Delayed Diagnosis of Tuberculosis in Patients with Human Immunodeficiency Virus Infection FRANCOISEKRAMER,M.D., TAMARAMODILEVSKY,M.D., ABDUL R. WALIANY,M.D., JOHN M. LEEDOM,M.D., PETERF. BARNES,M.D., LosAngeles, California
To determine the frequency with which the diagnosis of tuberculosis is delayed in patients with concomitant human immunodeficiency virus (HIV) infection, and to identify reasons for such delays. PURPOSE
PATIENTS
AND METHODS:
We reviewed
medical
records of 52 consecutive HIV-infected patients with culture-proven tuberculosis seen at a 1,900bed general hospital serving a predominantly indigent population in Los Angeles, where the prevalences of HIV infection and tuberculosis are high. The late-treatment (LT) group consisted of 25 patients in whom tuberculosis was untreated prior to death (n = 6) or treated more than 22 days after presentation (n = 19). The earlytreatment (ET) group comprised 27 patients in whom antituberculous therapy was begun less than 16 days after presentation. RESULTS: Symptoms, physical and laboratory findings, chest roentgenographic abnormalities suggestive of tuberculosis (hilar adenopathy, pleural effusion, miliary pattern, cavitation, predominant upper lobe infiltrate), and frequencies of concomitant nontuberculous disease were similar in LT and ET groups. Delayed diagnosis of tuberculosis was attributable to errors in management in 21 (84%) of 25 LT group patients. The most common error was failure to obtain at least three sputum samples for acidfast smear and mycobacterial culture in patients with clinical and chest roentgenographic fiidings compatible with tuberculosis (15 cases). Acid-fast sputum smears were positive in 25 (61%) of 41 cases of pulmonary tuberculosis. Acid-fast smears of stool were positive in eight From the Department of Medicine (FK, TM, ARW, JML. PFB) and Sections of infectious Diseases (FK, JML) and General Internal Medicine (TM, ARW. PFB). University of Southern California School of Medicine, Los Angeles, California. Computational assistance was provided by the National Institutes of Health, National Center for Research Resources, GCRC MO1 RR-43 CLINFO project. Requests for reprints should be addressed to Peter F. Barnes, M.D., HMR 904. University of Southern California School of Medicme. 2025 Zonal Avenue, Los Angeles, California 90033. Manuscript submitted February 15. 1990, and accepted in revised form July 12. 1990.
(42%) of 19 cases. Blood cultures yielded MYCObacterium tuberculosis in 18 (38%) of 48 CAMS. CONCLUSIONS: Delayed therapy of tuberculosis in HIV-infected patients at our medical center was common and was not due to atypical manifestations of tuberculosis. In most cases, delays could have been avoided if adequate numbers of sputum samples for acid-fast smear and mycobacterial culture had been obtained, and if empiric antituberculous therapy had been given to symptomatic patients in whom chest roentgenographic findings were suggestive of mycobacterial disease.
T
uberculosis is a common concomitant of human immunodeficiency virus (HIV) infection. Timely diagnosis and therapy of tuberculosis are important, not only for the patient’s benefit, but because Mycobacterium tuberculosis is one of the few HIV-associated pathogens that is moderately contagious by the respiratory route. Undiagnosed tuberculosis in HIV-infected patients thus poses a significant health hazard to patient contacts, including health care personnel. In order to determine the frequency with which the diagnosis of tuberculosis is delayed, and to identify reasons for such delays, we reviewed the clinical presentations and diagnostic evaluations of all HIV-infected patients with tuberculosis seen at our medical center during a ‘I-month period.
PATIENTS AND METHODS Patient Population We identified all patients with culture-proven tuberculosis evaluated at the Los Angeles CountyUniversity of Southern California Medical Center from October 1, 1988, through April 30, 1989. The medical records of these patients were evaluated and all HIV-infected patients were included in the present investigation. Because serologic testing for HIV infection is performed in approximately 90% of patients seen at our medical center, it is likely that almost all HIV-infected patients with tuberculosis were identified in this manner. Patients were considered to have pulmonary tuOctober
1990 The American
Journal
of Medicine
Volume
89
451
TUBERCULOSIS IN PATIENTS WITH HIV INFECTION / KRAMER ET AL
TABLEI Symptoms and Physical Findings in Patients with Concomitant Tuberculosis and HIV Infection LT Group* (n = 25)
ET Groupt (n = 27)
Symptoms Fever, number (%) Cough, number (%) Chest pain, number (%) Shortness of breath, number (%) Diarrhea, number (%) Weight loss, number (%) Mean duration of symptoms, weeks f SD
21 (84)
24 (92)
‘A 15 15 20 6.7
g;j (60) (60) (80) f 5.5
‘! I::; 14 (52) 9 (33)* 21 (77) 5.2 f 3.8
38.8* 1.4 10 (40)
39.3 f 1.2 ll(41)
Physicalfindings
.
Mean temperature, “C f SD Cachexia, number (%) Oral thrush, number (%) Generalized lymphadenopathy, number (%) Localized lymphadenopathy, number (%) Hepatomegaly, number (%) -
‘;g
; Is;{
6 (24)
4(15)
6 (25)
3(11)
‘atlent In the L I group were those In whom tubercuIosIS was diagnosed atter death or ., ^^C .1 more man LL oays aner presenranon t Patients in the ET group were those in whom tuberculosis was diagnosed less than 16 days after presentation. * p = 0.05, LTgroupcompared with ETgroup.
berculosis if (1) M. tuberculosis was isolated from pulmonary secretions or pulmonary tissue, or (2) chest roentgenographic abnormalities were suggestive of tuberculosis (hilar adenopathy, pleural effusion, miliary pattern, cavitation, predominant upper lobe infiltrate) and cultures from an extrapulmonary site yielded M. tuberculosis. Microbiologic Methods Clinical specimens were processed by standard methods [1,2]. Acid-fast stains were performed by the method of Truant, and standard procedures were used to isolate and identify M. tuberculosis [2]. The IsolatorTM lysis-centrifugation system (Du Pont Co., Wilmington, Delaware) was used to detect mycobacteremia. Data Collection and Analysis Fifty-four HIV-infected patients with concomitant tuberculosis were identified, and medical records were available in 52 cases. A standard form was used to collect information on clinical features, laboratory test results, and diagnostic evaluation. Patients were considered to have asymptomatic HIV infection prior to the diagnosis of tuberculosis if all their symptoms and signs were attributable to tuberculosis. Patients were considered to have symptomatic HIV infection before the diagnosis of tuberculosis if they had disease manifestations not attributable to tuberculosis, such as oral thrush or infectious diarrhea. The presence of acquired immunodeficiency syndrome (AIDS) prior to the diagnosis of tuberculosis was based on standard criteria 131. 452
October 1990 The American Journal of Medicine
Volume 89
Patients were divided into two groups: the latetreatment (LT) group, comprising those patients in whom tuberculosis was not treated in a timely fashion, and the early-treatment (ET) group, consisting of patients who received prompt antituberculous therapy. In the 46 patients who received antituberculous therapy, the number of days between presentation and initiation of antituberculous treatment was bimodally distributed. Antituberculous treatment was begun in 27 patients 0 to 15 days (median 3 days) after initial presentation and in 19 patients 23 to 112 days (median 57 days) after presentation. No patients were treated after a delay of 16 to 22 days, which provided a natural point to separate the two groups. Thus, the 27 patients treated less than 16 days after presentation comprised the ET group, whereas the 19 patients treated more than 22 days after presentation were included in the LT group. All six patients who died with untreated tuberculosis 10 to 69 days (median 28 days) after initial presentation were included in the LT group as a matter of definition. The LT group therefore consisted of 25 patients. For categorical variables, the chi-square or Fisher’s exact test was used to assess differences between groups. For continuous variables, the t-test was used for variables with a normal distribution, and the Wilcoxon rank-sum test for variables that were not normally distributed.
