661

A Prospective Study of Tuberculosis and Human Immunodeficiency Virus

Infection: Clinical Manifestations and Factors Associated with Survival
Peter L. Alpert, Sonal S. Munsiff, Marc N. Gourevitch, From the Division of Infectious Diseases, Department of Medicine, and
Barbara Greenberg, and Robert S. Klein the Department of Epidemiology and Social Medicine, Montefiore
Medical Center and the Albert Einstein College of Medicine, Bronx,
New York

We prospectively studied the effect of human immunodeficiency virus (HIV) infection on the
presentation and outcome of tuberculosis. A total of 216 patients with tuberculosis were identified;
162 (75%) of these patients were tested for antibodies to HIV; 92 (57%) were seropositive. The
patients who were seropositive for HIV were more likely to be male and Hispanic and to have been

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

homeless or incarcerated. Eighty-one percent of these patients had CD4 lymphocyte counts of
200/mm 3 . The seropositive patients had extrapulmonary tuberculosis more often than did the
seronegative patients (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.2-4.8). Smears for acid-
fast bacilli were positive more often for non-HIV-infected patients with pulmonary tuberculosis (74.5%)
than for HIV-infected patients (54.3%) [OR, 2.46; 95% CI, 1.01-6.02]—even those with focal or
cavitary disease. A delay in initiating therapy was associated with in-hospital mortality: the median
time from admission to the start of treatment was 4 days for patients who survived and 15 days for
those who died (P = .02). The median survival was 22.7 months for HIV-infected patients who did
not die during the initial hospitalization. Factors independently associated with reduced rates of
survival included the severity of immunodeficiency, nonuse of directly observed therapy, infection due
to drug-resistant Mycobacterium tuberculosis, and a history of injection drug use.

After decades of decline, the annual incidence of tuberculosis of patients with TB were included; these studies were restricted
(TB) in the United States rose each year from 1985 to 1992 according to gender, risk behaviors, anatomic site of tuberculo-
[1]. Factors contributing to this increased incidence included sis, drug susceptibility, or stage of HIV disease; data collection
the epidemic of HIV infection, an increase in the frequency was retrospective or not stated as being prospective; there was
of antibiotic-resistant Mycobacterium tuberculosis isolates, a either no information or limited information about risk factors
decline in the public health infrastructure, the spread of TB in for TB; no seronegative comparison group was included; there
congregate settings such as homeless shelters and jails, and was little or no information about drug susceptibility; or the
immigration from countries with high rates of TB [2, 3]. The sample size was limited (50 subjects) [2, 4, 6-20]. Although
presence of HIV infection greatly increases the risk that indi- there are substantial data showing that rates of mortality and
viduals with longstanding or recently acquired M. tuberculosis morbidity are higher among patients with active TB and con-
infection will develop active disease [4, 5]. comitant HIV infection than among non-HIV-infected patients
Numerous clinical studies of TB in HIV-infected patients in with tuberculosis [21-23], information about the factors that
the United States have been reported previously. However, influence survival among HIV-infected patients is limited [24,
many of these studies had substantial limitations: only subsets 25]. Knowledge of such factors offers the potential for optimiz-
ing interventions to improve survival.
To study the impact of HIV infection on the presentation of
TB and to identify factors affecting survival of coinfected pa-
Received 30 May 1996; revised 5 December 1996. tients at our institutions in the Bronx, New York City (a com-
This work was presented in part at the 1994 Infectious Diseases Society of
America annual meeting held on 4-7 October in Orlando, Florida, and in part munity beset by the converging epidemics of these two dis-
at the 3rd Conference on Human Retroviruses and Opportunistic Infections eases), we prospectively studied all patients who presented with
held in January 1996 in Washington, D.C. tuberculosis at our medical centers, and we followed up the
Informed consent was obtained from the patients and the guidelines of the
Institutional Review Board for Human Subjects of Montefiore Medical Center HIV-infected patients after hospital discharge.
were followed in the conduct of the research.
This work was supported by cooperative agreement No. U64/CCU200714
with the Centers for Disease Control and Prevention.
P.L.A. was supported in part by a training grant from the National Institutes Methods
of Health (5-T32-AI070183).
Reprints or correspondence: Dr. Peter L. Alpert, Division of Infectious From July 1992 through June 1995, we prospectively studied
Diseases, Montefiore Medical Center, 111 East 210th Street, Bronx, New York all patients at two hospitals in the Bronx, New York City, for
10467. whom cultures of any anatomic site yielded M tuberculosis.
Clinical Infectious Diseases 1997;24:661-8 The hospitals were the Moses Division of Montefiore Medical
© 1997 by The University of Chicago. All rights reserved.
1058-4838/97/2404 — 0017$02 .00 Center, a voluntary medical center that is the largest provider
662 Alpert et al. CID 1997; 24 (April)

of care to residents of the Bronx, and its physically adjacent from North Central Bronx Hospital against second-line antitu-
municipal affiliate, North Central Bronx Hospital. After identi- berculous drugs was performed either at the National Jewish
fication, hospitalized patients were interviewed by two (P.L.A. Center for Immunology and Respiratory Medicine (Denver) or
and S.S.M.) of the authors and followed up prospectively until at the Mycobacteriology Laboratory of the New York City
death or hospital discharge. HIV-infected patients were fol- Department of Health.
lowed up prospectively, as detailed below. Patients not known to have HIV infection were offered
A standardized interview was conducted to ascertain demo- HIV counseling and testing (EIA with western blot confir-
graphic factors, past behavioral risk factors, and medical histor- mation), as recommended for all patients with TB [28].
ies. For Spanish-speaking patients, interviews were conducted T cell subset determinations were recommended for all HIV-
in Spanish, and for children, interviews were conducted with infected patients. When patients had not had T cell subset
parents. We reviewed the medical records of patients pre- determinations made at the time that TB was diagnosed,
viously seen at our institutions for documentation of prior we ascertained the last recorded CD4 cell count before the

