INT J TUBERC LUNG DIS 20(4):524–529

Q 2016 The Union

http://dx.doi.org/10.5588/ijtld.15.0690

Mortality and predictors in pulmonary tuberculosis with

respiratory failure requiring mechanical ventilation

S. Kim,* H. Kim,* W. J. Kim,† S-J. Lee,† Y. Hong,† H-Y. Lee,† M-N. Lim,‡ S-S. Han†
*Department of Internal Medicine, Seoul Medical Center, Seoul, †Department of Internal Medicine, School of
Medicine, Kangwon National University, Chuncheon, ‡Environmental Health Center, Kangwon National University
Hospital, Chuncheon, Korea

SUMMARY

O B J E C T I V E : To analyse the predictors and mortality
rate among patients receiving mechanical ventilation
(MV) for respiratory failure due to pulmonary tubercu-
losis (TB).
D E S I G N : We retrospectively compared patients who
required MV for TB with patients who required MV for
community-acquired pneumonia (CAP).
R E S U LT S : In-hospital mortality was significantly differ-
ent between the two groups: 95.1% in TB vs. 62.7% in
CAP (P , 0.001 using the v2 test). TB patients had a
higher 30-day mortality (P ¼ 0.040 using log-rank test),
although the median sequential organ failure assessment
(SOFA) (7.0 vs. 6.0, P ¼ 0.842) and mean Acute
Physiology and Chronic Health Evaluation (APACHE)
II scores (20.0 6 6.7 vs. 21.2 6 6.7, P ¼ 0.379) for TB
and CAP patients were not different. TB patients were
more likely to have increased lung lesion intrusions (OR
1.307, 95%CI 1.042–1.641, P ¼ 0.021), and reduced
albumin (OR 0.073, 95%CI 0.016–0.335, P ¼ 0.001),
C-reactive protein (OR 0.324, 95%CI 0.146–0.716, P ¼
0.005) and CURB-65 score (confusion, uraemia, respi-
ratory rate, blood pressure and age 765 years) (OR
0.916, 95%CI 0.844–0.995, P ¼ 0.037).
C O N C L U S I O N S : TB patients showed identical SOFA
and APACHE II scores, but higher mortality than CAP
patients. The higher mortality was not related to
severity, but suggested an association with the extent
of destructive lung lesions.
K E Y W O R D S : Mycobacterium tuberculosis; multi-or-
gan failure; pneumonia

ABOUT 1–3% OF ALL PATIENTS with tuberculo-
sis (TB) have pulmonary TB (PTB) that requires
intensive care;1 however, TB-induced acute respira-
tory failure requiring mechanical ventilation (MV) is
associated with mortality rates of 25.9–100%.2–14
Factors contributing to in-hospital mortality among
TB patients requiring MV include multi-organ
failure, TB-destroyed lung and alveolar consolida-
tion, high Acute Physiology and Chronic Health
Evaluation (APACHE) II scores, sepsis, development
of nosocomial pneumonia while in the intensive care
unit (ICU) and delayed anti-tuberculosis treat-
ment.2,5–9,14 However, the predictive factors and
pathophysiology associated with the high in-hospi-
tal mortality of such patients remains unclear.
We analysed the in-hospital mortality rate of
patients who received MV for respiratory failure
caused by PTB. We also examined factors that
affect mortality by comparing South Korean
patients with respiratory failure requiring MV
due to TB or due to community-acquired pneu-
monia (CAP).
STUDY POPULATION AND METHODS
Study design
We reviewed the medical records of patients admitted
to the ICUs of a national university-affiliated hospital
and a public teaching hospital in South Korea.
Patients with diagnoses of acute respiratory failure
requiring MV secondary to active PTB from January
2011 to April 2014 were included; the control group
included patients hospitalised for respiratory failure
necessitating MV due to CAP. Only patients in whom
active PTB or CAP appeared to be the sole cause of
respiratory failure requiring MV were included.
Those with dyspnoea resulting from heart failure,
acute exacerbation of chronic obstructive pulmonary
disease (COPD), interstitial lung disease or other
sepsis were excluded.
Of 100 patients diagnosed with PTB who were
admitted to the ICU during the study period, 59 were
excluded: 22 were admitted for surgery or other
interventions, 16 had other causes of dyspnoea (8
with acute exacerbation of COPD, 5 with combined
pneumonia and stable TB, 3 with cardiac problems),
8 were not microbiologically confirmed as TB, 7 were

Correspondence to: Seon-Sook Han, Department of Internal Medicine, Kangwon National University Hospital, 156
Baengnyeong-ro, Chuncheon 200-722, Korea. Fax: (þ82) 33 258 2455. e-mail: ssunimd@kangwon.ac.kr
Article submitted 13 August 2015. Final version accepted 1 November 2015.