RESULTS The mean age of our patients was 35.5 years (range 22 to 60 years), and all but four were men. Twenty-three patients (44%) were Hispanic, 16 (31%) were black, and 13 (25%) were non-Hispanic whites. Thirty-three patients (63%) were homosexual or bisexual, 11 (21%) were intravenous drug users, five (10%) were both intravenous drug users and either homosexual or bisexual, and three (6%) had other risk factors for HIV infection. Prior to the diagnosis of tuberculosis, 25 patients (48%) had asymptomatic HIV infection, 14 (27%) had symptomatic HIV infection, and 13 (25%) had AIDS. Ten patients (19%) received zidovudine before the diagnosis of tuberculosis. Concomitant diseases in our patients included Pneumocystis carinii pneumonia (three cases), Kaposi’s sarcoma (two cases), cryptococcosis (two cases), histoplasmosis (two cases), toxoplasmosis (two cases), and lymphoma (one case). The distribution of the aforementioned demographic and clinical characteristics was similar in LT and ET groups (data not shown). We hypothesized that the diagnosis of tuberculosis might have been delayed in LT group patients because their clinical presentations were less frequently suggestive of tuberculosis. However, the frequencies of symptoms and abnormal physical
TUBERCULOSIS
findings were not lower in the LT group than in the ET group, most patients presenting with a subacute illness characterized by respiratory and systemic symptoms (Table I). Ten patients (19%) had pulmonary tuberculosis only, 38 (73%) had both pulmonary and extrapulmonary tuberculosis, and four (7%) had extrapulmonary tuberculosis. In patients with pulmonary tuberculosis, chest roentgenographic findings suggestive of mycobacterial infection (hilar adenopathy, pleural effusion, predominant upper lobe infiltrate, miliary pattern, or cavitation) were present in more than 80% of LT and ET groups (Table II). Thus, atypical chest roentgenographic features did not explain the delayed diagnoses in the LT group. Among the 52 patients, the mean hemoglobin level (k SD) was 100 f 24 g/L, the mean leukocyte count 6.0 f 2.7 X log/L, and the mean CD4 cell count 0.11 f 0.12 X log/L. The mean serum albumin was 30 f 7 g/L and the mean lactate dehydrogenase concentration was 396 f 242 U/L. All these laboratory values were similar in the LT and ET groups (data not shown). Tuberculin skin tests (5 tuberculin units) were performed in four (16%) of 25 LT group patients, compared to 11 (41%) of 27 ET group patients (p = 0.05). Significant skin test reactivity (more than 10 mm induration) was noted in only one case. Table III summarizes the reasons for delayed diagnosis of tuberculosis in the 25 LT group patients. Avoidable errors in management were noted in 21 (84%) of these cases. In 15 cases, fewer than three sputum samples were obtained for acid-fast smear and mycobacterial culture despite the presence of respiratory symptoms (15 of 15) and chest roentgenographic findings suggestive of tuberculosis (14 of 15). When sputum samples were eventually obtained, acid-fast smears were positive in eight of these 15 patients. In three additional cases, the chest roentgenogram suggested tuberculosis, although multiple sputum samples and bronchoscopic specimens did not reveal acid-fast bacilli on smear or granulomatous changes on biopsy. Nevertheless, nontuberculous pathologic processes were not identified in these cases, and we believe that empiric antituberculous therapy was warranted. In one case, the acid-fast sputum smear was positive, and in another, granulomata were present in a transbronchial biopsy specimen, but due to a failure in communication, the physicians were not made aware of these results in a timely fashion. A bone marrow examination was not performed in one patient who presented with a febrile illness and a hemoglobin level of 76 g/L. Autopsy revealed disseminated tuberculosis involving the bone marrow and multiple other organs. In four of the 25 cases, we considered the delayed diagnosis of tuberculosis to
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
TABLEII ChestRoentgenographicFindingsin Patientswith Pulmonary Tuberculosisand HIVInfection
Roentgenographic
Findings
Typical mycobacterial patterns Predominant upper lobe infiltrate Hilar adenopathy Pleural effusion Miliary pattern Cavitation Normal L Patients more than + Patients days after
LT Group* (n = 23) Number(%)
ET Groupt (n = 25) Number(%)
20 (87) 7 (30) z iy 4(17)
2 [:$I ;’ Iif’
2 03)
; [ii
In__.the LT group were those in whom tuberculosis was diagnosed after death c 22 days atter presentatron. in the ET group were those in whom tuberculosis was diagnosed less than 16 presentation.
TABLERI Reasonsfor DelayedDiagnosisof Tuberculosis Number of Patients ~ Avoidable Fewer than 3 sputum samples obtained for 15* mycobacterial culture at initial presentation Respiratory symptoms Mycobacterial chest roentgenographic patternt Mycobacterial chest roentgenographic pattern! 3 3 or more sputum samples, and bronchoscoptc specimens nondiagnostic Positive acid-fast smear or presence of 2 granulomata not reported to physician Bone marrow examination not performed 1 despite a hemoglobin of 76 g/L Unavoidable Atypical manifestations of tuberculosis* 3 Patient refused bronchoscopy 1 _ _. ’ tight Ot these patlentS had am-taSt b.SCIIIIon sputum smear when specimens were eventually obtarned. + The chest roentgenogram showed pleural effusion, hilar adenopathy. a miliary pattern, cavitation, or predominant upper lobe infiltrates. * One patient had tuberculous mastoiditis, one had a tuberculous pancreatic abscess, and one had M. fuberculosis isolated from the bone marrow, despite a normal hemoglobin, leukocyte count, and platelet count.
be unavoidable. In one case, the patient refused bronchoscopy, and in three others, the manifestations of tuberculosis were extremely atypical. One patient underwent mastoidectomy for mastoiditis, and acid-fast smears of the surgical specimens were negative, although cultures yielded M. tuberculosis. One patient had a tuberculous pancreatic abscess, and another had M. tuberculosis isolated from bone marrow despite a normal hemoglobin, leukocyte count, and platelet count. In 41 patients with pulmonary tuberculosis in whom specimens were obtained for mycobacterial culture, acid-fast sputum smears were positive in 61% of cases and cultures yielded M. tuberculosis in 88% of patients (Table IV). The yields of these diagnostic tests were similar in the LT and ET groups. In 12% of cases, acid-fast smears and cultures were negative, but chest roentgenographic ab-
October
1990
The American
Journal
of Medicine
Volume
89
453
TUBERCULOSIS
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
fast smear from any source or granulomatous inflammation on biopsy was found in 12 (48%) of 25 LT group patients, compared to 26 (93%) of 27 ET group patients (p = 0.0001). Among the 25 LT group patients, 20 had followup data for 0 to 13 months (mean 4.0 months) after diagnosis. During this period, 13 patients died, nine of tuberculosis and four from other complications of HIV infection. Among the 27 ET group patients, 21 had follow-up data available for 1 to 13 months (mean 6.5 months). Of these, 10 died, four of tuberculosis and six from nontuberculous processes. Antituberculous therapy was begun in the remaining five patients in the LT group and six patients in the ET group, but no further information on their clinical course during therapy was available.
TABLE IV Yield of Diagnostic Tests in Patients with Pulmonary Tuberculosis and HIV Infection Specimen
PulmonaryTuberculosis(n = 48) n positive/n tested (%)*
Sputum Acid-fast smear Mycobacterial culture Bronchoalveolar lavage Acid-fast smear Mycobacterial culture Transbronchial biopsy Acid-fast smear Mycobacterial culture Granulomas
25/41(61) 36/41(88) 2/21(10) 12/21(57) 3121 (14) 9/21(43) 4/21 (19)
1refers to the number of patients in whom specimens were obtained.