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

AIDS-defining clinical conditions and TB. diagnosis of TB was made.
Alcoholism was defined as the consumption of > 12 drinks The dates of hospital admission and initiation of antituberculous
per week, and homelessness was defined as the lack of a home therapy were recorded. Patients were treated for TB as directed
for _-2 weeks within the preceding 6 months. For patients who by their primary care physicians. For analysis of the interval
refused to be interviewed or who died before the diagnosis of between hospital admission and initiation of therapy, we used the
TB was made, alcoholism and homelessness were ascertained interval between admission and the thy after death for individuals
by reviewing the medical charts. We reviewed the medical who died without receiving any antituberculous therapy. All am-
records and recorded the results of tests including smears, his- bulatory patients with TB were offered directly observed therapy
tology, cultures, antimicrobial susceptibility testing, and chest (DOT). DOT was defined as the administration of antituberculous
radiographs. medication by a trained health care worker.
Specimens were processed for isolation of mycobacteria by Follow-up was arranged whenever possible in the clinics of
means of standard methods [26]. Concentrated sputum smears the medical center. If HIV-infected patients were being fol-
were stained with auramine-rhodamine, and those found to be lowed up by private physicians, these clinicians were contacted
positive for acid-fast bacilli were examined with a Kinyoun periodically for information about their patients' conditions. If
carbolfuchsin stain for confirmation. All cultures were inocu- patients were rehospitalized, we reviewed their medical records
lated into Bactec 12B bottles (Becton-Dickinson, Sparks, MD), to obtain information on possible relapses of TB and the devel-
incubated for 6 weeks at 37°C, and read on the basis of a 7- opment of other opportunistic infections.
day protocol, according to the recommendations of the manu- At the end of the study period, we again reviewed all records
facturer [27]. Specimens were also inoculated onto slant cul- at the two hospitals to ensure that no hospitalizations had been
tures containing LOwenstein-Jensen culture medium, incubated overlooked and to verify the T cell subset data. We searched
for 6-8 weeks, and examined weekly. All mycobacterial iso- the Death Registry and the Tuberculosis Registry of New York
lates were identified by using criteria described by the Centers City to obtain additional information on patients lost to follow-
for Disease Control and Prevention (CDC) [26], by using ge- up and to determine whether any of the patients had previously
netic probes, or by using both methods. been reported as having tuberculosis. All patients who survived
Isolates of M. tuberculosis complex were tested for suscepti- were censored from the study on 23 May 1995. Death was
bility to first-line antituberculous drugs (i.e., isoniazid, strepto- attributed to TB if a culture for M tuberculosis had been posi-
mycin, rifampin, ethambutol, and pyrazinamide [provided by tive in the month before death and there was no other identified
Becton-Dickinson]) with the Bactec 460 system, according to cause of death. Autopsies were not routinely performed.
the recommendations of the manufacturer [27]. The susceptibil- Data management and analysis were performed with dBASE
ity patterns of isolates that were resistant to either isoniazid or (Borland International, Scotts Valley, CA) software and Epi-
rifampin and of those that were resistant to two or more of any Info Version 6 software (CDC). Categorical variables were
first-line antituberculous drugs were confirmed by using the analyzed with use of the x 2 test or Fisher's exact test where
agar proportion technique [26]. appropriate; continuous variables were analyzed with use of
For isolates recovered at Montefiore Medical Center that the Kruskall-Wallis test. Findings were considered statistically
were resistant to one or more first-line agents, susceptibility to significant when the P value was .05. All tests were two-
second-line drugs was tested on Middlebrook 7H10 medium tailed. For selected analyses, we calculated odds ratios and
at the following concentrations: p-aminosalicylic acid, 2 p,g/mL 95% confidence intervals. We analyzed survival by the method
(Becton-Dickinson); ethionamide, 5 pg/mL, and kanamycin, of Kaplan-Meier; we performed multivariate analyses, which
6 ktg/mL (BBL Microbiology Systems, Cockeysville, MD); were based on the Cox regression model, with use of the SPSS
cycloserine, 10 p,g/mL before 1994 and 30 ug/mL since 1994; 6.1 for Windows statistics program (SPSS Inc., Chicago). Inter-
and capreomycin, 10 fig/mL (Sigma Chemical Co., St. Louis, action effects were evaluated, and the resulting hazard ratios
MO). Any additional testing of such isolates and of all isolates and 95% confidence intervals were obtained.
CID 1997;24 (April) Tuberculosis and HIV Infection 663

