989 to the Centers for Disease Control and Prevention (CDC). 27
Epidemiology A common misconception
IS
that Legionnaires, disease is a rare disease.
Studies that have tried to estimate the incidence of community-acquired Legionnaires, disease found that Legionella caused 2% to 16% of community-acquired pneumonia cases in industrialized countries (Figure 4).~x- 39 This makes Legionella the second- to fourth-most common cause of community-acquired pneumonia.*
Most of the
aforementioned studies were hospital-based and therefore tended to detect only the more severe cases of pneumonia that required hospitalization: little data is available on the incidence of less severe Legionnaires' disease. Prospective studies for evaluating the cause of community-acquired pneumonia invariably failed to identify the cause of 40°/o-60% of cases. 30
Therefore, estimates of Legionnaires, disease incidence are
likely to go up as more pathogenic strains of Legionella and Legionella-!ike organisms continue to be discovered.~ 526 .4° Less than 5% of legionellosis cases are eventually reported to public health authorities through passive surveillance; most are probably never diagnosed.
27
.4
1
Of
the cases that were reported to the Centers for Disease Control and Prevention (CDC) from 1980 to 1989, 25% died as a result of their Legionella infection.
27
The overall
fatality rate among all cases of Legionnaires' disease may be lower because severe or fatal cases are subject to more aggressive diagnostic testing and thus are more likely to be identified and reported. Although outbreaks of Legionnaires' disease have received a lot of attention, the majority (80% - 90%) of Legionnaires' disease cases are sporadic, meaning that they cannot be linked to other cases or to outbreaks.
Research has shown that the
sources of outbreaks and sporadic cases are similar. These sources are often hot water systems in hotels, homes, and hospitals -t 2-44; cooling towers 45 A 6 ; and whirlpool baths. 47 There are no data on the proportion of sporadic disease attributable to various sources.
4g
* The most./fequent~r ident(/ied cause o(community-acquired pneumonia in most studies H'as Streptococcus pneumoniae (usually ident(/ied in -15%-20%). Second place is (dien shared hy Mycoplasma pneumoniae. Legionella. Chlamydia pneunwniae. and (depending on local epidemiolog~· and vaccination practices) Haemophilus injluen::ae, each identified in -5%10% (~lCAP in many countries. lftestedfor. ~·ira/ etiologies H·ere qftenfound in a similar proportion o(cases.
II
Clwp1er I
Attack rates durin g ou tbrea ks of Leg ionn ai res' di sease are low: less th an 5° o of exposed perso ns de\c lop the disease. This like ly rd l ects di ffe rences in host susce pti bility. Import ant predi spos in g host fac tors are: smokin g, old age . male ge nder. high-d ose co rti costeroid medi ca ti on. chro ni c obstru cti\·e pulm onary di sease. immun osupprcss i\ e medi ca ti on as is used in conj un cti on with orga n or bone marrow tra nsp lant at ion. and oth er immun oco mpromi sin g cond iti o n s.~ 9 The in cuba ti on period of Leg ionna ires· disease is usuall y betwee n 2 days and I 0 days alth ough longe r peri ods have bee n re po rt e d . ~ Th e \\·ide geographi ca l \a ri at ion in in cid ence of Leg ionn aires ' di sease probabl y reflec ts diffe rences in epidemi o logy and qu alit y o f lab tec hniques as we ll as geographi ca l \'ariat ion in th e occ urrence of Legionelfa. Until rece ntl y thi s va ri ati on \\·as large ly attri buted to eco logica l fac tors such as clim ate, density of \\ ater-bascd coo lin g dev ices. and natural wa ter co mpos iti on. Muni cipal drinkin g water di sinfec ti on tec hnique \\·as recentl y fo und to hmc a signi fica nt impac t on occ u!Tencc of 1
Unknown Legionella
Mycoplasma Haemophilu Staphylococ
Figure 4: Etiological agents identified in several prospective studies of communityacquired pneumonia requiring hospitalization. Courtesy J. Butler, CDC.
12
General imroduclion
Legionnaires· disease (this thesis).'''·~~
Community-acquired Legionnaires' disease 7Y'Io of cases reported to the CDC arc community-acquired.~~ The true proportion of community-acquired Legionnaires' disease may be substantially higher because nosocomial cases arc more likely to be detected and subsequently reported to public health officials. Sporadic cases and outbreaks have been caused by sources such as air-conditioning cooling towers. evaporative condensers. whirlpools baths. home hot water systems. and hotel hot water systems. Hot water systems of larger buildings such as hotels arc at increased risk of Legionf!lla colonization ..~~ A large proportion of community-acquired cases appear to h:1\'c acquired their infection in association with overnight travel.' 4· " One explanation for this ob\ iously is the higher risk associated with hotel stays but one study also indicated that persons who tr< Q)
0
3 Legionella
2
growth possible
c::
0
J
~ I...
::J
c
1 min at 12JGF 30 sec at 130°F 5 sec at 2 sec at 1 sec at 140°F 150°F 158°F ~
lt
0
110
120
130
140
150
160
Water temperature (°F) Figure 6: Duration of exposure to hot water to cause full-thickness epidermal burns at different temperatures n
18
General introduction
have reported that corrosion could be controlled with addition of silicates. which would form a protective "coating".
°
8
Continuous supplemental chlorination of a
hospital water system for a duration of 10 years in combination with silicate for 113 corrosion control was reported to cost $334,994; an average cost of$33,500 per year. Copper-silver ionization has shown promising results in some hospitals.x-u.::; This method may be in use in approximately 30 hospitals in the US. 73 Electrodes made of a copper-silver alloy are installed in the water system and a controlled electrical current is run through them, resulting in electrolytically generated copper and silver ions. Some doubts about the effectiveness of this method remain. The first study on the effectiveness of copper-silver for Legionella control found that it was only effective in conjunction with a concentration of free chlorine of at least 0.4 mg/L. so It is conceivable that, given the right circumstances, this concentration of free chlorine could by itself have been adequate for control of Legionella. Subsequent publications reporting efficacy of copper-silver systems failed to report chlorine concentrations. A dose-effect relationship between copper-silver levels and Legionella 85 inhibition does not appear to exist, suggesting that the disinfecting action of coppersilver is not straightforward. Some hospitals have continued to recover Legionella or have identified ongoing transmission of Legionnaires disease while using copper. . . 61 ·sq7 . that L egwne . IIa may deve Iop a to Ierance s1'! ver 1omzat10n. -· ·88 A nother concern 1s 89
to silver, as was reported from a long-term follow-up study. The author of this thesis has the impression that copper-silver ionization in combination with supplemental injection of free chlorine to a concentration of least 0.5 mg/L can be effective but that use of copper-silver alone may often lead to failure. An advantage of copper-silver is that it is not corrosive to plumbing materials. A disadvantage is that it is expensive. In addition to the initial costs for purchase and installation of the equipment, these systems are costly to maintain and it is difficult and labor-intensive to regulate levels of copper and silver within the narrow range between minimal effective dose and maximum acceptable levels set by the Environmental Protection Agency. The cost of copper-silver ionization is usually reported to range between $30,000 and $200,000, but one 300-bed hospital, which experienced an outbreak that 0 will be discussed in this thesis/ later installed a copper-silver system at a cost of approximately $1 million in the first year. Legionella was successfully controlled in that hospital with the combination of copper-silver ionization and supplemental injection of free chlorine to a concentration of 0.5 mg/L (see addendum to Chapter 4 ). Other methods such as ultraviolet light, ozone, chlorine dioxide, potassium permanganate and the intermittent superheat-and-flush method are infrequently used
19
( lw;J!cr I hnspit~tl
and \\ ith \ ariahk rcsulh ..\
in Pittsburgh used thL·
Hush mdhod unsLH.:ccssfully for l .-;
year~:
111tcrmitt~nt
superheat-and-
an a\ cragc of 6 nosocomial casl's of
L:gionnaircs· disease continued to occur annually and Lcgioncl/u \\·as reccn ered hom distal\\ atcr sy~tcm sites during the entire p~riod.'·'
Diagnosis initial!~
It \\as
thnught that Legionnaires·
di~eumophila serogroup 1. for which only monovalent antiserum should he used. Because of all these drawbacks. Legionella serology has \cry limited ,-alue in clinical practice. but it can be useful for retrospective diagnosis and for epidemiological im·estigations. Direct fluorescent antibodies ( DFA J for L(_)gionella detection. also called immunofluorescence microscopy·. used to he the only rapid test a\ailablc. In the hands of an experienced microscopist it can ha\ e a sensiti\ ity of up to 75° o when used on
21
Chapter I sputum and up to 90% when used on lung biopsy specimens.
Sensitivity and
specificity can become very low, however, when inexperienced laboratory personnel do the test or if the test is used for detection of other species than L pneumophila. PCR testing of clinical specimens has been done experimentally with reports of varying sensitivity and good specificity. 103 - 105
Theoretically it could give a high
proportion of false-positive results since it can detect live Legionella as well as DNA fragments of dead bacteria (e.g. partly digested bacteria or bacteria that are inhaled after being killed by disinfectants). 3 DNA probe methods have a sensitivity and specificity similar to that of DFA. They have the theoretical advantage that they can detect infections caused by all Legionella species. Otherwise the probe test has no particular advantages over other 49
available tests. None of the available diagnostic tests are perfect and the diagnosis of Legionnaires' disease should not be ruled out on the basis of a negative result of one or more of these tests.
Treatment After the first recognized Legionnaires' disease outbreak, erythromycin became the de-facto standard of therapy. During that outbreak it was noted that ~-lactam antibiotics resulted in a poor outcome; patients treated with erythromycin showed the best overall survival. Doxycycline, trimethoprim-sulfamethoxazole, azithromycin, clarithromycin, ciprofloxacin, levofloxacin and otloxacin also have activity against
Legionella. However, very few, if any, controlled trials of antimicrobial therapy have been done. Case reports and studies with animal models suggest that azithromycin may be more effective for treatment of severe Legionnaires' disease, and it generally has less side effects and drug interactions than erythromycin. In very severe cases rifampin may be beneficial if added to a macrolide during the first few days of treatment. Recent publications have reported very good results with the newer . 106 . Iones sue h as Ievo fl oxacm. fl uoroqumo Delay of appropriate therapy results in a poor outcome. 107
Therefore it is
important that rapid tests for Legionella are requested whenever the diagnosis of Legionnaires' disease is a possibility. Empiric treatment of pneumonia should include an antimicrobial agent with activity against Legionella.
Recent guidelines from the
[nfectious Diseases Society of America for the treatment of community-acquired pneumonia recommend empiric treatment with an antibiotic such as azithromycin or levofloxacin, which have activity against Legionella, Chlamydia, Mycoplasma, as well
22
General introduction
as S pneumoniae.30 These guidelines also recommend that empmc pneumonia case management should not be based on the presumed likely etiological agent, but rather on age, severity and comorbidity.
Scope of this thesis The author worked as Epidemic Intelligence Service Officer at the CDC's Respiratory Diseases Branch, Division of Bacterial and Mycotic Diseases, National Center of Infectious Diseases, from 1996 to 1998, where he investigated four Legionnaires' disease outbreaks. The following chapters will present several useful observations that were made during these investigations as well as a case-control study of the effect of drinking water disinfection on incidence of Legionnaires' disease, which was set up by the author as a result of some of these observations. Optimal control policies will be discussed in the General Discussion, with emphasis on prevention of potable water-associated Legionnaires' disease and on implications of newly available diagnostic tests.
References l. 2.
3. 4. 5.
6. 7. 8. 9.
10. II.
Rowbotham TJ. Pontiac fever explained? Lancet 1980; 2: 969. Fields BS, Barbaree JM, Sanden GN, Morrill WE. Virulence of a Legionella anisa strain associated with Pontiac fever: an evaluation using protozoan, cell culture, and guinea pig models. Infect lmmun 1990; 58: 3139-42. Miller LA, Beebe JL, Butler JC. et al. Use of polymerase chain reaction in an epidemiologic investigation of Pontiac fever. J Infect Dis 1993; 168: 769-72. Fraser DW, Tsai TR, Orenstein W, et al. Legionnaires' disease: description of an epidemic of pneumonia. N Eng! J Med 1977; 297: 1189-97. McDade JE, Shepard CC, Fraser DW, Tsai TR. Redus MA. Dowdle WR. Legionnaires' disease: isolation of a bacterium and demonstration of its role in other respiratory disease. l'i Eng/ J Med 1977; 297: 1197-203. Tatlock H. Clarification of the cause of Fort Bragg fever (pretibial fever)- January 1982. Rev Infect Dis 1982; 4: 157-8. Osterholm MT, Chm TD, Osborne DO, et al. A 1957 outbreak of Legionnaires' disease associated with a meat packing plant. Am J Epidemio/1983; 117: 60-7. Thacker SB, Bennett JV, Tsai TF, et al. An outbreak in 1965 of severe respiratory illness caused by the Legionnaires' disease bacterium. J Infect Dis 1978; 138: 512-9. Glick TH, Gregg MB, Berman B, Mallison G, Rhodes WW. Jr.. Kassanoff I. Pontiac fever. An epidemic of unknown etiology in a health department: I. Clinical and epidemiologic aspects. Am J Epidemio/1978; 107: 149-60. Rowbotham TJ. Isolation of Legionella pneumophila from clinical specimens via amoebae, and the interaction of those and other isolates with amoebae. J Clin Patho/1983; 36: 978-86. Fields BS, Sanden GN, Barbaree JM, et al. Intracellular multiplication of Legionella pneumophila in amoebae isolated from hospital hot water tanks. Curr Microhiol 1989; 18: 131-7.
23
ChapTer 1 12. I J.
1-J..
15. 16. 17.
1X. 19.
20.
21. 22.
23.
24. 25.
26.
27. 2X.
29.
JO.
31.
24
Chen X. StC\\JI1 PS. Chlonnc penetration 1nto and!cial biutilm 1~ limited bv a reactlonJiffu~ion interaction. 1:"111 imn Sci Tc( luwl 1996; 30: 207X-KJ. Kih ington S. Price J Sun i\·al of f.eginnel!u fi!ICIIIII!!Jihilu \\·ithin cy-;ts of Acuntlwnwchu jinhjilwgu folkm ing chlorine e\pChure. J .lpj!/ Hucrerio! 1990; 68: 519-25. Bre11nan RF. Cozen \\. Field~ BS. et al. Role of air ~ampl111g in im e~t1gation of an outbreak of lcg10nn~ure~· d1~Chirlpool spas on cruise ships. Atlanta: Centers for Disease Control and Prevention. 1995. Department of Health and Social Secunty and the Welsh Office. The control of f.egionellae in health care premises. A code of practice. London: HMSO. I 989. Health & Safety Executive. The control of legionellosis including Legionnaires' disease. London: Health & Safety Executive. 1991. Anonymous. Guidelines for prevention of nosocomial pneumonia. MMH·'R Morh :Horta/ ~f'k(r Ri:'p 1997: 46: 1-79. Muder RR, Yu VL. McClure JK. Kroboth FJ. Kommos SD, Lumish RM. Nosocomial Legionnaires' disease uncovered in a prospective pneumonia study. JAA1A 1983; 249: 3!84-8. Fiore AE. Butler JC. Emori TG. Gaynes RP. A survey of methods used to detect nosocomial legionellosis among participants in the National Nosocomial Infections Sun·eillance System. Infect Control /Imp Epidemio/ 1999: 20: 412-6. Joseph C A, Watson JM, Harrison TG, Bartlett CL. Nosocomial Legionnaires' disease in England and Wales. 1980-92. Epidl:'miolfn/i!ct 1994; 112: 329-45. Mandel AS. Sprauer MA. Sniadack DH, Ostroff SM. State regulation of hospital water temperature. In(ect Control Hosp Epidemioll993; 14: 642-5. Marrie T, Green P, Burbridge S, et al. Legionellaceae in the potable water of Nova Scotia hospitals and Halifax residences. Epidemiolfn/f!ct 1994: 112: 143-50. Arnow PM, Chou T, Wei! D, Shapiro EN. Kretzschmar C. Nosocomial Legionnatres' disease caused by aerosolized tap water from respiratory devices. J Inlect Dis 1982; 146: 460-7. Graman PS, Quinlan GA, Rank JA. Nosocomial legionellosis traced to a contaminated ice machine. fn(ect Control!losp Epidemiol 1997: 18: 637-40. Venezia RA, Agresta MD. Hanley EM. Urquhart K. Schoonmaker D. Nosocomial legionellosis associated with aspiration of nasogastric feedings diluted in tap water. lnfi!Ct Control Hosp t'pidemiol 1994; 15: 529-33. Goetz A. Yu VL. Screenmg for nosocomial legionellosis by culture of the water supply and targeting of high-risk patients for specialized laboratory testing. Am J Inlecr Conrrol 1991: 19: 63-6. Goetz AM. Stout JE, Jacobs SL, et al. Nosocomial legionnaires' disease discovered in community hospitals following cultures of the w·ater system: seek and ye shall find. Am J fn(ect Control 1998: 26: X-I I. Yu VL, Beam TR, Jr., Lumish RM, et a!. Routine culturing for Legionella in the hospital environment may be a good idea: a three-hospital prospective study. Am .J Med Sci 1987; 294: 97-9. Boulanger CA. Edelstein PH. Precision and accuracy of recovery of Legionella pneumophila from seeded tap water by filtration and centrifugation. Appl Environ Microhiol 1995; 61: 1805-9. Bezanson G. Burbridge S. Haldane D. Yoell C. Marrie T. Diverse populations of Legionella pneumophila present in the water of geographically clustered institutions served hy the same water reservoir. J Clin Microhiol 1992; 30: 570-6. Lin YS. Stout JE. Yu VL. Vidic RD. Disinfection of water distribution systems for f.egionella. Seminars in Respiratmy Infections 1998: 13: 14 7-59.
