Thoracic Empyema in HIV-Infected Patients

Thoracic Empyema in HIV-Infected Patients* Microbiology, Management, and Outcome Jacinto Hernandez Borge, MD; Inmaculada Alfageme Michavila, MD; Jesrls Munoz Mendez, MD; Francisco Campos Rodriguez, MD; Nicolas Peiia G1iiian, MD; and Rafael Villagomez Cerrato, MD

Objectives: To evaluate etiology, bacteriology, stage of disease, treatment, and outcome of HIV-infected patients with thoracic empyema (TE) over a 9-year period at a hospital teaching center. Design: We have retmspectively reviewed the charts of all HIV-infected patients with a hospital discharge diagnosis of empyema between January 1985 and November 1993. Patients: Twenty-three patients were identified (22 male and 1 female). The average patient age was 28.7±5.3 years. All the patients wet·e injection-drug users, and 10 (43%) fulfilled criteria for an AIDS diagnosis. In 15 cases (65%), the empyema was the first cause of medical consultation, which then led to an HIV infection diagnosis in 11 of them (48%). Measurements: In each case, symptoms, chest studies, culture results, procedure timing, length of hospitalization, and outcome were reviewed. Results: Twenty-one patients (91 %) had developed an empyema secondary to communityacquired pneumonia. The cultures of pleural fluid were positive in 19 cases (83%). Anaembes were isolated from 6 patients and aerobes fmm 13. A single bacteria was isolated from 10 (52%), and multiple m·ganisms (average 2.66 per case) grew in the remaining 9 positive cultures. The most common organism culture growths were Staphylococcus aureus (23%) and Gram-negative bacilli (36% ). Length of hospitalization averaged 25.6 days ( ± 15). Intercostal tube drainage was necessary in 18 patients and none required surgery. Patients with AIDS diagnosis needed a longer period of hospitalization, and the presence of bacteremia and bronchopleural fistula was more frequent. However, this did not influence a patient's final outcome. A follow-up was available in 18 cases, with 4 deaths recorded (average survival, 35 months; range, 4 to 84 months). Conclusions: In our series, TE associated with HIV infection was often the primary cause leading to hospital admission and later HIV diagnosis. IV drug abuse was the predominant factor for HIV infection and was also related to clinical presentation and microbiological findings . The best approach to treatment is-as with other patient groups-a prompt drainage and appropriate antibiotic treatment, since a favorable outcome is expected. (CHEST 1998; 113:732-38) Key words: H IV-infected p atients; manageme nt; outcome; thoracic empyema Abbreviations: BP= bronchopleural fistula; IDU = injection-drug TNAB= transthoracic n eedl e aspirati on biopsy

T boracic e mpyema

seems rare in adult HIVrelated diseases, despite the increased incidence of respiratory infections, although there are no available data (to our knowledge) to estimate true incidence. In the general population, empyema is most common as a consequence of community-acquired pneumonia. Bacterial pneumonia is common in patients with AIDS ,1·2 but most clinical reviews of

*From the Department o f Internal M edicine, Respiratory Disease Unit, Valme Un iversity Hospital, Seville, Spain . Manuscript r eceived July 23, 1996; revision accepted August 28, 1997.

732

user; NS= not significant; SK = streptokinase;

infections associated with HIV are notable for the lack of empyema documentation. The reasons for this are not clear. Pneumococcal pneumonia occurs in 1-2.6 c ases/ 1,000 per year for the generalpopulation,3 compared with 18-46 cases/1,000 per year in AIDS patients. 4 Furthermore, the rate of pneumococcal bacteremia among these patients may be more than 100 times higher than that found in an age-matched population. In about 2% of cases, associated parapneumonic effusions fulfill empyema criteria in the HIVseronegative population with pneumococcal pneumonia; however, there are few cases of clearly Clinical Investigations