Results The anatomic sites of TB among HIV-infected patients

From July 1992 through June 1995, 221 patients had were as follows: the lungs, 40 patients (43.5%); sites other
M tuberculosis isolated from one or more specimens at our than the lungs, 20 (21.7%); or both sites, 32 (34.8%). Among
two institutions. Five patients were excluded from the analysis seronegative patients, the distribution was as follows: the
for the following reasons: only one culture was positive for lungs, 45 (64.3%); sites other than the lungs, 18 (25.7%); or
M tuberculosis; there was a strong suspicion that the results both sites, 7 (10.0%). HIV-infected patients were more likely
were erroneous (in two cases there was evidence of mislabeling to have extrapulmonary disease (OR, 2.3; 95% CI, 1.2-4.8).
of a sample from a roommate); the primary care physicians HIV-infected patients with extrapulmonary infection were
concluded that the isolate did not represent a true infection; more likely to have concomitant pulmonary disease than were
and although no treatment for TB was given, there was no non-HIV-infected patients with extrapulmonary infection
further suggestion that the patients were infected. (OR, 4.2; 95% CI, 1.3-14.3). CD4 cell counts did not differ
Of the 216 patients included in this report, most were either between HIV-infected patients with extrapulmonary tubercu-

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

Hispanic (44%) or African-American (37%). One hundred losis—with or without concomitant pulmonary involvement
ninety-seven patients (91.2%) were interviewed; 11 died in (median CD4 cell count, 75.5/mm3 ; range, 1 –424/mm 3 )—
the hospital before they were interviewed, three could not be and those with pulmonary disease alone (median CD4 cell
interviewed because of altered mental status, and five were count, 71/mm 3 ; range, 0-1,362/mm 3 ; P = .6).
discharged from the hospital and lost to follow-up before they Of the 124 patients with pulmonary tuberculosis, 117
could be interviewed. For these 19 patients, historical informa- (94.4%) had positive sputum cultures; bronchoscopic speci-
tion was abstracted from the medical records. mens, gastric aspirates, or lung biopsy specimens were positive
The median age of the patients was 38 years (range, for the seven remaining patients. Since none of these patients
<1-95 years). Two hundred eleven patients (97.7%) were at had pharyngeal or laryngeal tuberculosis alone, all were as-
least 15 years of age (median age, 39 years). Seventeen percent sumed to have pulmonary involvement regardless of the find-
of these patients had been homeless, 21% had a history of ings on chest radiographs. Focal infiltrates, upper-lobe infil-
alcoholism, and 19% had been incarcerated. A history of psy- trates, and cavitary disease were each significantly less likely
chiatric illness was reported by 5% of those interviewed. to be present in HIV-infected patients (table 2). Intrathoracic
Sixty-seven (31%) of the 216 patients had previously been lymphadenopathy was seen on the chest radiographs of HIV-
found to be infected with HIV. Of the remaining 149 patients, infected patients significantly more frequently than on the ra-
95 (63.8%) underwent testing for antibodies to HIV, 25 diographs of non-HIV-infected patients. The findings on chest
(26.3%) of whom were found to be seropositive. Therefore, of radiographs were normal for a small proportion of patients with
the entire sample, 92 patients (42.6%) were seropositive for pulmonary tuberculosis, with or without HIV infection.
HIV, 70 (32.4%) were seronegative, and 54 (25%) did not Acid-fast smears of sputum were performed for 64 (88.9%)
undergo testing. There were no significant differences with of 72 HIV-infected patients and 43 (82.7%) of 52 non-HIV-
respect to age, history of homelessness, or history of prior infected patients with pulmonary tuberculosis. Acid-fast stains
episodes of TB when patients of known HIV status were com- were positive significantly more often for HIV-seronegative
pared with those of unknown status. However, the former pa- patients with positive sputum cultures (74.5% of patients) than
tients were significantly more likely to be male, to have a for HIV-infected patients (54.3%) (OR, 2.46; 95% CI,
history of alcoholism, and to be Hispanic than were the latter 1.01-6.02).
patients. Among patients with pulmonary tuberculosis and positive
Demographic characteristics and possible risk factors for sputum cultures who had either focal infiltrates or cavities ap-
tuberculosis, according to HIV status, are shown in table 1. parent on chest radiographs, non-HIV-infected patients were
HIV-infected patients were more likely to be Hispanic, to have significantly more likely to have positive sputum acid-fast
been homeless or incarcerated, and to be male. They were not smears than were HIV-infected patients (34 of 41 patients
more likely to have a history of tuberculosis or alcoholism. [82.9%] vs. 20 of 39 [51.3%], respectively; OR, 4.6; 95% CI,
The HIV-infected patients were also more likely to have been 1.5 –14.9). Additional smears were performed for 34 (87.2%)
injection-drug users or to have had same-sex partners (data not of 39 HIV-infected patients with pulmonary TB vs. 20 (87.0%)
shown). of 23 non-HIV-infected patients with pulmonary TB; therefore,
Fifty HIV-infected patients (54.3%) had had their T cell differences in the rates of positive smears according to HIV
subsets measured at the time that TB was diagnosed. The me- status cannot be explained by the fact that more smears were
dian CD4 cell count was 71/mm 3 (range, 5-1,362/mm 3 ), and done for non-HIV-infected patients.
CD4 cell counts were .--_.200/mm 3 for 81% of these patients. The antimicrobial susceptibilities of the M tuberculosis iso-
Among an additional 35 HIV-infected patients whose T cell lates are shown in table 3. HIV-infected patients tended to be
subset levels were not determined when TB was diagnosed, infected somewhat more often with drug-resistant isolates, but
the median CD4 cell count before the diagnosis of TB was this difference approached statistical significance only for
49.5/mm 3 (range, 0-534/mm 3 ). rifampin (P = .09). Of the isolates that were resistant to both
664 Alpert et al. CID 1997;24 (April)

Table 1. Demographic characteristics and risk factors for tuberculosis according to HIV status.