General introduction 74.
75.
76. 77. 78. 79.
80. 81. 82.
83.
84.
85.
86.
87.
88. 89.
90.
91.
Snyder MB, Siwicki M, Wireman J, et al. Reduction in Legionel!u pneumophila through heat flushing followed by continuous supplemental chlorination of hospital hot water. J lnf(:ct Dis 1990; 162: 127-32. Ciesielski CA. Blaser MJ, Wang WL. Role of stagnation and obstruction of water flow in isolation of Legionel!u pneumophila from hospital plumbing. Appl Environ Microbial 1984; 48: 984-7. Liu WK, Healing DE, Yeomans JT, Elliott TS. Monitoring of hospital water supplies for Legionella. J Hosp lnfi!ct 1993; 24: 1-9. Katcher ML. Scald bums from hot tap water. JAMA 198 L 246: 1219-22. Makin T, Hart CA. The effect of a self-regulating trace heating element on Legionella within a shower. J Appl Bacterial 1991; 70: 258-64. Hoebe CJ, Cluitmans JJ, Wagenvoort JH, van Leeuwen WJ, Bilkert-Mooiman MA. Cold tap water as a source of fatal nosocomial pneumonia due to Legionella pneumophila in a rehabilitation center. Ned Tijdschr Geneeskd 1999; 143: I 041-5. Nafziger DA. Successful chlorination for Legionella at Henry Ford Hospital. lnfi!ct Dis Clin Pract 1997; 7: 118. Fiore AE, Kool JL, Carpenter J, Butler JC. Eradicating Legionella from hospital water Reply. JAMA 1997; 278: 1404-5. Helms CM, Massanari RM, Wenzel RP, rfaller MA, Moyer NP, Hall N. Legionnaires' disease associated with a hospital water system. A five-year progress report on continuous hyperchlorination. JAMA 1988; 259: 2423-7. Grosserode M, Wenzel R, Pfaller M, Helms C. Continuous hyperchlorination for control of nosocomial Legionella pneumophila pneumonia: a ten-year follow-up of efficacy, environmental effects, and costs. In: Barbaree JM, Breiman RF, Dufour AP, eds. Legionella: current status and emerging perspectives. Washington, DC: American Society for Microbiology, 1993: 226-9. Liu Z, Stout JE, Tedesco L, et al. Controlled evaluation of copper-silver ionization in eradicating Legionella pneumophila from a hospital water distribution system. J Infect Dis 1994; 169:919-22. Stout JE, Lin YS, Goetz AM, Muder RR. Controlling Legionella in hospital water systems: experience with the superheat-and-flush method and copper-silver ionization. Infect Control Ho,\p Epidemiol 1998; 19: 911-4. Landeen LK. Yahya MT, Gerba CP. Efficacy of copper and silver ions and reduced levels of free chlorine in inactivation of Legionella pneumophila. Appl Environ Microbial 1989; 55: 3045-50. Biurrun A, Caballero L, Pelaz C. Leon E, Gaga A. Treatment of a Legionella pneumophilacolonized water distribution system using copper-silver ionization and continuous chlorination. lnfi!ct Controlllosp Epidemiol 1999; 20: 426-8. Anonymous. Sustained transmission of nosocomial Legionnaires disease-- Arizona and Ohio. MMWR Morb Mortal Wkly Rep 1997; 46: 416-21. Rohr U, Senger M, Se1enka F, Turley R, Wilhelm M. Four years of experience with silvercopper ionization for control of legionella in a german university hospital hot water plumbing system [In Process Citation]. Clin lnf(xt Dis 1999; 29: 1507-11. Granados A, Podzamczer D, Gudiol F, Manresa F. Pneumonia due to Legionella pneumophila and pneumococcal pneumonia: similarities and differences on presentation. Eur Respir J 1989; 2: 130-4. Macfarlane JT, Miller AC, Roderick Smith WH. Morris AH, Rose DH. Comparative radiographic features of community acquired Legionnaires' disease. pneumococcal pneumonia. mycoplasma pneumonia. and psittacosis. Thorax 1984; 39: 28-33.
27
lJ2.
Yu \.L. ldarston B.l. Plouffe .IF. Fik T.l. et al. Incidence of community-acquired pneumonia requiring lwspitali/ation. Results of a population-based acti\e sun eillance study in Ohio. The Community-Based Pneumonia Incidence Study (iroup .·I rei! lnrern \fed 1097: 157: 1709-1 X. 1\larston B.l. Lipman liB. Brcim~m RF. Sun·eillance for Legionnaires' disease. Risk 1~JCtnrs for morbidity and mortality Arch lntcm \fed 191.J.f: 154: 2.f 17-22 . .i llogc C\\. Brieman R F. Alh ances in the epidemiology and \.'Ontrol of Legionellu infe\.'tlons. l:jndemiol Rn 199 L 13: 3.20--lO. -l. Tison DL. Seidler RJ. Legionella inciden\.'c and density in potabk drinking'' ater suppllcs . .·lfJfJ/ Lnl·iron \!icrohiol II.JX3: 45: 337-9. :'. Fidds BS. Sanden Ci'J. 13arharee Jl\1. et al. lntral..'cllular multiplication ofLcgionella JmCIIIIIofJI!ila in amoebae isolated from hospital hot water tanks. C·urr .\licmhiol 1981.): 18:
I.
131-7.
6. 7
X 9. 10. II.
12.
13. 1-t
I:'. 16.
17.
I X. 19. .20.
Arno\\ P\1. Chou T. \\'eil D. Shapiro E:-..i. Kret/sl..'hmar C. i\osocomial Legionnaires' d1seasc caused by aerosolized tap ,,·atcr from respiratory de\icl..'s . .1/n/ecr /)is 14K:?:: 146:460-7. Brl..'iman RF. Fil..'lds BS. Sanden Ci.'\1. \"olmcr L. \le1cr :\. Sp1ka JS. Association of shmwr usc with Legionnaires' disease. Possibk role of amol..'bae. J I:\!. I 1990: 263: .292-l-6. Garbe PL. Da\ is BJ. Weisfeld JS. ct al. Nosocomial LL"glonnaires' disease. Epid..:miologic demonstration of cooling to\\'ers as a sourl..'c. Jl .\!A l9X5: 254: 521--l. Mastro TD. Fields BS. Breiman RF, Campbell .1. Plikaytis !3D. Spika JS i\osocomial Legionnaires' disease and usc ofmcdication nebulizers. J ln/i'ct Dis 1991:163:667-71. Mahoney F.l. lloge CW. Farley T A, ct al Communityv,:ide outbreak of Legionnaires' disease associated with a grocery store mist machine . .! fn(ect /)is 199.2: 165: 736-9. Hlady WG, Mullen RC Mintz CS. Shelton BG. Hopkins RS. Daikos GL. Outbreak of Legionnaire's disease linked to a dcwrati\e fountain by rnokcular epidemiology. Am.! l:jJidemiol 1993: 138: 555-62. Anonymous Legionnaires disease assol..'iatcd with a whirlpool spa display-- V1rgi111a. September-October. 19% [published erratum appears in MMWR Morb Mortal Wkly Rep 1997 Apr I ~:-l6( 15 ):336] . .\1.\JHR \!orh .\!orral Wk/r Rep 1997: 46: X3-6. Jernigan DB. Hofmann J. Cetron \IS. et al. Outbreak of Legionnaires' disease among cruise ship passengers cxposl..'d to a contaminated whirlpool spa. l.anccr 1996: 347: 494-9. Johnson JT. Yu \"L Best \Hi. et al. ~osocomiallegionellosis in surgical patients \\ith headand-neck l..'anl..'cr: impii\.'ations for cpidem10logi1..'al rescr\oir and mode of transmission. l.ancct 19X5: 2: 29~-300. Steck TW. Lanser .1, Sangster N. Isolation of l.cgioncllu longheaclwc serogroup I from potting mixes . . ljipl Fnlimn.\licmhin/1990: 56: -l9-53. Brcim Sc\ era! t~1ctors might explain the higher ratio at the transplant center. First. transplant patients arc more likely than nonimmunosuppressed patients to have a f~1tal outcome of Legionnaires' discasc_.:::.Js Second. the hospital did not have a routine system of screening for Legionnaires· disease before 1996, and it is likely that tests for legionellosis were performed on patients \\·ho were most acutely ill and or not responding to treatment. Finally. eight of the cases were identified as the result of tests routinely performed on autopsy specimens. and this may have blsely increased the case-f~1tality ratio. The cascf~ttality ratio f(Jr patients \Vho had onset of Legionnaires' disease in ]996 \Vas 25°o. and the last person \\ ith Legionnaires· disease to die had onset in February 1996. This may signify more timely diagnosis and treatment as a result of increased a\\ arcness on the part of physicians. Case-control study results Most risk f~1ctors found in the case-control study (i.e. intubation. corticosteroid therapy. and not \valking outside of the room) may he surrogates fix the general susceptibility of the patients to infection. \Valking in the hall during I to 3 days out of the total hospital stay appeared to he protective. likely because bedridden patients arc more at risk to de\ clop pneumonia. Hm,·c\ cr. walking in the hall during more than 3 days appeared not to he protecti \c. possibly because patients that spent much time outside of their room were at increased risk of exposure to aerosols from the contaminated carpet clcann that \\as used in the halh\ay. Risk htctors rdated to
7X
Legionnu ires 'diseuse among tra wpla n t /)(/ rien ''
location and specific activities were not identified. possibly because int(xmation about specific activities was not consistently recorded in the medical records and routes of transmission may have differed O\er time.
Intubation had been a risk factor among
patients hospitalized before 1996. However. intubation probably was not associated with transmission of Legionella in Jl)l)6. because most case-patients that \vere hospitalized in
Jl)l)6 had not been intubated. respiratory equipment cleaning
procedures were adequate at the time of the investigation. and no Legionella was recovered from respiratory therapy equipment. Control measures This study highlights the need to systematically monitor parameters such as chlorine concentration to identify unexpected problems during decontamination. Like many other hospitals. this hospital has a complex water system. To control the growth of Legionella on a long-term basis. such water systems may require redesign and simplification.
Even atkr successful intencntions the water system should be
monitored for Legiondfa for an extended period and decontamination procedures may need repeating to maintain an undetectable level of concentration of Legione//a and to prevent recurrence of nosocomial transmission. In this hospitaL a 5-minute tlush with water with I 0-20 mg/L was effective once all the other necessary adjustments had been made (removal of dead legs. of electrical \Vater heaters. and of chlorine-depleting devices: installation of continuous chlorine injection devices).
Conclusion The results of this investigation suggest that. during Jl)X7 -ll)96. nosocomial acquisition of Legionnaires· disease occurred at this hospital though multiple routes of transmission. These routes include inhalation of aerosols from showers. inhalation or aspiration associated with intubation. and possibly inhalation of aerosols from carpet cleaning devices. Control of nosocomial Legionnaires' disease in similar large institutions is complicated and im·oh·es coordination between many departments including facilities and maintenance personnel. infection control staff. nurses. and physicians.
Control
efforts im oh e targeted sun eillancc f()r cases. extcnsi\ e en\,ironmental im estigations. and comprehensive decontamination procedures.
This example illustrates that after
attainment of such coordination the problem of nosocomial Legionnaires· disease can be resolved. Transplant patients arc at particular risk for morbidity and mortality from
79
Chapter 4
Legionnaires· disease. Hospitals specializing in the care of these patients should make it a priority to establish suneillance for nosocomial Legionnaires' disease.
Acknowledgments The authors thank Concepcion Moore. other infection control staff: and medical records staff of the hospital for help with the epidemiological investigation. We thank facilities and maintenance staff of the hospital and the adjacent university for assistance with the environmental investigation. In addition we thank Dr. Margarette Kolczak of CDC's Division of Bacterial and Mycotic Diseases for advice on the statistical analysis and Dr. Cecilia Rosales of the County Health Department for assistance in reviewing medical records and providing other helpful information.
References I.
2. 3.
4.
5. 6. 7.
Marston BJ. Plouffe JF. File TM. Hackman BA. Salstrom SJ. Lipman HB et al. Incidence of community-acquired pneumonia requiring hospitalization: population-based active sur.eillance in Ohio. Arch Intern Med 1997; 157:1709-1717. Marston BJ, Lipman BH. Breiman RF. Surveillance for Legionnaires' disease: risk factors for morbidity and mortality. Arch Intern Med 1994: 154:2417-2422_ Matulonis U. Rosenfeld CS, Shadduck RK. Prevention of Legionellu infections in a bone marrow transplant unit: multifaceted approach to decontamination of a water system. Infect Control Hosp Epidemiol 1993:14:571-5. Prodinger WM, Bonatti H. Allerberger F. Wewalka G. Harrison TG. Aichbcrgcr C. et al. Legioncl/a pneumophila in transplant recipients: a cluster of cases of eight years duration_ J Hosp Infect 1994:26:191-202. Singh N. Gayowski T. Wagener M, Marino JR. Yu VL. Pulmonary infections in liver transplant recipients receiving tacrolimus. Transplantation 1996:61:396-40 I. Harrington RD. Woolfrey A. Bowden R. McDowell MG. Hackman RC. Legionellosis in a hone marrow transplant center. Bone Marrow Transplantation 1996: I X:361-X. Kugler JW. Armitage JO, Helms CM. Klassen LW. Goeken NE. Ahmann GB. et al. Nosocomial Legionnaires· disease: occurrence in recipients of hone marroy., transplants. Am J Med 19S3:74:2X 1-8.
8.
Harbarth S, Pittet D. Romand J. fatal concomitant nosocomial Legionnaires· disease and cytomegalovirus pneumonitis after card1ac transplantation. Intensive Care Medicine 1996:22:1133-4.