documented HIV-associated pneumococcal empyema, despite both an increased incidence and severity of infections :5 Bacteria other than pneumococcus, which commonly causes empyema in the general population, include Staphylococcus aureus, Streptococcus pyagenes, Pseudomonas species, other Gram-negative organisms and anaerobes. There are, however, few accounts of empyema in HIV-infected individuals due to any of these organisms. 6 The aim of this study was to analyze, retrospectively, our experience with thoracic empyema in HIV-infected patients throughout a 9-year period in a 500-bed hospital teaching center, giving special attention to HIV infection risk factors, stage of the infection (as reflected by CD4 counts), and the microbiological diagnosis. Likewise, we evaluated the procedures, length of hospitalization, and outcome.

MATERIALS AND METHODS

We reviewed computerized medical records of HIV-infected patients admitted to Valme University Hospital between Janumy 1985 and November 1993. Of the 419 patients identified as HIV positive, 23 (5.4%) had a hospital discharge diagnosis of thoracic empyema and all were available for review. The diagnosis was confirmed by one of following critelia: (1) pleural fluid culture or Gram's stain showing organisms; (2) documentation of gross purulent fluid at thoracentesis; or (3) bioche mical evidence of empyema defined as pH <7.10 and either lactate dehydrogenase level > 1,000 IU/L or glucose level <40 mg/dL. Charts were reviewed for patient age, sex, known risk factors for HIV infection, and symptoms. We obtained the following information about the HIV disease stage at the time of diagnosis of empyema: (1) years from HIV infection diagnosis; ( 2)presence of AIDS diagnosis; 7 and (3) blood tests, including CD4 counts, within a period of 3 months p1ior to or follO\ving the hospital admission date. Chest radiographs and CT studies were reviewed. The site and extent of pleural involvement and the presence of parenchymal disease and cavity formation were noted. We also investigated as to whether the pleural fluid was loculated or free. The bacteriologic diagnosis, by microbiological exam ination of pleural fluid and other samples, was recovered, paying special attention to antimicrobial sensitivity studies. Bactere mia was defin ed as the isolation of a bacte1ial pathogen from two or more blood culture samples. Endocarditis required either a demonstration of valvular vegetations on echocardiography or evidence of septic emboli. The following data about the timing of procedures were noted: (1) days plior to hospital admission from the outset of symptoms; (2) days from hospital admission to chest tube drainage, days of drainage, use of fibrinolytic agents, and volume of pleural fluid drained; and (3) success of each procedure, length of hospitalization, complications, and outcome. Antibiotic therapy and cavity drainage with closed thoracostomy commenced immediately. Success was measured by clinical and radiologic (or CT) status improvemen t within a 24- to 48-hr period. Drainage was maintained until the daily fluid yield dropped to <50 mL and improvement in the chest radiograph was evident. Indication for intrapleural thrombolytics (streptoki-

nase [SK]) was persistent fluid collection that had been inadequately drained by thoracostomy tubes. SK (250,000 Uld) was instilled into the pleural cavity until a complete radiologic recovery was detected. Results are expressed as mean±SD (or range) and percentages. Either the Student's test or th e nonparametric K-Wallis test, as needed, was used for quantitative variables; and either the x2 or Fisher test was used for qualitative variables. The correlation of continuous variables was determined by simple regression. A p value < 0.05 was considered to be a significant difference.