HIV status

Seropositive Seronegative Unknown

Variable (n = 92) (n = 70) (n = 54) P value*

Gender
Male 66 (71.7) 39 (55.7) 23 (42.6)
Female 26 (28.3) 31 (44.3) 31 (57.4) .051
Race
White 5 (5.4) 6 (8.6) 8 (14.8)
Black 36 (39.1) 25 (35.7) 18 (33.3)
Hispanic 51 (55.4) 28 (40.0) 16 (29.6)

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

Other 0 11 (15.7) 12 (22.2) <.001
Nationality
American 46 (50.0) 31 (44.3) 17 (31.5)
Puerto Rican 25 (27.2) 6 (8.6) 6 (11.1)
African 3 (3.3) 3 (4.3) 5 (9.3)
Asian 0 2 (2.9) 10 (18.5)
Other/unknown 18 (19.6) 28 (40.0) 16 (29.6) .000
History of homelessness
No 55 (69.6) 60 (88.2) 40 (88.9)
Yes 24 (30.4) 8 (11.8) 5 (11.1) <.02
History of alcoholism
No 62 (73.8) 48 (72.7) 40 (88.9)
Yes 22 (26.2) 18 (27.3) 5 (11.1) NS
History of incarceration
No 45 (58.4) 60 (88.2) 41 (95.3)
Yes 32 (41.6) 8 (11.8) 2 (4.7) <.001
History of prior tuberculosis
No 78 (86.7) 57 (81.4) 46 (86.8)
Yes 12 (13.3) 13 (18.6) 7 (13.2) NS
Median age in y (range) 37 (21-55) 40.5 (15-83) 41.5 (<1-95) NS

NOTE. Data are number of patients (%) unless otherwise indicated. NS = not significant.
* For comparing HIV-seropositive patients with HIV-seronegative patients.
HIV-seronegative patients more commonly reported nationalities other than American or Puerto Rican.

isoniazid and rifampin (i.e., multidrug-resistant isolates), 15 profloxacin and capreomycin; in contrast, the proportion of
(93.8%) of 16 were susceptible to three or more alternative these isolates that were resistant to the first-line antimicrobial
antimicrobials, and the remaining isolate was susceptible to agents (streptomycin, ethambutol, and pyrazinamide) equaled
two alternative drugs. The small number of multidrug resistant or exceeded 50 percent. Multidrug-resistant isolates were re-
isolates tested in this study were uniformly susceptible to ci- covered from five (20.0%) of 20 patients with a history of TB
vs. 11 (8.1%) of 136 without a history of TB (P = .08).
Factors associated with in-hospital mortality are shown in
Table 2. Radiographic findings for patients with pulmonary tuber- table 4 for patients hospitalized at the time that TB was diag-
culosis, according to HIV status. nosed. A trend towards increased mortality was observed for
both HIV-infected patients and those with multidrug-resistant
HIV status TB. A longer delay between admission and institution of anti-
microbial therapy was also associated with increased mortality.
Seropositive Seronegative
Finding (n = 72) (n = 52) P value
We included three outpatients and 74 HIV-infected patients
who had enrolled in the study by 1 March 1995 and who survived
Focal infiltrate 38 (53) 46 (89) <.01 initial hospitalization in the analyses of survival of patients who
Upper-lobe infiltrate 19 (26) 32 (62) <.01 did not die during the initial hospitalization. There were no sig-
One or more cavities 5 (7) 23 (44) <.01 nificant demographic differences between these patients and the
Hilar or mediastinal HIV-infected patients who died during hospitalization. The esti-
lymphadenopathy 28 (39) 6 (12) <.01
8 (11) 3 (6) NS
mated median survival was 22.7 months for these 77 patients
Normal
(figure 1). Thirty-two patients (41.6%) died during follow-up; at
NOTE. Data are number of patients (%). NS = not significant. least 11 (34.4%) of these deaths were attributable to TB. Univari-
CID 1997;24 (April) Tuberculosis and HIV Infection 665

Table 3. Antimicrobial susceptibilities of Mycobacterium tubercu- 1.0

losis isolates recovered from patients with tuberculosis, according to
HIV status. o

c
*+-
0
0.8 -
Ca
No. of isolates not susceptible to
CS
)
N
To
indicated drug/total no. of isolates (%)
an P- an
co -

Seropositive Seronegative O2 0
E ?,
Drug patients patients Co .c 76
CD F. 0.4 -
.c
cn
Isoniazid 21/92 (22.8) 13/70 (18.6) c a)
o
Rifampin 16/92 (17.4) 5/70 (7.1) 0.2 -
Isoniazid + U)

