Comparative Study of Legionella pneumophila and Other Nosocomial-Acquired Pneumonias

Comparative Study of Legionella pneumophila and Other NosocomialAcquired Pneumonias* jorge Roig, M.D.; Xavier Aguilar; M.D.; Juan Ruiz, M.D.; Christian Domingo, M.D.; Eduardo Mesalles, M.D.; jose Manterola, M.D.; and Jose Morera, M.D.

We studied, in a prospective way, the characteristics of de6nitively diagnosed nosocomially acquired pneumonias in our hospital over 36 months. Out of 55 cases, 27 were due to LegioneDa pneumophila and 28 to other, nonLegionella bacteria. The cases of IegioneDosis concentrated in July, August, and December. The only risk factors that showed signi6cant differences (p
that showed signi6cant difference were pleural effusion, more frequent in the non-LegioneUa group (p
isolation in 1977,1 the importance of Since its first pneumophila as a causative agent of

pose of contributing to the evaluation of this subject, and taking into account the existence of previous endemic nosocomial legionellosis in our hospital, we decided to undertake a prospective study comparing the characteristics of noo-Legionella pneumonia and Legionella pneumonia, both acquired in our medical center.

l.£gioneUa

both community and hospital-acquired pneumonias has increased continually.2-4 Its role as a possible, and sometimes unexpected, etiologic agent of nosocomialacquired pneumonia has been well established for more than a decade}-13 Moreover, the knowledge of the variability and severity of involvement of various organs14,15 has been expanding to the present date. The problem of differentiating Legionella pneumonia from other types of non-Legionella pneumonias has been a matter of controvers~2-1 Some references have focused on the comparison between communityacquired pneumonia due to Legionella and that due to other bacteria. 16-18 To our knowledge, the setting of nosocomial pneumonia has not been approached, from this point of vie~ to a similar extent. There are just a few references in the English literature. Some ofthem consider together cases of hospital-acquired as well as community-acquired Legionella pneumonia,s and others report various species other than L pneumophila, most frequently L micdadei. 19•20 With the pur*From the Servei de Pneumologia, Servei de Cures Intensive~ Servei de Microbiologia, Hospital Germans Trias i Pujol, Badalona (Barcelona), Spain. Manuscript received May 16; revision accepted July 26. &print requests: Dr. Roig, Hospital Germans 1Has i Pujol, Badalona (Barcelona), Spain 08916 344

=

=

NLP non-LegioneDa pneumonia; LP Legionella pneumonia; DFA = direct 8uorescent antibody staining; IFA = immuno8uorescent antibody test; YfC = plugged telescopic catheter

MATERIAL AND METHODS

Fifty-five patients with a definite diagnosis of nosocomially acquired pneumonia were identified in a prospective study performed in our hospital over 36 months, from August 1985 to September 1988. The hospital had been inaugurated 2Jh years before. The number of beds increased slowly from 294 at the beginning ofthe study to 419 at the end. The level ofbed occupation approached 80 percent. No organ transplant had been performed in our hospital until the year 1985, when just a few kidney transplants began to take place. To accept a patient with proven pneumonia for this study, at least one of the following three criteria was required: (1) patients that had been admitted to the hospital at least 48 hours prior to the time of onset of pneumonia; (2) patients that had been discharged from the hospital within eight days of the onset of symptoms or signs of pneumonia; and (3) outpatients or visitors of hospitalized patients that, for whatever reason, had remained in the hospital for many hours a day, and at least 48 hours prior to the onset of illness. Out of the 55 cases of nosocomial pneumonia, 27 were due to L pneumophila and 28 to other, non-Legionella bacteria. To include a patient in the Legionella pneumonia group, at least one of the following three criteria was required: (1) a positive Legionella culture from respiratory secretions. The specimens were inoculated in a selective buffered charcoal yeast extract medium. 21 The specific Comparative Study of Nosocomial-Acquired Pneumonias (Roig st 81)