admitted for observation of haemoptysis, 4 had their
diagnoses changed after admission, and 2 recovered
without MV.
Data collection
We investigated clinical features, laboratory results,
radiographic features and ICU presentation in the
patients’ records. Two pulmonologists (SK and SSH)
evaluated radiological data independently, and then
reached a consensus based on both the findings from
chest radiograph (CXR), with or without chest
computed tomography (CT), and on the formal
interpretation by chest radiologists, with respect to
the following features: cavities, pleural effusion,
alveolar consolidation, nodules, ground-glass opacity
and TB-destroyed lung.15 The designation of TB-
destroyed lung was based on a clear history of present
or past TB, coupled with radiological findings of
destroyed lung parenchyma, lung volume loss or
secondary bronchiectatic changes, verified using
CXR or CT scan by a chest radiologist.
The degree of lung involvement was evaluated
semi-quantitatively from CXRs and categorised into
three groups (minimal .0%, ,33.3%; moderate
733.3%, ,66.7%; advanced 766.7%), and was
qualitatively scored for each lung quadrant using the
Northern score (0–20), with a higher score reflecting
more severe radiological change.16 To quantify the
extent of disease, we used a ‘modified CT score’ based
on the lung area involved based on the chest CT scans.
A score of 0–3 was recorded for each lobe, with a
total possible score of 15, for cavities, alveolar
consolidation, nodules and ground-glass opacity; this
was a simplified version of the method used in studies
of severe acute respiratory syndrome.17
Clinical and laboratory data included age, sex,
coexisting medical problems, including Charlson
Comorbidity Index, blood gas analyses, blood chem-
istry, CURB-65 (confusion, uraemia, respiratory rate
blood pressure and age 765 years) and pneumonia
severity index (PSI). The presence of active PTB was
defined as the identification of Mycobacterium
tuberculosis in acid-fast bacilli (AFB) of sputum or
tracheal aspiration culture, or a positive AFB smear
accompanied by a positive nucleic acid amplification
(NAA) test result for M. tuberculosis. Acute respira-
tory distress syndrome (ARDS) was diagnosed using
the ‘Berlin definition’, and was classified as mild,
moderate or severe according to the partial arterial
oxygen pressure/fraction of inspired oxygen ratio
with a positive end-expiratory pressure or continuous
positive airway pressure value of at least 5 cmH2O.18
The sequential organ failure assessment (SOFA) and
APACHE II scores were calculated within 24 h of ICU
admission to evaluate severity of disease and provide
an estimate of in-hospital mortality.
TB with respiratory failure 525
Statistical analysis
We compared the categorical characteristics of TB and
CAP patients using Fisher’s exact and v2 tests.
Continuous variables were compared using indepen-
dent t-tests or Mann-Whitney U-tests, and expressed
as medians with their interquartile ranges (IQRs) or as
means 6 standard deviation (SD). Prognostic factors
were examined using logistic regression analysis.
Cumulative survival probabilities were estimated using
the Kaplan-Meier method and compared using the log-
rank test. All statistical analyses were performed using
SPSS, version 21.0 (IBM, Chicago, IL, USA) and SAS
software, version 9.4 (Statistical Analysis System,
Cary, NC, USA). All tests were two-tailed; P , 0.05
was considered statistically significant.
Ethical considerations
The institutional ethics review board of the Kangwon
National University Hospital, Chuncheon, and the
Seoul Medical Center, Seoul, South Korea, approved
the study.
RESULTS
Characteristics, clinical features and laboratory results
Forty-one patients with PTB (TB group) and 59
patients who received MV for CAP during the same
period (CAP group) were included. The median age of
the TB group was significantly lower than that of the
CAP group (56.3 vs. 74.0 years, P , 0.001). Of the 59
CAP patients, 13 (22%) had underlying lung disease,
significantly more than in the TB group (2/41, 4.9%, P
¼ 0.018; Table 1). The Charlson Comorbidity Index
was not significantly different between the TB and
CAP groups (0.76 6 1.28 vs. 1.05 6 1.15, P ¼ 0.232).
Symptoms of general weakness (85.4%), cough
(68.3%), dyspnoea (65.9%) and weight loss (61.0%)
were more common than fever (36.6%) and haemop-
tysis (4.9%), with general weakness (P ¼ 0.001) and
weight loss (P , 0.001) being significantly higher in
the TB group. The mean time from onset of the major
complaint to hospital admission was longer in the TB
than in the CAP group (10.2 vs. 3.3 days, P ¼ 0.003).
Compared to the CAP group, both the body mass
index (BMI) and laboratory parameters reflecting
nutritional status were significantly lower in the TB
group, including total cholesterol, total protein,
albumin and haematocrit levels (Table 2). CURB-65
score was higher in the CAP group than in the TB
group; however, the PSI was not different between the
two groups.
The sputum specimens of 37/41 (90.2%) TB patients
were AFB-positive on smear: 29 (70.7%) had 73þAFB
smears (Appendix Table A.1).* Anti-tuberculosis treat-
* The appendix is available in the online version of this article, at
http://www.ingentaconnect.com/content/iuatld/ijtld/2016/
00000020/00000004/art00018
526 The International Journal of Tuberculosis and Lung Disease

