Nosocomial Legionnaires ~ disease : Clinical and radiographic patterns

TJ MARRIE, D HALDANE, G BEZANSON. Nosocomial Legionnaires' disease : Clinical and radiographic patterns. Can J Infect Dis 1992;3(5):253·260. From 1981 to 1991. 55 patien ts (33 males. 22 females. mean age 58.6 years) \v:ilh nosocomial Legionnaires· disease were studied. The mortal ity rate was 64%. One-half of the patients developed nosocomial Legionna ires· disease wiU1in three weeks of admission. A surprising clinical feature was U1e low rate of findings of consolidation on physical examination, despite the fact that 52% of patients had lhis finding on chest radiograph. More U1an one-half of patients had pre-existing lung disease. renclerinO" a radiographic diagnosis of pneumonia clue to Legionella pnewnophila impossible in 16% of cases despite microbiological confim1alion. Nineteen per cent of patients who had blood cultures clone had a pathogen olher lhan L pnewnophila isolated. suggesting dual infection in at least some of lhe patients. When lhe clinical and radiographic findings were combined it was notecllhat 40% of patients had one of lhree patterns suggestive of nosocomial LeO"ionnaires· disease: rapidly progressive pneumonia. lobar opacity and multiple peripheral opacities . However. in 60% of patients U1ere were no distinctive features.

S HORTLY AFTER ITS ISOLATION IN 1977 ( I ) IT BECAM E EVI - de nt.tha t.Legionella pneumophila ca used nosoco mial pneumonia (2 -7).Its a da ption to an a q u a tic ecological environme nt h a s led. in part., lo its s u ccess as a n osocomia l pa thogen (8)(9)(10).Contamin a ted pola ble water h as been s hown to be a so urce of legion ell a in m a ny h ospitals (ll -14).a nd erad ica tion of the microo rganis m from th e wa ter s upply has resulted in cessation of o utbreaks (14).Desp ite m a ny desc rip ti ons of the epidem iology of n osocomi a l legio n e ll osis ( 11 -14).there h ave been very few s tud ies that.have llied lo cha racterize d istinctive cl inical presen ta tion s of th is form of nosocomial pneumonia.
Th e a u th ors h ave identified cases of nosocomi a l Legionnaires• d isea s e at th eir ins ti tution since 198 1.In the presen t paper they describe their experien ce wiU1 55 cases oftilis illne sand emphasize recognition of its different.clinical presentati on s .

M ATERIAlS AND METHODS
Case det e ctio n : From 1983 lo 1987 (1 5) U1e present a u thors performed serological testing for L p neumophila on all patients wi th nosocom ia l pn e umonia.In a dd ition .fro m 1981 to 199 1. pa ti ents wi U1 nosocomia l pn e umon ia h ave been investigated fo r L p neumophila a t.Lh e d isc retion of their attending physician s a n d infec ti ou s d isease co n s u lta nts.Knowl edge of th e prese n ce of legion ella in the hosp itar s pota b le water is widespread a nd p romp ts efforts to id entify cases of nosocomi a l Legion naires• disea e .Using th e d efin iti on s given below.55 cases of n osocomi al Legionnaires• d isease we re diagnosed.For ty of these were s tudied pros p ec tive ly .Da ta on the remainder we re collected throu gh a retros pective ch a r t. review.For each p a ti e nt.th e symptom s p resen t. a l on set. of p ne um oni a we re record ed .Case definit ion: Legionna ires• d isease was diagnosed if on e o r m ore of th e fo ll owing c rite r ia we re fu lfi ll ed : L pneumophila isola ted from respiratory sec re tio ns .pulmonary tissue or p le ura l fl u id: acute a nd conva lescen t. ser um sam ples wiUl at least a fourfold rise in a ntibody   Just over one-half of the patients (56.3o/o) acquired Legionnaire • disease within iliree weeks of hospitalization (Figure 1): however, two patients had been hospitalized for more than 71 days before they acquired Legionnaire • disease.The symptoms and signs that these patients exhibited are given in Table l.Selected demographic features are given ln Table 2. Of note is U1e large number of patien ts wh o were receiving corticosteroid therapy-69o/o.An additional 16o/o received orne oU1er form ofln1munosuppressive therapy.1\venty-nineper cent had a malignan cy and 60o/o required assisted ventilation.Observed aspiralion was surprisingly co mmon.affecting 20o/o of patients.
Results of blood cultures are given in Table 3. 1\venty-eight per cent of those who had blood cultures done had a pailiogen isolated.Staphylococcus aureus was mo t common.followed by Escherichia coli.A vari-  ely of microorganisms was isolated from respiratory secretions (Tab le 4).Eleven per cent of the sputum samples showed no growth.SLaph aureus was iso lated from 23o/o of palienls.Fifty per cent had an aerob ic Gram-necrauve microorgan ism isolated.The auU1ors were unable to distinguish colonization from infection.Table 5 correlates the APACIIE score ( 18) with mortality.While the numbers in each category are small.there is an increase in mortality with increasing score.A comparison of patients who lived with those who died revealed U1at those who died were more likely to Twentyix of the 50 patients (52%) who h ad chest radiographs available for review at the end of U1e study showed consolidation .Patchy opacities were present in 15 (30%) and rounded opacities were seen in eight (16%).Diffuse alveolar disease was present in eight (16%) and a pleural effusion in five (10%).Cavitation was observed in two (4%).A major finding was U1 e presence of pre-existing lung disease in 29 patients (58%).Chronic obstructive pulmonary disease was most common (14 or 48%).followed by interstitial fibrosis.pulmonary edema and pulmonary hemon-hage (U1ree each).Other finding included pleural effusion 256 and pre-existing pneumonia (two each) .and one each of bronchial stenosis and metastatic carcin oma.When clinical and radiographic features were combined .the following patterns emerged: nonspecific pattern.14 (30%): lobar pneumonia.ll (20%): rapidly progressive pneumonia, eight (16%): aspiration.three (6% ): pulmona1y embolus, two (4%): and multiple peiipheral opacities.four (8%).The authors were unab le to determine a specific pattern due to p re-existing lung disease in eight (16%).Examples of the various radiographic patterns are given in Figures 2 to 6. Figures 7  and 8 illustrate the clinical course and radiographic features of one patient.
Twenty-eight patients had acute and at least four week convalescent serum samples collected.Twenty-one (75% ) showed at least a fourfold rise in antibody li lre.