ingredients of the culture medium (bioMerieux. France) were yeast extract. activated cbarcoaI and granulated agar buffered with ACES (2 acetamido. 2 aminoethane sulfonic acid) and with an enrichment supplement of ferric pyrophosphate and cysteine bYcirocbloride. Tbe mixture for selective isolation was composlid:()t vancomYcin and colistin. (2) Demonstration ofthe organism by direct 8uorescent antibody staining of respiratory secretions. II The DFA test was performed with a 8uorescein-labeled monoclonal antibody that reacts with L pneumophila, without serogroup specificity. Tbe specimens obtained. from part of the patients, to be processed for DFA and/or culture. included one or more oftbe following: sputum, tracbeal aspirate, broncboalveolar lavage. brushing with doublelumen plugged telescopic catheter,&> and transthoracic needle pulmonary aspiration samples. Almost all diagnoses obtained by DFA were subsequently confirmed by culture or serology. As we specify in tbe "Results" section, only one of the 27 diagnoses of legionellosis was exclusively based on the DFA test. (3) A fourfold rise in seric IFA titer. Due to its noninvasive nature. Legionella serology was systematically performed wbenever nosocomial pneumonia was suspected, provided that the patient survived long enough to draw a second blood sample at least four weeks after the first. Taking into account the high diagnostic yield of IFA serology for L pneumophila infection,'" our 27 Legionella cases probably include the majority of Legionella pneumonias that occurred in our center for the period of time considered. To include a patient in the non-Legionella pneumonia group, at least one of the following criteria was required: (1) A positive blood, pleural 8uid or pulmonary aspiration culture. Samples obtained for culture were processed according to standard methods. (2) A quantitative culture greater than 100 colony-forming units per ml from samples obtained by PTC brusb. We specifically rejected both transtracbeal aspirates and sputum cultures, whatever their Murray's grade, wben deciding to include a patient in this non-Legionella group." By means of this strict bacteriologic confirmation of the diagnosis, 28 cases of non-Legionella pneumonia were included in the control group to be compared with the 27 pneumonia cases due to L pneumophila. Other possible non-Legionella bacteria cases that could not be confirmed by isolation of a determined etiologic agent from a non-rontaminated sample were excluded from this series since the present study hinges upon rigorous diagnostic criteria. The decision of wbether or not an invasive diagnostic procedure was to be performed was based on the following two factors: (1) The underlying general condition, and especially the cardiorespiratory state. of the patient. which often ruled out transparietallung puncture or even bronchoscopy because of unacceptable morbidity; and (2) The immediate availability of a pneumologist and a microbiologist to perform a determined invasive procedure and process the sample quickly and properly. Risk factors and clinical. roentgenographic. and analytic data were evaluated by the authors according to a preestablished protocol. All these measurements were obtained during the first 24 to 48 hours of illness. X-ray films were evaluated by two observers, and in addition to posteroanterior views, lateral views were available in most cases. Urinary abnormalities were excluded from analytic evaluation because the frequent presence of a bladder catheter at the moment of diagnosis of pneumonia could have produced misleading results. Statistical evaluation was performed with Student's t-test and cbisquare analysis, with Yates's correction for small numbers. RESULTS

During a period of 36 months, we diagnosed 55 cases of nosocomial pneumonia. Out of these 55 cases, 27 were due to L pneumophila serogroup 1 and 28 were due to other types of bacteria. The breakdown

of this second group of cases was the following: Gram-negative bacilli 17 Staphylococcus aureus 6 'St~tlJOOCCUS pneumoniae 5 Miscellaneous 5 Mixed infection 4 Most of the Gram-negative bacilli were Pseudomonas aeruginosa. The miscellaneous group was formed by Streptococcus viridans and anaerobic bacteria (two cases each) and Streptococcus faecalis (one case). Three patients showed double infection. Three different bacteria were isolated in a fourth patient. According to the criteria previously specified, 27 cases of Legionella pneumonia were diagnosed. In 16 cases a positive Legionella culture was the method of diagnosis. Of these 16 positive cultures, three were obtained from sputum, 13 from bronchoaspirate or bronchoalveolar lavage samples, and three from transthoracic lung aspirates. In three patients, both sputum and bronchoalveolar lavage samples, obtained by fiberbronchoscopy, were positive. The DFA test was performed in a minority of patients and there was only one case in the Legionella group in which the diagnosis was obtained by a clearly positive DFA test from a bronchial sample without subsequent confirmation by culture or serology. In ten patients, serology was the only method of diagnosis. Five of 27 Legionella cases and two of 28 nonLegionella cases were patients that had been discharged from the hospital a few days before the onset of pneumonia. Two patients of the Legionnaires' disease group and none of the non-Legionella pneumonia group were visitors that spent many hours a day in the hospital during the days prior to the onset of illness. None of these differences showed statistical significance. The age distribution of both groups of pneumonia is shown in Figure 1. There was a predominance in