Breiman RF. Fields BS. Sanden G, Volmer L. Meier A. Spika .1. An outbreak of Legionnaires' disease associated with shower use: possible role of amoebae. JAMA 1990;263:2924-6. I 0. Centers for Disease Control and Prevention. Sustained Transmission of Nosocomial Legionnaires· Disease-arizona and Ohio. MMWR 1997:46:416-21. II. Heimberger T. Birkhead G. Bornstein D. Same K. Morse D. Control of Nosocomial Legionnaires· Disease through Hot Water flushing and Supplemental Chlorination of Potable Water. J Infect Dis 1991;163:143. 12. Tobin JO. Beare J. Dunnill MS. Fisher-Hoch S. french M. Mitchell RG. et al. Legionnaires· Disease in a Transplant Unit: Isolation of the CausatiYe Agent from Shower Baths. Lancet
9_
80
Legionnaires 'disease among tramp/ant patients 1980;2: 118-21. 13. Hanrahan JP, Morse DL, Scharf VB, Debbie JG. Schmid GP. McKinney RM, et al. A community hospital outbreak of legionellosis. Am J Epidemiol 1987;125:639-49. 14. Menne] LA, Josephson SL, Giorgio CH, Dempsey J, Parenteau S. Association of Legionnaires' Disease with Construction: Contamination of Potable Water? Infect Control Hosp Epidemiol 1995; 16:76-81. 15. Colville A. Crowley J. Dearden D. Slack RCB, Lee JV. Outbreak of Legionnaires' Disease at University Hospital, Nottingham. Epidemiology, Microbiology and Control. Epidemiol Infect 1993;110:105-116. 16. Helms CM, Massanari RM, Wenzel RP, Pfaller MA. Moyer NP, Hall N. Legionnaires' Disease Associated with a Hospital Water System. JAMA 1988;259:2423-2427. 17. Best M, Yu VL, Stout J, Goetz A, Muder RR, Taylor F. Legionellaceae in the Hospital Water Supply. Lancet 1983;2(8345):307-31 0. 18. Dondero TJ, RendtorffRC, Mallison GF, Weeks RM, Levy JS, Wong EW, et al. An Outbreak of Legionnaires' Disease Associated with a Contaminated Air-conditioning Cooling Tower. N Eng! J Med 1980;302:365-370. 19. Klaucke DN, Vogt RL, Larue D, Witherell LE, Orciari LA, Spitalny KC, et al. Legionnaires' Disease: the Epidemiology of Two Outbreaks in Burlington, Vermont. Am J Epidemiol 1984; 119:382-91. 20. Garbe PL, Davis BJ, Weisfeld JS, Markowitz L, Miner P, Garrity F, et aL Nosocomial Legionnaires' disease: epidemiologic demonstration of cooling towers as a source. JAMA 1985;254:521-524. 21. Mastro TO, Fields BS, Breiman RF, Campbell J, Plikaytis BD, Spika JS. Nosocomial Legionnaires' disease and use of medication nebulizers. J Infect Dis 1991;163:667-671. 22. Arnow PM, Chou T, Wei! D, Shapiro EN, Kretzschmar C. Nosocomial Legionnaires' disease caused by aerosolized tap water from respiratory devices. J Infect Dis 1982;146:460-467. 23. Centers for Disease Control and Prevention. Guidelines for prevention of nosocomial pneumonia. MMWR 1997;46 (RR-1 ); 28-34, 54-57, and 74-79. 24. Joly JR, McKinley RM, Tobin JO, Bibb WF, Watkins 10, Ramsey D. Development of a standardized subgrouping scheme for Legionel/a pneumophila serogroup I using monoclonal antibodies. J Clin Microbial 1986;23:768-771. 25. Breiman RF, Cozen W, Fields BS, Mastro TO, Carr SJ, Spika JS, et al. Role of air sampling in investigation of an outbreak of Legionnaires' disease associated with exposure to aerosols from an evaporative condenser. J Infect Dis 1990; 161: 125 7-1261. 26. Copeland J, Wieden M, Feinberg W, Salomon N, Hager 0, Galgiani J. Legionnaires' disease following cardiac transplantation. Chest 1981 ;79:669-71. 27. Fuller J, Levinson MM, Kline JR. Copeland J. Legionnaires disease after heart transplantation. Ann Thoracic Surgery 1985;39:308-11. 28. Edelstein PH, Legionnaires' disease. Clin Infect Dis 1993; 16:741-9.
81
Chapter 4
Addendum to Chapter 4 The hospital has continued to do active surveillance for cases of Legionnaires' disease and to test water for presence of Legionella.
No nosocomial cases were
detected since September 1996 and no Legionella were recovered after February 1997. Free chlorine v·.:as injected into the hot side as V•/ell as the cold side of the water system to achieve a concentration of 2 to 4 mg/L at all points of use. Because of corrosion problems that \vere attributed to the high chlorine concentration. the management decided to install a copper-silver ionization system in the summer of 1997. The total cost in the first year to get the system up and running was approximately S 1 million (emails to J. Kool from C. Glasby. infection control practitioner, University Medical Center. Tucson. AZ. July 22. November 15. and November 16. 1999 ). Major expenditures during the first year were for: • •
Study and design (-S 100,000 ). Separation of the water system from the adjacent university campus and installing a recirculation system for the cold water ( -S200,000).
•
Installation and maintenance of the copper-silver system ( -$600,000). The coppersilver electrodes needed to be replaced regularly. A problem of black discoloration of sinks necessitated a change in the composition of the copper-silver alloy and filters were installed to remove silver particles.
These filters arc replaced every
month and are also tested for Legionella colonization. •
Approximately one full-time staff member ( 1.0 FTE) was needed to balance and monitor the system in the first year, f(Jilovv·ed by approximately 0.5 FTE in following years.
•
Cost for installation and maintenance of the chlorine injection devices v.·as similar to that described in the General Introduction for other hospitals. After installation of the copper-silver ionization system the hospital continued to
inject free chlorine but the concentration \Vas reduced to 0.5 mg/L.
Corrosion
problems diminished. \Vater samples have remained negative for Legione//a. and no nosocomial cases of Legionnaires' disease ha\c been detected despite intensive clinical surveillance.
82
Chapter 5: Hospital characteristics associated with colonization of water systems by Legionella and risk of nosocomial Legionnaires' disease: a cohort study of 15 hospitals
Jacob L. KooL MD, MS; David Bergmire-Sweat, MPH; Jay C. Butler. MD; Ellen W. Brown; Deborah J. Peabody. MD; Daniel S. Massi, MD: Joseph C. Carpenter, PE; Janet M. Pruckler. BA; Robert F. Benson. MS; Barry S. Fields. PhD
From the Respiratory Diseases Branch, Division of Bacterial and Mycotic Diseases (Drs. Kool. Butler. Peabody. Massi. Fields, Ms. Brown. Ms. Pruckler, Mr. Benson): Hospital Infections Program (Mr. Carpenter). National Center for Infectious Diseases: Epidemic Intelligence Service (Dr. Kool ); Epidemiology Elective Program (Drs. Peabody and Massi), Epidemiology Program Office. Centers for Disease Control and Prevention, Atlanta, Georgia: Infectious Disease Epidemiology and Surveillance Division. Texas Department of Health (Mr. Bcrgmire-Sweat). Austin. Texas.
fnji>ction Control and Hospital Epidemiolog\· 1999: 20: 798-805.
Chapter 5
Abstract Objective: To investigate increased reporting of Legionnaires' disease in patients of multiple hospitals in San Antonio, Texas, and to study risk factors for nosocomial transmission of Legionnaires' disease and determinants for Legione!fa colonization of hospital hot water systems. Setting: The 16 largest hospitals in the cities of San Antonio, Temple, and Austin, Texas. Design: Review of laboratory databases to identify patients with Legionnaires' disease in the 3 years prior to the investigation and to determine the number of diagnostic tests for Legione!fa performed; measurement of hot water temperature and chlorine concentration and culture of potable water for Legionella. Use of exact univariate calculations, Poisson regression, and linear regression to statistically determine factors associated with water system colonization and transmission of Legionella. Results: Twelve cases of nosocomial Legionnaires' disease were identified; eight of these occurred in 1996. The rise in cases occurred shortly after physicians had started requesting Legionella urinary antigen tests. Hospitals that frequently used Legionella urinary antigen tests tended to detect more cases of Legionnaires' disease. Legionella was isolated from the water systems of 11 of 12 hospitals in San Antonio; the 12th had just experienced an outbreak of Legionnaires' disease and had implemented control measures. Nosocomial legionellosis cases probably occurred in five hospitals. The number of nosocomial Legionnaires' disease cases in each hospital correlated better with the proportion of water system sites that tested positive for Legionella (P=. 07) than with the concentration of Legionella bacteria in water samples (P=.23 ). Hospitals in municipalities where the water treatment plant used monochloramine as a residual disinfectant (n=4) and the hospital that had implemented control measures were Legionella-free. The hot water systems of all other hospitals (n= 11) were colonized with Legionella. These were all supplied with municipal drinking water that contained free chlorine as a residual disinfectant. In these contaminated hospitals, the proportion of sites testing positive was inversely correlated with free residual chlorine concentration (P=. 01 ). In all hospitals, hot water temperatures were too low to inhibit Legionella growth. Conclusions: The increase in reporting of nosocomial Legionnaires' disease was attributable to increased use of urinary antigen tests; prior cases may have gone unrecognized. Risk of Legionnaires' disease in hospital patients was better predicted by the proportion of water system sites testing positive for Legionella than by the measured concentration of Legionella bacteria (colony forming units per ml ). Use of
84
Hospital characteristics and Legionnaires 'disease
monochloramine by municipalities for residual drinking water disinfection may help prevent Legionnaires' disease.
Introduction Between 8,000 and 18.000 cases of Legionnaires' disease occur annually in the United Statesu and thousands more occur in other countries. Of cases reported to the Centers for Disease Control and Prevention (CDC), up to 25% are acquired in a hospita1.
1
7
Fatality rates in nosocomial outbreaks have exceeded 30%?- The disease is
caused by the bacterium Legionella, which lives in the biofilm covering the inside of water-containing pipes and tanks. Legionella thrives in hot water systems in large buildings such as hospitals. The most frequently described route of transmission appears to be inhalation of contaminated aerosol (e.g. from showers or cooling towers). It has also been reported that transmission can occur through aspiration of contaminated water, especially in patients with swallowing disorders. A study in Quebec found that 68% of 84 hospitals were contaminated with Legionella on at least one occasion during quarterly sampling over one year. Yet transmission to patients apparently occurred in some but not all of these hospitals.x This paradox may be explained by many complex factors, including differences in susceptibility of patients, differences in virulence between Legionella species, great variety in water system configurations, respiratory equipment disinfection methods, concentration of the bacteria in the water, number and size of 'sloughed off pieces ofbiofilm containing the bacteria, and frequency of diagnostic testing for Legionnaires' disease among patients with nosocomial pneumonia. At the end of 1996, five hospitals in San Antonio, Texas, reported Legionella infection in one or more of their patients with onset of disease between July and November that year. The number of reported cases was higher than in previous years. We did a cohort study of the largest hospitals in San Antonio and in two other municipalities to determine the extent of the problem and to study risk factors for nosocomial transmission and factors associated with colonization of hospital water systems with Legionelfa.
Methods We selected all acute-care hospitals with an average daily census greater than I 00 in the three counties that contain the cities of San Antonio, Austin. and Temple.
85
( '/wptcr 5
Case findin!.! lnkction control staff nf each hosrital \\ere asked for a list of knm\n cases of Legionnaires· disease occurring from January 1994 to January I997. and im cstigators rc\ iewcd the microbiology laboratory database of each hosrital for positi\c Legiondla test results from the same period. \1cdical records of idcnti lied patients \\·ere re\ icwed to confirm the diagnosis and to determine if the patient had acquired the infection in the hospital. :\case of Legionnaires' disease was dctined as a patient with a clinical diagnosis of pneumonia confirmed by chest X-ray. and at least one ofthe following: isolation of Legionclla from respiratory secretions or from lung tissue. detection of Legione//a pncumophila scrogroup I antigen in urine. detection of /_ JmCtllllOJJhilo in respiratory secretions or lung tissue by direct fluorescent antibody (OF:\). detection of a f()urfold or greater rise in titers of antibodies against L JmeumoJJhila in acute- and com alcscent-phase serum. to a Yalue of I: I 2g or higher. Because the incubation time of Legionnaires· disease usually lies between 2 to I 0 days. we defined nosocomial Legionnaires' disease as Legionnaires' disease in a patient hospitalized f()r the entire period of 10 to 1 days bcl()re onset of symptoms; \Ve dctined possibly nosocomial Legionnaires' disease as Legionnaires' disease in a patient hospitalized for part of that period. For example, a patient hospitalized continuously for 14 days and discharged less than 4R hours bet()rc the onset of symptoms of Legionnaires· diseasc was considered definitely nosocomial. A patient hospitalized tor li\ e days prior to onset of symptoms. or a patient who was hospitalized for most of the 2 to I 0 days hut Vv'hO was discharged during part or that period. ''as classilied as possibly nosocomial. Very rc,, clinical isolates had been sa\ed hy thc hospitals. making it impossible to systematically compare clinical and en\ ironmcntal isolates hy subty·ping techniques to further distinguish nosocomial Jl·om community-acquired case-patients. Lm·ironmcntal imcsti!.!ation In each hospital. \\ater \\as collected f()J· f_egionella culture from the bottom of all hot \\ater storage tanks and all \\·ater heating tanks that supplied patient areas. If no tanks were present. water was taken from all hot water return Iines from patient areas. \\'ater and s\\·ah samples \\·ere obtained from shm\·erheads and f~!llect aerators in those patient rooms that were located furthest Jl·om the hot water source. In each hospitaL het\\ een I\\ o and four patient rooms \\·ere sampled. depending on the size and number of\\ater di~tribution S\stems. The rooms were .-;elected to achie\e an e\cn distribution among all hot \\ater distribution ~ystems that supplied patient areas \\ithin a hospital.
!-lospi ta I char([( ·tcristics and Legion 11a ires 'cli.\euse
For example, if there ""·ere four different hot water distribution systems in a hospital, one patient room served by each was selected: if there were only tv,o distribution systems. two patient rooms in each were selected. preferably each in a different wing. We obtained one liter of water from the faucet and the show·er in each selected patient room. Half a milliliter of a 0.1 nom1al solution of sodium thiosulphate was added to each water sample to neutralize the residual free chlorine or monochloramine. The faucet's aerator and the showerhead were then removed and a swab sample was taken from the inside of each fitting. All samples were processed at the CDC's Respiratory Diseases Laboratory. The samples were plated on buffered charcoal-yeast extract supplemented with a-ketoglutarate ( BCYE-o.): on BCYE-a plates supplemented with polymyxin.
anisomycm.
and
vancomycin
supplemented with glycine and PAY.
(PAY):
and
on
BCYE-a
plates
They were plated directly without acid
treatment or after acid treatment as necessary. or concentrated by filtration. The filter 1
\Vas resuspended in sterile water. \Ortexed and then plated.'
After the samples were
obtained. the hot water faucet was flushed over a thermometer until the water temperature qid not change more than 0.1° F in I 0 seconds. and this temperature was then recorded.
The free residual chlorine concentration was measured immediately
after the flush.
lnfonnation on pH of the water was obtained from municipal water
authorities. Survey of hospital characteristics A questionnaire was administered to hospital infection controL microbiology laboratory, and facilities maintenance staff. Questions asked to infection control and laboratory staff included the size of the hospital (number of beds and average daily census). the number of immunocompromised patients. the number and types of laboratory tests for Legionella that were requested in the last 5 years for inpatients. Facilities maintenance personnel were asked about age. number. and configuration of water heaters and hot water storage tanks. the source of the water (municipal supply or private welL ground or surface water), hot water temperature setting. maintenance done to the \Vater system in the last tin~ years. changes made to the water system. and
if the hospital additionally treated v.:ater. e.g. v.,:ith water softeners or additional disinfection. Statistical analysis Responses to questionnaires, results of case finding. results of environmental cultures. and temperature and chlorine measurements were entered in \1 icrosoft Excel and analyzed using Epi-lnfo (CDC. Atlanta. GA) and SAS (SAS Institute. Cary. NC). The unit of analysis was the hospital. not individual patients. We did t\vo separate
X7
Chapter 5
ana lyses: one with the number of Legionnaires' disease cases as the outcome. and o ne with the proportion of sites in the hospital water system from which Legionella was recovered as the outcome.
Although the number of observations was small, we
attempted multivariate Poisson regress ion analysis to assess co nfou ndin g.