Table 1 Clinical characteristics of patients with pulmonary TB and CAP requiring mechanical ventilation for respiratory failure
TB group (n ¼ 41) CAP group (n ¼ 59)
n (%) n (%) P value
Age, years, median [IQR] 56.3 [47.0–73.0] 74.0 [67.0–82.0] ,0.001
Male sex 35 (85.4) 40 (67.8) 0.046
Types of insurance 0.558
Medical insurance 29 (70.7) 46 (78.0)
Medical assistance, homeless 12 (29.3) 13 (22.0)
Living alone without family 21 (51.2) 10 (16.9) ,0.001
HIV infection 1 (2.4) 0 0.228
History of anti-tuberculosis treatment 10 (24.4) 12 (20.3) 0.631
Completion of TB medication 2 (20.0) 11 (91.7) 0.002
Comorbidities
Underlying lung disease* 2(4.9) 13 (22.0) 0.018
Hypertension 6(14.6) 28 (48.3) 0.001
Diabetes 5(12.2) 21 (35.6) 0.009
Cerebrovascular disease 1(2.4) 10 (16.9) 0.023
Chronic kidney disease — 4 (6.8) 0.089
Liver cirrhosis 4 (9.8) 2 (3.4) 0.187
Malignancies 3 (7.3) 4 (6.8) 0.917
Charlson Comorbidity Index, mean 6 SD 0.76 6 1.28 1.05 6 1.15 0.232
* Included chronic obstructive pulmonary disease, idiopathic pulmonary fibrosis.
TB ¼ tuberculosis; CAP ¼ community-acquired pneumonia; IQR ¼ interquartile range; HIV ¼ human immunodeficiency virus.

ment in the TB group was initiated within a median of 1
hospital day; 40/41 (97.6%) TB patients were started
on treatment within 4 days of hospitalisation.
Radiological features
Major findings in the TB group included alveolar
consolidation in 41 (100%), centrilobular nodules in
39 (95.1%), cavities in 33 (80.5%), ground-glass
opacity in 33 (80.5%) and pleural effusion in 15
(36.6%) of the 41 patients. Alveolar consolidation
(100% vs. 88.1%, P ¼ 0.022), centrilobular nodules
(95.1% vs. 5.3%, P , 0.001), cavities (80.5% vs.
8.5%, P , 0.001), TB-destroyed lung (48.8% vs.
13.6%, P , 0.001) and pneumothorax (17.1% vs.

Table 2 Laboratory and radiological findings in the TB and CAP groups

Subjects
TB group (n ¼ 41) CAP group (n ¼ 59)
median [IQR] median [IQR] P value
Biological parameters
BMI, kg/m2, mean 6 SD 18.0 6 3.3 21.8 6 4.3 0.000
PaO2/FiO2, mmHg (ICU admission day) 135.0 [85.0–184.0] 113.2 [80.0–182.0] 0.451
PaO2, mmHg 69.0 [59.0–82.3] 55.6 [47.0–73.0] 0.027
PaCO2, mmHg 32.0 [29.0–35.0] 37.3 [30.0–52.0] 0.003
Haematocrit, mean 6 SD 31.9 6 4.8 34.9 6 8.5 0.030
White blood cell count (3109/l), mean 6 SD 11.3 6 6.9 13.5 6 7.1 0.119
Neutrophil 89.4 [86–93] 83.2 [74–89] ,0.001
Lymphocyte 5.7 [4.0–8.6] 9.5 [4.9–17.2] 0.004
Serum creatinine, mg/dl 0.8 [0.6–1.2] 1.1 [0.7–1.6] 0.012
C-reactive protein, mg/dl 13.8 [10.0–18.1] 15.7 [5.1–25.6] 0.372
Lactate dehydrogenase, IU/l 731.0 [453.0–1128.0] 479.5 [283.0–783.0] 0.008
Total protein, g/dl, mean 6 SD 5.8 6 0.9 6.3 6 0.9 0.006
Albumin, g/dl, mean 6 SD 2.2 6 0.6 2.9 6 0.6 ,0.001
Total cholesterol, mg/dl, mean 6 SD 90.0 6 39.1 111.3 6 42.1 0.043
PSI score, mean 6 SD 111.0 6 31.7 121.6 6 35.1 0.123
CURB-65 1.7 [2.0–3.0] 2.4 [2.0–3.0] 0.001
Radiological findings
Degree of lung involvement grossly evaluated (chest radiograph), n (%) 0.013
Minimal (.0%, ,33.3%) 0 6 (10.2)
Moderate (733.3%, ,66.7%) 10 (24.4) 25 (42.4)
Advanced (766.7%) 31 (75.6) 28 (47.5)
Northern score (chest radiograph), mean 6 SD 13.6 6 4.0 11.14 6 4.68 0.007
Lobes involved (CT scans) 4.8 [3.0–5.0] 4.2 [2.5–5.0] 0.001
Modified CT score 11.4 [9.0–14.0] 9.1 [6.0–12.0] 0.003
Combined TB-destroyed lung, n (%) 20 (48.8) 8 (13.6) ,0.001
Combined pneumothorax, n (%) 7 (17.1) 2 (3.4) 0.019
TB ¼ tuberculosis; CAP ¼ community-acquired pneumonia; IQR ¼ interquartile range; BMI ¼ body mass index; SD ¼ standard deviation; PaO2 ¼ partial arterial
oxygen pressure; FiO2 ¼ fraction of inspired oxygen; ICU ¼ intensive care unit; PaCO2 ¼ partial arterial CO2 pressure; IU ¼ international unit; PSI ¼ pneumonia
severity index; CURB-65 ¼ confusion, uraemia, respiratory rate, blood pressure and age 765 years; CT ¼computed tomography.