TABLE V Yield of Diagnostic Tests in Patients with Tuberculosis and HIV Infection
COMMENTS Depressed cell-mediated immunity is associated with poor outcomes in patients with tuberculosis [4]. Individuals with HIV-induced immunosuppression are thus at high risk for complications of tuberculosis, and require prompt diagnosis and therapy. We found that 48% of HIV-infected patients with tuberculosis were not expeditiously treated. Among the patients in whom follow-up was available, nine (45%) of 20 LT group patients died of tuberculosis, compared to four (19%) of 21 ET group patients, suggesting that delayed therapy resulted in increased mortality. We cannot exclude the possibility that this difference in mortality rates was due to differences in patient compliance with antituberculous medications, as this variable could not be systematically evaluated in a retrospective manner. Delayed diagnosis was rarely due to atypical clinical features, but may have been related to a lower frequency of positive acid-fast smears or granulomatous changes in clinical specimens, which in turn may have reflected a less thorough evaluation for tuberculosis in LT group patients. This explanation is supported by the finding that tuberculin skin tests were performed significantly less frequently in the LT group. Delayed diagnosis of tuberculosis was due to potentially avoidable errors in most cases (Table III). The most frequent problem was that inadequate numbers of sputum samples were obtained for mycobacterial studies in patients with respiratory symptoms and chest roentgenographic findings suggestive of mycobacterial disease (hilar adenopathy, pleural effusion, a miliary pattern, predominant upper lobe infiltrates, or cavitation). These findings, particularly hilar adenopathy or pleural effusion, are not features of P. carinii pneumonia [5], cytomegalovirus pneumonitis [6], or persistent
Tuberculosis (n = 49) n positive/n tested (%)’
Specimen Lymph node biopsy Acid-fast smear Mycobacterial culture Granulomas Bone marrow biopsy Acid-fast smear Mycobacterial culture Granulomas
5/10 (50) 7/10 (70) 3/7 (43)’ 3/12 3/12 6/12
(25) (25) (50)
8/19 11/19
(42) (58)
18/48
(38)
Stool
Acid-fast smear Mvcobacterial culture Blood Mycobacterial culture
1 refers to the number of patients in whom specimens were obtained. rlssue was inadequate for histologic examination in three fine-needle asplratlon blopsies.
normalities were suggestive of pulmonary tuberculosis and cultures of extrapulmonary sites yielded M. tuberculosis. The yield of acid-fast smears of bronchoscopic specimens was relatively low, and granulomas were noted on transbronchial biopsy in only 19% of cases. The frequencies of positive acid-fast smears, histologic findings, and mycobacterial cultures in extrapulmonary specimens are shown in Table V, and were similar in the LT and ET groups. Mean hemo‘globin levels and mean leukocyte counts were similar in patients with and without granulomata on bone marrow biopsy (data not shown). Acid-fast smears of stool were positive in six of 13 patients who complained of diarrhea, and in two of six patients who did not have diarrhea. Of 48 patients who had a mean of 3.5 f 2.4 blood samples obtained for mycobacterial culture, mycobacteremia was found in 18 cases. Blood was the only extrapulmonary site from which M. tuberculosis was isolated in 10 cases. When results of all acid-fast smears and histologic studies were reviewed, a positive acid-
454
October
1990
The American
Journal
of Medicine
Volume
89
TUBERCULOSIS
generalized lymphadenopathy [7], and should suggest the possibility of tuberculosis. We believe that empiric antituberculous therapy is warranted in all HIV-infected patients in whom chest roentgenographic abnormalities suggestive of mycobacterial infection are unexplained by diseases other than tuberculosis. Although these roentgenographic characteristics can be due to nontuberculous etiologies, clinical features and laboratory test results often suggest these diagnoses. Hilar adenopathy and pleural effusion+ may be manifestations of Kaposi’s sarcoma or lymphoma, but there is usually extrapulmonary evidence of these diseases [8,9]. Coccidioidomycosis and histoplasmosis can cause chest roentgenographic abnormalities mimicking tuberculosis, but diffuse infiltrates are more typical, extrapulmonary manifestations are prominent, and results of serologic tests for fungal infection are often positive [lo-121. Although pulmonary Mycobacterium auium complex infection may be roentgenographically indistinguishable from tuberculosis, other clinical features can separate these two infections [13]. Empiric treatment of tuberculosis does not substitute for a diagnostic evaluation, but is warranted until a definitive diagnosis is made. Antituberculous medications are well tolerated by HIV-infected patients [14], and a therapeutic trial is often diagnostically useful. Eightyfive percent (22 of 26) of HIV-infected patients with tuberculosis defervesce during the first week of therapy (Francoise Kramer, M.D., unpublished data), similar to findings in immunocompetent patients [15]. Because tuberculosis has declined in prevalence, most physicians have little experience with the illness, and are therefore often reluctant to treat tuberculosis in the absence of definitive evidence of mycobacterial disease. However, the frequency of positive acid-fast sputum smears is lower in HIVinfected patients than in immunocompetent adults [16]. At our hospital, acid-fast sputum smears are positive in 84% of non-HIV-infected patients with pulmonary tuberculosis [17], compared to 61% in the present series. This is not surprising, as roentgenographic manifestations of tuberculosis in HIVinfected patients are generally those of primary tuberculosis, in which acid-fast sputum smears are rarely positive. Therapy of primary tuberculosis in children is based on clinical features, and we favor a similar approach in HIV-infected patients. In order to facilitate the diagnosis of tuberculosis, it is important to determine the yield of diagnostic tests for mycobacteriosis in HIV-infected patients. Acid-fast sputum smears are positive in 45% to 82% of HIV-infected patients with pulmonary tubercu-
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
losis [13,16] and should be obtained in all patients with clinical evidence of pulmonary disease. The frequency of positive sputum smears does not correlate with chest roentgenographic findings, and positive smears are seen in some patients with normal chest roentgenograms [16]. In the present study, the yield of acid-fast smears and histologic findings of bronchoscopic specimens was low, compared to that in non-HIV-infected patients [18,19]. Nevertheless, bronchoscopy was useful in excluding nontuberculous processes. Although acid-fast stool smears are often used to diagnose M. auium complex infection in HIV-infected patients [20], their value in diagnosing tuberculosis has not been emphasized. Acid-fast smears of stool are positive in 40% of HIV-infected patients with tuberculosis ([ 131, present study). Although the number of patients evaluated was small, we found no correlation between positive acid-fast stool smears and a history of diarrhea. Thus, stool smears for acid-fast bacilli may be a useful noninvasive means to diagnose tuberculosis, even in the absence of diarrhea. Blood cultures yielded M. tuberculosis in 38% (18 of 48) of our patients and should be performed in all HIV-infected patients in whom tuberculosis is suspected. When the Isolator lysis-centrifugation system, combined with the BACTECTM radiometric system (Johnston Laboratories, Towson, Maryland),for mycobacterial culture and identification, is used, M. tuberculosis can be identified in less than 3 weeks [21]. Diagnostic delays would probably have been reduced in our patients if nucleic acid probes had been used for identification of mycobacteria. Blood was the only extrapulmonary source of M. tuberculosis in 10 cases. In HIV-infected patients, it is important to document extrapulmonary tuberculosis because this condition constitutes an AIDS-defining illness, whereas pulmonary tuberculosis does not. Tuberculin skin testing was performed in only 29% (15 of 52) of our patients, probably because many physicians believe that HIV-infected patients rarely develop reactivity on skin tests. However, 11% to 57% of HIV-infected patients with tuberculosis have significant tuberculin skin test results (more than 10 mm induration) [22-251. The frequency of reactive tests is highest (71%) when tuberculosis occurs more than 2 years prior to the diagnosis of AIDS, declining to 33% when tuberculosis occurs at or after the time of AIDS diagnosis [22]. Only one of our 15 patients had a reactive skin test, perhaps reflecting the mean CD4+ cell count of only 0.11 X log/L (110/mm3). We used 10 mm of induration as the criterion for a significant tubercu-
October
1990
The American
Journal
of Medicine
Volume
89
455
TUBERCULOSIS
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
lin skin test result because values of less than 10 mm were often not quantified in the medical record. However, in HIV-infected patients, 5 mm of induration should be considered significant [26], and application of this criterion would increase the test’s sensitivity. The most important finding of this study is that the diagnosis of tuberculosis in HIV-infected patients was delayed in 48% of cases. It is unlikely that this is because physicians at our institution are relatively inept at recognizing tuberculosis. First, 12% (six of 52) of our HIV-infected patients with tuberculosis were undiagnosed prior to death, compared to 18% (32 of 175) of those in Los Angeles County (Paul Davidson, M.D., personal communication) and 26% to 28% of those reported from other institutions [24,27]. Second, of 186 non-HIV-infected patients with pulmonary tuberculosis evaluated at our hospital, a diagnostic delay of 3 weeks or more occurred in only 13 patients (7%) and only five patients (2.6%) died with untreated tuberculosis (Peter Barnes, M.D., unpublished data). These percentages compare favorably with published reports on the accuracy of diagnosing tuberculosis [28,29]. Thus, we believe that the high frequency of delayed diagnosis of tuberculosis in HIV-infected patients is not unique to our institution, and is likely to be representative of the experience at other hospitals and in private-practice settings. Increased awareness of the clinical manifestations of tuberculosis and wider use of empiric antituberculous therapy should decrease morbidity and mortality from this treatable complication of HIV infection.