DISCUSSION
There have been many reports of nosocomial Legionnaires• disease.Korvick and co-•workers (19) reviewed 16 reports of nosocomial legionellosis published from 1965 lo 1983.Only four of U1e reports described more U1an 40 patients.The mortality rate ranged from 17 to 66% .The major underlying diseases were immunocompromised stale.maJicrnancy and chronic lung disease .Legionella was isolated from U1e environment in 14 of the 16 reports and in 12 of these 14. isolation was from potable water.Korvick and Yu (20) in another report noted that surgical patients represent 23 to 50% of all patients with nosocomial Legionnaires• disease.Fifty-three per cent of the present patients h ad undergone surgery within the two weeks prior to the onset of nosocomial pneumonia.
Table 6 summarizes selected data from two other studies of nosocomiallegionellosis (21.22) so that ready The chronology of his course in hospital is shown in Figure 7 comparison can be made with the present.patients.The study by Kirby el al (21) involved an outbreak in a Ve terans• Hospital in Los Angeles.California; the oU1er two sludie involved patients wiU1 sporadic nosocomial Legionnaires• disease.There are differences in mortality rates and in the percentage of patients who were immunosuppressed in Ule three studies.A major difference in the clinical manifestations i U1e low rate of hemoptysis in the present study.compared with about.one-lliird or patients in Kirby's study (21) .
Of note is the low number of patients in Lhe present.study wiU1 bronchial breathing (nine 116% ]).However.26 (52%) had an air bronchogram visualized radiographically.This is the clas ical ra diographic sign of pneumonic consolidation (23).
Nine of the present.patients had positive blood cultures suggesting that.at.lea t 16% or patients had dual infections .Isolation of a vatiety of paU1ogens other U1an 258 legionella from the sputum in most instances probably refl t.s colonization of the respiratory tract of seriously ill patients.Pneumocystis carinii, however, was responsible for pneumonia in the patient who had U1e organism in U1e respiratory secretion.and it is likely that at lea t some of the other pathogens also caused the pneumonia.lt is equally possible that in some instances legionella may have been a colonizer and not a paU1ocren.
The APACHE ll score is a severity of disease classification U1at uses the initial values of 12 routine physiolocrical measurements, age and previous health status to provide a general measure of severity of disease (18).Knaus and co-workers (18) validated this system for 5815 intensive care admissions from 13 hospitals.While the maximum score possible is 71.no patient has exceeded 55.A key finding in Knaus•s study (18) was the significant increase in death rate for each five-point  increase in score.In general.nonoperative patients had higher mortality rates at each score than d id postoperative patients.The present patients had higher mortali ty rates in each interval U1an did patients wiU1 various diagnoses in U1e report by Knaus et a l (18) .This observation holds true even wh en the s u bset of U1eir patients with resp iratory fa il u re from lnfection is compared with the present cases of nosocomial Legionnaires' disease.Whi le U1e num ber of patients in the present study is small.it seems that a pneu mon ia-specific sco ring system shou ld be developed.The rad iographic mani festations of nosocom ial Legionn ai res• disease h ave been reported by a number of investigators (2 1,22).However, they did n ot report on pre-existin g lung disease.Fifty-e igh t per cen t of th e p resent patien ts h ad pre-existing disease.often lo s uc h an extent that the a u thors were unable lo differen ti ate the changes cl ue to legionella infection from the primary d isease process.Wh en the radiogra phic pictures we re combined with the clinical data.three patterns em erged that were suggestive of nosocom ial Legionnaires• disease: lobar opacity.rapidly progress ive pneum onia a n d mu ltiple peripheral opacities.However.these three patterns accounted for only 40% of cases.Thus.there is nothing to distinguish most cases of nosocomial Legionnaires• disease fro m other types of nosocom ia l pneumonia.
Nu mber of days followi ng admission that LD occurredF igure 1) Nwnber of d ays af ter admiss ion to hospital that nosocom ia l Leg ionnaires • d isease (LD) was cliag nosecl.The percentage at variotLS time imer va ls as w ell as the cumulative percentage is shown 254