Figure Survival curves for 30-day mortality among patients with respiratory failure requiring
mechanical ventilation due to TB (dotted line) or CAP (solid line) as calculated by the Kaplan-Meier
method and compared using the log-rank test (P ¼ 0.040). CAP ¼ community-acquired
pneumonia; TB ¼ tuberculosis.
3.4%, P ¼ 0.019) were significantly more frequent in
the TB group than in the CAP group.
Of the 41 TB patients, 31 (75.6%) demonstrated
severe alterations, including extensive bilateral infil-
trates on CXR (P ¼ 0.013). The Northern scores for
each lung quadrant on CXR were significantly more
severe in the TB than in the CAP group.16 In 82
patients who underwent initial chest CT (38 with TB,
44 with CAP), the number of involved lobes was
significantly higher in the TB than in the CAP group
(median 4.76 vs. 4.17, P ¼ 0.001). The modified CT
score differed significantly between the TB and CAP
groups (median 11.38 vs. 9.14, P ¼ 0.003; Table 2).17
Intensive care and significant differences between the
two groups
The median intervals from initial admission to
transfer to the ICU were respectively 1.14 (IQR 0–
10) and 0.63 (IQR 0–2) days in the TB and CAP
groups (P ¼ 0.054). The median duration in the ICU
and hospital, and the duration of MV, as well as the
median SOFA scores (TB 7.00, IQR 4.00–9.00; CAP
6.00, IQR 4.00–8.00, P ¼ 0.842) and mean APACHE
II scores (TB 20.0 6 6.7, CAP 21.2 6 6.7, P ¼ 0.379)
were not significantly different between the groups.
Nineteen (46.3%) TB patients and 29 CAP patients
(49.2%) developed ARDS (P ¼ 0.784). During ICU
admission, 15 (36.6%) TB and 20 (33.9%) CAP
patients were diagnosed with ventilator-associated
pneumonia (VAP). The incidence of sepsis and of
acute kidney injury was slightly, but non-significantly,
higher in the TB group. However, multi-organ failure
(65.9% vs. 44.1%, P ¼ 0.032) and shock (92.7% vs.
TB with respiratory failure 527
67.8%, P ¼ 0.003) were significantly higher in the TB
than in the CAP group (Appendix Table A.2).
Thirty-nine patients in the TB group and 37
patients in the CAP group died in hospital (median
survival 12.5 vs. 13.0 days). The overall mortality
rates were respectively 95.1% and 62.7% in the TB
and CAP groups (P , 0.001; v2 test). The Kaplan-
Meier survival curves revealed a higher 30-day
mortality rate in the TB than in the CAP group (log-
rank test, P ¼ 0.040; Figure). Multivariate analysis
showed that a wider extent of lung lesion intrusions
(odds ratio [OR] 1.307, 95% confidence interval [CI]
1.042–1.641, P ¼ 0.021), lower albumin (OR 0.073,
95%CI 0.016–0.335, P ¼ 0.001) and C-reactive
protein levels (OR 0.324, 95%CI 0.146–0.716, P ¼
0.005) and lower CURB-65 scores (OR 0.916,
95%CI 0.844–0.995, P ¼ 0.037) were more likely
to be independently associated with TB than with
CAP (Table 3).
DISCUSSION
In-hospital mortality among TB patients with respi-
ratory failure was markedly higher (95.1%) in our
study than in previous reports (approximately 60%),
although some of these included patients who did not
require MV.19 A 59% mortality rate was reported in a
similar study of patients with respiratory failure
requiring MV.6 However, the median APACHE II
score of patients in that study was 16,6 while our
patients had more severe APACHE II scores (20.0 6
SD 6.7); our patients also had extensive disease
involving more than three lobes in all those with
528 The International Journal of Tuberculosis and Lung Disease
Table 3 Multivariate analysis for predicting prognostic factors
for mechanical ventilation*
Beta P value OR (95%CI)
Modified CT score 0.268 0.021 1.307 (1.042–1.641)
Albumin –2.617 0.001 0.073 (0.016–0.335)
C-reactive protein –1.128 0.005 0.324 (0.146–0.716)
CURB-65 –0.087 0.037 0.916 (0.844–0.995)
* Factors derived from the univariate analysis as potentially different between
the tuberculosis group and the community-acquired pneumonia group (age,
sex, body mass index, C-reactive protein, neutrophil, lymphocyte, lactate
dehydrogenase, total protein, albumin, creatinine, PaO2/FiO2 ratio, PaCO2,
pneumonia severity index, CURB-65, the presence of chronic lung diseases,
Charlson Comorbidity Index, duration from onset of major complaints to
admission, corticosteroid treatment, SOFA score, multi-organ failure, North-
ern score, modified CT score, pneumothorax and absence of family support)
were included in this multivariate analysis.
CI ¼ confidence interval; CT ¼computed tomography; CURB-65 ¼ confusion,
uraemia, respiratory rate, blood pressure and age 765 years; PaO2 ¼ partial
arterial oxygen pressure; FiO2 ¼ fraction of inspired oxygen; PaCO2 ¼ partial
arterial carbon dioxide pressure.
available chest CTs, as compared to 53% in the
previous study.6