6. Drew WL. Cytomegalovlrus infection in patients with AIDS. J Infect Dis 1988; 158: 449-56. 7. Abrams DI. Lewis BJ. Beckstead JH, Casavant CA, Drew WL. Persistent diffuse lymphadenopathy in homosexual men: endpoint or prodrome? Ann Intern Med 1984; 100: 801-8. 8. Meduri GU, Stover DE, Lee M, Myskowski PL. Caravelll JF. Zaman MB. Pulmonary Kaposi’s sarcoma in the acquired immune deficiency syndrome, Clinical, radiographic and pathologic manifestations. Am J Med 1986; 81: 11-8. 9. O’Brien RF, Cohn DL. Serosanguineous pleural effuslons in AIDS-associated Kaposi’s sarcoma. Chest 1989; 96: 460-6. 10. Bronnimann DA, Adam RD. Galgiani JN. et a/. Coccidioldomycosis in the acquired immunodeficiency syndrome. Ann Intern Med 1987; 106: 372-9. 11. Wheat LJ, Slama TG, Zeckel ML. Histoplasmosis in the acquired immune deficiency syndrome. Am J Med 1985; 78: 203-10. 12. Johnson PC, Khardori N, Najjar AF, Butt F, Mansell PWA. Sarosi GA. Progressive disseminated histoplasmosis in patients with acquired lmmunodeficiency syndrome. Am J Med 1988; 85: 152-8. 13. Modilevsky T, Sattler FR, Barnes PF. Mycobacterial disease in patients with human immunodeficiency virus Infection. Arch Intern Med 1989; 149: 2201-5. 14. Pitchenik AE. Fertel D. Bloch AB. Mycobacterial disease: epidemiology, diagnosis, treatment, and prevention. Clin Chest Med 1988; 9: 425-41. 15. Barnes PF, Chan LS, Wong SF. The course of fever during treatment pulmonary tuberculosis. Tubercle 1987; 68: 255-60.
16. Klein NC, Duncanson FP. Lenox TH Ill, Pitta A, Cohen SC, Wormser GP. Use of mycobacterial smears in the diagnosis of pulmonary tuberculosis in AIDS/ ARC patients. Chest 1989; 95: 1190-2. 17. Barnes PF, Verdegem TD, Vachon IA, Leedom JM, Overturf GD. Chest roentgenogram in pulmonary tuberculosis. New data on an old test. Chest 1988; 94: 316-20. 18. So SY, Lam WK. Yu DYC. Rapid diagnosis of suspected pulmonary tuberculoSIS by fiberoptic bronchoscopy. Tubercle 1982; 63: 195-200. 19. Chan JC, So S-Y. Lam WK, Ip MSM. High incidence of pulmonary tuberculoSIS in non-HIV infected immunocompromised patients in Hong Kong. Chest 1989; 96: 835-9. 20. Kiehn TE. Edwards FF. Brannon P, et al. InfectIons caused by Mycobacterium avium complex in immunocompromised patients: diagnosis by blood culture and fecal examination, antimicrobial susceptibility tests, and morphological and seroagglutination characteristics. J Clin Microbial 1985; 21: 168-73. 21. Witebsky FG. Keiser JF. Conville PS. et al. Comparison of BACTEC 13A and Du Pont isolator for detection of mycobacteremia. J Clin Microbial 1988; 26: 1501-5. 22. Rieder HL, Cauthen GM, Bloch AB. et al. Tuberculosis and acquired Immunedeficiency syndrome-Florida. Arch Intern Med 1989; 149: 1268-73. 23. Pitchenik AE, Burr J, Suarez M. Fertel D. Gonzalez G. Moas C. Human T-cell lymphotroplc wrus-III (HTLV-III) seropositivity and related disease among 71 consecutive patients in whom tuberculosis was diagnosed. A prospective study. Am Rev Respir Dis 1987; 135: 875-9. 24. Louie E. Rice LB, Holzman RS. Tuberculosis in non-Haitian patients with acquired tmmunodeficiency syndrome. Chest 1986; 90: 542-5. 25. Chaisson RE. Schecter GF. Theuer CP. Rutherford GW, Echenberg DF, Hopewell PC. Tuberculosis in patients with the acquired lmmunodeficiency syndrome. Am Rev Respir Dis 1987; 136: 570-4. 26. Centers for Disease Control. Tuberculosis and human immunodeficiency virus infection: recommendations for the Advisory Committee for the Elimination of Tuberculosis (ACET). MMWR 1989; 38: 236-8. 27. Sunderam G. McDonald RJ. Maniatis T, Oleske J. Kapila R. Reichman LB. Tuberculosis as a manifestation of the acquired immunodeficiency syndrome (AIDS). JAMA 1986: 256: 362-6. 28. Greenbaum M, Beyt BE Jr, Murray PR. The accuracy of diagnosing pulmonary tuberculosis at a teaching hospital. Am Rev Respir Dis 1980; 121: 477-81. 29. Counsel1 SR, Tan JS. Dittus RS. Unsuspected pulmonary tuberculosis In a community hospital. Arch Intern Med 1989; 149: 1274-8.
ACKNOWLEDGMENT We thank Linda Vachon. M.D., for her expert review of the chest roentgenograms, Ken Anderson, MS., CLINFD systems manager, for assistance in data analysis, William Lewis, Ph.D.. for assistance in identifying the patients included in this report, and Joseph Indenbaum. M.D., and Om Sharma. M.D., for their thoughtful review of the manuscript.
REFERENCES 1. Barnes PF, Arevalo C. Six cases of Mycobacterium Infect Dis 1987; 156: 377-9.
tuberculosis
bacteremia.
J
2. Sommers HM. Good RC. Mycobacterium. In: Lennette EH. Balows A, Hausler WJ Jr, Shadomy HJ. eds. Manual of clinical microbiology. Washington: American Society for Microbiology, 1985: 216-48. 3. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987; 36(1S): 3-155. 4. Barnes PF. Leedom JM. Chan LS, et a/. Predictors of short-term prognosis in patients with pulmonary tuberculosis. J Infect Dis 1988; 158: 366-71. 5. Hopewell PC. Pneumocystis carinii pneumonia: diagnosis. J Infect DIS 1988; 157: 1115-9.