However,
most of the ana lyses described in the following paragraphs are univariate . One hospital had experi enced an outb reak of Legionnaires ' disease two years earlier and had implemented contro l mea sures and intensive surveillance for new cases: thi s hospita l was exc luded from the analysis. The relationship between colonization of the hosp ital water system and hospital characte ri st ic s such as size. hot water temperature. and size of hot water tank s was analyzed using linear regress ion , with the proportion of sites testing positive for
Legione/la as the outcome variab le. Poi sson regression (univariate and multivariate) was used to ana lyze associat ion of hospital characterist ics with occurrence of cases of Legionnaires' disease for those hospitals in which Legion e/la was detected in the water system.
Because some
possibly nosocomial case-patients may have acquired their infection outside the hospital , we counted definitely no soco mial cases as I and possibly nosocomial cases as Yz in the Poi sson regress ion analysis. For each ho sp ital , the sum of th ese adjusted 'cases' was used as the o utcome. To further account for uncertainty of nosocomial transmission, hospita ls with only one possibly nosocomial case (n= 2) received a statistica l weighting of ~ -
•
4
LJ
Definitely nosocomial
Possibly nosocomial
3
"'~
2
u"'
0
LJ....__
__L__ 2_
3 4
1994
2
3
1995
4
2
3
4
1996
Year and quarter
Figure 1: Identified cases of nosocomial Legionnaires' disease by year and quarter of onset, San Antonio , 1994 to 1996. Hospitals are indicated by letters A through E.
88
Ho!>pitul characteristics and Legionnaires 'disease
Results Epidemiological findings Sixteen of 18 eligible hospitals agreed to participate in the study. Twelve of these were in San Antonio, 3 were in Austin and one was in Temple.
The two
hospitals that did not participate were located in San Antonio and Temple. Twelve patients with definitely or possibly nosocomial Legionnaires' disease were identified; eight had onset in 1996 (Figure l ). Six cases were definitely nosocomial and six were possibly nosocomiaL the case-patients were inpatients of five different hospitals in San Antonio. The cases occurring in 1994 and 1995 in the hospital indicated in Figure 1 with the letter A had been recognized as a recurrence of an outbreak, which had first been detected in 1989.
10
Hospital A had implemented control measures and continued
to do intensive surveillance for new cases at the time of our investigation. The five hospitals with identified cases were not clustered geographically except that they were all located within Bexar County (San Antonio).
Eleven of the 12 San Antonio
hospitals were supplied with the same municipal water; hospital A was supplied from a private well.
Area San Antonio {not including hospital A) San Antonio (hospital A)
Austm 1
Type of test UAG Serology DFA Culture UAG Serology DFA Culture Culture
1994 Tests/ hospitals' Average 718 07 30/3 70 812 3413 0/1 10 438/7 43817 10
4 77
0 438 438
7995 Tests I hospitals· Average 3718 3.9 433 I 5 87 718 I 4 32216 711 10 399 I 1 399 I 7 3717
30 54 7 399 399 31
1996 Tests I hospitals· Average 30618 38 49115 98 709/8 28216 27717 10 39011 39017 3517
14 47 217 390 90 35
Table 1: Frequency of use of diagnostic tests for Legionella in hospitals in San Antonio and Austin. Footnotes: *,we were not able to obtain information one some tests in some of the hospitals; the number of hospitals that were able to supply us with the data is shown after each slash; t, we could only get information from one hospital in Austin and this hospital supplied us only with the number of cultures performed in 1995 and 1996; no information was received from the hospital in Temple. Abbreviations: UAG, urinary antigen test; DFA, direct fluorescent antibodies test.
89
Chapra 5
The amount of information that ,,.e could obtain on the frequency of diagnostic testing in pre,·ious years '
c: [
.;,
V>
u"'
60
6
50
5 4
V>
;;;
40 .!!! 30
3
0
10
1 0 10
2
8
11
2
5
8
11
2
8
5
11
Year and Month
-
Defm1tely nosocom1al
c:::::::J Poss1bly nosocomial
c:::::::J Commun1ty acqUired
-
Unnary antigen tests
Figure 2: Temporal relationship between increase in use of the urinary antigen test by 12 San Antonio hospitals and detection of cases of nosocomial and community-acquired Legionnaires · disease, October 1993 to January 1997. Most of the co mmunity-acquired cases and all nosocomial cases in 1994 and 1995 were patients of hospital A.
90
0
20 z
2
Hospira/ characrerisrics and Legionnaires 'disease
20
15 C/)
.,
'E
:;
a. 10 d.>
C/)
"'
(..)
5
I I
0 1992
1993
B
1994
I 1995
1996
Year of onset •
0 D
U r~ne antigen DFA Culture
Figure 3: Identified cases of Legionnaires ' disease (nosocomial and community-acquired) among patients of 12 large hospitals in San Antonio by diagnostic test used to confirm the diagnosis, by year, 1992 to 1996. Some of the cases of 1996 that are shown as cultureconfirmed were also confirmed by urinary antigen test; these are shown as cultureconfirmed cases.
Legionellaceae (L anisa. L dumoffii. Lfeefeii. L pneumophi/a serogroups I. 3. 6. 8. and I 0. and two unknown Legione//a-like organisms) were iso lated from sa mpl es from I I of 12 San Antonio hospitals. L pneumophi/a se rogro up I and L anisa were recove red from the hi ghest number of different hosp itals (eac h in four hos pital s). The on ly hospi tal in San Antonio th at was Legione//a-free was hospital A. whi ch had ju st experienced an outbreak of potable water- assoc iated Legionnaires' di sease and had impl emen ted co ntrol meas ures. No legione ll aceae we re recovered from the four hosp itals in Austin and Templ e (Tabl e 2) . Apart from hosp ital A. none of th e hospitals had changed the water system confi gurati on or impl emented suppl emental disinfection methods aim ed at co ntrol of Legionnaires· disease. Hosp ital A had insta ll ed a coppe rsi lve r ion izatio n system. rep laced all its hot water storage tanks with instantan eo us water heaters. increased the temperat ure of the hot water to 135°F. and removed '·dead legs .. (areas in the wate r system where water stagnates). Ri sk of wa ter system co lon izat ion by Leg ionell a Table 2 gi\·es an overview of the location and water characterist ics of the hospitals. Assoc iat ion of se lected water parameters with ri sk of water system
91
('/wprer 5
County San Antonio San Antonio (hosp. A) Austin Temple
Water source Ground (municipal) Ground (private well) Surface (municipal) Surface (municipal)
Drinking water disinfection method and additional control measures against Legionella Free Chlorine
pH of water 7.2
Free chlorine, redesigned water
7.2
Number of hospitals 11
Hospitals colonized with Legionel/a 11 0
system. metal ionization, water temperature 135oF* Monochloramine
9.5 - 10
Monochloramine
7.4
3
0 0
Table 2: risk factors for colonization of hospital water systems with legionella, by county. Footnote: *,this hospital had experienced an outbreak of water system associated nosocomial Legionnaires' disease 2 years earlier and had implemented control measures against Legionella; automatic mixing valves lowered water temperature to 11 OaF in patient rooms.
colonization is shown in Table 3. In the I I Legionella-compromised hospitals in San 2
Antonio there was a statistical correlation (P=. 0 L R =.52) between the number of sites testing positive for Legionella and the average free chlorine concentration that was measured in patient rooms (Figure 4 ).
The hospitals in Austin and Temple were
Legionella-free, and the most striking ditTcrence between these two municipalities and San Antonio was that they used monochloramine for residual drinking water disinfection while San Antonio used free chlorine (?=.0007). Other differences were that the Legionella-free hospitals received water with a higher pH than the water of the other two municipalities, and that their water was from a surface source while the San Antonio municipal water was ground water.
No statistically significant association
was found between the proportion of sites from which Legionella was recovered and all other parameters, including hospital size, hot water temperature. age of the hospital water system, and size and configuration of the water heaters. Risk of Legionnaires disease in hospital patients In univariate Poisson regression analysis (Table 4 ), a correlation was found between the number of nosocomial Legionnaires' disease patients and the proportion of sites in a hospital's potable water system testing positive for Legionel/a (P=.O? ). but not with the average, lowest. median, or highest estimated number of colonyfanning units ( CFU) in \\iater samples (highest CFU came closest to statistical
92
Hospital characteristics and Legionnaires 'disease
Characteristic Type of residual drinking water disinfectant used by municipality pH of the water Water source
Hospitals with Legion ella in water system (n=11)
Hospitals without Legione/la (n=4)
Cl Mono
11 0
28
:;:;
'Vi
0.8
0
a. (/)
~ 'Vi 0.6
0 c 0
t
0
0.4
a.
e
c..
0.2
0 0
0. 1
0.2
0.3
0.4
0. 5
Chlorine concentration (mg/L)
Figure 4: Average free residual chlorine concentration as measu red in patient room tap water and proportion of sites positive for Legionella, 11 hospitals in San Antonio . Each dot represents one hospital ; a reg ression line is shown . Linear regression statistics : R2 =.52 , P= .01 . One hospital in San Antonio (hospital A) had implemented control measures against Legionella several years before our investigation ; this hospital is not included . No Legionella was recovered from water in all four hospitals in Temple and Austin , where monochloramine was used instead of free chlorine for drinking water di sinfecti on.
P=.OO l ). contain ed onl y th e ave rage dail y ce nsus (P =.03) and th e proporti on of water system sites that tested pos iti ve for Legionella (P = 14 ) (Tabl e 4 )
Di sc ussion Coloni zation o f hospital hot water systems by Legionella was doc um ented in l l of 12 hospital s sa mpl ed in one metropolitan area durin g thi s in\·esti ga ti on of an apparent mu lti-h ospital "outbrea k" of nosoco mi al Leg ionn aires di sease. Th e one Legionella-free hospital had prev iously detec ted th e bacterium in it s water system. bnn gin g the proporti on of co mpromi sed hospitals ove r a 2-yea r period to I 00°'0.
94
1/ospira/ characteristics and l.egionnaires 'disease
Our study illustrates the impact of the introduction of new and improved diagnostic tests. Introduction of the urinary antigen test led to recognition of nosocomial cases. This was evidenced by the temporal relationship bet\veen the introduction of this test in San Antonio hospitals and the rise in detected legionellosis cases (Figure 2 ). and the fact that the urinary antigen test led to the diagnosis of the majority of cases in 1996 (Figure 3 ). This conclusion was further supported by our statistical analysis finding an association that was almost statistically significant (?=0.09. Table 4). There may have been some bias in that some hospitals may have started to use more diagnostic tests in response to the rise in identified cases. We Univariate
Multivariate
Hospital characteristic
RR (CI95)
p
RR (CI95)
p
Average daily census
1.02 (1.00- 1.03)*
0.05*
1.02 (1.00- 1.04)
0.03
No. of transplant patients
1.03 (1.00- 1.06)"'
0.07*
t
t
No. of urinary antigen tests done in 1996t
1.03 (1.00- 1.06)
0.09
t
t
14.3 (0.80- 256)
0.07
598 (0.11 - 3·1 06 )
0.14
1.05 (0.98-1.15)
0.23
t
t
Proportion of water system sites from which Legionel/a was recovered Highest no. of Legionella colony-forming units (CFU) per ml. isolated from any water sample
Table 4: Risk of nosocomial Legionnaires' disease and hospital characteristics, by univariate and multivariate poisson regression analysis. The unit of analysis was the hospital. The outcome variable was the number of nosocomial Legionnaires' disease cases in a hospital. Definitely nosocomial cases were counted as 1 "case", possibly nosocomial cases were counted as IS "case'', and for each hospital these were added up. In addition, a statistical weight of IS was given to hospitals that had only one possibly nosocomial case. One hospital in San Antonio that had experienced an outbreak of Legionnaires' disease 2 years earlier (hospital A) was not included in the analysis. Unless stated otherwise, the calculation included all 15 hospitals in San Antonio, Temple, and Austin. Footnotes:*, Calculation done for the 11 hospitals in San Antonio where Legione/la was recovered from the potable water system; t, Did not contribute significantly to the multivariate model;
:t, data on urinary antigen testing available for only
eight hospitals, which all had water systems contaminated with Legionel/a. Abbreviations: RR, relative risk (hazard ratio); Cl95, 95% confidence interval.
95
Chapter 5
The apparent impact of the introduction of the urinary antigen test suggests that cases may have been occurring in San Antonio earlier. which have gone unrecognized. Urinary antigen tests have a sensitivity of 80-99% and a specificity of 99%. 11 This, in combination with its ease of use and the very short time needed to perform the test, makes it an extremely useful tool for diagnosis of Legionnaires' disease. Currently available urinary antigen tests can only detect L. pneumophila serogroup 1. This test should therefore always be used in conjunction \vith culture for Legionella from clinical specimens (e.g. respiratory secretions). Culture has the additional advantage that environmental and clinical isolates can be compared to help identify the source of the infection. Use of the urinary antigen test is increasing in the United States and in other countries and this will likely lead to increased numbers of reported cases of Legionnaires' disease and to identification of outbreaks that would otherwise go unrecognized. 12 The fact that serology did not contribute at all to the identification of cases is not surprising because convalescent-phase serum is required for confirmation of a case; a single titer has very low specificity.
13
Convalescent-phase serum is often not
collected because, after the required six to eight weeks, many patients have either been discharged, have died, or have already been diagnosed by means of other tests. The small number of observations ( 15 hospitals with only 7 nosocomial cases) in our cohort study was a weakness since it decreased the statistical power and made assessment of confounding difficult. Because this was a retrospective study we had to rely on available laboratory and surveillance data and we do not know how many cases of Legionnaires' disease were missed.
In addition, we were not able to obtain
background information on the number of clinical tests for Legionnaires' disease that were performed on inpatients in Temple and two hospitals in Austin. Risk of colonization of hospital water systems by Legionella No Legionella was recovered from water samples in Austin and Temple. We found a strong statistical association between use of monochloraminc and absence of Legionella (?=.0007). However, at the time of the investigation we were not aware that the water services in these cities were using monochloramine. Neutralization of monochloramine requires a higher addition of sodium thiosulphate than is required to 14 neutralize free chlorine and we cannot be certain that enough thiosulphate was added to completely neutralize the disinfectant. Thus, the absence of Legionella from these specimens could in part be due to residual monochloramine killing the organisms during transport to the laboratory. Nonetheless. this does not account for the absence of cases in these cites, and a recent case control study also supports the role of
96
1/mpi fa 1clw racterist ics and l.cgionna ires 'disease
monochloramine in municipal water systems for pre\enting transmission of Legionnaires' disease. 1 ' The etTect of monochlorarnine on Legionella is biologically plausible: it is better at reaching distant points in a water system and it penetrates better into biotilm than free chlorine. 1 ~ The pH of water supplied to Lcgionclla-free hospitals was higher (Tables I and 2). This was not a causal association but rather a confounder because monochloramine requires a higher pH than free chlorine f()r optimal disinfection. 1 ~ \Ve believe that the fact that all Lcgione!la-frce hospitals were supplied with surface water is a chance tinding since many outbreaks have occurred in municipalities supplied with surface water: in fact. one study found hospitals recei\ ing surt~lCe water to be more at risk of 11 Legion ella colonization. ' In hospitals supplied with chlorine-containing water there was a linear correlation between the free residual chlorine concentration and the proportion of sites testing positive for Legiondla (Figure 4 ). This means that e\·en low chlorine concentrations have an impact on survival of Lcgionella in water systems. Chlorine penetrates poorly into hiofilm. However. in real-world situations the more important weakness of chlorine may be that it often does not reach distal sites in water systems. as illustrated by our findings. free chlorine can disappear quickly and when distances between the water treatment plant and the points of use are large the concentration often drops to undetectable levels. Chlorine is also used up rapidly when water passes through water softeners or heaters. I' Our findings underscore the importance of measuring chlorine concentration in patient rooms rather than at a central location to assess risk of Legion ella colonization. As in many States. the Texas Department of Health's Hospital Licensure Division required that water supplied to patient care areas was not warmer than II oo F to reduce the risk of scalding. 1~ Water temperatures of 95° F to I I(p F arc ideal for growth of Legionella species. Our measurements confirmed that water temperatures in all hospitals were too low for control of Lcgionella. Changing regulations to require a hot water temperature of at least 122° F in patient rooms (not at the heater) should only slightly increase the risk of scalding '') but it may significantly reduce the incidence of nosocomial Legionnaires' disease. Risk of nosocomial Legionnaires' disease for hospital patients Risk of nosocomial transmission was predicted better by the proportion of water system sites that tested positive for I~egionella than the bacterial concentration of Legionella (CFU per ml) in the positive samples (Table 4). This observation was 211 made once before in a single hospital. lt may seem counter-intuitive that Lcgionclla
97
( 'hupter 5
bacterial concentrations would not be an important determinant of risk of transmission. J lowever. bacterial concentrations can \ ary significantly 0\ cr time ~ and current methods for estimation of bacterial counts can he \cry inaccurate.~~ Quantitative counts arc intlucnced by sampling technique and laboratory methods such as concentration. acid treatment. choice of culture media. and plating techniques, \1oreover. since only a small proportion of exposed persons develop Legionnaires· disease. the number of Legionc//a-disscminating points in a \\·ater system (in other words. the number of potentially exposed individuals) may be a more important detenninant of transmission than the infectious dose each person is exposed to. It may therefore be unnecessary to usc quantitative microbiological methods to assess risk of 211 f_egionella transmission from potable \Vater systems. Based on similar observations. one cOLinty health department recommends implementing control measures \\·henc\ cr the proportion of positi\e sites in a hospital surpasses ~:woo in settings \\here nosocomial Legionnaires' disease has not been observed.~' However. nosocomial transmission has occurred when less than 30° o of sites tested positi \c.,- II ighly susceptible patient populations. in particular transplant patients. arc at risk at significantly lower levels of contamination. On the other hand. in smaller general hospitals with a less susceptible patient population the risk of nosocomial transmission likely is much lower. Ideally, policies and guidelines for control of nosocomial Legionnaires' disease should take into account the size of the hospitaL the number of highly susceptible patients (in particular transplant patients). the municipal water disinkction method, and the disinfectant concentration measured in patient rooms. It may not be costeffective f()r lower-risk hospitals to test their drinking \\. Case-fatality rates in outbreaks typically are 20%-40%.