Many previous studies reported that high (.18–
20) APACHE II scores had some independent
predictive value for in-hospital mortality. However,
the mean APACHE II scores did not differ statistically
significantly between the groups in our study, despite
significantly higher mortality in the TB group. The
mean APACHE II score in the TB group was 20,
which is typically associated with a mortality rate of
30–40% in patients with respiratory failure due to
infection, which was markedly lower than that in our
study. The APACHE II score may thus consistently
underestimate mortality among PTB patients,6,8,12,20
and the initial single APACHE II might not reflect all
of the variables that change after ICU admission or
during disease progression.21 Patients who develop
respiratory failure due to PTB may have longer
treatment response times, and aggravation of the
clinical presentation and worsening of CXR findings
due to parenchyma extension may be observed in the
first 3 months of treatment.22,23
TB patients requiring ICU care may have high rates
of ICU-related complications, which may increase the
risk of mortality; in particular, nosocomial pneumo-
nia, including VAP, has been suggested to predict
mortality in such patients.5–8,10 Lin et al. reported an
incidence of nosocomial pneumonia of around four-
fold higher in non-survivors than in survivors among
PTB patients with respiratory failure.14 Nevertheless,
in our study, the incidence of VAP in the two groups
was not significantly different.
ARDS is a rare complication of TB that is
associated with high in-hospital mortality rates.24 In
our study, less than 50% of patients in both groups
were diagnosed with ARDS. Another rare complica-
tion, severe TB sepsis, associated with septic shock
with multi-organ dysfunction, has been reported
almost exclusively among immunocompromised
hosts.1 In our study, which included only one human
immunodeficiency virus infected patient without
disseminated TB, the incidence of sepsis was not
significantly higher; however, multi-organ failure in
the TB group was higher than in the CAP group
(65.9% vs. 44.1%), although the difference was not
statistically significant. Mycobacterial sepsis is due to
host factors such as depressed cell-mediated immu-
nity, rather than to the virulence of the pathogen,25
and in vitro studies have shown that a cell wall
component of M. tuberculosis, lipoarabinomannan,
activates macrophages.13,26
The presence of alveolar consolidation on CXR
was a strong independent factor contributing to in-
hospital mortality among patients with PTB requiring
MV (hazard ratio 7.731).8 Earlier studies have
suggested that consolidation indicates an inadequate
immune response to a heavy bacillary burden, and
that death results from poor immunity.8,27 Although
the ‘modified CT score’ was the only significant factor
in multivariate analysis, lung involvement was more
extensive in the TB than in the CAP group,.
Radiological manifestations of PTB depend on
several host factors, including prior exposure to TB,
age and underlying immune status;22 we thus
hypothesised that the rapid destruction of lung
parenchyma28 may contribute to the higher mortality
of PTB patients with respiratory failure. Malnutrition
and underlying comorbidities may predispose pa-
tients to TB by immune suppression. Previous studies
have reported that M. tuberculosis infection reduces
monocyte expression and immunity, especially in
patients aged .60 years, those with a lower initial
body weight (,50 kg), those with coexisting medical
conditions and those with extensive disease on
radiography.29 Study patients with TB were younger;
however, their BMI and albumin levels were lower
than among the CAP patients, suggesting that the
extremely poor nutritional status of the TB patients
may have caused immune suppression.
The mean interval from hospital admission to
initiation of anti-tuberculosis treatment was 1 day,
which is shorter than reported previously for other
countries (14.9 and 7.2 days).8,11 Nevertheless,
treatment may have been delayed because of the high
proportion of homeless patients of low socio-eco-
nomic status and those without family support; the
longer time from onset of symptoms to admission
may also be a potential confounder for disease
severity in the TB group.
The limitations of the study are its retrospective
nature and small sample size. We compared the
mortality and clinical features between two different
diseases, but failed to determine the direct cause of
the extremely high mortality in PTB patients requir-
ing MV. Moreover, we used only the SOFA and the
APACHE II scores to estimate severity. This study
used patients from two institutions who may have