456
October
1990
The American
Journal
of Medicine
Volume
of
89
To determine the frequency with which the diagnosis of tuberculosis is delayed in patients with concomitant human immunodeficiency virus (HIV) infection, and to identify reasons for such delays. PURPOSE
PATIENTS
AND METHODS:
We reviewed
medical
records of 52 consecutive HIV-infected patients with culture-proven tuberculosis seen at a 1,900bed general hospital serving a predominantly indigent population in Los Angeles, where the prevalences of HIV infection and tuberculosis are high. The late-treatment (LT) group consisted of 25 patients in whom tuberculosis was untreated prior to death (n = 6) or treated more than 22 days after presentation (n = 19). The earlytreatment (ET) group comprised 27 patients in whom antituberculous therapy was begun less than 16 days after presentation. RESULTS: Symptoms, physical and laboratory findings, chest roentgenographic abnormalities suggestive of tuberculosis (hilar adenopathy, pleural effusion, miliary pattern, cavitation, predominant upper lobe infiltrate), and frequencies of concomitant nontuberculous disease were similar in LT and ET groups. Delayed diagnosis of tuberculosis was attributable to errors in management in 21 (84%) of 25 LT group patients. The most common error was failure to obtain at least three sputum samples for acidfast smear and mycobacterial culture in patients with clinical and chest roentgenographic fiidings compatible with tuberculosis (15 cases). Acid-fast sputum smears were positive in 25 (61%) of 41 cases of pulmonary tuberculosis. Acid-fast smears of stool were positive in eight From the Department of Medicine (FK, TM, ARW, JML. PFB) and Sections of infectious Diseases (FK, JML) and General Internal Medicine (TM, ARW. PFB). University of Southern California School of Medicine, Los Angeles, California. Computational assistance was provided by the National Institutes of Health, National Center for Research Resources, GCRC MO1 RR-43 CLINFO project. Requests for reprints should be addressed to Peter F. Barnes, M.D., HMR 904. University of Southern California School of Medicme. 2025 Zonal Avenue, Los Angeles, California 90033. Manuscript submitted February 15. 1990, and accepted in revised form July 12. 1990.
(42%) of 19 cases. Blood cultures yielded MYCObacterium tuberculosis in 18 (38%) of 48 CAMS. CONCLUSIONS: Delayed therapy of tuberculosis in HIV-infected patients at our medical center was common and was not due to atypical manifestations of tuberculosis. In most cases, delays could have been avoided if adequate numbers of sputum samples for acid-fast smear and mycobacterial culture had been obtained, and if empiric antituberculous therapy had been given to symptomatic patients in whom chest roentgenographic findings were suggestive of mycobacterial disease.
T
uberculosis is a common concomitant of human immunodeficiency virus (HIV) infection. Timely diagnosis and therapy of tuberculosis are important, not only for the patient’s benefit, but because Mycobacterium tuberculosis is one of the few HIV-associated pathogens that is moderately contagious by the respiratory route. Undiagnosed tuberculosis in HIV-infected patients thus poses a significant health hazard to patient contacts, including health care personnel. In order to determine the frequency with which the diagnosis of tuberculosis is delayed, and to identify reasons for such delays, we reviewed the clinical presentations and diagnostic evaluations of all HIV-infected patients with tuberculosis seen at our medical center during a ‘I-month period.
PATIENTS AND METHODS Patient Population We identified all patients with culture-proven tuberculosis evaluated at the Los Angeles CountyUniversity of Southern California Medical Center from October 1, 1988, through April 30, 1989. The medical records of these patients were evaluated and all HIV-infected patients were included in the present investigation. Because serologic testing for HIV infection is performed in approximately 90% of patients seen at our medical center, it is likely that almost all HIV-infected patients with tuberculosis were identified in this manner. Patients were considered to have pulmonary tuOctober
1990 The American
Journal
of Medicine
Volume
89
451
TUBERCULOSIS IN PATIENTS WITH HIV INFECTION / KRAMER ET AL
TABLEI Symptoms and Physical Findings in Patients with Concomitant Tuberculosis and HIV Infection LT Group* (n = 25)
ET Groupt (n = 27)
Symptoms Fever, number (%) Cough, number (%) Chest pain, number (%) Shortness of breath, number (%) Diarrhea, number (%) Weight loss, number (%) Mean duration of symptoms, weeks f SD
21 (84)
24 (92)
‘A 15 15 20 6.7
g;j (60) (60) (80) f 5.5
‘! I::; 14 (52) 9 (33)* 21 (77) 5.2 f 3.8
38.8* 1.4 10 (40)
39.3 f 1.2 ll(41)
Physicalfindings
.
Mean temperature, “C f SD Cachexia, number (%) Oral thrush, number (%) Generalized lymphadenopathy, number (%) Localized lymphadenopathy, number (%) Hepatomegaly, number (%) -
‘;g
; Is;{
6 (24)
4(15)
6 (25)
3(11)
‘atlent In the L I group were those In whom tubercuIosIS was diagnosed atter death or ., ^^C .1 more man LL oays aner presenranon t Patients in the ET group were those in whom tuberculosis was diagnosed less than 16 days after presentation. * p = 0.05, LTgroupcompared with ETgroup.
berculosis if (1) M. tuberculosis was isolated from pulmonary secretions or pulmonary tissue, or (2) chest roentgenographic abnormalities were suggestive of tuberculosis (hilar adenopathy, pleural effusion, miliary pattern, cavitation, predominant upper lobe infiltrate) and cultures from an extrapulmonary site yielded M. tuberculosis. Microbiologic Methods Clinical specimens were processed by standard methods [1,2]. Acid-fast stains were performed by the method of Truant, and standard procedures were used to isolate and identify M. tuberculosis [2]. The IsolatorTM lysis-centrifugation system (Du Pont Co., Wilmington, Delaware) was used to detect mycobacteremia. Data Collection and Analysis Fifty-four HIV-infected patients with concomitant tuberculosis were identified, and medical records were available in 52 cases. A standard form was used to collect information on clinical features, laboratory test results, and diagnostic evaluation. Patients were considered to have asymptomatic HIV infection prior to the diagnosis of tuberculosis if all their symptoms and signs were attributable to tuberculosis. Patients were considered to have symptomatic HIV infection before the diagnosis of tuberculosis if they had disease manifestations not attributable to tuberculosis, such as oral thrush or infectious diarrhea. The presence of acquired immunodeficiency syndrome (AIDS) prior to the diagnosis of tuberculosis was based on standard criteria 131. 452
October 1990 The American Journal of Medicine
Volume 89
Patients were divided into two groups: the latetreatment (LT) group, comprising those patients in whom tuberculosis was not treated in a timely fashion, and the early-treatment (ET) group, consisting of patients who received prompt antituberculous therapy. In the 46 patients who received antituberculous therapy, the number of days between presentation and initiation of antituberculous treatment was bimodally distributed. Antituberculous treatment was begun in 27 patients 0 to 15 days (median 3 days) after initial presentation and in 19 patients 23 to 112 days (median 57 days) after presentation. No patients were treated after a delay of 16 to 22 days, which provided a natural point to separate the two groups. Thus, the 27 patients treated less than 16 days after presentation comprised the ET group, whereas the 19 patients treated more than 22 days after presentation were included in the LT group. All six patients who died with untreated tuberculosis 10 to 69 days (median 28 days) after initial presentation were included in the LT group as a matter of definition. The LT group therefore consisted of 25 patients. For categorical variables, the chi-square or Fisher’s exact test was used to assess differences between groups. For continuous variables, the t-test was used for variables with a normal distribution, and the Wilcoxon rank-sum test for variables that were not normally distributed.