RESULTS

Twenty-three HIV-infected patients were diagnosed as having empyema during the period of study. The group included 22 men and one woman, with an average age of 28.7:±:5.3 years (range, 21 to 43 years). All of them were injection-drug users (IDUs), and 10 (43%) fulfilled criteria for an AIDS diagnosis when they manifested empyema. Empyema was the primary cause for medical consultation in 15 patients (65%) and in 11 of them (48%), the diagnosis of HIV infection was established during hospitalization. In 12 cases, the diagnosis of HIV infection was prior to hospitalization, with a range from 1 to 8 years (average, 4.1:±:2.4 years). Community-acquired infections were manifest in 21 patients and 2 were due to nosocomial infections . The chest findings and characteristics of pleural fluid are described in Table 1. The pleural fluid had a purulent aspect in 12 cases and none were fetid. Eleven samples were nonpurulent: nine were serofibrinous and two were bloody. In six patients, there was evidence of thrombophlebitis and in five of them, a diagnosis of endocarditis was established. The predominant factor leading to infection was parenteral drug abuse (91%) and in five cases, associated factors coexisted (alcoholism [three] ; airway instrumentation [one]; and severe malnutrition [one]) . All the patients had symptoms attributable to their empyema, with fever and chest pains being the most common symptoms (96%). Other symptoms included dyspnea (53.5%), hemoptysis (35%), cough (39%), and septic shock (9%). Gram's stains from pleural fluid samples were negative in 12 cases (52%); however, organisms were later cultured from 8 of these cases. Of the 19 positive cultures, 13 (68%) had exclusively aerobic bacteria, 2 (11 %) had exclusively anaerobic bacteria, and 4 (21%) had both aerobic and anaerobic bacteria. All the isolated organisms are listed in Table 2. In the nine cases in which a single organism was isolated, more aerobic bacteria than anaerobic bacteria were found (seven vs two). Multiple organisms were cultured from the remaining 10 cases (average, 2.66 per patient). In 12 cases, blood cultures were positive (Table 2) and the most common organism CHEST I 113 I 3 I MARCH, 1998

733

Table !-Chest Radiographs and Characteristics of Pleural Fluid* Global (n =23) Site Right Left Bilateral Size < 113 113 > 1/2 Locu lated effusion Consolidation SI MI

cv Pyopneumothorax pH Proteins , giL Glucose , giL LDH, UIIL ADA, UIIL Leukocytes, cells/mm

16 5 2 4 16 3 14 (61%) 21 (91%) 15 5 9 (39%) 6 (26%) 6.78:!:0.36 5.35:!:0.75 0.16:!:0.15 6,813:!: 10,632 116.4:!: 115.4 72,926:!: 181,573

AIDS (n =10) 7 2 1 3 6 1 6(60%) 10 (100%) 7 2 4(40%) 4(40%) 6.6:!:0.4 4.8:!: 1.8 0.16:!:0.16 5,430:!:6,566 116.2:!:110 41,005:!:102,072

Non-AIDS (n = l3 )

9 3

10 2

8 (62%) 11 (85%) 8 3 5 (38%) 2 (15%) 6.8:!:0.1 5.3:!:0.7 0.15:!:0.7 7,613:!: 1,271 116.2:!:125 85,139:!:215,120

*p values > 0.05. SI=single infiltrate; lv!I=m ultiple infiltrates; CV=cavitation; LDH=Iactate dehydrogenase; ADA=adenosine deaminase.

culture growth was S aureus (5 cases). In five patients, a transthoracic needle aspiration biopsy (TNAB) was performed and nine bacteria were cultured with Gram-negative bacilli predominant (Table 2). Sixty-one organisms, including 46 aerobic bacteria and 15 anaerobic bacteria, were recovered, considering all samples (pleural fluid, blood, and TNAB samples). Only in three patients were all the cultures negative, including blood, pleural fluid, and TNAB samples; in seven cases, a single organism was isolated, and in the remainder, multiple organisms were cultured (range, 2 to 9; average, 3.38±2). Sputum specimens were recovered in 13 patients, and only 3 were found to be positive. Isolated organisms never conformed to pleural isolation (1 Streptococcus pneumoniae, 1 Haemophilus influenzae, and 1 Mycobacterium. tuberculosis) . No drainage was instituted in five patients due to the scarcity of fluid (in one of these patients, repeated thoracentesis was sufficient). In the remaining 18 patients, closed chest drainage was maintained for an average of 14± 14 days (range, 3 to 48 days ), with an average drainage of 1,916±1,883 mL (range, 140 to 7,000 mL). Hospital stays (34±19 vs 19.3±7 days; p=0.053) and drainage time (18±19 vs 5.8±4.4 days; p=0.1) vvere longer for empyemas with polymicrobial flora. In these patients, the diagnosis of endocarditis or thrombophlebitis (60% vs 31 %; p=not significant [NS]) and bacteremia (60% vs 46% ; p=NS) were more common. Six patients had bronchopleural fistula (BP) and none required surgery. This was more common in 734