been different from TB patients at other hospitals.
Large-scale prospective studies are thus required to
investigate these issues further.
CONCLUSIONS
PTB patients with respiratory failure requiring MV
had higher mortality than CAP patients, despite
similar SOFA and APACHE II scores, and a similar
level of complications, including ARDS and VAP. A
wider extent of lung lesion intrusions, lower albumin
and C-reactive protein levels, and lower CURB-65
scores were independently associated with TB,
suggesting a presumptive association between the
extent of destructive lung parenchymal lesions and
the high mortality observed in TB patients with
respiratory failure.
Conflicts of interest: none declared.
References
1 Bridges D A, Bedimo R G. Severe tuberculosis sepsis in an
immunocompetent patient. Am J Med 2006; 119: e11–e14.
2 Kim Y J, Pack K M, Jeong E, et al. Pulmonary tuberculosis with
acute respiratory failure. Eur Respir J 2008; 32: 1625–1630.
3 Kwon Y S, Kim Y H, Song J U, et al. Risk factors for death
during pulmonary tuberculosis treatment in Korea: a
multicenter retrospective cohort study. J Korean Med Sci
2014; 29: 1226–1231.
4 Silva D R, Gazzana M B, Dalcin Pde T. Severe tuberculosis
requiring ICU admission. J Bras Pneumol 2012; 38: 386–394.
5 Silva D R, Menegotto D M, Schulz L F, Gazzana M B, Dalcin P
T. Mortality among patients with tuberculosis requiring
intensive care: a retrospective cohort study. BMC Infect Dis
2010; 10: 54.
6 Ryu Y J, Koh W J, Kang E H, et al. Prognostic factors in
pulmonary tuberculosis requiring mechanical ventilation for
acute respiratory failure. Respirology 2007; 12: 406–411.
7 Erbes R, Oettel K, Raffenberg M, Mauch H, Schmidt-Ioanas M,
Lode H. Characteristics and outcome of patients with active
pulmonary tuberculosis requiring intensive care. Eur Respir J
2006; 27: 1223–1228.
8 Lee P L, Jerng J S, Chang Y L, et al. Patient mortality of active
pulmonary tuberculosis requiring mechanical ventilation. Eur
Respir J 2003; 22: 141–147.
9 Levy H, Kallenbach J M, Feldman C, Thorburn J R,
Abramowitz J A. Acute respiratory failure in active
tuberculosis. Crit Care Med 1987; 15: 221–225.
10 Frame R N, Johnson M C, Eichenhorn M S, Bower G C,
Popovich Jr J. Active tuberculosis in the medical intensive care
unit: a 15-year retrospective analysis. Crit Care Med 1987; 15:
1012–1014.
11 Heffner J E, Strange C, Sahn S A. The impact of respiratory
failure on the diagnosis of tuberculosis. Arch Intern Med 1988;
148: 1103–1108.
12 Zahar J R, Azoulay E, Klement E, et al. Delayed treatment
contributes to mortality in ICU patients with severe active
TB with respiratory failure 529
pulmonary tuberculosis and acute respiratory failure. Intensive
Care Med 2001; 27: 513–520.
13 Penner C, Roberts D, Kunimoto D, Manfreda J, Long R.
Tuberculosis as a primary cause of respiratory failure requiring
mechanical ventilation. Am J Respir Crit Care Med 1995; 151:
867–872.
14 Lin S, Wang T, Liu W, et al. Predictive factors for mortality
among non-HIV-infected patients with pulmonary tuberculosis
and respiratory failure. Int J Tuberc Lung Dis 2009; 13: 335–
340.
15 Park J H, Na J O, Kim E K, et al. The prognosis of respiratory
failure in patients with tuberculous destroyed lung. Int J Tuberc
Lung Dis 2001; 5: 963–967.
16 Conway S P, Pond M N, Bowler I, et al. The chest radiograph in
cystic fibrosis: a new scoring system compared with the
Chrispin-Norman and Brasfield scores. Thorax 1994; 49: 860–
862.
17 Chang Y C, Yu C J, Chang S C, et al. Pulmonary sequelae in
convalescent patients after severe acute respiratory syndrome:
evaluation with thin-section CT. Radiology 2005; 236: 1067–
1075.
18 Ferguson N D, Fan E, Camporota L, et al. The Berlin definition
of ARDS: an expanded rationale, justification, and
supplementary material. Intensive Care Med 2012; 38: 1573–
1582.
19 Sohn J W. Pulmonary tuberculosis and acute respiratory failure.
Korean J Med 2011; 81: 455–457.
20 Papadakis M A, Browner W S. Prognosis of non-cardiac
medical patients receiving mechanical ventilation in a veterans
hospital. Am J Med 1987; 83: 687–692.
21 Knaus W A, Draper E A, Wagner D P, Zimmerman J E.
APACHE II: a severity of disease classification system. Crit
Care Med 1985; 13: 818–829.
22 Leung A N. Pulmonary tuberculosis: the essentials. Radiology
1999; 210: 307–322.
23 Lim S J, Lew D, Song H N, et al. Development of acute
respiratory failure on initiation of anti-tuberculosis medication
in patients with pulmonary tuberculosis: clinical and radiologic
features of 8 patients and literature review. Korean J Crit Care
Med 2013; 28: 108–114.
24 Lee K, Kim J H, Lee J H, et al. Acute respiratory distress
syndrome caused by miliary tuberculosis: a multicentre survey
in South Korea. Int J Tuberc Lung Dis 2011; 15: 1099–1103.
25 Jog S, Pawar B, Patel D. Mycobacterial sepsis and multiorgan
failure syndrome. In: Vincent J-L, ed. Annual update in
intensive care and emergency medicine, 2011. Berlin, Germany:
Springer-Verlag, 2011: pp 531–542.
26 Ahuja S S, Ahuja S K, Phelps K R, Thelmo W, Hill A R.
Hemodynamic confirmation of septic shock in disseminated
tuberculosis. Crit Care Med 1992; 20: 901–903.
27 Barnes P F, Leedom J M, Chan L S, et al. Predictors of short-
term prognosis in patients with pulmonary tuberculosis. J Infect
Dis 1988; 158: 366–371.
28 Cardona P J. Revisiting the natural history of tuberculosis. The
inclusion of constant reinfection, host tolerance, and damage
response frameworks leads to a better understanding of latent
infection and its evolution towards active disease. Arch
Immunol Ther Exp 2010; 58: 7–14.
29 Wang C S, Chen H C, Yang C J, et al. The impact of age on the
demographic, clinical, radiographic characteristics and
treatment outcomes of pulmonary tuberculosis patients in
Taiwan. Infection 2008; 36: 335–340.
 TB with respiratory failure i