Age Distribution 12r---------------------, 10 8 6 4

2

Age

10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 .LP

~NLP

FIGURE 1. Age distribution of patients with hospital-acquired Legionella pneumophila pneumonia (LP) and other. non-Legionella pneumonias (NLP). CHEST I 99 I 2 I FEBRUARY, 1991

34S

Monthly Distribution

Table 2-Praenting Clinical &mira in ~

with

Legionella pneumophila Pneumonia (LP) and Other, Non-ugionelJa Pneurnoniaa (NLP)

H~-AequiTed

8 7

No. of Patients and Percentage

6

5

Clinical Features

4

Chest pain Cough Hemoptysis Absence of respiratory symptoms Lack of localizing signs in chest Fever >38"C Chills Abdominal pain Diarrhea Changes in mental state Headache

3 2

o

Jan

I 11 Feb

Mar

"" Apr

May

Jun

.LP

Jul

Aug

Sep

fl Ocl

Nov

Dec

mmNLP

FICURE 2. Monthly distribution of patients with hospital-acquired LegWneIUJ pneumophila pneumonia (LP) and other, non-Legionella pneumonias (NLP).

both groups of the 50- to 80-year-old age bracket; however, both types of hospital-acquired pneumonia were found at all ages. The mean age of the Legionnaires' disease group was 65 and that of the nonLegionella group was 58 (p<0.05). The male-female ratio was 18:9 for the Legionnaires' disease group (2:1) and 25:3 for the non-Legionella pneumonia group (8:1) (p<0.02). We were interested in 6nding out if there was a seasonal pattern ofdistribution to these groups, so we looked at the monthly distribution (Fig 2). The higher incidence ofLegionnaires' disease cases in the months ofJuly, August, and December was statistically significant when compared with the incidence reported in other months (p<0.05). This seasonal pattern was not found in the non-Legionella group. The.only risk factors that showed significant differences (Table 1) were general anesthesia and surgery and immunosuppressive disease (p<0.05). Chronic liver disease and the presence of lowering of the consciousness level, prior to the diagnosis of pneumonia, were also more frequent in the non-Legionella group. The 6ve immunosuppressive disease cases found in the Legionnaires' disease group were the following: kidney transplant (three); lung cancer under Table I-Rilk fiJdorI in ~

with Hoapital-Acquired

LegioaeUa pneumophila Pneumonia (LP) and Other,

Non-ugioMlJa PtWUmOfIiaI (NU)

No. of Patients Risk Factors

LP(27)

NLP(28)

General anesthesia Surgery Immunosuppressive disease Chronic liver disease Lowering of consciousness level

5 5 5

n* n*

*p
o

o

13*

5 5

LP(27)(%)

NLP(28)(%)

5 (18.5)

5 (17.8) 8 (28.5)

n

(40.7)

2 (7.4) 12 (44.4)

o (.)

24 (88.8) 4 (14.8)

13 (46.4) 14 (50) 26 (92.8) 2 (7.1)

o (.)

(7.1)

10

(37)

o (-) o (.) I

(3.7)

I I 2 I

(3.5) (3.5)

(3.5)

cytostatic therapy (one); and hematologic malignancy (one). The 13 cases of immunosuppressed condition that were found in the non-Legionella pneumonia group were the following: lung cancer or other solid malignancies under cytostatic therapy (6ve cases); hematologic malignancies (three); systemic disease under high-dosage corticosteroid therapy (two); and miscellaneous diseases under long-term corticosteroid therapy (three). Other risk factors that showed no statistical difference between both groups were the following: prior antibiotic treatment, 6berbronchoscopy, gastroduodenoscopy, tracheostomy, smoking, alcoholism, chronic obstructive pulmonary disease, diabetes mellitus, chronic heart disease, chronic renal failure, the use of nebulization equipment, and mechanical ventilation. Six out of27 Legionnaires' disease cases and two out of 28 non-Legionella pneumonia cases lacked any severe underlying disease, that is, they did not show any of the pathologic conditions that we have included in the previous list of analyzed risk factors.