patients with an AIDS diagnosis (40% vs 15%), endocarditis or thrombophlebitis (60% vs 17%; p=0.08) , bacteremia (83% vs 40% ; p=NS), and those with polymicrobial flora in the pleural fluid (67% vs 35%; p= NS). Both hospital stays (36.6±21.2 vs 21.7:±10.4 days; p=0.07) and drainage time (20.5±20.8 vs 7.7:±9 days; p=0.09) were more prolonged for these patients. Clinical and radiologic improvement was observed in nine patients following SK treatment (average dose, 2.3±1.5; range, 1 to 6). The delay time from closed drainage to the use of SK was 4.2±4.9 days (range, 0 to 13 days ). Drainage improvement following SK treatment was notable (average, 672 mL before SK vs 1,627 mL after SK), and there was no significant difference in length of hospitalization (25.7±11.8 vs 25.5±17.2 days) or drainage time (12.3±10.3 vs 10.2±16 days). The delay time for arrangement of closed thoracostomy drainage was correlated to hospital stays (r=0.52; 95% confidence interval, 0.14 to 0.77) and drainage time (r=0.55; 95% confidence interval, 0.17 to 0.78) . There was no significant difference between loculated and free-flO\ving pleural fluid, taking into consideration the drainage time and length of hospitalization. Patients with an AIDS diagnosis needed a longer hospitalization time (p=0.03), and bacteremia was more common (60% vs 46%; p=NS). Nevertheless, the presence of thrombophlebitis-endocarditis, polymicrobial empyemas (30% vs 54%; p=NS ), and patients \vith multiple-isolated organisms was less Clinical Investigations

Table 2-Bacterial Isolation in Pleural Fluid, Blood, and TNAB Bacte ria Aerobic bacteria Gram-positive cocci S aureus S pneumoniae S pyogenes S viridans Staph coagul ase negative Gram-negative bacilli P aen tginosa Pseudomonas cepacia A hydrophila H injluen:.ae Haemophilus pamirifluenzae Escherichia coli Enterobacte1· cloacae Klebsiella pneumoniae Unidentified Gram-negative Other organisms Branhamella catarrhalis A lwolfii M tuberculosis Anaerobic bacteria Gram-positive cocci S intermedius Streptococcus salivarius Streptococcus sanguis Peptococcus sp. Peptostreptococcus sp Gram-negative bacilli Prevotella melaninogenica Bacteroides uniformis F usobacterium sp Total

Pleural Fluid* Blood 1 T NAB ' (12/23) (515) (19123 )

8 (3) 3 (1) 2 (1)

5 (4)

1 (l )

1 5 (l ) 3 (l) 1 1 (l) 1 2 1

1 (1)

l (1)

1 (1)

1 (1)

1 1

tant to ampicillin were isolated in 8 of 61 cases (13%). These included two S pneumoniae, three Haemophilus s pecies, one Streptococcus intermedius, one Streptococcus viridans, and one Aeromonas hydrophila. Pathogens resistant to cloxacillin were identified in two cases (one S aureus and one Staphylococcus coagulase negative). Bacte1ial pathogens resistant to second- and third-generation cephalosporins and piperacillin were isolated in 7 of 61 cases (11 %) and all of them involved aerobic Gramnegative bacilli (5 Pseudonw nas aeruginosa and one Acinetobacter lwolfii ). All the patients survived and w eredischarged from the hospital. Long-term follow-up was available in 18 patients for an average of 36::±::28 months (range, 4to 84 months ). Four patients died, with an average survival of 35 months (range, 4 to 84 months ) and all had an AIDS diagnosis when they presented with the empyema. Of the remainder, e ight fulfilled AIDS criteria (5 7%) after an average follow-up of 39::±::21 months ( range, 10 to 84 months ).