rifampin* 11/92 (12.0) 5/70 (7.1) 2c

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

Ethambutol 5/92 (5.4) 3/70 (4.3) 0
Streptomycin 12/92 (13.0) 5/70 (7.1) 0 5 10 15 20 25 30 35
Pyrazinamide 5/43 (11.6) 3/22 (13.6) No. of months after initial hospitalization
Ethionamide 3/23 (13.0) 2/9 (22.2)
Figure 1. Kaplan-Meier estimate of the proportion of survivors
p-Aminosalicylic acid 2/23 (8.7) 1/10 (10.0)
among 77 HIV-seropositive patients with tuberculosis who did not
Cycloserine 2/11 (18.2) 1/4 (25.0)
die during an initial hospitalization.
Amikacin 3/9 (33.3) 0/3
Kanamycin 3/9 (33.3) 0/3
Ciprofloxacin 0/9 0/9
use of antiretrovirals at the time of hospitalization or during
Capreomycin 0/7 0/4
follow-up was noted.
* Multidrug resistant. We constructed a model for multivariate analysis based on
age (-35 years or >35 years), antimicrobial susceptibility
(fully susceptible vs. resistant to isoniazid and/or rifampin),
ate analysis showed that decreased survival was associated with CD4 cell count (--.100/mm 3 , 101-200/mm 3 , and >200/mm 3 ),
the following factors: a history of injection drug use; infection use of DOT, and history of injection drug use. A CD4 cell count
with isolates resistant to isoniazid, rifampin, or both; nonuse of of .-100/mm 3 , nonuse of DOT, recovery of an M tuberculosis
DOT; low CD4 cell counts; and the presence of AIDS prior to isolate that was resistant to isoniazid and/or rifampin, and a
the diagnosis of TB. history of injection drug use were each independently associ-
Injection drug users did not differ significantly from non- ated with decreased survival (table 5). Age was not associated
injection drug users in terms of the anatomic site of TB, receipt with survival in the final model.
of DOT, antimicrobial resistance of the M tuberculosis isolate,
use of antiretrovirals, baseline CD4 cell counts, or site of fol- Discussion
low-up care. No difference in survival according to age, gender, Although previous studies have described series of patients
race, anatomic site of TB, current injection drug use, or the with TB and HIV infection, the design of our study, which

Table 4. Survival among hospitalized HIV-infected patients and non-HIV-infected patients with tuber-
culosis.

No. of patients with indicated outcome/

total no. of patients (%)

Variable Discharged Died P value

Susceptibility of Mycobacterium
tuberculosis isolate
Susceptible* 155/172 (90.1) 17/172 (9.9)
Multidrug resistant 14/18 (77.8) 4/18 (22.2) .12
HIV status
HIV seronegative 57/60 (95.0) 3/60 (5.0)
HIV seropositive 76/89 (85.4) 13/89 (14.6) .11
Median delay in therapy in
d (range)t 4 (0-63) 15 (0-77) .02

* M tuberculosis isolate susceptible to isoniazid, rifampin, or both.

I The number of days between admission and institution of therapy for tuberculosis for hospitalized patients,
excluding those with multidrug-resistant infections.
666 Alpert et al. CID 1997;24 (April)

Table 5. Factors independently associated with survival among 77 among our patient population and the fact that many of our
HIV-infected patients with tuberculosis who did not die during their HIV-infected patients received routine TB prophylaxis and thus
initial hospitalizations. did not develop TB until the onset of severe immunodeficiency
Hazard ratio [34]. Finally, we designed our study to include only patients
Factor for death 95% CI with TB that was confirmed by culture at our medical centers,
which likely resulted in the exclusion of some patients with
CD4 cell count of -100/mrn 3 2.07 1.11-3.86 TB and less advanced HIV disease.
Nonuse of directly observed therapy 1.94 1.23-3.06
The prompt diagnosis of TB is important for optimal treat-
Recovery of resistant
Mycobacterium tuberculosis 1.87 1.19-2.95 ment as well as for institution of necessary infection control
History of injection drug use 1.68 1.14-2.47 procedures, to protect both health care workers and other pa-
tients [35]. We found that TB involving the lungs was equally
frequent among patients with HIV infection and those who

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

were seronegative for the virus. However, acid-fast smears of
included a substantial sample as well as women, patients with sputum were less likely to be positive for HIV-infected pa-
varying risk factors, HIV-seronegative patients for comparison, tients—even those with focal or cavitary disease potentially
and prospective evaluation and follow-up of patients, renders contributing to a delay in diagnosis and treatment.
its conclusions more rigorously scientific. Patients with HIV infection were more likely to have ex-
We were able to document HIV status for three-quarters of trapulmonary disease, alone or in combination with pulmo-
our sample. Forty-three percent of all patients (57% of those nary tuberculosis. Regardless of HIV status, >20% of our
whose HIV status was known) were infected with HIV. HIV- patients had extrapulmonary infection without accompa-
infected patients were more likely to be male and Hispanic; nying lung involvement. Such patients often present with
both of these characteristics are associated with HIV infection nonspecific signs and symptoms, and the diagnosis of TB
in the Bronx [29]. Seronegative patients were more than three may be difficult. Clinicians should suspect TB in all patients
times as likely to identify themselves as a nationality other with compatible clinical syndromes; this is especially true
than American or Puerto Rican than were HIV-seropositive for those with HIV infection because they are more likely
patients; this finding suggests that the association between TB to have atypical presentations.
and foreign birth is less important for HIV-infected persons, as A recent report from the health department in Miami sug-
has been noted previously [12]. A history of prior incarceration, gests that patients with TB and HIV infection are less infectious
known to be associated with both the risk for TB [3] and illicit to contacts than are patients without HIV infection [36]. Al-
drug use, was also significantly more common among HIV- though the results of several studies have suggested that the
infected patients. rates of positive acid-fast smears of sputum are lower among
TB commonly occurs earlier in the course of HIV infection HIV-infected patients with pulmonary TB [8, 37], the results
than do many other opportunistic infections, and this finding of other studies have not demonstrated this finding [7, 9, 38].
has been ascribed to the greater virulence of TB [6, 7, 9, 30]. In the present study we found a significantly higher rate of
Median CD4 cell counts in previous reports of tuberculosis in positive sputum smears for HIV-seronegative patients whose
HIV-infected patients have ranged from < 150/mm 3 to sputum cultures were positive for M. tuberculosis than for HIV-
>300/mm 3 [9, 13, 17, 19, 31]. Lower median CD4 cell counts infected patients with positive sputum cultures. This finding
have been described for HIV-infected persons with dissemin- could not be explained by different frequencies of sputum
ated or multidrug-resistant TB [10, 15, 20]. In the present smear examinations between the two groups. A possible expla-
series, we did not observe differences in CD4 cell counts among nation is that HIV-related immunosuppression impairs the nor-
HIV-infected patients with extrapulmonary TB and those with- mal cellular response to TB, making pulmonary consolidation,
out extrapulmonary TB. Our patients generally had advanced necrosis, and cavitation (with its associated high mycobacterial
HIV disease; > 80% of those with documented HIV infection load) less likely to occur. The lower rates of focal and cavitary
met the 1993 revised case definition of AIDS that was pub- disease observed among HIV-infected patients are consistent
lished by the CDC [32], even when we did not take the presence with this hypothesis.
of TB into account. However, even among patients with focal or cavitary disease,
There are several potential explanations for the finding of those infected with HIV were significantly less likely to have
advanced disease in such a large proportion of the patients in positive sputum smears than were the non-HIV-infected pa-
our study. One explanation may be recent acquisition of tients. This finding suggests that there is an additional explana-
M tuberculosis infection, with rapid progression to disease tion for the observed differences that is not readily apparent.
among patients with advanced HIV infection; such progression Even if HIV-seropositive patients with pulmonary TB are less
was described previously for more than one-third of patients infectious than are seronegative patients, a high index of suspi-
with TB at our institutions [33]. Other explanations for this cion must be maintained for HIV-infected patients who develop
finding may be the relatively long duration of the HIV epidemic signs or symptoms suggestive of TB.
CID 1997;24 (April) Tuberculosis and HIV Infection 667