Clinical Features We considered the presence of chest pain, cough, hemoptysis, the absence of respiratory symptoms, the lack of localizing signs in chest examination, the presence of fever > 38°C, chills, abdominal pain, diarrhea, changes in mental state, and headache. None of these parameters showed a statistically signi6cant difference between the two groups (Table 2). Fever was the most constant clinical feature in both types of pneumonia. We want to emphasize the high percentage of patients (40 to 50 percent), in both groups, who did not show any respiratory symptoms or specific signs on chest examination.

Roentgenographic Features We analyzed the following roentgenographic data

Table 3-Preserating Roentgenographic FeatureB in Patient, with Hospital-Acquired LegioneUa pneumophila Pneumonia (LP) and Other, Non-ugionella PneurnoniaB (NLP)

glyco~ide~ No. of Patients

Roentgenographic Features Extent of shadowing: One lung Both lungs One lobe Two or more lobes Homogeneous consolidation Cavitation Pleural effusion

LP (27)

NLP (28)

17 10 15 12

19

24 2 4

patients ofthe non-Legionella group were treated with a combination of erythromycin, 4 g1day, and a thirdgeneration cephalosporine, or ·sometimes, an amino-

9 17 11 21 3 14*

*p
(Table 3): (1) extent of shadowing, involving one lung, both lungs, one lobe or two or more lobes; (2) homogeneous consolidation; (3) cavitation; and (4) pleural effusion. We looked specifically for the presence of blunting of the costophrenic angle which was considered as evidence of pleural effusion and was subsequently confirmed by thoracocentesis in each case. Unilateral and unilobar involvement was the most frequent initial roentgenographic finding in both groups. The only feature that showed a statistically significant difference was pleural effusion (p
'Ireatment and Outcome Since the discovery of the first cases of Legionella pneumonia in. our hospital, erythromycin has been included in the initial therapeutic approach to nosocomial pneumonia. So, all patients belonging to the Legionella group were treated with 4 g1day of erythromycin from the moment of diagnosis of pneumonia.\ Patients of the non-Legionella group were treated according to culture and' antibiogram results. Before the isolation of the causative bacteria was performed,

Out of the 27 Legionnaires' disease patients, four Ten of died despite appropriate parenteral therap~ the 28 non-Legionella pneumonia patients also died. Progressive respiratory failure was the most common cause of death in both groups. The comparison between the mortality rate of the Legionella group (14.8 percent) and that of the non-Legionella group (35.7 percent) showed a statistically significant difference (p
The aim ofour study was to obtain clinically relevant conclusions from the comparison, in a prospective wa~ of the characteristics of hospital-acquired pneumonia due to L pneumophila and other, Don-Legionella bacteria. It has been suggested10 that community- and hospital-acquired legionellosis tend to be similar in many aspects. However, it seems logical to expect that the population admitted to a hospital usually has a compromising underlying condition that could modify clinical response to this infection and induce some significant discrepancies between the features ofthese two forms of LegioneUa pneumonia. That is why we have restricted this comparative study to the nosocomial setting. Previous references to this subject show some drawbacks25 that we have tried to eliminate. Most published series are retrospective reviews and do not differentiate between nosocomial and community acquisition of the Legionnaires' disease,s,8,9 so their results might not be as reliable as expected. Other references have focused on species .of Legionella different from L pneumophila, usually L micdadei, also known as Pittsburgh pneu~onia agent}9,?A) It is stated that, in spite of some similarities between L pneumophila and L micdadei, there are also some Table 4- Analytical FeatUreB in PatientB with HospitalAcquired LegioneUa pneumopbila Pneumonia (LP) and Other, Non-ugionella PneurnoniaB (NIP). Mean Value ± SD

LP (27)

Laboratory Data Natremia Gam maglutamyltranspeptidase Total bilirubin Natremia <130 mEqIL Bilirubinemia>1.2 mgldl Creatinine >1.3 mgldl Arte~ hypoxemia <60 mm Hg

NLP(28)

134.2±6.2 135.8±4.8* 50.2 ± 39.8 98.7 ± 140.8* 0.9± 1.2 1.9±2.5* No. of Patients 5 2