2
The disease is caused by Legionel/a, a bacterium that lives in
symbiosis with amoebae in the biotilm that covers the inside ofv,'ater-containing pipes and tanks.
LC!giondla thrives in warm (25-42 °C) water. especially in areas \A/here
\Vater stagnates.
Transmission occurs through inhalation of a Legione//a-containing
aerosol or by aspiration of contaminated water. Most outbreaks have been traced to either potable water or cooling towers and ecological and laboratory studies show that sporadic cases are caused by the same sources.'--"
Most nosocomial outbreaks have
been Iinked to a hospital potable water system that \vas compromised by Legion ella. Current Centers for Disease Control and Prevention (CDC) guidelines recommend hospitals that have identitied nosocomial transmission from their water system to decontaminate the water and then implement control measures aimed at prC\cnting
104
Effect ofmonochloramine on Legionnaires 'disease
regrowth.
These long-term control measures include increasing the hot water
temperature and/or continuously injecting additional chlorine.f,
Although these
methods are effective. a high \Vater temperature increases the risk of scalding and a continuous high chlorine concentration can cause corrosion of plumbing material. Other methods for control of Legionella are more expensive and have not proved to be more etTective than supplemental chlorination. Monochloramine has been used for drinking water disinfection since 1916_u: It is forrned when ammonia and free chlorine are mixed in water. Municipal drinking water disinfection has two stages: initial disinfection to kill organisms in the water coming into the treatment plant, and residual disinfection to maintain biocidal activity in the end product throughout the water distribution system.
Monochloramine 's
disinfecting action is slower than that of free chlorine, so it is less useful for initial disinfection. On the other hand it is more stable than free chlorine, so a disinfecting residual can be maintained over long distances in a distribution system, which can reduce cost.~ Monochloramine penetrates better into biofilm than free chlorine and it is better able to kill sessile biofilm bacteria such as some Pseudomonas spp.
LJ-ll
The US Environmental Protection Agency has issued regulations to reduce adverse health effects, including cancer, associated with disinfectants and disinfection by-products [http://www.epa.gov/OGWDW/mdbp/dbp 1.html, December 4.
1998}.
Many water treatment plants are considering using monochloramine instead of free chlorine as a residual disinfectant because monochloramine usage minimizes the formation of disinfection by-products such as trihalomcthanes and haloacetic acids. Consensus is that monochloramine probably poses a lower risk of cancer than free chlorine when used for residual disinfection. A survey in 1989 and 1990 of municipal water utilities in the USA that serve populations greater than 50 000 found that 23% were using monochloraminc as residual disinfectant and that others were considering switching to monochloramine.~'
Currently. a typical monochloramine-using water
treatment plant uses free chlorine for initial disinfection and monochloramine for residual disinfection. Very little is known about the effect of monochloramine on Legionella. In the only study that attempted to measure this effect, monochloramine was added to a model water system containing planktonic (free-floating) Legionella in sterile water.
12
Monochloramine was more effective than free chlorine in this situation. There appear to be no data about the effect of monochloramine on amoebae. In recent investigations in which CDC participated. we noticed that Legionella was isolated from many drinking water samples in three municipalities that used free
105
( '/wprer ()
chlorine for residual disinfection hut that all sampled drinking water from four municipalities that used monochloramine \\as Lcgione//a free. ~-~- All samples had 1
been treated \\ith sodium thiosulphate immediately after collection to neutralize the disinfectant. To assess whether municipal \\·ater disinfection practices \\:ere associated with risk of legionellosis. we conducted a case-control study of hospitals.
Methods Identification of outbreaks and case definition A case was defined as a hospital in the United States that experienced at least one potable water-associated Legionnaires· disease outbreak. which \vas reported in a peer-revicv,·ed journal. at a scientific meeting, or as an official CDC report. V/c did a MEDLII\;E literature search for all articles with '"Lcgionclla'". "Lcgioncl/a pneumophilc/', '"legionellae ". ..Legionellaceae ", "legionellosis". or '"Legionnaires· disease·· as a key\vord or as a textword in the abstract, in combination with "nosocomial", "hospital". '"nosocomial infection". "hospital infection". or "'hospitalacquired" to identify all published nosocomial outbreaks of Legionnaires' disease in the United States between 1977 and I 997. Whenever an abstract suggested eligibility for inclusion in our study, the article \vas retrieved. Bibliographies of all retrieved articles were searched for additional pertinent articles or scientific presentations. We also reviewed CDC archives for publications. presentations. and unpublished CDC reports. We selected only those outbreaks in which the potable water system was implicated as the source by epidemiological study and confirmed by isolation of identical Lcgionc//a strains from patients and potable water. We excluded outbreaks associated \\·ith cooling towers or other aerosol-producing devices not directly associated with potable water. hut we included outbreaks associated with devices that had been contaminated directly from potable water. Information on location and timeperiod of the outbreak was extracted from the publication. \Vhcne\ er the published information was not sufficient for our study. one of the authors or the pertaining hospital infection control staff was contacted. Selection of controls Controls were defined as hospitals not reported to have had potable v·.:atcrassociated nosocomial Legionnaires· disease. \Ve randomly selected hospitals from the Arnerican Hospital Association (AHA) Guide to the Health Care Field. 1 ~ The outbreak-hospitals were predominantly large ( ,
CJ:) CJ:)
.....
Figure 1: Potable water-associated nosocomial Leg ionnaires' disease outbreaks identified through literature review in the United States , by year. Solid bars represent hospitals supplied with free chlorine-containing water, the shaded bar is the hospital that was supplied with monochloramine-containino wa ter.
108
Ej('eCI ofmonochlommine on Legionnaires 'disease
res idual di sinfecti on since 19 177 Thi s outbreak in vo lved 3 pat ients in I 98 1: one additi o nal case-pati ent was identifi ed in I 983 . c~ Th e main interve nti on co nsisted of remo val of fi ve large water tanks in whi ch wann water stag nated for peri ods of up to one week. and in stallati on of in stantaneo us water hea ters. Case-co ntrol stud y results We rand oml y se lec ted 50 co ntrol-hospitals. One control hospital was suppli ed with potabl e water th at was di sinfected with chlorine and mon oc hl oramine alternatingly, another was suppli ed with water di sin fec ted with chl o rine di ox ide. Th ese two hospitals were ex cluded from the stati sti ca l analys is. Pri vate well s suppli ed one case-hospital and one co ntrol-h os pital: hospital fac ility staff prov ided info rm ati o n on the se water suppli es . Th e geographi c di stributi on of cases and co ntrols is show n in Figure 2.
Residual disinfecta nt
• • ~
Free chlorine Monochloramine Other
hospitals
0 0 11
Figure 2: Geographical distribution of hospitals with rep orted Legionnaires' disease outbreaks associated with potable water and of randoml y selected control-hospitals. Some overlapping points were dispersed to improve legibility .
109
( "lwpter
(j
Characteristic
Hospital characteristics Number of municipal water distribution systems 200 beds or more Transplant program _rv1_edian year {range} Residual disinfectant Free chlorine Monochloramine Median free chlorine concentration (range)~ Median monochloramine _c:~.smcentration (range) 11 Initial disinfectant
Outbreakhospitals (n=32)
Controlhospitals (n=48*)
26
45
31 23 (72%) 1988.5 {79-97)
35 (73%)
46 1988{79~
31
36
10.2 ( 1.4 - 460)
1
12 +
1.0§
0 55 (0.0- 1 8)
0.6 (0.02.1) 2.7(1.54.3}
1.0(1.0-1.0)
Free chlorine
31
43
Monochloramine Alternating chlorine/ monochloramine
1 0
3 1
Ozone Water source Surface Ground __lli1ix of surface/ground Median no. of persons supplied by water utility (range)
0 22 5 5 242.500 (5,000-~Q.O_Q_QL__
Median pH of finished water
Adjusted OR (95% Cl}
7.6 (7.0- 10.0)
-----
--------
2.1 (0.16-114) 1
o§
Excluded Excluded
33 10 5
-----
198.000 (3,5009,000,0001__ 7.8(7.010.5
Table: Characteristics of case- and control-hospitals and their municipal water suppliers. Not including one hospital supplied with water containing chlorine dioxide and one hospital supplied with chlorinated water as well as chloraminated water from separate water treatment plants that supplied one water distribution system. Abbreviations and footnotes: Cl. confidence interval; OR, odds ratio; t, Mann-Whitney U test; ::j:, including one hospital supplied with water that contained free chlorine for two weeks per year; §, reference category; ~. values are calculated for those water systems that used this disinfectant and are the water treatment specialist's estimate of the concentration at the location of the hospital. Free chlorine concentration IS 1n mg/L, monochloramine concentration is expressed as mg/L combined chlorine; 11. analyzing the proportion surface water as a continuous variable.
110
!:'flee! o/!nonoc/IIorulllillc on Legionnain:.\ 'diseuse
\Vater supplied to case- and control-hospitals \\·as similar \\·ith respect to \\atcr source. size of population supplied by the water plant. pH. and all other water treatment parameters (Table). Case-hospitals were more Iikely than controls to he sup pi icd with water containing free chlorine as a residual disinfectant (crude OR = I 0.1: 95°1) confidence in ten al [Cl] = I A - 456. adjusted OR I 0.2: 95 11 o Cl = I A- 460 ). In the logistic regression analysis the following f~tctors did not contribute significantly to the model and were eliminated: time of the outbreak (or randomly assigned year t()r controls). population supplied hy the water plant. amount of \\·ater produced by the plant. type of initial disinfection. type of source \\·ater ( surl~tce or ground). concentration of the residual disinfectant. the interaction term of '"type of residual disinfectant" and concentration. pH of the finished \\ater. and the interaction term of pH and type of residual disinfectant. The only bctor that was signiticant was type of residual disinfectant. Although neither hospital size nor presence of a transplant program contributed significantly. these tactors were kept in the model to account for the matched study design. Using this logistic regre..;sion model. we found an adjusted odds ratio of I 0.8 (9Y~'o CI 1.9 to 203) f()r use of free chlorine vs. monochloramine as residual disinfectant. Because in the stratified analysis some cell \a lues were small or zero. we consider the adjusted odds ratio derived using the maximum likelihood estimation and exact confidence interval as the best estimation. The attributable proportion estimated from this odds ratio was 0.90 (95°;(1 Cl = 0.29- 0.99R). suggesting that approximately
90°/o of nosocomial potable water-associated Legionnaires· disease outbreaks could he prevented if every municipality would usc monochloramine as a residual disinfectant.
Discussion Our results show that use of monochloramine f()r residual drinking water disinfection \\as associated with a lower likelihood of potable \\ater-related Legionnaires· disease outbreaks. \1ore precisely. we h)lmd that hospitals supplied with water containing only free chlorine were I 0.2 times more likely to experience a potable water-associated outbreak that \\·as subsequently investigated and reported in a peer-re\ ie\\ed journal. at a scientific meeting. or as a CDC report. Our finding is supported by laboratory tests frequently recm ering Legion ella 1 1 from chlorinated municipal water 1'· '· 1 ' but not from ch loraminated \\·atcr 14 1' r 111 recent CDC-assisted im estigations.
\1onochloramine diffuses better into areas of
Ill
Chapter
n 1
stagnant water because of its slower decay. and it penetrates better into biofilm.'
Therefore. it may he more effccti,·e in the biological niche of Legionel/u. ,,·hilc its eftccti' eness against planktonic Lcgionello has also been reported. L'
Acti\ ity of
monochloramine against amoebae has not been studied but may be important. Studv limitations Hospitals \\ere classified as cases or controls depending on \\hdher outbreaks had been detected. im estigated. and reported. Some of the control municipalities may h'n e actually experienced legionellosis outbreaks that ,,·ere ne\'er detected or reported. resulting in misclassi fication. Ho,,·e, er it is reasonable to assume that. for any gi\'en outbreak. the likelihood of a hospital detecting and reporting this outbreak is independent of the municipal disinfection method because. to our kno,,·lcdge. this publication is the first report of an association between municipal disinfection and risk of Legionnaires' disease. Therefore.
non~detection
and non-publication of outbreaks
leads to non-differential misclassitication and can only introduce a bias toward the null.
In other words. it \vill lead to underestimation of the protecti\'e effect of
monochloramine.
The same is true for misclassi fication of case-hospitals due to
incorrect identification of potable water as the source of transmission.