APPENDIX

Table A.1 Clinical and microbiological involvement in TB

patients
Total (n ¼ 41)
Variables n (%)
Pattern of pulmonary TB
Pulmonary only 34 (82.9)
Miliary þ pulmonary 7 (17.1)
Extra-pulmonary TB
Negative 36 (87.8)
Gastrointestinal 3 (7.3)
Spinal 1 (2.4)
Urogenital 1 (2.4)
Presence of drug resistance
Not checked* 20 (48.8)
Pansusceptible TB 17 (41.5)
MDR-TB 1 (2.4)
Single resistance to first-line drug 3 (7.3)
Sputum AFB smear†
Negative 4 (9.8)
1þ 3 (7.3)
2þ 5 (12.2)
3þ 6 (14.6)
4þ 23 (56.1)
NAA for Mycobacterium tuberculosis
Negative 2 (5.1)
Positive 37 (94.9)
Sputum AFB culture for M. tuberculosis
Negative 2 (4.9; all PCRþ)
Positive 39 (95.1)
* Patients died before AFB culture results were reported.
†
1þ¼ 1~9/100 fields; 2þ¼ 1~9/10 fields; 3þ ¼ 1~9/field; 4þ 7 9/field.
TB ¼ tuberculosis; MDR-TB ¼ multidrug-resistant TB; AFB ¼ acid-fast bacilli;
NAA ¼ nucleic acid amplification; PCR ¼ polymerase chain reaction; þ ¼
positive.
ii The International Journal of Tuberculosis and Lung Disease

Table A.2 ICU features and outcomes in the TB and CAP groups
TB group (n ¼ 41) CAP group (n ¼ 59)
n (%) n (%) P value
Clinical features in the ICU
Temperature, 8C, median [IQR] 36.7 [36.5–37.7] 36.9 [36.6–38.0] 0.231
Glasgow coma scale, median [IQR] 13.4 [9–15] 13.4 [10–15] 0.828
Systolic blood pressure, mmHg, median [IQR] 100.7 [94.0–128.0] 94.3 [80.0–110.0] 0.141
Heart rate, beats/min, median [IQR] 116.0 [99.0–140.0] 111.8 [101–132] 0.362
Respiratory rate, breaths/min, mean 6 SD 22.8 6 6.1 28.2 6 7.1 0.752
APACHE II in the ICU,* mean 6 SD 20.0 6 6.7 21.2 6 6.7 0.379
SOFA score in the ICU,* median [IQR] 7.00 [4.00–9.00] 6.00 [4.00–8.00] 0.842
Complications in the ICU
ARDS severity 0.784
Mild 1 (2.4) 2 (3.4)
Moderate 10 (24.4) 11 (18.6)
Severe 8 (19.5) 16 (27.1)
VAP 15 (36.6) 20 (33.9) 0.782
Sepsis 30 (73.2) 34 (57.6) 0.111
Shock 38 (92.7) 40 (67.8) 0.003
Acute kidney injury (CRRT needed) 12 (29.3) 11 (18.6) 0.214
Multi-organ failure† 27 (65.9) 26 (44.1) 0.032
Outcomes
Mortality 39 (95.1) 37 (62.7) ,0.001
Days to death, median [IQR] 12.5 [7.0–24.0] 13.0 [5.0–24.0] 0.509
ICU length of stay, days, median [IQR] 7.8 [3.0–17.0] 9.0 [4.0–18.0] 0.676
Hospital length of stay, days, median [IQR] 13.2 [7.0–28.0] 16.75 [8.0–34.0] 0.654
Duration of mechanical ventilation, days, median [IQR] 6.3 [3.0–14.0] 6.8 [3.0–15.0] 0.601
Cause of death
Hypoxaemia 9 (23.1) 11 (29.7) 0.463
Septic shock 16 (41.0) 10 (27.0)
Multi-organ failure 14 (35.9) 15 (40.5)
Other cause 0 (0.0) 1 (2.7)
* Worst value within 24 h of admission to ICU.
†
More than two failed organs.
ICU ¼ intensive care unit; TB ¼ tuberculosis; CAP ¼ community-acquired pneumonia; IQR ¼ interquartile range; SD ¼ standard deviation; APACHE ¼ Acute
Physiology and Chronic Health Enquiry; SOFA ¼ sequential organ failure assessment; ARDS ¼ acute respiratory distress syndrome; VAP ¼ ventilator-associated
pneumonia; CRRT ¼continuous renal replacement therapy.