The incidence of multidrug-resistant TB in New York City for M tuberculosis or whose cultures were not performed in
has risen dramatically in recent years [39, 40], in large part our medical center laboratories may limit the extent to which
because of suboptimal compliance with therapy among patients our findings can be generalized. Patients with TB that was
with TB [2, 39]. In the present study, a history of TB was diagnosed clinically or whose positive cultures were performed
more common among patients who had multidrug-resistant TB. elsewhere were not included. Furthermore, the HIV epidemic
This difference approached statistical significance and is con- has existed longer in New York City than in many other areas
sistent with the finding that prior suboptimal therapy is a major of the country.
contributor to the development of multidrug-resistant TB. Al- Therefore, the findings of this study should be applied cau-
ternatively, this difference could have been due to a higher tiously to patients in other locations. In addition, although the
relapse rate (and subsequent eligibility for this study) among patients with TB and unknown HIV statuses were similar in
previously treated patients with multidrug-resistant disease than terms of age, history of homelessness, and history of tuberculo-
among patients with susceptible infections. sis to those with known HIV statuses, the former patients were

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

Although the rates of drug resistance were slightly higher more likely to have been male, have a history of alcoholism,
for HIV-seropositive patients than for HIV-seronegative pa- and be Hispanic. Although this raises the possibility of some
tients, the differences did not reach statistical significance with bias in our observations, the data are consistent with those from
our sample size. In the treatment of multidrug-resistant tubercu- prior reports.
losis, the best response is obtained with the administration of The results of the present study suggest that several interven-
multiple agents to which the isolate is susceptible [16]. tions have the potential for improving survival among HIV-
Fortunately, all multidrug-resistant isolates in this series infected patients with TB. Prompt diagnosis and aggressive
were susceptible to two or more alternative antimicrobial treatment of TB are important for improving survival. DOT
agents. Although the in-hospital mortality was somewhat should be implemented whenever possible, particularly for pa-
higher for patients with multidrug-resistant infections, a delay tients with drug-resistant TB. As improved antiretroviral thera-
in instituting therapy was the factor that was significantly asso- pies become available, maintaining or improving the immune
ciated with death. Therefore, a reduction in in-hospital mortal- status of patients is likely to have an important effect. Finally,
ity should be possible if a high index of suspicion leads to the treatment of substance abuse could mitigate the adverse
early diagnosis and initiation of appropriate chemotherapy. impact of the resultant complications on survival.
We found that several factors affected long-term survival.
Decreased survival was associated with severity of immuno-
deficiency, as would be expected for HIV-infected patients, Acknowledgments
even those without TB. In contrast to a recent somewhat smaller The microbiological studies were performed at Montefiore Med-
retrospective study in which the investigators demonstrated that ical Center under the direction of John McKitrick, Ph.D. and Mi-
severity of immunodeficiency was the only factor indepen- chael Levi, Ph.D. and at North Central Bronx Hospital under the
dently associated with decreased survival of HIV-infected pa- direction of Mr. Neville Trowers. The infection control staff at
tients with TB [25], our study showed that there were other both hospitals assisted in early identification of patients.
independent predictors of decreased survival, even when we
controlled for level of immunodeficiency. The association be-
tween DOT and improved survival confirms that compliance References
with treatment is important [23-25, 41]. 1. Cantwell MF, Snider DE Jr, Cauthen GM, Onorato IM. Epidemiology of
Although the injection drug users were similar to other tuberculosis in the United States, 1985 through 1992. JAMA 1994; 272:
patients in terms of demographics and in location and duration 535-9.
2. Brudney K, Dobkin J. Resurgent tuberculosis in New York City: human
of follow-up, survival was decreased among injection drug
immunodeficiency virus, homelessness, and the decline of tuberculosis
users independent of antimicrobial susceptibility and the use control programs. Am Rev Respir Dis 1991; 144:745-9.
of DOT. Previous studies have shown that the survival rates 3. Bellin EY, Fletcher DD, Safyer SM. Association of tuberculosis infection
are lower among injection drug users with HIV infection than with increased time in or admission to the New York City jail system.
among patients who are not injection drug users [42, 43], while JAMA 1993; 269:2228-31.
other studies have shown that with similar medical care, sur- 4. Daley CL, Small PM, Schecter GF, et al. An outbreak of tuberculosis
with accelerated progression among persons infected with the human
vival does not differ [44]. The reason for decreased survival immunodeficiency virus: an analysis using restriction-fragment-length-
among injection drug users with TB in the present study is polymorphisms. N Engl J Med 1992; 326:231-5.
not clear but might be the fact that the incidence of bacterial 5. Selwyn PA, Hartel D, Lewis VA, et al. A prospective study of the risk of
infections was higher among these patients [45]. Unfortunately, tuberculosis among intravenous drug users with human immunodefi-
ciency virus infection. N Engl J Med 1989; 320:545-50.
we did not prospectively assess for incident bacterial infections.
6. Pitchenik AE, Cole C, Russell BW, Fischl MA, Spira TJ, Snider DE Jr.
We recognize several limitations to the present study. While Tuberculosis, atypical mycobacteriosis, and the acquired immunodefi-
the population of the Bronx is similar to that of other inner ciency syndrome among Haitian and non-Haitian patients in south Flor-
cities, our exclusion of patients whose cultures were negative ida. Ann Intern Med 1984; 101:641-5.
668 Alpert et al. CID 1997; 24 (April)