Figure 4 )Figure 5 )Figure 6 )
Figure 4 ) Chest radiog raphs s howing consolidation of the entire right lung due to Legionella pneu mophil a

C
AN J INFECT DIS V OL 3 No 5 SEPTEMBER/OCTO BER 1992

C
AN J INFECT DIS V OL 3 No 5 SEPTEMBER/OCTOBER 1992

TABLE 2
(17)tracheal secretions.lungtissue and pleural fluid were examined for L pneumophiLa serogroups 1 to 6 us ing the direct Ouorescent antibody technique desc1ibed by Cherry et al(16).Antibody titres to L pneumophila: Anti body titres to L pneumophila serogroup 1 were performed on acute and convalescent serum samples usincr an indirect fluorescent antibody lechniqu(17).Positive and negative controls were included with each run.All reagents for this test were obtained from U1e Centers for Disease Control in Atlanta.Georgia Review of c h est radiogr aphs: All 55 patients had pneumonia diagnosed radiographically by a radiologist.Fifty had chest radiographs available at the end of the study: these were reviewed by one of the authors during July and Augustl990 Thirty-eigh t of the iso late were L pneumophiLa serogroup l. one was serogroup 6 and one was Legionella micdadei.
pneumophlla serogroup l antisera (Centers for Disease Control in AUanla .Georgia).Direct fluorescence antibody studies for L pneumaphila: ln order to correlate radiographic patterns with clinical presentation.Rapidly progressive pneumonia was defined as an increase in the size of the opacity at lea l 50o/o in 48 h .Chronic obstructive pulmonary disea e. pulmonary edema.pulmonary hemorrhage and interstitial fibrosis were diacrnosed on the basis of a combination of clinical and radiological criteria.Chronic obstructive pulmonary disea e is strictly a clinical rather than a radiological diagnosis.This term was used when the chest radiograph demonstrated flattening of the diaphragm and increased anteriorposterior diameter of tl1e chest.

TABlE 3
Results of blood cultures in 55 patients with nosocomial legionnaires' disease

TABLE 6 Comparison of selected features of nosocomial Legion- noires' disease from three reports Present Feature Korvick et al* Kirb et alt study
Numbers in brackets refer to percentage; 'Reference 20; 1 Reference 21 NS Not stated In-hospital course of a 69-year-old male admiLLed February21.1989.withanexacerbation of chronic lung disease.An admission chest radiograph showed only chronic obstructive lung disease.The patient became febrile on March 7. 1989.andLegione lla pneumophila was isolated from a sputum sample obtained onMarch 9. 1989.The infectious d isease consultant who saw the patient on March 9 made a clinical diagnosis of nosocomial Legionnaires• disease.At autopsy.there were severe emphysema.bilateral lobar pneumonia and bilateral microabscesses.Gram-positive cocci and yeast were seen on m icroscopic exam ination of the lungs.C Candida: N Intravenous: L Legionella: P aeru Pseudomonas aeruginosa: PO Per ora (by mouth): S Staphy lococcus : S epi S laphy lococcus epiclermidis: WBC White blood cell count