1 DISCU SSION 3 (2) 2

1 (1)

2 (2) 1 (1)

1 l

35

17

9

*Nu mber of single isolated cultures enclosed in parentheses. 1 Number of si olated organi sms coincident with other samples fro m the same patient enclosed in p aren theses.

frequent (considering all samples). Other characteristics are summarized in Table 3 and we did not find differences b etween the groups, except in lymphocytes and CD4 counts. During hospitalization, we found 11 complications in 9 patients (39%) : 5 were due to chest tubes (1due to an obstruction and a r esidual empyema, 1 b ceause of a r elapse after withdrawal, 1 due to a splenic laceration, 1 b ceause of a r eversible obstruction, and lastly, a residual pneumothorax). The remainder were infectious (two bilateral empyemas, one persistent BP, one peritoneal septic embolization, and one infected central venous catheter), with the exception of the last case, which was a f ever due to antibiotics . The antimicrobial agents administered arelisted in Table 4. Based on microbiological findings, antibiotics were changed in 83% (19/23) of the cases (by a more sensitive or less toxic agent). Pathogens resis-

Bacterial infections are a well-documented complication in patients with HIV and AIDS diagnosis, but most of the clinical reviews are noted for their lack of empyema documentation. 1·2 Bacterial pneumonia is a common source of morbidity and mortality among these patients. The annual incidence of pneumococcal pneumonia and other r espiratory pathogens, such as H injluenzae, are considerably higher when compared with the general population.8·9 There are few cases of clearly documented HIV-associated thoracic empyema, despite both increased incidence and severity of infections. Ambrosi et aP 0 report 16 empyemas and the outcomes w ere successful in all cases with percutaneous drainage (CT guided ). Mouroux et al,U in a study of chest surgical management of thoracic manifestations in HIV-infected patients, described 18 empyemas. Nine cases were treated b y drainage, with the remaining nine requiring decortication; one death was recorded. Only 10 cultures were positive, with a predominance of Streptococcus species (3) and Gram-negative bacilli (3). However, IDUs account for an increasing proportion of the cases of HIV infection and AIDS in North America and Europe . In our series, all the cases had this risk factor and several studies have reported certain distinctive clinical characteristics in these patients.l 2 ·13 IDUs without HIV infection have been noted to suffer an increased risk of bacterial pneumonia;14 HIV infection appears to heighten that risk. Several authors 8·9·13 have reported a higher rate of bacterial pneumonia among HIV-seropositive IDUs CHEST I 113 I 3 I MARCH, 1998

735

Table 3-Characteristics of Hospitalization and Laboratory Tests in Patients With and Without AIDS Diagnosis*

Hospitalization Admission delay' Dminage Drainage delay Drainage time Days of fever Complications Endocarditis Thromboph. 1 > 1 bacteria in pleural fluid Bacteremia > 1 bacteria in all samples Hemoglobin, gldL Leukocytes, cell s/mm 3 Lymphocytes, cells/mm 3 CD4 counts, cells/mm 3

Global (n = 23)

AIDS (n = lO )

Non-AIDS (n = 13)

25.6 :!: 15 8.9 :!: 7.5 18 (78%) 4.7:!:4.8 14.1 :!: 14.1 5.6 :!:4.8 9 (39%) 5 (22%) 6 (26%) 10 (43%) 12 (52%) 13 (57%) 12.1:!: 1.9 7,420 :!:4, 725 1,458 :!: 656 366 :!: 217

31.4:!: 6.2 8.7:!: 9.1 7 (70%) 5.4:!:5.5 14.2:!:15.2 5.8:!:4.9 4 (40%) 2 (20%) 2 (20%) 3 (30%) 6 (60%) 4 (40%) 11.7:!:1.8 6,698 :!:4,929 1,171 :!: 323 171:!:118