7. Pitchenik AE, Burr J, Suarez M, Fertel D, Gonzalez G, Moas C. Human T- 26. Kent PT, Kubica GP. Public health mycobacteriology, a guide for the level
cell lymphotropic virus-III (HTLV-III) seropositivity and related disease III laboratory. Atlanta: Division of Laboratory Training and Consulta-
among 71 consecutive patients in whom tuberculosis was diagnosed: a tion, Laboratory Program Office, U.S. Department of Health and Human
prospective study. Am Rev Respir Dis 1987; 135:875-879. Services, Public Health Service, Centers for Disease Control, 1985.
8. Klein NC, Duncanson FP, Lenox TH III, Pitta A, Cohen SC, Wormser 27. Siddiqi S. Bactec 460TB system product and procedure manual. Sparks,
GP. Use of mycobacterial smears in the diagnosis of pulmonary tubercu- Maryland: Becton Dickinson, 1995.
losis in AIDS/ARC patients. Chest 1989; 95:1190-2. 28. Centers for Disease Control. Tuberculosis and human immunodeficiency
9. Theuer CP, Hopewell PC, Elias D, Schecter GF, Rutherford GW, Chaisson virus infection: recommendations of the Advisory Committee for the
RE. Human immunodeficiency virus infection in tuberculosis patients. Elimination of Tuberculosis (ACET). MMWR Morb Mortal Wkly Rep
J Infect Dis 1990; 162:8-12. 1989; 38:236-8; 243-50.
10. Clark RA, Blakley SL, Greer D, Smith MHD, Brandon W, Wisniewski 29. Friedland GH, Saltzman B, Vileno J, Freeman K, Schrager LK, Klein RS.
TL. Hematogenous dissemination of Mycobacterium tuberculosis in Survival differences in patients with AIDS. J Acquir Immune Defic
patients with AIDS. Rev Infect Dis 1991; 13:1089-92. Syndr 1991; 4:144 -53.
11. Small PM, Schecter GF, Goodman PC, Sande MA, Chaisson RE, Hopewell 30. Rieder HL, Cauthen GM, Kelly GD, Bloch AB, Snider DE Jr. Tuberculosis

Downloaded from https://academic.oup.com/cid/article/24/4/661/439971 by guest on 03 April 2024