RESULTS The mean age of our patients was 35.5 years (range 22 to 60 years), and all but four were men. Twenty-three patients (44%) were Hispanic, 16 (31%) were black, and 13 (25%) were non-Hispanic whites. Thirty-three patients (63%) were homosexual or bisexual, 11 (21%) were intravenous drug users, five (10%) were both intravenous drug users and either homosexual or bisexual, and three (6%) had other risk factors for HIV infection. Prior to the diagnosis of tuberculosis, 25 patients (48%) had asymptomatic HIV infection, 14 (27%) had symptomatic HIV infection, and 13 (25%) had AIDS. Ten patients (19%) received zidovudine before the diagnosis of tuberculosis. Concomitant diseases in our patients included Pneumocystis carinii pneumonia (three cases), Kaposi’s sarcoma (two cases), cryptococcosis (two cases), histoplasmosis (two cases), toxoplasmosis (two cases), and lymphoma (one case). The distribution of the aforementioned demographic and clinical characteristics was similar in LT and ET groups (data not shown). We hypothesized that the diagnosis of tuberculosis might have been delayed in LT group patients because their clinical presentations were less frequently suggestive of tuberculosis. However, the frequencies of symptoms and abnormal physical
TUBERCULOSIS
findings were not lower in the LT group than in the ET group, most patients presenting with a subacute illness characterized by respiratory and systemic symptoms (Table I). Ten patients (19%) had pulmonary tuberculosis only, 38 (73%) had both pulmonary and extrapulmonary tuberculosis, and four (7%) had extrapulmonary tuberculosis. In patients with pulmonary tuberculosis, chest roentgenographic findings suggestive of mycobacterial infection (hilar adenopathy, pleural effusion, predominant upper lobe infiltrate, miliary pattern, or cavitation) were present in more than 80% of LT and ET groups (Table II). Thus, atypical chest roentgenographic features did not explain the delayed diagnoses in the LT group. Among the 52 patients, the mean hemoglobin level (k SD) was 100 f 24 g/L, the mean leukocyte count 6.0 f 2.7 X log/L, and the mean CD4 cell count 0.11 f 0.12 X log/L. The mean serum albumin was 30 f 7 g/L and the mean lactate dehydrogenase concentration was 396 f 242 U/L. All these laboratory values were similar in the LT and ET groups (data not shown). Tuberculin skin tests (5 tuberculin units) were performed in four (16%) of 25 LT group patients, compared to 11 (41%) of 27 ET group patients (p = 0.05). Significant skin test reactivity (more than 10 mm induration) was noted in only one case. Table III summarizes the reasons for delayed diagnosis of tuberculosis in the 25 LT group patients. Avoidable errors in management were noted in 21 (84%) of these cases. In 15 cases, fewer than three sputum samples were obtained for acid-fast smear and mycobacterial culture despite the presence of respiratory symptoms (15 of 15) and chest roentgenographic findings suggestive of tuberculosis (14 of 15). When sputum samples were eventually obtained, acid-fast smears were positive in eight of these 15 patients. In three additional cases, the chest roentgenogram suggested tuberculosis, although multiple sputum samples and bronchoscopic specimens did not reveal acid-fast bacilli on smear or granulomatous changes on biopsy. Nevertheless, nontuberculous pathologic processes were not identified in these cases, and we believe that empiric antituberculous therapy was warranted. In one case, the acid-fast sputum smear was positive, and in another, granulomata were present in a transbronchial biopsy specimen, but due to a failure in communication, the physicians were not made aware of these results in a timely fashion. A bone marrow examination was not performed in one patient who presented with a febrile illness and a hemoglobin level of 76 g/L. Autopsy revealed disseminated tuberculosis involving the bone marrow and multiple other organs. In four of the 25 cases, we considered the delayed diagnosis of tuberculosis to
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
TABLEII ChestRoentgenographicFindingsin Patientswith Pulmonary Tuberculosisand HIVInfection
Roentgenographic
Findings
Typical mycobacterial patterns Predominant upper lobe infiltrate Hilar adenopathy Pleural effusion Miliary pattern Cavitation Normal L Patients more than + Patients days after
LT Group* (n = 23) Number(%)
ET Groupt (n = 25) Number(%)
20 (87) 7 (30) z iy 4(17)
2 [:$I ;’ Iif’
2 03)
; [ii
In__.the LT group were those in whom tuberculosis was diagnosed after death c 22 days atter presentatron. in the ET group were those in whom tuberculosis was diagnosed less than 16 presentation.
TABLERI Reasonsfor DelayedDiagnosisof Tuberculosis Number of Patients ~ Avoidable Fewer than 3 sputum samples obtained for 15* mycobacterial culture at initial presentation Respiratory symptoms Mycobacterial chest roentgenographic patternt Mycobacterial chest roentgenographic pattern! 3 3 or more sputum samples, and bronchoscoptc specimens nondiagnostic Positive acid-fast smear or presence of 2 granulomata not reported to physician Bone marrow examination not performed 1 despite a hemoglobin of 76 g/L Unavoidable Atypical manifestations of tuberculosis* 3 Patient refused bronchoscopy 1 _ _. ’ tight Ot these patlentS had am-taSt b.SCIIIIon sputum smear when specimens were eventually obtarned. + The chest roentgenogram showed pleural effusion, hilar adenopathy. a miliary pattern, cavitation, or predominant upper lobe infiltrates. * One patient had tuberculous mastoiditis, one had a tuberculous pancreatic abscess, and one had M. fuberculosis isolated from the bone marrow, despite a normal hemoglobin, leukocyte count, and platelet count.
be unavoidable. In one case, the patient refused bronchoscopy, and in three others, the manifestations of tuberculosis were extremely atypical. One patient underwent mastoidectomy for mastoiditis, and acid-fast smears of the surgical specimens were negative, although cultures yielded M. tuberculosis. One patient had a tuberculous pancreatic abscess, and another had M. tuberculosis isolated from bone marrow despite a normal hemoglobin, leukocyte count, and platelet count. In 41 patients with pulmonary tuberculosis in whom specimens were obtained for mycobacterial culture, acid-fast sputum smears were positive in 61% of cases and cultures yielded M. tuberculosis in 88% of patients (Table IV). The yields of these diagnostic tests were similar in the LT and ET groups. In 12% of cases, acid-fast smears and cultures were negative, but chest roentgenographic ab-
October
1990
The American
Journal
of Medicine
Volume
89
453
TUBERCULOSIS
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
fast smear from any source or granulomatous inflammation on biopsy was found in 12 (48%) of 25 LT group patients, compared to 26 (93%) of 27 ET group patients (p = 0.0001). Among the 25 LT group patients, 20 had followup data for 0 to 13 months (mean 4.0 months) after diagnosis. During this period, 13 patients died, nine of tuberculosis and four from other complications of HIV infection. Among the 27 ET group patients, 21 had follow-up data available for 1 to 13 months (mean 6.5 months). Of these, 10 died, four of tuberculosis and six from nontuberculous processes. Antituberculous therapy was begun in the remaining five patients in the LT group and six patients in the ET group, but no further information on their clinical course during therapy was available.
TABLE IV Yield of Diagnostic Tests in Patients with Pulmonary Tuberculosis and HIV Infection Specimen
PulmonaryTuberculosis(n = 48) n positive/n tested (%)*
Sputum Acid-fast smear Mycobacterial culture Bronchoalveolar lavage Acid-fast smear Mycobacterial culture Transbronchial biopsy Acid-fast smear Mycobacterial culture Granulomas
25/41(61) 36/41(88) 2/21(10) 12/21(57) 3121 (14) 9/21(43) 4/21 (19)
1refers to the number of patients in whom specimens were obtained.