26.7 :!:3.7 9.1 :!:5.5 11 (85%) 4.3:!:4.6 14:!: 14.2 5.4 :!:4.9 5 (38%) 3 (23 %) 4 (3 1%) 7 (54%) 6 (46%) 9 (69%) 12.3 :!:2 7,976:!:4,685 1,678 :!: 768 517:!: 139

p Value 0.03 NS NS NS NS NS NS NS NS NS NS NS

s

0.07 0.005 0.005

*Procedure timings are expressed in days :!: SD. NS = not sign ificant. 1From the outset of symptoms. 1Superfici al thrombophlebitis.

with respect to HIV-seronegative IDUs and to the rate observed in other HIV-infected groups. It is also well recognized that community-acquired pneumonia in HIV-infected patients is frequently caused by encapsulated organisms such as S pneumoniae and H injluenzae. 8 ·9 Thus, it appears that the B-cell defects in these patients may explain the increased incidence of these infections. In our series, community-acquired bronchopulmonary infections were the greatest cause of empyema (91 %). In 48% of them, HIV infection was diagnosed concurrently (11/23) and was the primary cause of medical attention in another 17% (4/23). Nevertheless, neither the

Table 4-Antibiotic Therapy and Sensitivity Studies Antibiotics Aminoglycosides Cephalosporins, 2nd-3rd generation Antistaphylococcal penicitlin Clindamycin Antipseudomonal p enicillin Penicill in-ampicillin Others* Monothe rapy Duration, d Hesistance (10/20 patients) Amp icillin Cloxacillin Cephalosporin, 2nd-3rd generation Piperacillin

Initial

Definitive

13 12 10 4 0 0 5 3/23 7:!:3.9

12 8 8 4 3 4 2 2/22 17.4:!:9.6

No. 8 2 6 1\

*Including trimethoprim in four; erythromycin, ciprofloxacin, and imipenem in one case e ach. 1Number of isolated bacteiial resistance.

736

community origin nor the early manifestation during the course of infection was related to the pathogens commonly isolated among the general population or HIV-infected patients in other series.l·2 ·8 ,9,13 The greatest incidence of S aureus appears to be related to IV drug abuse. We found the presence of thrombophlebitis in six cases, five of which were diagnosed as endocarditis. Other organisms, such as anaerobes and Gram-negative bacilli, represented 57% of total isolation in pleural fluid and this finding was probably due to aspiration risk of oropharyngeal flora relative to the same practice. Although other predominant conditions were present in five patients, in three of them, the above-mentioned was coexistent. The finding of S pneumoniae has been described in from 2% 15 to 21% 16 of cases in other series of empyema. In our series, it was found within the expected range (9% ). Among Gram-negative bacilli, H influenzae was isolated in only 3% of our patients, whereas Pseudomona species was more frequent (9%). The percentage of sterile fluids in our series (17% ) was greater than that reported by other authors (7%)_17 This finding was probably due to the early administration of broad-spectrum antibiotics. Bacteremia was detected in 52% of the cases and, once more, S aureus was the bacteria most frequently encountered (29%). In 53% of these episodes, the pathogens isolated were different from those isolated in pleural effusions. Multiple organisms were cultured in 56% of our patients (average, 3.38±2 per patient) and only in three cases (13%) were all the cultures negative. Clinical Investigations