Likewise.
individual hospital characteristics associated with risk of contamination of hospital water systems. such as temperature setting of the hospital water heaters and additional water treatment (e.g. copper-silver systems) by the hospital. are all independent of the choice of disinfectant by the municipal water authority and will result in a bias toward the null. We preferred to work with this nondifferential misclassification rather than interview each hospitaL because that would have introduced a reporting bias. A potential confounder could be the association between the size of the municipal water system and the size and number of hospitals it ser\'es. The need to maintain adequate levels of residual disinfectant O\'er longer distances sometimes is a reason for large-area water utilities to choose monochloramine. At the same time. risk factors for nosocomial outbreaks include hospital size 1~.~ 1 and the presence of 1
immunocompromiscd patients. especially transplant recipients. 2 · ~ Large tertiary-care hospitals are also more likely to be university-at1iliated and to have staff interested in publishing in peer-reviewed journals. Howe\'er. as a result of the matched design of the study. case- and control-hospitals were similar with regard to hospital size and "having a transplant program". and municipal water treatment plants supplying casehospitals were of similar size as those supplying controls (Table). The proportion of municipalities using monochloramine \aries by geographic location and may be highest in the Southeast.~ although none of our control-hospitals
112
f}!'cct o(monochloramine
011
l.egionnaires 'disease
in the Southeast \Verc suppl icd \Vith chloraminated water (Figure 2 ). There also arc regional differences in the incidence of Legionnaires' disease outbreaks. \\'ith the lowest incidence in the Southeast.~ \\'e were concerned that matching by geographic area could have resulted in ovcm1atching because the geographic variation in disease incidence may actually be caused by regional differences in drinking water disinfection practices. Excluding the Southeast from our analysis would evidently have resulted in a higher odds ratio estimate. In addition. regulation of hospital water temperature for control of scalding injuries. by State health departments. likely is not correlated with choice of drinking water disinfectant by municipal water authorities and therefore should not have been a confounder in this study. Implications Using monochloramine can be inexpensive: it involves installation of automated injection devices for ammonia and chlorine. and training of staff. A 1990 survey of four municipalities found that switching from free chlorine to monochloramine cost $5.000 to S I gs,OOO.x The higher of these four costs was incurred because that facility's building had to be expanded to house the additional equipment. Ammonia was purchased at a cost of $0.75 to $6.16 per million gallons of water treated. Chloramination of individual hospital water systems for control of nosocomial legionellosis is an option that deserves further evaluation. It is possible to install a chloramination booster station in a chlorine-using water system to chloraminate only a small part of that system. such as a hospital campus. This method may compare favorably to other options: it will cause less corrosion than continuous supplemental chlorination.x will not increase risk of scalding. and may he simpler. more eflective. and cheaper than other long-term disinfection methods. To our knowledge. this is the first report of a protective effect of monochloramine on potable water-associated Legionnaires· disease. This finding will need to be confirmed by other studies. The decision by municipal water authorities to switch to monochloramine depends on many complex factors; prevention of Legionnaires' disease may become an additional argument in fa\'Or of this switch. Most cases of Legionnaires' disease arc sporadic, but. like outbreaks. many sporadic cases have been linked to potable water. [t is estimated that 8.000 to 18.000 cases of 1 Legionnaires' disease occur in the United States annually. [f we assume that at least half of all Legionel/a infections arc acquired from potable \A/atcr. 90{Yo of these. or 3.600 to 8.100 cases. would be pre\,ented annually in the United States if all
113
( 'fwprt:r
(J
municipal itics would use monochloraminc. If O\ era II mortality 1s 25° o."' this could result in prc\cntion of900 to 2.025 deaths per year.
Contributors Jacob Kool noticed the reduced occurrence of Iegionellosis outbreaks in municipalities that use monochloramine. designed the study. collected data. did the analysis. and wrote the paper. Joseph Carpenter designed the study. colkcted data. contributed to the analysis and critically revie\ved the paper. Barry Fields contributed to study design and analysis. and critically re\iC\\·ed the paper.
Acknowledgments We thank Sharon Kooi-Hiemcke for general suppot1. Professor Da\ id Klcinbaum. Emory University School of Public Health. and Brian Plikaytis. CDC. reviewing the statistical analysis: Dr. Jay C. Butler. Dr. Robert F. Breiman. Cynthia G. Whitney. and Dr. Anne Schuchat. CDC. t()r critically rc\·Jewmg manuscript.
I l.f
G. f()r Dr. the
f}/t'cl 0/1110/IIJCizfoi"UllliiiC
rJ/1
f.cgl!!/11/Uil"n 'dtSt'U\t'
References I.
-+. 5. 6.
7. X.
'-1.
I 0. II.
12.
13. 14 I 'I
16.
17.
I X. ]l)_
.20.
\lar:-;ton BJ. Plouffe .11-. hk T\1. eta!. ln ..·idence of CllllltllUnit~ -acljuired pneunwnia requinng hospitali;auun. Results of a populatton-h:Lsed acti\e :-.une!llance study in ()hin . . lrc!J l/llan .\fed 1997: 157 1709-1 X. \1ar~ton BJ. Ltpman liB. Bre11nan RF. Suncillance fur Legionnaires· disea-.c Risk factors for mnrbtdity and mortality . . lrch lntcm _\/n/ltJli-+: 15-t: ~-+17-~.2. Bhopal RS. Fallon R.I. Buist IT. Black R.I. Lrquhart .ID. Pw:\llllity of th'-' home to a cooling lO\\l:r and risk of non- outbreak Legionnaires' disease. H.\!J dondon) 1991:302: ~7X-X~Stout JL Yu \'L. \1uraca P. Joly J. Troup\. Tompkins LS_ Potable'' ater as a caLhl' of sruradtc cases of community-acquired Legionnaires' disease .\ l:'ngi.J \!c·d 199~: .U6: 151-5. Straus \VL Plouffe .IF. File TM. Jr.. ct al. Risk t:tctors tix domestic acquisition of Legionnaires' disease . . lrch Intern \led 1996: 156: lliX)-92 Centers for Disease Control and Pre\ ention_ ( iuidelincs for pre\ entinn of nnsoeotmal pncumotlla . .\1.\IIIR .\lorh \lortu!ll'kh· Rl'fl 1997: -l6 (RR-1): I-7Y_ Dice J. Derner's se\·en decades nf e:-;perience \\ ith chloraminatton. J ./111 Jl'urer IJ'ork.\ .lssoc 19X5: 77: 3~-7. Kirmeyer CiJ. FDust ( i\\·. Picrsnn CiL Simmkr J.l. LeChe\ alier \1\\. Opttmtzmg chloramine treatment. Demcr: /\merican Water Works Research Foundation. ]lJ93. LcChC\alicr !\1Vv'. Cmthon CD. Lee !Hi lnacti\ation of hitl!llm bacteria. .IJijJ/ Ln1·iwn .\licrohiol 19XX: 5-l: 2~9~-9. Chen X. Stewart PS_ Chlorine penetration into artificial btotllm Is limited by a react1on-dt!Tusiun interaction. l:'m·iron .\ci Techno! 1996: 30 2078-X3. Samrakandi Mrvt Ruqut:s C. l\1ichel G. Influence of trophic conditions on exopolysacchande production: bacterial biotilm susceptibility to cblonne and monochloramine. Can ./ .\licmhiol 1'-197: -l3: 751-:-;. Cunliffe Di\. Inactivation of regionc!fa f111CIInwphila by monochloramine. J .1f1JII Hacrcriol 1990; 68: ~53-9. Centers for Disease Control and Pre\ cntion. Sustained transmission of nosocomial Legionnaires' disease-- Arizona and Ohio. \!\!fiR .\fnrh _\fnrru/lt'kf\· Nl'fJ 1997: -l6: ~ 16-~ I. Kool JL. Warwick !\1C. Pruckkr JM. Brown EW. Butler JC. Outbreak of Legionnaires' disease at a har alter basement flooding. l.unccr 199X: 351: l 0.10. Kool JL Bergmtre-Sweat D. Bnmn l:W. et al. The tepid /one: risk ftL·tors for nosocomial transmission of Legionnmre< disease and for colonrzation of hospital water systems by regionella. In: The Lighth Annual Meeting of the Society fur I kalthcare Epidemiology uf America. Orlando. \1t. Royal. \.1: Society for Healthcare Epidemwlog: of .\mcrica. 199~: 39. Koul .IL Flore :\I .. Kiuskt C\1. et al. \lor.: than ten ~car-. of unrecognized rwsocomi~tl transmission of Legtonnaires' disease among transplant patients. ln!'-'c! ( ·untmlllmjl Fpulemiol I99X: 19:XYX-90~. Buchhol/ L'. Peterson C. Kool J. ct al. 1-XOO-D'-.:.-\-TYPI - :-.trengths and lirmtations uf molecular subtyping in a community outbreak of Legionnaire:-.· disease in Lus A.ngclcs Count~. In: 3~th lnterscience Conference on Antimicrobial Agent.-; and Chemotherapy. San Diego Washington: American Society ti.lr \licrob10logy. (lJlJX: 5:-13. .\mcrican I lo-..pital .-\ -.,snL·iation The .-\II.-\ guide to the health car.: tiel d. Chicago: \mcncan llosprtal .-\s:-.oeiatton. llJlJ3 .\rnow Pl\t Chou T. \\eil D. Shapiro I N. Kret;schmar C '-.:osocomial Legionnam:~ · disL·as ..· caused hy aero-;oli/cd tap water from rc:-.piratury de\ icc-, .J ln/ct! /)/\ I9S~: l-l6: ~hll-7_ Be-,t \1. Yu \'L. Stout .1. Goct; -\. \luder RR. Ta> lor 1- Lcgronellaccae in the hospttal \\akrsupply. Lpidemiological link \\ ith dtscase and e\ aluatrun of a metlwd for cuntrol of rwsocomial Legionnaire-,· di-,easc and Pithburgh pneumonia. lw!cl'l l9X3: 2: 307-10.
115
Clwpta fJ 21. ')')
Blatt SP. Parkinson MD. Pace E. et al. Nosocomial Legionnaires· disease: aspiration as a primary mode of d1sease acquisition. Am .I .\fed 1993: 95: 16-22. Brady MT. 1\osocomial Legionnaires· d1sease in a children'.' s hospital. J Pediatr 19X9: 115: -+6-50.
Breiman RF. Fields BS. Sanden GN. Volmer L Meier A. Spika JS. Association of shower use with Legionnaires· disease. Possible role ofamoebae. JAJIA 1990:263:2924-6. 24. Brown i\. Lema M. Ciesielski CA. Blaser MJ. Combined plasmid and peptide analysis of clinical and em mmmental Legionella pneumophila strains associated with a small cluster of Legionnaires' disease cases. Infection 1985: 13: 163-6. 25. Ciesielski CA. Blaser MJ, Wang WL. Role of stagnation and obstruction of water tlow in isolation of Legionclla pneumophila from hospital plumbing. Appl Dn·iron lificrohiol 19X4: 48: 984-7. 26. 27. 2X.
29.
30. 31.
32.
33. 34.
35.
36. .1,7
Cordes LG. Wiesenthal AM. Gorman GW. et al. Isolation of Legionella pneumophila from hospital shower heads. Ann lmern ,\fed J9X I: 94: 195-7. Doebbeling BN. Ishak MA. Wade BH. et al. Nosocomial Legionella micdadei pneumonia: I 0 years experience and a case-control study. J /losp Jn(cct 1989: 13: 289-98. Graman PS. Quinlan GA. Rank JA. Nosocomial legionellosis traced to a contaminated ice machine. lnj'ect Control Hosp l:j;idemioll997: 18: 637-40. Green M. Wald ER. Dashefsky B. Barhadora K, Wadowsky RM. Field inversion gel electrophoretic analysis of Lcgionella pneumophila strains associated with nosocomial legionellosis in children. J Clin .\ficrohiol 1996: 34: 175-6. Hanrahan JP. Morse DL, Scharf VB, et al. A community hospital outbreak of legionellosis. Transmission by potable hot water. Am J l:jJidemiol 19R7: 125: 639-49. Heimberger T, Birkhead G. Bomstein D. Same K. Morse D. Control of nosocomial Legionnaires' disease through hot water flushing and supplemental chlorination of potable water . ./Infect Dis 1991: 163: 413. Helms CM, Massanari RM, Wenzel RP, Pfaller MA, Moyer NP, Hall N. Legionnaires' disease associated with a hospital water system. A five-year progress report on continuous hyperchlorination . .lAMA 1988: 259: 2423-7. Johnston JM, Latham RH, Meier FA. et al. Nosocomial outbreak of Legionnaires· disease: molecular epidemiology and disease control measures. Infect Control 1987; 8: 53-8. Johnson JT. Yu VL, Best MG. et al. Nosocomial legioncllosis in surgical patients with headand-neck cancer: implications for epidemiological reservoir and mode of transmission. l.ancl.!f 1985: 2: 298-300. Lowry PW. Blankenship RJ. Gridley W, Troup NJ, Tompkins LS. A cluster of Legionella sternal-wound infections due to postoperative topical exposure to contaminated tap water. N Engl.! .\led 199 L 324: I 09-13. \1astro TD. Fields BS. Breiman RF. Campbell J. Plikaytis BD, Spika JS. Nosocomial Legionnaires· disease and use ofmedicauon nebulizers . ./ ln/ecr Dis 1991:163:667-71. Matulonis U. Rosenfeld CS, Shadduck RK. Prevention of Lt'gionella infections in a hone marrow transplant unit: multifaceted approach to decontammation of a water system. lnji:ct Control /Imp /:jJidcmioll993: 14: 571-5.
JX.
Mermel LA. Josephson SL, Giorgio CH. Dempsey J, Parenteau S. Assoc1ation of Legionnmres· disease with construction: contamination of potable water? ln(ecr Control limp Epidemiol 1995:16:76-81.
40
116
Nafziger DA. Successful chlorinat1on for l.egionella at Henry Ford Hospital. ln(ecr Dis Clin Pracr 1997: 7: 118. Nolte FS. Conlin CA. Roisin AJ, Redmond SR. Plasmids as epidemiological markers in nosocomial Legionnaires' disease . ./Infect /)is 1984: 149: 251-6.
E.ffect (~(monochloramine on Legionnaires 'disease 41.
42. 43.
44_
45. 46.
47.
48.
49_
50.
51.
Parry MF. Stamplcman L Hutchinson JH. Folta D. Steinberg MG. Krasnogor LJ Waterborne Legionclla ho::enwnii and nosocomial pneumonia in immunosuppressed patients. Ann hucm Med 1985: l 03: 205-10. Plouff(: JF. Para MF. Maher WE. Hackman B. Webster L. Subtypes of [,l!gionl'lla pncumophila serogroup I associated \vith different attack rates. Lancer 19X3: 2: 649-50. Plouffe JF. Webster LR. Hackman B. Relationship betv.. een colonization of hospital building with Legionella pneumophila and hot water temperatures. .4.ppl l:'nl·inm .\ficrohiol 1983: 46: 769-70. Schoonmaker D. Heimberger T. Birkhead G. Comparison of ribotyping and restriction enzyme analysis using pulsed-field gel electrophoresis for distinguishing Legiondla pneumophila isolates obtained during a nosocomial outbreak. J Clin Microhiol 1992: 30: 1491-X. Shands KN, Ho JL Meyer RD. et al. Potable water as a source of Legionnaires' disease . .JA.\1.-1 1985; 253: 1412-6. Snyder MB. Siwicki M. Wireman J, et al. Reduction in Legionella pneumophila through heat flushing followed by continuous supplemental chlorination of hospital hot water. .J ln/i..'cl Dis 1990: 162: 127-32. Venezia RA, Agresta MD, Hanley EM. Urquhart K. Schoonmaker D. Nosocomiallegionellosis associated with aspiration of nasogastric feedings diluted in tap water. ln/f..'cl Control f!osp Epidemio/1994; 15: 529-33. Hanrahan JP, Morse DL, Scharf VB, et al. Community hospital legionellosis outbreak linked to hot-water showers. In: Thornsberry C, Balows A. Feeley JC. eds. /.egionella: Proceedings of the 2nd International Symposium, Washington DC: American Society for Microbiology, 19R4: 2245. Lepine L, Jernigan DB. Wyatt B. et a!. Use of urinary antigen testing to deteL"t an outbreak of nosocomial Legionnaires' disease. In: Abstracts of the 35th lnterscience Conference on Antimicrobial Agents and Chemotherapy. Washmgton DC: American Society for Microbiology, 1995: 267. Sellick JA, Mylotte JM. Nosocomial Legionella pneumophila pneumonia in a hospital with an instantaneous hot water heater. In: Barbaree JM, Breiman RF. Dufour AP, eds. Legione!la: Current Status and Emerging Perspectives. Washington, DC: American Society for Microbiology, 1993:43-5. Alary M, Joly JR. Factors contnbuting to the contamination of hospital water distribution systems by legione1lae . .J lnj'ec£ /)is 1992; 165: 565-9
117
C"hapra tJ
Addendum to Chapter 6 Th~
authnr of this
th~sis
has collahoratcd
a
111
numh~r
C\ aluat~ the potential \ aluc of monoch loraminc for
pr~\
of follm\
~up
studies to
ention of Legionnaires·
disease: •
The CDC's
R~spiratory
Di~~ases
Branch and llospitcll
Infections Program
collaborated \\ith the Society l()r Healthcare Epidemiology of i\merica in a sur,·ey of more than -lOO hospitals. I Iospitals \\ere asked about occurrence of cases of Legionnaires· disease among their patients and this data ,,·as correlated with municipal \\akr disinkction practices. The results confirmed the protecti\e effect of monochloraminc against Legionnaires· disease (J. llcfkltinger. S.K. Fridkin . .J.L. 1\.ool. \·. Fraser. J. llag~man. B. Kupronis. L Zell. C.G. V..'hitncy. Association hct\\Cen nwnochloramin~ usc hy municipal ,,·ater treatment plants and nosocomial Legionnair~s · disease. lit Deccnniu/ lntemurionu/ Cun{crencc on \"osocnlllia/ und llculthcarc~.·lssociurcd
•
/nf('criuns. Atlanta. !\larch
5~9.