O B J E C T I F : Analyser les facteurs de prédiction et la
mortalité des patients bénéficiant d’une ventilation
mécanique (MV) pour détresse respiratoire causée par
une tuberculose (TB) pulmonaire.
S C H É M A : Nous avons comparé rétrospectivement des
patients TB qui ont eu besoin d’une MV à des patients
qui ont bénéficié d’une MV pour une pneumonie
communautaire (CAP).
R É S U L T A T S : La mortalit é intrahospitali ère a ét é
significativement différente entre les deux groupes :
95,1% pour les patients TB et 62,7% pour les patients
CAP (P , 0,001 par le test du v2). Les patients TB ont eu
une mortalité de 30 jours plus élevée (P ¼ 0,040 par test
de log-rank), mais les scores médians SOFA (Sequential
Organ Failure Assessment) (7,0 contre 6,0 ; P ¼ 0,842) et
OBJETIVO : Analizar los factores pronósticos y la
mortalidad de los pacientes que reciben ventilación
mecánica (MV) en caso de insuficiencia respiratoria
causada por la tuberculosis (TB) pulmonar.
M É T O D O S : Se llevó a cabo un análisis retrospectivo de
pacientes con diagnóstico de TB que precisaron MV y de
pacientes que necesitaron MV por neumonı́a de origen
extrahospitalario (CAP).
R E S U L T A D O S : La mortalidad intrahospitalaria fue
significativamente diferente en ambos grupos, a saber,
95,1% en el grupo con TB y 62,7% en el grupo con CAP
(P , 0,001 según la prueba de la v2). En los pacientes
con TB la mortalidad fue más alta en los primeros 30
dı́as (P ¼ 0,040 según la prueba de orden logarı́tmico),
pero no se observó una diferencia en la puntuación
promedio de las escalas SOFA (Sequential Organ Failure
Assessment) (7,0 contra 6,0; P ¼ 0,842) y APACHE II
TB with respiratory failure iii
RESUME
APACHE II moyen (Acute Physiology and Chronic
Health Evaluation) (20,0 6 6,7 contre 21,2 6 6,7 ; P ¼
0,379) n’ont pas été différents entre les groupes de
patients TB et CAP. Les patients TB ont eu plus souvent
des survenues de lésions pulmonaires (OR 1,307 ;
IC95% 1,042–1,641 ; P ¼ 0,021) et une diminution du
taux d’albumine (OR 0,073 ; IC95% 0,016–0,335 ; P ¼
0,001), de la protéine C-réactive (OR 0.324 ; IC95%
0,146–0,716 ; P ¼ 0,005) et du CURB-65 (OR 0,916 ;
IC95% 0,844–0,995 ; P ¼ 0,037).
C O N C L U S I O N : Les patients TB ont eu des scores SOFA
et APACHE II identiques, mais une mortalité plus élevée
que les patients CAP. La mortalité plus élevée n’a pas été
liée à la gravité, mais a suggéré une association avec
l’extension des lésions destructives des poumons.
RESUMEN
(Acute Physiology and Chronic Health Evaluation) (20,0
6 6,7 contra 21,2 6 6,7; P ¼0,379). En los pacientes con
TB fue más probable observar más lesiones pulmonares
(OR 1,307; IC95% 1,042–1,641; P ¼ 0,021) y una
disminución de la albuminemia (OR 0,073; IC95%
0,016–0,335; P ¼ 0,001), la proteı́na C reactiva (OR
0,324; IC95% 0,146–0,716; P ¼ 0,005) y de la
puntuación en la escala CURB-65 (OR 0,916; IC95%
0,844–0,995; P ¼ 0,037).
C O N C L U S I Ó N : Ambos grupos presentaron una
puntuación idéntica en las escalas SOFA y APACHE II,
pero la mortalidad fue más alta en los pacientes con TB
que en los pacientes con CAP. La mortalidad más alta no
se asoció con la puntuación de las escalas de gravedad,
pero hubo indicaciones de una asociación con la
magnitud de las lesiones destructivas del pulmón.