The length of hospitalization, duration of pleural drainage, and the empyemas associated with BP were similar to those reported in other series 17-20 of empyema in HIV-seronegative patients. However, the most notable difference between our series and other studies was that none of the patients needed surgical procedures to control the infection. None of them, once closed chest drainage was instituted, required other types of intervention, and no adverse secondary reactions to intrapleural administration of SK were noted in any patient. The greater success rate in our series with respect to other authors could be due to the immediate onset of pleural drainage (including nonpurulent fluid with a low pH) , early use of fibrinolytics , and the absence of long-standing pleuropulmonary alterations in our patients. Thoracoscopy was not used in any of our patients. Several investigators have reported a successful lysis of adhesions and a debridement of empyema cavities via thoracoscopy. 21 ·22 A success rate of only 60%, together with the inability to perform an adequate decortication by thoracoscopy, limits the use of this technique in some cases. However, it appears to be a reasonable approach to try to minimize the morbidity and mortality associated with thoracic empyema. As in other series, polymicrobial and BP-associated empyemas required longer periods of hospitalization and pleural drainage. Moreover, the presence of endocarditis-thrombophlebitis and bacteremia was more common. The delay in arranging closed drainage was correlated to a longer period of hospitalization and a longer drainage time. However, there was no significant difference between loculated and nonloculated effusions or in the use of intrapleural SK with respect to length of hospitalization or drainage time. Patients with an AIDS diagnosis required a significantly longer hospitalization period. However, on analyzing the delay in hospital admission, the delay or duration of pleural drainage , the days with fever and complications, as severity indicators , no differences were found, although the number of patients studied was small. Other factors , such as the presence of BP (40% vs 15%) and bacteremia (60% vs 46%) , were more common in this group and were probably responsible for prolonged hospitalization. Although pleural effusions are thought to be relatively uncommon, several authors 23 have found a high incidence in AIDS patients (27%) admitted to hospital. Bacterial parapneumonic effusions were the most common cause of this, but reports of thoracic empyema are unusual in these series. 2 3. 24 It seems that the impact of HIV on mononuclear cells (lymphocytes and monocytes), their subsequent cy-

tokine production, and their indirect impact on neutrophils impair the development of empyema despite the increased incidence of bacterial infections in these patients. 6 In our series, aminoglycosides (57%), cephalosporins (52%), and antistaphylococcal penicillins (43%) comprised the empiric therapy commonly used. Among patients in whom at least one pathogen was identified (20/23), resistant organisms were isolated in 10 (SO%). Our findings suggest that ampicillin and cloxacillin would be an inappropriate empiric therapy, if given in a nonselective fashion to these patients, since nearly 16% of isolated organisms were resistant to these drugs. The resistant organisms that we identified are commonly associated with nosocomial infections. This could explain the substantial proportion of more resistant organisms, such as staphylococci and enteric Gram-negative bacilli. Therapy for thoracic empyema in the general population requires appropriate antibiotics, prompt drainage of the infected pleural cavity, and lung reexpansion. In HIV-infected patients, the treatment should be as in other patient groups. For a small number of HIV-positive cases with empyema, minimally invasive methods have been employed, with a success rate similar to that found in our series.l 0 · 11 In our experience, the prevalence of the thoracic empyema in HIV-infected patients was low (5.4%) and particularly frequent among IDUs. In these patients, the empyema either preceded a diagnosis of HIV infection or was the primary cause of hospital admission. The identified organisms differed from the bacterial pathogens isolated in community pneumonia. Factors other than HIV such as drug abuse or poor nutrition may increase host susceptibility to these organisms. Because of the large number of potential pathogens, an aggressive diagnostic approach is warranted to establish a specific diagnosis, thus aiding direct therapy. An AIDS diagnosis was present in 43%, but this did not influence a patient's final outcome. We believe that the best approach to treatment would be an immediate administration of closed thoracostomy and a prompt administration of appropriate antibiotics, just as in other patient groups. This treatment was found to have a high success rate and a low morbidity and mortality rate in our series. ACKNOWLEDGMENT: We thank Marfa Rajo and Isabel Vazquez for their help in preparing the manuscript.