3 12

2 9*

1 14

*p
347

special characteristics of L micdadei, such as its peculiar roentgenographic abnormalities26-28 and a greater tendency to subclinical manifestation. 29 So the conclusions obtained in this type of study can not be extended to the specific case of L pneumophila infections. In some series, the control, non-Legionella group is formed by nosocomial pneumonia cases diagnosed on the basis of cultures from samples obtained without avoiding contamination with the saprophyte oropharyngeal 80ra, such as sputum or tracheal aspirates. 5. 12 In our opinion and that of others,24,30 these diagnoses are not completely reliable, so we used the strictest criteria for inclusion of a case in our non-Legionella group. We were also concerned about the unusual but wellreferenced possibility of a false positive DFA Legionella test,13.!:5,31 especially that due to some Pseudomonas species. 32 Therefore, we confirmed almost all our Legionella cases by positive culture or serology. Some series of legionellosis show signi6cant percentages of their diagnoses based exclusively on the DFA test. 10 We focused on the first two days of illness, due to the fact that, in the therapeutic approach to nosocomial pneumonia, it is known that an earlier diagnosis implies a better prognosis. 33 Those studies that do not concentrate on the early stage of the disease probably give information that is not so relevant. The reported relative incidence of nosOC'Omial legionellosis in hospital-acquired pneumonia is characterized by its variability.25 A hallmark of many references is that Legionella can remain easily unsuspected or underdiagnosed until specific diagnostic methods are routinely used in the initial diagnostic approach to nosocomial pneumonia.5-13.25 A common point to all hospital-acquired legionellosis series has been the discovery of contamination of the water supply system,i-13.:J4..36 which was also observed in our hospital. Routine culturing of the water supply system was not performed, but L pneumophila was frequently isolated at different locations. The irregular basis of our environmental cultures and the absence of molecular subtyping oflinked patient and environmental isolates hampered our ability to draw conclusions on this particular point. The clear predominance of Gram-negative bacilli in the non-Legionella group is in agreement with most published nosocomial pneumonia series. 33 The occurrence of mixed infection has also been well recognized. 33,:rr The importance ofconsidering visitors and recently discharged patients as susceptible to nosocomial pneumonia has been especially stressed in the case of legionellosis, with a reported incidence as high as 45 percent in Kirby's series. 8 In relation to this point, we have found no significant differences between Legion348

ella pneumonia and non-Legionella pneumonia patients. Whatever its causative agent is, hospital-acquired pneumonia usually shows a predilection for patients of advanced age. 33 ,37 Our series shows that this tendency was common to both groups. The real utility of mean age and sex differences between Legionella and non-Legionella pneumonia cases, despite their statistical significance, can probably be considered low from a practical point of vie~ A seasonal pattern of distribution has been described in some outbreaks of nosocomial legionellosis. 3.•.9 ,10 A preference for the summer months has generally been reported. 38 The water temperature during these months can be most favorable to the amplifying mechanisms of Legionella. A Legionellafavoring in8uence of certain algae, aquatic protozoa and plants such as MyriophiUum spicatum (which tends to show a seasonal relationship with legionella), has been hypothesized. 3.38 Our outbreaks in July and August suggested a correlation with the operation of the air conditioning system that would have acted as a disseminator of the infection, but we were at a loss to find such an explanation for the high incidence in December. At any rate, it seems that the presence of significant monthly differences in the incidence of nosocomial pneumonia could raise the suspicion of Legionella pneumonia and even support this etiology when approaching a differential diagnosis. Specific problems sometimes found in an expanding, new hospital-such as medical and surgical patients, as well as different specialties, sharing the same ward, or frequent movement of patients due to the opening of new wards-have made impossible any reliable analysis of the in8uence of these factors in our study. Hospital-acquired Legionnaires' disease is said to occur very infrequently in the absence of underlying disease,2.13 but this statement can also be applied to other etiologies of hospital-acquired pneumonia. 33.:rr In fact, in our study, the lack of severe underlying disease was more frequent in Legionella pneumonia patients. The high frequency with which conditions classically associated with aspiration appear as significant risk factors for the non-Legionella group (general anesthesia, surgery) lowering of consciousness level) supports the concept of aspiration of previously colonized oropharyngeal contents as the usual pathogenic way of acquiring nosocomial pneumonia due to nonLegi.onella bacteria. 33,:rr Some references have emphasized that direct aspiration or tracheal inoculation of Legionella-contaminated water could be a mode of transmission of Legionnaires' disease.7.u-13.34 However, colonization of the oropharynx by Legionella has not been demonstrated to the present date.38--40 Serious immunosuppressive disease occurred in CornparaINe Study of NosocomiaJ-Acquir8d Pneumoni88 (Roig et 81)