TABLE V Yield of Diagnostic Tests in Patients with Tuberculosis and HIV Infection
COMMENTS Depressed cell-mediated immunity is associated with poor outcomes in patients with tuberculosis [4]. Individuals with HIV-induced immunosuppression are thus at high risk for complications of tuberculosis, and require prompt diagnosis and therapy. We found that 48% of HIV-infected patients with tuberculosis were not expeditiously treated. Among the patients in whom follow-up was available, nine (45%) of 20 LT group patients died of tuberculosis, compared to four (19%) of 21 ET group patients, suggesting that delayed therapy resulted in increased mortality. We cannot exclude the possibility that this difference in mortality rates was due to differences in patient compliance with antituberculous medications, as this variable could not be systematically evaluated in a retrospective manner. Delayed diagnosis was rarely due to atypical clinical features, but may have been related to a lower frequency of positive acid-fast smears or granulomatous changes in clinical specimens, which in turn may have reflected a less thorough evaluation for tuberculosis in LT group patients. This explanation is supported by the finding that tuberculin skin tests were performed significantly less frequently in the LT group. Delayed diagnosis of tuberculosis was due to potentially avoidable errors in most cases (Table III). The most frequent problem was that inadequate numbers of sputum samples were obtained for mycobacterial studies in patients with respiratory symptoms and chest roentgenographic findings suggestive of mycobacterial disease (hilar adenopathy, pleural effusion, a miliary pattern, predominant upper lobe infiltrates, or cavitation). These findings, particularly hilar adenopathy or pleural effusion, are not features of P. carinii pneumonia [5], cytomegalovirus pneumonitis [6], or persistent
Tuberculosis (n = 49) n positive/n tested (%)’
Specimen Lymph node biopsy Acid-fast smear Mycobacterial culture Granulomas Bone marrow biopsy Acid-fast smear Mycobacterial culture Granulomas
5/10 (50) 7/10 (70) 3/7 (43)’ 3/12 3/12 6/12
(25) (25) (50)
8/19 11/19
(42) (58)
18/48
(38)
Stool
Acid-fast smear Mvcobacterial culture Blood Mycobacterial culture
1 refers to the number of patients in whom specimens were obtained. rlssue was inadequate for histologic examination in three fine-needle asplratlon blopsies.
normalities were suggestive of pulmonary tuberculosis and cultures of extrapulmonary sites yielded M. tuberculosis. The yield of acid-fast smears of bronchoscopic specimens was relatively low, and granulomas were noted on transbronchial biopsy in only 19% of cases. The frequencies of positive acid-fast smears, histologic findings, and mycobacterial cultures in extrapulmonary specimens are shown in Table V, and were similar in the LT and ET groups. Mean hemo‘globin levels and mean leukocyte counts were similar in patients with and without granulomata on bone marrow biopsy (data not shown). Acid-fast smears of stool were positive in six of 13 patients who complained of diarrhea, and in two of six patients who did not have diarrhea. Of 48 patients who had a mean of 3.5 f 2.4 blood samples obtained for mycobacterial culture, mycobacteremia was found in 18 cases. Blood was the only extrapulmonary site from which M. tuberculosis was isolated in 10 cases. When results of all acid-fast smears and histologic studies were reviewed, a positive acid-
454
October
1990
The American
Journal
of Medicine
Volume
89
TUBERCULOSIS
generalized lymphadenopathy [7], and should suggest the possibility of tuberculosis. We believe that empiric antituberculous therapy is warranted in all HIV-infected patients in whom chest roentgenographic abnormalities suggestive of mycobacterial infection are unexplained by diseases other than tuberculosis. Although these roentgenographic characteristics can be due to nontuberculous etiologies, clinical features and laboratory test results often suggest these diagnoses. Hilar adenopathy and pleural effusion+ may be manifestations of Kaposi’s sarcoma or lymphoma, but there is usually extrapulmonary evidence of these diseases [8,9]. Coccidioidomycosis and histoplasmosis can cause chest roentgenographic abnormalities mimicking tuberculosis, but diffuse infiltrates are more typical, extrapulmonary manifestations are prominent, and results of serologic tests for fungal infection are often positive [lo-121. Although pulmonary Mycobacterium auium complex infection may be roentgenographically indistinguishable from tuberculosis, other clinical features can separate these two infections [13]. Empiric treatment of tuberculosis does not substitute for a diagnostic evaluation, but is warranted until a definitive diagnosis is made. Antituberculous medications are well tolerated by HIV-infected patients [14], and a therapeutic trial is often diagnostically useful. Eightyfive percent (22 of 26) of HIV-infected patients with tuberculosis defervesce during the first week of therapy (Francoise Kramer, M.D., unpublished data), similar to findings in immunocompetent patients [15]. Because tuberculosis has declined in prevalence, most physicians have little experience with the illness, and are therefore often reluctant to treat tuberculosis in the absence of definitive evidence of mycobacterial disease. However, the frequency of positive acid-fast sputum smears is lower in HIVinfected patients than in immunocompetent adults [16]. At our hospital, acid-fast sputum smears are positive in 84% of non-HIV-infected patients with pulmonary tuberculosis [17], compared to 61% in the present series. This is not surprising, as roentgenographic manifestations of tuberculosis in HIVinfected patients are generally those of primary tuberculosis, in which acid-fast sputum smears are rarely positive. Therapy of primary tuberculosis in children is based on clinical features, and we favor a similar approach in HIV-infected patients. In order to facilitate the diagnosis of tuberculosis, it is important to determine the yield of diagnostic tests for mycobacteriosis in HIV-infected patients. Acid-fast sputum smears are positive in 45% to 82% of HIV-infected patients with pulmonary tubercu-
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
losis [13,16] and should be obtained in all patients with clinical evidence of pulmonary disease. The frequency of positive sputum smears does not correlate with chest roentgenographic findings, and positive smears are seen in some patients with normal chest roentgenograms [16]. In the present study, the yield of acid-fast smears and histologic findings of bronchoscopic specimens was low, compared to that in non-HIV-infected patients [18,19]. Nevertheless, bronchoscopy was useful in excluding nontuberculous processes. Although acid-fast stool smears are often used to diagnose M. auium complex infection in HIV-infected patients [20], their value in diagnosing tuberculosis has not been emphasized. Acid-fast smears of stool are positive in 40% of HIV-infected patients with tuberculosis ([ 131, present study). Although the number of patients evaluated was small, we found no correlation between positive acid-fast stool smears and a history of diarrhea. Thus, stool smears for acid-fast bacilli may be a useful noninvasive means to diagnose tuberculosis, even in the absence of diarrhea. Blood cultures yielded M. tuberculosis in 38% (18 of 48) of our patients and should be performed in all HIV-infected patients in whom tuberculosis is suspected. When the Isolator lysis-centrifugation system, combined with the BACTECTM radiometric system (Johnston Laboratories, Towson, Maryland),for mycobacterial culture and identification, is used, M. tuberculosis can be identified in less than 3 weeks [21]. Diagnostic delays would probably have been reduced in our patients if nucleic acid probes had been used for identification of mycobacteria. Blood was the only extrapulmonary source of M. tuberculosis in 10 cases. In HIV-infected patients, it is important to document extrapulmonary tuberculosis because this condition constitutes an AIDS-defining illness, whereas pulmonary tuberculosis does not. Tuberculin skin testing was performed in only 29% (15 of 52) of our patients, probably because many physicians believe that HIV-infected patients rarely develop reactivity on skin tests. However, 11% to 57% of HIV-infected patients with tuberculosis have significant tuberculin skin test results (more than 10 mm induration) [22-251. The frequency of reactive tests is highest (71%) when tuberculosis occurs more than 2 years prior to the diagnosis of AIDS, declining to 33% when tuberculosis occurs at or after the time of AIDS diagnosis [22]. Only one of our 15 patients had a reactive skin test, perhaps reflecting the mean CD4+ cell count of only 0.11 X log/L (110/mm3). We used 10 mm of induration as the criterion for a significant tubercu-
October
1990
The American
Journal
of Medicine
Volume
89
455
TUBERCULOSIS
IN PATIENTS
WITH HIV INFECTION
/ KRAMER
ET AL
lin skin test result because values of less than 10 mm were often not quantified in the medical record. However, in HIV-infected patients, 5 mm of induration should be considered significant [26], and application of this criterion would increase the test’s sensitivity. The most important finding of this study is that the diagnosis of tuberculosis in HIV-infected patients was delayed in 48% of cases. It is unlikely that this is because physicians at our institution are relatively inept at recognizing tuberculosis. First, 12% (six of 52) of our HIV-infected patients with tuberculosis were undiagnosed prior to death, compared to 18% (32 of 175) of those in Los Angeles County (Paul Davidson, M.D., personal communication) and 26% to 28% of those reported from other institutions [24,27]. Second, of 186 non-HIV-infected patients with pulmonary tuberculosis evaluated at our hospital, a diagnostic delay of 3 weeks or more occurred in only 13 patients (7%) and only five patients (2.6%) died with untreated tuberculosis (Peter Barnes, M.D., unpublished data). These percentages compare favorably with published reports on the accuracy of diagnosing tuberculosis [28,29]. Thus, we believe that the high frequency of delayed diagnosis of tuberculosis in HIV-infected patients is not unique to our institution, and is likely to be representative of the experience at other hospitals and in private-practice settings. Increased awareness of the clinical manifestations of tuberculosis and wider use of empiric antituberculous therapy should decrease morbidity and mortality from this treatable complication of HIV infection.