PC. Treatment of tuberculosis in patients with advanced human immu- in the United States. JAMA 1989; 262:385-9.
nodeficiency virus infection. N Engl J Med 1991; 324:289-94. 31. Shafer RW, Chirgwin KD, Glatt AE, Dandouh MA, Landesman SH, Suster
B. HIV prevalence, immunosuppression, and drug resistance in patients
12. Onorato IM, McCray E, the Field Services Branch. Prevalence of human
with tuberculosis in an area endemic for AIDS. AIDS 1991; 5:399-405.
immunodeficiency virus infection among patients attending tuberculosis
32. Centers for Disease Control and Prevention. 1993 revised classification
clinics in the United States. J Infect Dis 1992; 165:87-92.
system for HIV infection and expanded surveillance case definition for
13. Stoneburner R, Laroche E, Prevots R, et al. Survival in a cohort of human
AIDS among adolescents and adults. MMWR Morb Mortal Wkly Rep
immunodeficiency virus-infected tuberculosis patients in New York
1992; 41(RR-17):1-19.
City: implications for the expansion of the AIDS case definition. Arch
33. Alland D, Kalkut GE, Moss AR, et al. Transmission of tuberculosis in
Intern Med 1992; 152:2033-7. New York City: an analysis by DNA fingerprinting and conventional
14. Pearson ML, Jereb JA, Frieden TR, et al. Nosocomial transmission of epidemiologic methods. N Engl J Med 1994; 330:1710-6.
multidrug-resistant Mycobacterium tuberculosis: a risk to patients and 34. Gourevitch MN, Wasserman W, Panero MS, Selwyn PA. Successful adher-
health care workers. Ann Intern Med 1992; 117:191-6. ence to observed prophylaxis and treatment of tuberculosis among drug
15. Shriner KA, Mathisen GE, Goetz MB. Comparison of mycobacterial users in a methadone program. J Addict Dis 1996; 15:93-104.
lymphadenitis among persons infected with human immunodeficiency 35. Centers for Disease Control and Prevention. Guidelines for preventing the
virus and seronegative controls. Clin Infect Dis 1992; 15:601-5. transmission of Mycobacterium tuberculosis in health-care facilities,
16. Fischl MA, Daikos GL, Uttamchandani RB, et al. Clinical presentation 1994. MMWR Morb Mortal Wkly Rep 1994; 43(RR-13):1 -132.
and outcome of patients with HIV infection and tuberculosis caused by 36. Cauthen GM, Dooley SW, Onorato IM, et al. Transmission of Mycobacte-
multiple-drug-resistant bacilli. Ann Intern Med 1992; 117:184-190. rium tuberculosis from tuberculosis patients with HIV infection or
17. Jones BE, Young SMM, Antoniskis D, Davidson PT, Kramer F, Barnes AIDS. Am J Epidemiol 1996; 144:69-77.
PF. Relationship of-the manifestations of tuberculosis to CD4 cell counts 37. Long R, Scalcini M, Manfreda J, et al. Impact of human immunodeficiency
in patients with human immunodeficiency virus infection. Am Rev Re- virus type 1 on tuberculosis in rural Haiti. Am Rev Respir Dis 1991;
spir Dis 1993; 148:1292 -1297. 143:69-73.
18. Given MJ, Khan MA, Reichman LB. Tuberculosis among patients with 38. Smith RL, Yew K, Berkowitz KA, Aranda CP. Factors affecting the yield
AIDS and a control group in an inner-city community. Arch Intern Med of acid-fast sputum smears in patients with HIV and tuberculosis. Chest
1994; 154:640-645. 1994; 106:684-6.
19. Whalen C, Horsburgh CR, Hom D, Lahart C, Simberkoff M, Ellner J. 39. Frieden TR, Sterling T, Pablos-Mendez A, Kilburn JO, Cauthen GM,
Accelerated course of human immunodeficiency virus infection after Dooley SW. The emergence of drug-resistant tuberculosis in New York
tuberculosis. Am J Respir Crit Care Med 1995; 151:129-35. City. N Engl J Med 1993; 328:521-6.
20. Turett GS, Telzak EE, Torian LV, et al. Improved outcomes for patients 40. Neville K, Bromberg A, Bromberg R, Bonk S, Hanna BA, Rom WN. The
with multidrug-resistant tuberculosis. Clin Infect Dis 1995; 21: third epidemic -multidrug-resistant tuberculosis. Chest 1994; 105:
45-8.
1238-44.
41. Kassim S, Sassan-Morokro M, Ackah A, et al. Two-year follow-up of
21. Perriens JH, St. Louis ME, Mukadi YB, et al. Pulmonary tuberculosis in
persons with HIV-1 and HIV-2 associated pulmonary tuberculosis
HIV-infected patients in Zaire: a controlled trial of treatment for either
treated with short-course chemotherapy in West Africa. AIDS 1995; 9:
6 or 12 months. N Engl J Med 1995; 332:779-84.
1185-91.
22. Ackah AN, Coulibaly D, Digbeu H, et al. Response to treatment, mortality, 42. von Overbeck J, Egger M, Smith GD, et al. Survival in HIV infection: do
and CD4 lymphocyte counts in HIV-infected persons with tuberculosis sex and category of transmission matter? AIDS 1994; 8:1307-13.
in Abidjan, COte d'Ivoire. Lancet 1995; 345:607-10. 43. Selwyn PA, Sckell BM, Alcabes P, Friedland GH, Klein RS, Schoenbaum
23. Elliott AM, Halwiindi B, Hayes RJ, et al. The impact of human immunode- EE. High risk of active tuberculosis in HIV-infected drug users with
ficiency virus on mortality of patients treated for tuberculosis in a cohort cutaneous anergy. JAMA 1992; 268:504-9.
study in Zambia. Trans R Soc Trop Med Hyg 1995; 89:78-82. 44. Chaisson RE, Keruly JC, Moore RD. Race, sex, drug use, and progression
24. Alwood K, Keruly J, Moore-Rice K, Stanton' DL, Chaulk CP, Chaisson of human immunodeficiency virus disease. N Eng J Med 1995; 333:
RE. Effectiveness of supervised, intermittent therapy for tuberculosis 751 -6.
in HIV-infected patients. AIDS 1994; 8:1103-8. 45. Hirschtick RE, Glassroth J, Jordan MC, et al. Bacterial pneumonia in
25. Shafer RW, Bloch AB, Larkin C, et al. Predictors of survival in HIV- persons infected with the human immunodeficiency virus. N Eng J Med
infected tuberculosis patients. AIDS 1996; 10:269-72. 1995; 333:845-51.