both groups but was more frequent in the nonLegionella group. In fact, immunocompromised hosts are known to be a most susceptible population to nosocomial pneumonia, either due to Legj~eJJa <>.~-JQ~" other bacteria. 33 •37 Other comparative studies have not found significant differences in the prevalence of any risk factors for Legionella vs non-Legionella pneumonia. S.12 Our relatively low incidence of severe underlying immunosuppression in the Legionella pneumonia group, when compared to other series of nosocomial legionellosis,8.10.13 can have inBuenced the mortality rate, as we discuss later. Chronic liver disease was more common in the oon-Legionella group, and one can speculate, to suggest an explanation for this finding, on the possibility of subclinical encephalopathy favoring aspiration of oropharyngeal Bora in non-Legionella pneumonia patients. There was an evident overlap of the clinical features of pneumonia due to L pneumophila and to other bacteria. Our results agree with those of two previous prospective, nosocomial studies. s.l i Clinical data that have been suggested as distinctive of legionellosis, such as diarrhea or neurologic symptoms, are not helpful in the nosocomial setting. 7 As is known, fever is the most constant clinical feature. 2-13 The high incidence (40 to 50 percent) of patients in both groups who did not complain of any respiratory symptoms and who did not show rales or other consolidation signs on chest examination has to be emphasized when considering the first 24 to 48 hours of disease. More distinctive symptoms and signs of pneumonia appear later in the course of the illness with progressive frequenc~8.10 The appearance of fever >38°C in a hospitalized patient, without any other obvious cause, should raise a high index of suspicion of pneumonia. The early evaluation of chest x-ray abnormalities showed a predominance of unilateral and unilobar involvement that was common to both groups of pneumonia. This presenting roentgenographic feature has usually been reported in most legionellosis series.2-4·8.10.13.28 However, it is worthwhile to emphasize that, even in such an early stage, 10 out of 27 Legionella pneumonia patients and 9 out of 28 nonLegionella pneumonia patients showed initial bilateral lung involvement. Cavitation has been stressed2-4·13.28 as more frequent in legionellosis when involving seriously immunocompromised hosts. In fact, our two cases of cavitation were found in patients with severe immunosuppressive disease. Pleural effusion was significantly more frequent in the non-Legionella group. Although pleural effusion has been reported to be relatively common in legionellosis, usually it is not a presenting roentgenographic feature.8.10.38.•1 So, it seems that the finding of pleural effusion on the initial chest x-rayl8 could be a quite valuable roentgenographic clue in the initial differential diagnostic ap-

proach to hospital-acquired pneumonia. Laboratory data in nosocomial legionellosis are generally considered as nonspecmc. 2-5.10 Yu et als s~ggeste.d~tPat hyponatremia <130 mEq/L during the first days ~f illness could be a helpful analytic feature to support the diagnosis of legionellosis. In spite of a tendency toward lower mean natremia values in our Legionella pneumonia patients, the number of cases that showed natremia < 130 mEq/L did not show a significant differenCE} when compared to the nonLegionella group. The association of underlying chronic liver disease as a more common risk factor for the non-Legionella group with the higher incidence of abnormal hepatic tests in these patients seems to be evident. The finding of severe hypoxemia in almost 50 percent of the patients in both groups is in agreement with respiratory failure as being the usual cause of death in both Legionella and non-Legionella pneumonia patients. 8 The striking difference in mortality between the two groups, 14.8 percent in legionellosis patients vs 35.7 percent in non-Legionella patients, can be attributed to the high efficacy of the treatment with 4 glday of erythromycin against legionellosis.2-4·8 The mortality rate ofmost series ofnosocomiallegionellosis Two factors can ranges from 25 to 70 percent.S.13.25.~ jus~ at least in part, this reported variability in the case-fatality rate: first, the percentage ofimmunocompromised patients and the degree of severity of their underlying disease at the moment of diagnosis of legionellosis.2-5·25.~ In our series, the number of immunocompromised hosts is lower than that reported in many references that have a higher mortality rate than ours; and second, whether or not an effective antibiotic treatment is administered. An inappropriate antibiotic therapy, in Legionella pneumonia cases, usually implies a high mortality rate.2-13.25 Moreover, a delay in the proper treatment also implies a worse outcome. 38 Erythromycin remains the drug of choic~·8.10.13.25 for the treatment of legionellosis, both hospital-acquired and community-acquired. We conclude that, due to the overlap of clinical, roentgenographic, and analytic features between nosocomial pneumonia due to L pneumophila and that due to other bacteria, it is advisable to include routine, specific diagnostic methods for Legionella in the initial diagnostic approach to hospital-acquired pneumonia. If nosocomial legionellosis, sometimes unexpectedly, is discovered, the outcome can be improved by including erythromycin in the initial therapeutic approach to nosocomial pneumonia, until the efficacy of Legionella eradicating measures is finally proven. REFERENCES 1 McDade JE, Shepard CC, Fraser D~ Tsai TR, Redus MA, Dowdle WH. Legionnaires' disease: isolation of a bacterium. CHEST I 99 I 2 I FEBRUAR'(, 1991