6. Drew WL. Cytomegalovlrus infection in patients with AIDS. J Infect Dis 1988; 158: 449-56. 7. Abrams DI. Lewis BJ. Beckstead JH, Casavant CA, Drew WL. Persistent diffuse lymphadenopathy in homosexual men: endpoint or prodrome? Ann Intern Med 1984; 100: 801-8. 8. Meduri GU, Stover DE, Lee M, Myskowski PL. Caravelll JF. Zaman MB. Pulmonary Kaposi’s sarcoma in the acquired immune deficiency syndrome, Clinical, radiographic and pathologic manifestations. Am J Med 1986; 81: 11-8. 9. O’Brien RF, Cohn DL. Serosanguineous pleural effuslons in AIDS-associated Kaposi’s sarcoma. Chest 1989; 96: 460-6. 10. Bronnimann DA, Adam RD. Galgiani JN. et a/. Coccidioldomycosis in the acquired immunodeficiency syndrome. Ann Intern Med 1987; 106: 372-9. 11. Wheat LJ, Slama TG, Zeckel ML. Histoplasmosis in the acquired immune deficiency syndrome. Am J Med 1985; 78: 203-10. 12. Johnson PC, Khardori N, Najjar AF, Butt F, Mansell PWA. Sarosi GA. Progressive disseminated histoplasmosis in patients with acquired lmmunodeficiency syndrome. Am J Med 1988; 85: 152-8. 13. Modilevsky T, Sattler FR, Barnes PF. Mycobacterial disease in patients with human immunodeficiency virus Infection. Arch Intern Med 1989; 149: 2201-5. 14. Pitchenik AE. Fertel D. Bloch AB. Mycobacterial disease: epidemiology, diagnosis, treatment, and prevention. Clin Chest Med 1988; 9: 425-41. 15. Barnes PF, Chan LS, Wong SF. The course of fever during treatment pulmonary tuberculosis. Tubercle 1987; 68: 255-60.
16. Klein NC, Duncanson FP. Lenox TH Ill, Pitta A, Cohen SC, Wormser GP. Use of mycobacterial smears in the diagnosis of pulmonary tuberculosis in AIDS/ ARC patients. Chest 1989; 95: 1190-2. 17. Barnes PF, Verdegem TD, Vachon IA, Leedom JM, Overturf GD. Chest roentgenogram in pulmonary tuberculosis. New data on an old test. Chest 1988; 94: 316-20. 18. So SY, Lam WK. Yu DYC. Rapid diagnosis of suspected pulmonary tuberculoSIS by fiberoptic bronchoscopy. Tubercle 1982; 63: 195-200. 19. Chan JC, So S-Y. Lam WK, Ip MSM. High incidence of pulmonary tuberculoSIS in non-HIV infected immunocompromised patients in Hong Kong. Chest 1989; 96: 835-9. 20. Kiehn TE. Edwards FF. Brannon P, et al. InfectIons caused by Mycobacterium avium complex in immunocompromised patients: diagnosis by blood culture and fecal examination, antimicrobial susceptibility tests, and morphological and seroagglutination characteristics. J Clin Microbial 1985; 21: 168-73. 21. Witebsky FG. Keiser JF. Conville PS. et al. Comparison of BACTEC 13A and Du Pont isolator for detection of mycobacteremia. J Clin Microbial 1988; 26: 1501-5. 22. Rieder HL, Cauthen GM, Bloch AB. et al. Tuberculosis and acquired Immunedeficiency syndrome-Florida. Arch Intern Med 1989; 149: 1268-73. 23. Pitchenik AE, Burr J, Suarez M. Fertel D. Gonzalez G. Moas C. Human T-cell lymphotroplc wrus-III (HTLV-III) seropositivity and related disease among 71 consecutive patients in whom tuberculosis was diagnosed. A prospective study. Am Rev Respir Dis 1987; 135: 875-9. 24. Louie E. Rice LB, Holzman RS. Tuberculosis in non-Haitian patients with acquired tmmunodeficiency syndrome. Chest 1986; 90: 542-5. 25. Chaisson RE. Schecter GF. Theuer CP. Rutherford GW, Echenberg DF, Hopewell PC. Tuberculosis in patients with the acquired lmmunodeficiency syndrome. Am Rev Respir Dis 1987; 136: 570-4. 26. Centers for Disease Control. Tuberculosis and human immunodeficiency virus infection: recommendations for the Advisory Committee for the Elimination of Tuberculosis (ACET). MMWR 1989; 38: 236-8. 27. Sunderam G. McDonald RJ. Maniatis T, Oleske J. Kapila R. Reichman LB. Tuberculosis as a manifestation of the acquired immunodeficiency syndrome (AIDS). JAMA 1986: 256: 362-6. 28. Greenbaum M, Beyt BE Jr, Murray PR. The accuracy of diagnosing pulmonary tuberculosis at a teaching hospital. Am Rev Respir Dis 1980; 121: 477-81. 29. Counsel1 SR, Tan JS. Dittus RS. Unsuspected pulmonary tuberculosis In a community hospital. Arch Intern Med 1989; 149: 1274-8.
ACKNOWLEDGMENT We thank Linda Vachon. M.D., for her expert review of the chest roentgenograms, Ken Anderson, MS., CLINFD systems manager, for assistance in data analysis, William Lewis, Ph.D.. for assistance in identifying the patients included in this report, and Joseph Indenbaum. M.D., and Om Sharma. M.D., for their thoughtful review of the manuscript.
REFERENCES 1. Barnes PF, Arevalo C. Six cases of Mycobacterium Infect Dis 1987; 156: 377-9.
tuberculosis
bacteremia.
J
2. Sommers HM. Good RC. Mycobacterium. In: Lennette EH. Balows A, Hausler WJ Jr, Shadomy HJ. eds. Manual of clinical microbiology. Washington: American Society for Microbiology, 1985: 216-48. 3. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987; 36(1S): 3-155. 4. Barnes PF. Leedom JM. Chan LS, et a/. Predictors of short-term prognosis in patients with pulmonary tuberculosis. J Infect Dis 1988; 158: 366-71. 5. Hopewell PC. Pneumocystis carinii pneumonia: diagnosis. J Infect DIS 1988; 157: 1115-9.
456
October
1990
The American
Journal
of Medicine
Volume
of
89