2000 ).
Assisted by Pathcon Laboratories of Norcross. Cieorgia. a hospital in \Vashington DC did a pilot study '' ith a prototype small-scale monochloraminc injector de\ icc (Donegan N. Pic~i\luas L. Witherell L. Shelton B. Kool J. Flanders D. Short trial of rnonochloraminc for control of f_cgionclla in a hospital setting. ..;.rh Dcn:nnial lntemationul Con(i.>rcnce on .Vosocnmiol and
Atlanta. March •
5~9.
llcalthcore~Associotcd
ln(cctions.
2000) ..
CDC is setting up a \\·ater sampling study to compare pre\alcncc of Lcgionclla contamination of water systems (of hospitals as well as private homes and hotels) bel()rc and after implementation of residual disinfection \\ ith monochloramine. Sc\ era I large metrnpolitan \\ater systems ha\C hecn identified that arc planning to S\\ itch hom th:-c chlorine to monochloraminc. \\"atcr \\ill be sampled systcmatically before and alter the S\\itch and \\ill he tested for presence of /_cgitlliel/u.
II X
Chapter 7: General discussion Control of Legionnaires· disease has been hampered by two deeply rooted misconceptions among clinicians and public health officials: Misconception #1: "Legionnaires' diseasl! is rarl!" Legionnaires' disease is, in fact, among the three or four most common causes of community-acquired pneumonia in many industrialized countries.
Studies that
evaluated the etiologies of se\·ere community-acquired pneumonia (CAP) found
Legionella to be the causative agent in about X'% of cases (range: 2'~-16%). 1 - 11 Moreover. in about half of all CAP cases. an etiologic agent is not identified even \vith comprehensive diagnostic testing. The steady stream of reports of newly discovered pathogenic Legion ella species and Ll!gionella-like organisms ( LLOs) is an indication that many more strains are not yet known and therefore cannot be diagnosed at this time. Therefore it is very well possible that this proportion may in reality he higher than X0/c1. In a study to assess the role of LLOs in community-acquired pneumonia, 19'Yr) of patients with pneumonia of unidentified etiology demonstrated a ~ fourfold
rise in antibody titer to~ I :128 to at least one of the LLOs tested*.l.2.u In the United States. 2-3 million cases of CAP occur each year, of which 500.000 are so severe that they need to be hospitalized. 2 Little data arc available on the proportion of less severe CAP caused by Legionella: proportions ranging from O'Yr1 to 12'~1o have been rcported.
14 1 - :-.:
It seems reasonably conservative to assume that at
least X0I!J of severe CAP and 4'% of less severe CAP are caused by Ll!gionella or
Legionella-like organisms.
This v.:ould imply an annual incidence of I 00.000 to
140.000 cases of Legionnaires· disease. including 40.000 severe cases requiring hospitalization.' The number of Legionnaires' disease cases detected by passive surveillance is invariably much lower, likely because most cases arc never diagnosed.
19
By deciding
* This 1ras a non-hlinded serological inn'stigation 1rith the inherent potentwl o(cm.'>srcaction and /ahoratorian hias: tlu:re/iJre the exact proportion o(C.4.P caused h.1· Lcgionc!lalike organisms remains W1ct>rfain. ·r .\!any CDC puhlicarions. including some clwprcn in rhis thesis. use estimates o(8.f){J019 /8.00() 1 or I 7.f)(}{J-:!3.0()() case\ o/.\LTere communi~\·-acctuired Legionnaires· disea\C re(j/1 iring hospi fa li::.a lion: i 11 1·inr of' rhc ahoH'-prcsen red inj()rma tion this mar he an undcrcsr i /Jill tc.
119
that it is not necessary to request diagnostic tests for /_egioni!lla because they assume that Legionnaires· disease is rare. physicians arc unconsciously perpetuating the \·icious circle. \lisconception #2: "In most cases the diagnosis o(Lcgionnaires ·disease can he made or excluded on !he husis o/tlw clinical picture ..
\\'hen Legionnaires· disease had just been discm ered. it was thought to he one of the '"aty-pical pneumonias"'. characterized by a distinct clinical syndrome and accompanied by specitic electrolyte disturbances and X-ray presentation. In later years it has become apparent that this is not correct. Numerous well-designed studies looking at predictive value of\·arious combinations of clinical symptoms and chest radiograph presentations have found that Legionnaires' disease cannot he distinguished from pneumonia due to other pathogens on clinical grounds: specialized ; -\ 'II~"' Iahoratory tests arc necessary. · - -· These misconceptions probably result in unnecessary morbidity and mortality: it is Iikely that many patients arc not adequately treated for their Legionnaires' disease and that outbreaks of Legionnaires' disease arc missed. In this thesis I have presented an outbreak that continued unrecognized, probably for more than 18 years, with a btality rate of 48°;().~~> In another investigation reported in this thesis. ongoing transmission in multiple hospitals was only recognized after introduction of the /_cgionclla urinary antigen test.~~
What should be done to reduce morbidity and mortality due to Legionella? Surveillance f\1ost important is that diagnostic tests for regione/lu should routinely he requested whene\er an effort is made to identify the causati\C agent in communityacquired- as \\ell as nosocomial pneumonia. This will assure appropriate treatment timely recognition of outbreaks. and a realistic impression of the burden of disease caused by Legionella. Medical schools and pub Iic health authorities should play a leading role in changing physicians· knowledge and attitudes. Many physicians still arc unfamiliar with the advantages that the urinary antigen test offers. In th1s thesis we hmc shown that introduction of this test can lead to recognition ofprC\ iously unrecognized outbreaks and cases_:- The test is rapid and
120
General di.'>cus·sion it can detect approximately 70% of prevalent Legion ella strains with a sensitivity of 80% to 9(YYo and a specificity of>99°1r); this probably makes it the most reliable test available today.~:;.:-; An immediate bedside test for Legionella urinary antigen (BINAX NOW), which will soon become commercially available. has similar sensitivity and speciticity.
2 24 :-.:.
Culture of respiratory secretions on special media has
the advantage that it can detect more species and serogroups of Legion ella but unfortunately the sensitivity of this technique varies significantly depending on the level of expertise of the clinical laboratory and the quality of growth media. Thirty percent of otherwise sophisticated laboratories are not able to recover Legion ella from a pure culture,
25
and one study reported a sensitivity of culture of routinely handled
specimens of only 11 °!rJ.
30
Nevertheless. culture is a valuable diagnostic test,
especially when used in combination with the urine antigen test. Culture also has the advantage that Legionella strains can be compared to environmental isolates during outbreak investigations to help identify the source. Physicians should be aware that Legionella serology has limited value in clinical practice. Public health authorities should promptly investigate clusters of Legionnaires' disease. Such investigations may uncover instances where guidelines are not properly adhered to, as illustrated in Chapter 3 of this thesis. This thesis has also presented two investigations in which unexpected and previously unknown sources were implicated (a sump pump and a carpet cleaner).
3126
Even when all Legionella prevention
guidelines are followed, such unknown sources will continue to cause disease until they are implicated by an epidemiological and microbiological investigation. This thesis also illustrated the importance of epidemiology and the limitations of microbiology in the investigation of legionellosis outbreaks (Chapter 3 ). Pneumonia case management When a physician decides that it is not necessary to determine the etiologic agent of CAP, as happens in the majority of cases, patients should receive empiric antibiotic treatment that covers all common causes of CAP, and this includes Legionella.
2
This rules out empiric treatment with ~-Jactam antibiotics unless they are
combined with a macrolide or a quinolone. Environmental control measures. Many texts about optimal environmental prevention strategies have been written. The aim is to reduce the prevalence of pathogenic Legionella strains in the environment. One approach could be widespread testing of drinking water and of high-risk devices such as wet air-conditioning devices and whirlpools for presence of Legionella and then implementing control measures whenever the bacterium is found.
121
Chapter 7
This approach has significant practical drawbacks. Lcgionc/la is a ubiquitous organism: it is found in most natural waters, in up to 32°/;) of residential home water systems, and in up to 1ooxpcriencc is gained ~rirh practical application ofthis m/e_
124
General discussion
important design rcquirements.
1 Y
In addition to maintaining a high water temperature.
these hospitals should ensure and monitor an adequate residual disinfectant concentration in their water system. Adequate residual concentrations are :;:=: 1.0 mg/L free chlorine and (probably):;:=:: 1.5 mg/L monochloramine. 27 · 43 · 4 ~>.-I?.S-I These concentrations should be attained at all points of use (showers, faucets, etc.). Many hospitals supplied with chloraminated municipal water would not need to do anything to achieve the required concentration, but most hospitals located in free chlorine-using municipal water systems would need to install chlorine injection equipment. A copper-silver ionization system in combination with a minimum free chlorine concentration of 0.5 mg/L probably is also appropriate. Limited data is available about minimal inhibitory levels of other disinfectants such as chlorine dioxide and ozone; additional studies may be necessary for these disinfectants. Hospitals in category 2 might also be required to test their water for Legionella at regular intervals (e.g. 2 to 4 times per year) as one method to evaluate the effectiveness of the control measures. When Legionella is found in these hospitals then adherence to the aforementioned guidelines will need to be evaluated and corrected. However, negative environmental microbiology results do not mean that
Legionella is totally absent and microbiological testing should never replace regular monitoring of water temperatures and disinfectant concentrations in patient rooms. as was also illustrated in this thesis_.::n.::>? Laboratories examining environmental samples for Legion ella should be subject to a program of quality assurance. The above measures are simple, less expensive and probably more effective than other strategies, and will undoubtedly result in a considerable decrease of nosocomial Legionella infections in the United States. They should become part of standard hospital licensing regulations.
Legionella in The Netherlands Incidence of Legionnaires' disease In the Netherlands, the proportion of Legionella-caused CAP probably is similar to that in the US and other industrialized countries. A prospective study of CAP requiring hospitalization in 1991-1993 found Legione!la to be the causative agent
125
C'/wpter . .,
in approximately 5° c) of those cases that were tested for /_cgionella infection, .
111
Howe\ cr. this paper only used serology for detection of Legionnaires' disease and the time bet\\·een collection of acute- and com alcscent phase sera' aried from 2 to 6 weeks. It can take X weeks or longer for f_f!gionella serocom ersion to occur. Therefore this study probably did not detect all Legionella infections. :\ retrospecti\ e anal.ysis of patients hospitalized for pneumonia in The Netherlands found l_cgionellu 11 to be the causative agent in three (5 11 o) of62 cases ofse\ere CAP. That study also relied on serology for the diagnosis of Legionnaires· disease. and apparently this \\·as not perf(m11ed for every patient. The results of the abm e studies suggest that f_egionella causes a similar proportion of CAP as in other countries. which is generally estimated around X11 o (sec also the C.ienerallntroduct1on). About 110.000 cases of CAP occur each year in The Netherlands. of which approximately 12.000 are hospitalized." Follm,ing the same calculation as that used for the LS in the first part of this chapter.' in The :--.Jethcrlands there would be appro.ximatcly 5.300 cases of Legionnaires· disease per year. including about I.XOO se\ ere cases. This is a conscn arive estimate. /\not her way to estimate the true incidence of Legionnaires· disease is by 6 looking at the rate of underreporting. ~ Approximately 0.26 cases of Legionnaires' disease per I 00.000 persons arc reported through passi\'c sur\'eillance in The 1 Netherlands. This is similar to the reported rate in the US ( ~0.2 per I 00,000 ). Prospecti\'e studies from the US ha,·e indicated that the true incidence may be 30-50 times higher because most cases arc probably never diagnosed::'. I
'
C\'en more cases may be missed because. until recently. Dutch physicians were trained to only test for Lcgionella when there \\as '"clinical C\ idence" of Legionnaires' disease or if the patient had recently returned from foreign tra\'cl.' Introduction of the urinary
!he 111uin rnu!rs roh!e n/ that J)(IJh'l' lists the Jii'Oj)()rJion cuu,cd h1· /,egione//u U\ :!.'' "· 1/mnTcr. the de!WIIIinatnr fi;r that Jh'n enrage includes jJilffcllls 11/w trerc not rnrcd/nr l.egiunc//a in/ccrion. Fmm rhc rcxr it can he in/erred thor among puricllfs thor 11 ere tested Jnr tcgionc//a. 5"" lwd u posirire result. This Jiroportion mts con/inned hy one o/thc mrthon o( the Jhlfh'l' rpcnrma/ communicurirm R. Bohle . .\'on'mlwr I i)l)iJ). When colllfJuring the /Jutchjigurcs to the L ·.)' 1111111hcn one notices that in The .\'etha/und' uhour /()''" o/C.·IP case' arc hos;ntuli::cd. 11 hi/e this is UJiJWrninwrclr }.()",,in the L ·.\. Thi.\ jimhuhlr rc/lecrs u /om'!' t/11'(.'.\/wld/or hmpirali::ution in rhc L S. not a higher Jimportinn o/ \('\·ere cases. ,{nee in the l ·.\the /atalfl\·rutiu of( ·.IP f., ilfJJn·u.rimurefl· }.",, .~ cnll!j){trcd to 5",, in The .\ctherlund\." For rhe Jnllpwc n/thi' culculorirm. it 11 w O\SIIIIIcd that in hurh cnuntrin ahuut :!O''u o/( ··I P Cil'l'' cun he c lussi/icd u., \C\ ere ( ·.J/ 1 und thur tht• ( nntnhutinn of l.cginnd lo lor l hi' grnu;J is S" "· For rhe rcmu in i ng le" sc\ ·en' C·. I P ut.\n. ir 1111' 11 "tllnnl thur the jil'()fn!Uinn cU/1\cd h1· l.cgirmdla /1 ../''u.