349

N Engl J Med 1977; 297:1197-1203 2 Bartlett CLR, MacRae Ad, MacFarlane J1: Legionella infections. 1st ed. London: Edward Arnold Publisher, 1986; 37-55 3 Davis GS, Wino JrWC. Legionnaires' disease: respiratory infections caused by Legionella bacteria. Clin Chest Med 1987; 8:419-39 4 Edelstein PH, Meyer RD. Legionella pneumonias. In: Pennington JE, ed. Respiratory infections: diagnosis and management. New York: Raven Press, 1989:381-402 5 Yu VL, Kroboth FJ, Shonnard J, Brown A, McDeannan S, Magnussen M. Legionnaires' disease: new clinical perspective from a prospective pneumonia stud~ Am J Moo 1982; 73:35761 6 Muder RR, Yu VL, McClure JIe, Kroboth FJ, Kominos SO, Lumish RU. Nosocomial Legionnaires' disease uncovered in a prospective pneumonia study: implications for underdiagnosis. JAMA 1983; 249:3184-88 7 Johnson JT, Best MG, Goetz A, Wicker H, Yu VL, Vickers RM, et aI. Nosocomial legioneUosis in surgical patients with bead and neck cancer: implications for epidemiologic reservoir and mode of transmission. Lancet 1985; 2:298-300 8 Kirby BA, Snyder KM, Meyer RD, Finegold SM. Legionnaires' disease: report of sixty-6ve nosocomiaUy acquired cases and review of the literature. Medicine 1980; 59:188-205 9 Helms CM, Massanari RM, Zeitler R, Streed S, Gilchrist MJR, Hall N, et aI. Legionnaires' disease associated with a hospital water system: a cluster of 24 nosocomial cases. Ann Intern Med 1983; 99:172-78 10 Helms CM, Viner J~ Weisenburger' DO, Chiu Le, Renner ED, Johnson W Sporadic Legionnaires' disease: clinical observations on 87 nosocomial and community-acquired cases. Am J Med Sci 1984; 288:2-12 11 Johnson JT, Yu VL, Wagner RL, Best MG. Nosocomial ugioneUa pneumonia in a population of head and neck cancer patients. Laryngoscope. 1985; 95: 1468-71 12 Brennen C, Vickers RM, Yu VL, Puntereri A, Yee GY. Discovery of occult Legionella pneumonia in a long stay hospital: results of prospective serological surve~ Br Med J 1987; 295:306-07 13 Korvick JC, Yu VL. Legionnaires' disease: an emerging surgical problem. Ann Thorac Surg 1987; 43:341-47 14 Domingo CH, Roig J, Seres J. Pericardial effusion as a clinical sign of Legionnaires' disease. Intern J Cardioll989; 23:407-09 15 Finegold SM. Legionnaires' disease: still with us. N Engi J Med 1988; 318:571-73 16 MacFarlane Tf, Miller AC, Roderide Smith WH, Morris AH, Rose DH. Comparative radiographic features of community acquired Legionnaires' disease, pneumococcal pneumonia, Mycopltuma pneumonUJ, and psittacosis. Thorax 1984; 39:28-33 17 Woodhead MA, MacFarlane JT. Comparative clinical and laboratory features of Legionella with pneumococcal and Mycoplasma pneumonitis. Br J Dis Cbest 1987; 81:133-38 18 Granados A, Podzamczer D, Gudiol F, Manresa F. Pneumonia due to ugionella pneumophila and pneumococcal pneumonia: similarities and differences on presentation. Eur Respir J 1989; 2:130-34 19 Rudin JE, Wmg EJ. A comparative study of ugWneUa micdtulei and other nosocomial acquired pneumonia. Chest 1984; 86:67580 20 Doebbeling BN, Ishak MA, Wade BU, Pasquale MA, Gerszten RE, Groschel OHM, et aI. Nosocomial Legionella micdadei pneumonia: 10 years experience and a case-control study.