REFERENCES 1 Polsky B, Gold JNM, Whimbey E, eta!. Bacterial pneumonia in patients with acquired immunodeficiency syndrome. Ann Intern Med 1986; 104:38-41 CHEST /113 I 3/ MARCH, 1998

737

2 Witt OJ, Craven DE , McCabe WR. Bacterial infections in adult patients with the acquired immunodeficiency syndrome (AIDS ) and AIDS-related complex. Am J Med 1987; 82: 900-06 3 Centers for Disease Control. Pneumococcal polysaccharide vaccine. MMWR Morb Mortal Wkly Rep 1981; 30:400-12, 417-19 4 Janoff EN, Breiman RF, Daley CL, et al. Pneumococcal disease during HIV infection. Ann Intern Med 1992; 117: 314-24 5 Bawadus MCR, Musher DM , Hamill RJ, et al. Unusual manifestations of pneumococcal infection in human immunodeficiency virus-infected individuals: the past revisited. Clin Infect Dis 1992; 14:192-99 6 Coker RJ. Empyema thoracis in AIDS . J R Soc Med 1994; 87:65-67 7 1993 revised CDC HIV classification system and expanded surveillance definition for adolescents and adults. MMWR Morb Mortal Wkly Rep 1992; 41;RR-17 8 Hirschitick RE, Glassroth J, Jordan ML, et al. Bacterial pneumonia in persons infected with the human immunodeficiency virus. N Engl J Med 1995; 33:845-51 9 Wallace JM , Rao AV, Glassroth J, e t al. Respiratory illness in persons with human immunodeficiency virus infection. Am Hev Hespir Dis 1993; 148:1523-29 10 Ambrosi C, Cianni H, Perona F , et al. The percutaneous treatment of purulent intrathoracic fluid collections in HIVpositive patients. Hadiol Med (Torino) 1994; 87:833-36 11 Mouroux J, Riquet M, Podovani B, et al. Surgical management of thoracic manifestations in human immunodeficiency virus positive patients: indications and results. Br J Surg 1995; 82:39-43 12 Selwyn PA, Alcabes P, Hartel D, et al. Clinical manifestations and predictions of disease progression in drug users with

738

13

14 15 16

17

18

19 20 21 22 23 24

human immunodeficiency virus infection. N Eng! J M ed 1992; 327:1697-1702 Selwyn PA, Feingold AR, Hartel D , e tal . Increased risk of bacterial pneumonia in HIV-infected intravenous drugs users without AIDS. AIDS 1988; 2:267-72 O'Donnell AE , Pappas LS. Pulmona1y complications of intravenous drug abuse. Chest 1988; 94:251-54 Bartlett JG, Gorback SL, Thadepalli H , e t al. Bacteriology of empyema. Lancet 1974; 1:338-40 Brook I, Frazier EH. Aerobic and anaerobic microbiology of empyema: a retrospective review in two military hospitals. Chest 1993; 103:1502-07 Alfageme I, Munoz F, Pefia N, et al. Empyema of the thorax in adults: etiology, microbiological discoveries and management. Chest 1993; 103:839-43 Lemmer JH, Botham MJ, Orringer MB. Modern management of adult thoracic empyema. J Thorac Cardiovasc Surg 1985; 90:849-55 Asbaugh DC. Thoracic empyema: factors influencing morbidity and mortality. Chest 1991; 99:1162-65 Le Mense GP, Strange C, Salm SA. Thoracic empyema: therapeutic management and outcome. Chest 1995; 107: 1532-37 Hidley PO, Baimbridge MV. Thoracoscopic debridement and pleural irrigation in the management of thoracic empyema. Ann Thorac Surg 1991; 51:461-64 Kern JA, Hodgers BM . Thoracoscopies in the management of empyema in children. J Pediatr Surg 1993; 28:1128-32 Joseph J, Strange C, Sahn SA. Pleural effusions in hospitalized patients with AIDS. Ann Intern Med 1993; 118:856-59 Labadibi HMS , Gupta K, Newman T, et al. A retrospective analysis of pleural effusions in human immunodeficiency virus infected patients [abstract]. Chest 1994; 106:86S

Clinical Investigations