126
General dis(·ussi1J11
antigen test has not been as quick as in some other countries.·'~ A guidebook published by the main health insurance organization in The :\etherlands in 1997 did not mention the urine antigen test: it recommended the use of paired serology for diagnosis of Legionnaires' disease. but only in case of clinical suspicion for the disease. It stated that a period of 2 to 3 weeks between collection of acute- and con\ alescent-phase sera 9 was sufficient ' howen:r. 6 to 9 weeks is what is generally recommended by other texts.~·' As discussed in the introduction. serology rarely· leads to diagnosis of Legionnaires' disease cases and this \vas also illustrated in one of the investigations reported in this thesis ..::~ During a large outbreak of Legionnaires' disease at a flower show in February 1999, none of the local hospitals \Vhere patients were admitted had 1 11 the urinary antigen test at hand. ' It is therefore clear that the rate of under-diagnosis in the Netherlands can he expected to be similar or worse than in the United States. Again the conclusion is that se\·eral thousand cases of Legionnaires' disease occur in The Netherlands each year. In 1998, the number of reported cases of Legionnaires· disease per I 00,000 persons was 8 to 30 times higher in Ciem1any. Denmark. the United Kingdom. and France than in The Netherlands.(l 1 It appears that these countries have been quicker to adopt the urinary antigen test as a routine diagnostic tooLA.::.h.~ and physicians may have a higher index of suspicion for the disease. It seems unlikely that climate. geography. drinking water quality. or local ecology of Legiondla \Vould account for the large differences between these European countries. Diagnosis and treatment of community-acquired pneumonia in The Netherlands As mentioned. acceptance of the Legionella urinary antigen test has been slow 1 among Dutch physicians. '' .611 likely resulting in under-diagnosis of cases. This thesis has illustrated the importance of this test.~~ It should become one of the tests of first choice for patients with community-acquired as \\·ell as nosocomial pneumonia. A new rapid bedside urinary antigen test has been developed, which has a sensitivity and specificity similar to the "classic" ELISA and RIA urinary antigen 1 tests.~x ..::• \Vidcsprcad usc of this test by general practitioners could significantly reduce the problem of under-diagnosis and prevent therapy with ineffecti\ e antimicrobial drugs. Guidelines for empiric treatment of CAP in The Netherlands. published in 2 1998. are different from published US and L'K guidelines. The Dutch guidelines recommend drugs (such as amo.xycillin) that ha\e no activity against Legionella and suggest that clinicians can still consider the diagnosis Legionnaires' disease \\hen 114 there is no improvement after 4X hours. Delay of appropriate treatment of
127
( "/wpter
7
11
Legionnaires' disease. h(m:cver, is associated \\'ith higher mortality. ~ The aforementioned guidelines may ha\ e been based on the incorrect premise that legionellosis would be rare in The Netherlands. Because Legionella spp .. Chlamydia pneumoniac. and .\(n·oplasma Jmeumoniae cause a significant proportion of CAP in The Netherlands. guidelines for empiric treatment should be amended to include drugs with activity against these pathogens as well as against pneumococci and llaemophilus injlucn::ae. Good drug choices include macrolides (azithromycin or erythromycin: may also be combined with a ~-lactam antibiotic). the newer quinolones ( levofloxacin ). and doxycycline.::: A prospective population-based study should be done to determine with more certainty the relative importance of various infectious agents in the etiology of CAP in The Netherlands. and this study should employ the latest diagnostic methods available. Legionella in Dutch drinking water
Community-based and nosocomial outbreaks as well as sporadic cases have been reported in The Netherlands. and identified sources included the usual suspects: potable water. cooling towers. whirlpool spas.r>"-'' The majority of the Dutch population is supplied with drinking water that docs not contain any residual disinfectant. This is done to minimize formation of potentially unhealthy disinfection by-products and to address consumers' demands for a good taste. 74 Dutch drinking water authorities believe that residual disinfection is not necessary provided that good engineering practices are used to produce "biostable" water (i.e. water with such a low nutrient content that bacterial growth is not 4 supported) and that plumbing materials arc selected that minimize biofilm formation.~ Other experts in the field of drinking water treatment haYe argued that it is safer to always to use a residual disinfectant.'~ In an interview published early 1999. Dutch drinking water treatment authorities stated that "a little more than 45" cases of Legionnaires' disease occur annually in The Netherlands. implying that Dutch drinking water is much safer when it 11 comes to Legiondla than, for example. United States water. ~ As discussed in this thesis. the number 45 is based on passive surveillance, which suffers from significant under-reporting: it is more likely that several thousand cases of Legionnaires· disease occur annually in The Netherlands and indications are that incidence of Legionnaires· disease is at least as high as in the United States. The author of this thesis is skeptical about the notion that the Dutch drinking water policies arc safe with regards to Legiondla. Sessile biotilm bacteria such as IA:gionc/la arc capable of sur. iYi ng in very nutrient-poor em ironmcnts. Potable
128
General discussion
water-associated outbreaks occur in The Netherlands.""-
711 7 7 _ .:. !-i
implying that Legionella
can enter Dutch water systems and that it can amplify to dangerous concentrations. A large-scale sampling study of Dutch potable water systems in 1988 found that approximately 30rYo contained Legiondla.'ct Control 1998: 26: 8-11. Fiore A E. Butler JC, Emon TG. Gaynes RP. A surny of methods used to detect nosocomial legionellosis among participants in the National Nosocomial Infections Surveillance System. Infect Control Hosp l:/ndemiol 1999; 20: 412-6. Kool JL Carpenter JC. Fields BS. Effect of monochloramine disinfection of munic1pal drinking \Vater on risk of nosocomial Legionnaires' disease. Lancet 1999; 353: 272-7. Heffelfinger J, Fridkin SK. Kool JL et al. Association between monochloramine use by municipal water treatment plants and nosocomial Legionnaires' disease. 4th Decenniul lntenwtional Conf(Tence on Nosocomial and Healthcare-Associated Infections 2000. National Center for Em·ironmental Health, National Center for Infectious Diseases. Final recommendations to minimize transmission of Legionnaires' disease from whirlpool spas on cruise ships. Atlanta: Centers for Disease Control and Prevention. 1995. Anonymous. Legionnaires disease associated with a ~Ahirlpool spa display-- Virginia, September-October. 1996. /14A1WR Morh Mortal Wkh· Rep 1997; 46: 83-6. Steenbergen .IE, Slijkerman FAN, Speelman P. The first 48 hours of investigation and intervention of an outbreak of legionellosis in the Netherlands. EuroswT 1999: 4: 112-5. Anonymous. Legionellosis. trade fair- Belgium (2). ProMED muil. 18-11-1999. Redd SC, Schuster OM, Quan J, Plikaytis BD, Spika JS, Cohen ML. Legionellosis m cardiac transplant recipients: results of a nationwide survey. J Infect Dis 1988: 158: 651-3. Anonymous. The AHA guide to the health care field. Chicago: American Hospital Association, 1993. Cunliffe DA. Inactivation of Lcgione/la pneumophila by monochloramine . .J Appl Bacteriol 1990: 68: 453-9. Hustinx W. lnfecties van de lagere luchtwegen. In: Maas lAM, Gijsen R. Lobbezoo IE. Poos MJJC eds. Volksgezondheid toekomst verkenning 1997. 1: de gezondheidstoestand: een actualisering. Amsterdam: Elsevier. 1997: 482-9. Bosman A. [l.egione/la in public shower systems]. 1996: r-.1cenhorst PL. [Lower respiratory tract infections in adults]. In: Furth Rv. cd. Leerboek infectieziekten [Textbook of infectious diseases]. Houten: Bohn Stafleu van Loghum, 1992: 179-98. Bosman A. Kessel R. [Legionnaires'discase in Europe, 1996]. Injecue::iekren Bulle/in 1997: 8: 216. Anonymous. Diagnostisch Kompas. Amsteheen: Zickenfondsraad. 1997. Kohler. W Bactcrietlora: Legione//a is een gcwone water- en grondbacterie. NRC Hande/.1hlad 28-~-1999: 45. Anonymous. Legionnatres' disease. Europe. 1998. Weckh· Epidemiological Record 1999: 74: 273-80. Uldum S. Legionella infections 1998. !:"pi-.\"eu s Denmark 2000; I. lnfuso A. Hubert B. Etienne J Underreporting of Legionnaires'disease in France: the case for more active surveillance. !:"umsuiT 19YX: 3: 48-50.
General discussion 64.
Kasteren MEv. WiJnands WJ. Stobbcnngh EE. Jankncgt R. ,·an der \1ecr JW. [Optimization of the antibiotics policy m the Netherlands. II. SWAB guidelines for the antimicrobial therapy· ofpneumonia in patients at home and as nosocomial infections. The Netherlands Antibiotic Policy Foundation] . .Vcd T!Jdschr Gl!nceskd 199g; 142: 952-6. 65. Heath CH. Grove DI. Looke DF. Delay in appropriate therapy of legionella-pneumonia associated with increased mortality. Eur J Clin Microhiollnji.xt Dis 1996: 15: 286-90. 66. Leverstein-van Hall MA. Verbon A. Huisman MY. Kuijper EJ. Dankert J. Reinfection with Legionella pneumophila documented by pulsed- field gel electrophoresis. Clin ln(ect Dis 1994; 19: 1147-9. 67. Meenhorst PL. van der Meer JW. Borst J. Sporadic cases of Legionnaires' disease in the Netherlands. Ann lntt'rn !vied 1979; 90: 529-32. 68. Meenhorst PL. Reingold AL. Groothuis DG. ct al. Water-related nosocomial pneumonia caused by Legiondla pneumophila serogroups I and I 0. J ln/i!ct Dis 19X5; 152: 356-64. 69. Ketcl RJv. Rictra PJ. Zanen-Lim OG. Zanen HC [An epidemic of pneumonia caused by Legion(:'/ fa pneumophila in a Dutch hospital]. l'ied Tijd. abO\C.
142
Bibliography I.
Kool JL Buchholz U. Peterson C. Bro\\·n E\V. Benson RF. Prucklcr JM. Fields
BS. Sturgeon J. Lehnkering E. Cordova R. Mascola LM. Butler JC. Strengths and limitations of molecular subtyping in a community outbreak of Legionnaires' disease. Submitted. 2.
3.
4.
5.
6.
7.
X.
9.
Buchholz U. Richards C. Peterson C. Kool JL Pon D. Murthy R. Boghossian N. Mascola LM. Jarvis \\'. The Big Chill: An outbreak of endotoxin-like reactions associated with intravenous gentamicin- Los Angeles. llJ9lJ. Submitted. Heffelfinger JD. Fridkin SK. Kool JL. Fraser V. Carpenter J. Hageman JC. Zell ER. Kupronis B. Whitney CG. Association between monochloramine use by municipal water treatment plants and nosocomial Legionnaires' disease. 4th Decennial International Conference on Nosocomial and Hcalthcare-Associatcd Infections, Atlanta. March 2000. Donegan N. Pic-Aiuas L. Witherell L. Kool J. Flanders D. Shelton B. Short trial of Monochloramine for Legionel/a Disinfection in a Hospital Setting. 4th Decennial International Conference on Nosocomial and Hcalthcare-Associated Infections, Atlanta. March 2000 (Late-breaker session). Nuorti JP. Butler JC. Gelling L. Kool JL. Reingold AL. Vugia OJ. Epidemiological relation bet\vecn HIV and invasive pneumococcal disease in San Francisco County. Cali fomia. Anna/..,· of Internal Afedicine 2000: 132: 182-90. Kool JL. Bergmirc-Sweat D. Brown E\\/. Fields BS. Benson RF. Pruckler JM. Carpenter JC. Kolczak M. Butler JC. Hospital characteristics associated with colonization of water systems by /"egionella and risk of nosocomial Legionnaires· disease: a cohort study of 15 hospitals. ln(ec!ion Control and 1/ospira/ LjJidemiolog~· 1999:20: 798-X05. Kool JL. Carpenter JC. Fields BS. Effect of monochloramine disinfection of municipal drinking water on risk of nosocomial Legionnaires' disease. Lancer 19Y9:353 :272-77. Kool JL. Fiore /\E. Kioski CM. Brown L\V. Benson RF. Pruckler JM. Glasby C. Butler JC. Cage GO. Carpenter JC. Mandel R!\1. England B. Breiman RF. More than ten years of unrecogni.t:ed nosocomial transmission of Legionnaires· disease among transplant patients. Jnf'ecrion Control and Hospital Epidemiolog_1· 1998: llJ:89X-904. Sclnvartz B. Kolczak MS. \Vhitney CG. Kool JL. Schuchat A. LJ.S. counties with higher rates of antibiotic usc ha\ e significantly higher proportions of heta-lactam
143
Bihliograpln·
and macrolide nonsusceptible .)'. pneumoniae antimicrobial resistance. ICAAC
199:-\. 10. Gomez L Moore A. Sanchez J. Kool J. Castellanos PL. Feris J. Kolczak M. Levine OS. The epidemiology of Haemophilus influen::ae type B carriage among infants and young children in Santo Domingo. Dominican Republic.
The
Pediatric !nf'ectious Disease Journal 199R Sep: 17(9 ):782-6.
11. Gorkom J van. Leentvaar-Kuijpers A. Kool JL. Coutinho RA. [Annual epidemics of hepatitis A in four large cities related to holiday travel among immigrant children]. Nr!dr!rlands Tijdschrift voor Genf!eskundf!. 1998: 142(34 ): 1919-23.
12. Kool JL,
Warn·ick MC, Pruckler JM. Brown EW. Butler JC.
Outbreak of
Legionnaires· disease at a bar after basement flooding. Lancet 1998;351: I 030.
13. Fiore AE, Kool JL, Carpenter JC. Butler JC. Eradicating Legionella from hospital water: author's reply . .lAMA 1997: 278:1404-5.
14. Fiore AE. Kool JL. Carpenter JC. Butler JC. et a!.
Sustained transmission of
nosocomial Legionnaires· disease--Arizona and Ohio.
AfAHiR 46( 19 ):416-21.
1997 May 16. 15. Anonymous. Progress toward poliomyelitis eradication--Peoples Republic of China, 1990-1996. !vfMWR 45(49):1076-9. 1996 Dec 13.
16. Anonymous.
Progress toward global poliomyelitis eradication.
1985-1994.
MMWR 44( 14 ):273-5. 281. 1995 Apr 14. 17. Anonymous.
Progress toward global eradication of poliomyelitis. 1988-1993.
lvfMWR 43(27 ):499-503. 1994 Jul 15. 18. Leentvaar-Kuijpers A, Kool JL. Veugelers P J. Coutinho RA. van Griensven GJ. An outbreak of hepatitis A among homosexual men in Amsterdam, 1991-1993. International Journal ojEpidemiolots'Y 1995:24( I ):218-222.
19. Kool JL. Buth 1. [Choice of Material for Supragenual Femoro-poplitcal Bypass Grafts]. [Abstract]. Nederlands TUdschr(lt \'(JOr Geneeskundf! 1990:134:1574.
144
Stellingen behorend bij bet proefschrift
''Preventing Legionnaires' disease''
Jacob Kool
I.
Desinfectie van gemeentelijk drinkwater met monochloramine kan de meerderheid van drinkwater-geassocieerde veteranenziektegevallen helpen voork6men ( dit proefschrift).
2.
Veteranenziekte is niet zeldzaam: Legionella is een van de drie of vier meest frequente verwekk.ers van pneumonie in geindustrialiseerde Ianden. In Nederland krijgenjaarlijks enkele duizenden personen een legionella-pneumonie.
3.
Onderdiagnose van veteranenziekte kan leiden tot inadequate behandeling van patienten, verhoogde sterfte en tot het niet herkennen van outbreaks.
4.
Veel artsen zijn zich onvoldoende bewust dat veteranenziekte niet gediagnosticeerd of uitgesloten kan worden op grond van klinische of radiologische presentatie. Speciale diagnostische tests dienen te worden aangevraagd.
5.
lnvoering van de Legionella urine antigeen test kan lei den tot een sterke toe name van het aantal gediagnosticeerde gevallen van veteranenziekte (dit proefschrift).
6.
Omdat Legionella vrijwel overal in het drinkwater aanwezig is, dienen allereerst maatregelen genomen te worden om de vermeerdering van deze bacterie tegen te gaan. Hieronder vallen het verwijderen van delen van waterinstallaties waar stagnatie optreedt, temperatuur aanpassingen en adequate desinfectie. Pas daarna dient overwogen te worden om het zin heeft om het water microbiologisch te testen (dit proefschrift).
7.
De concentratie van Legionella in watermonsters heeft slechts een beperkte voorspellende waarde voor het risico van besmetting van personen. De proportie Legionella-positieve tappunten in een waterinstallatie is waarschijnlijk een betere maat (dit proefschrift).
8.
De meerderheid van de Nederlandse bevolking wordt voorzien van drinkwater dat geen residueel desinfectiemiddel bevat. Dit bevordert de smaak zowel als het risico op veteranenziekte (dit proefschri ft ).
9.
Empirische therapie (d.w.z. therapie die ingesteld wordt v66rdat de verwekker bekend is) van niet-in-het-ziekenhuis-opgelopen longontsteking dient te gebeuren met antibiotica die actief zijn tegen Chlamydia, Mycoplasma, en Legionella, omdat deze samen in de meeste Ianden de meerderheid van de bacteriele longinfecties veroorzaken. Dit maakt ~-lactam antibiotica ongeschikt als middel van eerste keus (diverse auteurs).
10. Het beleid bij niet-in-het-ziekenhuis-opgelopen longontsteking dient gebaseerd te worden op leeftijd, ernst van de longontsteking, en onderliggende ziekten (Bartlett JG et al. Community-acquired pneumonia in adults: guidelines for management. The Infectious Diseases Society of America. Clin Infect Dis 1998; 26: 811-38). 11. Het betrouwbaar voork6men van Legionella-verrneerdering in bubbelbaden vereist zeer vergaande maatregelen die bij prive-gebruik moeilijk te implementeren zijn. Totdat de wetenschap eenvoudigere en afdoende maatregelen ontwikkelt, dienen deze apparaten beschouwd te worden als inherent onveilig.
Amsterdam, 10 Mei 2000.