350

J Hosp Inf 1989; 13:289-98 21 Jones GL, Hebert GA. Legionnaires': the disease, the bacterium and methodology. 1st ed. Atlanta: Center for Disease Control, 1979 22 Broome ~ Cherry WB, Winn WC Jr, MacPherson BR. Rapid diagnosis of Legionnaires' disease by direct immun08uorescent staining. Ann Intern Med 1979; 90:1-4 23 Wimberley N, FalingJC, Bartlett JG. A 6beroptic bronchoscopic technique to obtain uncontaminated lower airway secretions for bacterial culture. Am Rev Respir Dis 1979; 199:337-43 24 Lentino JR, Lucks DA. Nonvalue of sputum culture in the management of lower respiratory tract infections. J Clin Microbioi 1987; 25:758-62 25 Meyer RD. Legionnaires' disease: aspects of nosocomial infection. Am J Med 1984; 76:657-63 26 Lee Pope Jr T, Annstrong ~ Thompson R, Donowitz GR. Pittsburg pneumonia agent: chest film manifestations. AJR 1982; 138:237-41 27 Muder RR, Reddy SC, Yu VL, Kroboth FJ. Pneumonia caused by Pittsburg pneumonia agent: radiologic manifestations. Radiology 1984; 150:633-37 28 Muder RR, Yu VL, Parry MF. The radiologic manifestations of Legionella pneumonia. Semin Respir Infect 1987; 2:242-54 29 Fang GO, Yu VL, Vickers RM. Disease due to the legionellaceae (other than ugionella pneumophila): historical, microbiological, clinical, and epidemiological review Medicine 1989; 68: 116-32 30 Tobin MJ. Diagnosis of pneumonia: techniques and problems. Clin Chest Med 1987; 8:513-27 31 Winn WC Jr. Legionella and legionnaires' disease: a review with emphasis on environmental studies and laboratory diagnosis. CRC Crit Rev Clin Lab Sci 1985; 21:323-82 32 Tenover FC, Edelstein PH, Goldstein Le, Sturge JC, Plorde JJ. Comparison of cross-staining reactions by Pseudomontu spp and 8uorescein-labeled polyclonal and monoclonal antibodies against Legionella pneumophila. J Clin Microbioll986; 647-49 33 Toews GB. Nosocomial pneumonia. Clin Chest Med 1987; 8:467-79 34 Neill MA, Gorman G~ Gibert C, Roussel A, Hightower A\V, McKinney RM, et aI. Nosocomial legionellosis, Paris, France: evidence for transmission by potable water. Am J Med 1985; 78:581-88 35 Harper D. Legionnaires' disease outbreaks -the engineering implications. J Hosp Infect 1988; II(supplement A):201-oB 36 Ezzeddine H, Van Ossel C, Delmee M.Wauters G. Legionella spp in a hospital hot water system: effect of control measures. J Hosp Infect 1989; 13:121-31 37 Bartlett JG, O'Keefe ~ Tally F~ Louie Tj, Gorbach SL. Bacteriology of hospital-acquired pneumonia. Arch Intern Moo 1986; 868-71 38 Skerrett SJ, Locksley RM. Legionellosis: ecology and patho~ In: Sande MA, Hudson LD, Root RK, eds. Respiratory infections. New York: Churchill Livingstone, 1986; 161-90 38 Bridge JA, Edelstein PH. Oropharyngeal colonization with ugioneUa pneumophila. J Clin Microbioll983; 1108-12 40 Moder RR, Yu VL, Woo AH. Mode of transmission of ugioneUa pneumophila: a critical revi~ Arch Intern Med 1986; 146:160712 41 Fairbank JT, Mamourian AC, Dietrich PA, Girod JC. The chest radiograph in Legionnaires' disease: further observations. Radiology 1983; 147:33-34

Comparative Study of Nosocomial-Acquired Pneumonias (Roig et aI)