Post-bronchoscopy sputum: Improving the
diagnostic yield in smear negative pulmonary TB
Peter M. Georgea,d, Meera Mehtaa,d, Jaideep Dhariwala,
Aran Singanayagama, Claire E. Raphaela, Mohammad Salmasia,
David W. Connella, Philip Molyneauxa, Melissa Wickremasinghea,
Annette Jepsonb, Onn Min Kona,c,*
aChest and Allergy Department, St Mary’s Hospital, Imperial College Healthcare NHS Trust, Praed Street,
London W2 1NY, UK
bDepartment of Microbiology, St Mary’s Hospital, Imperial College Healthcare NHS Trust, Praed Street, London W2 1NY, UK
cCentre for Respiratory Infection, St Mary’s Hospital, Imperial College, Norfolk Square, London W2 1PG, UK
Received 6 March 2011; accepted 23 July 2011
Available online 15 August 2011
Introduction: Patients with suspected active Pulmonary Tuberculosis (PTB) who are Acid-Fast
Bacilli (AFB) smear negative or non-productive of sputum may undergo bronchoalveolar lavage.
However, post-bronchoscopy sputum (PBS) sampling is not routine. The aim of this study was to
establish the potential diagnostic value of PBS sampling.
Methods: A retrospective study of patients attending a London University hospital with micro-
biologically confirmed PTB between January 2004 and December 2010. Patients who were AFB
smear negative or non-productive of sputum were eligible if sputum sampling was performed
within 7 days of bronchoscopy.
Results: Over the study period, 236 patients had microbiologically confirmed smear negative
PTB of which 57 patients were eligible for the study. 15 patients (26.3%) were infected with
HIV. 19 patients (33.3%) converted to AFB sputum smear positivity post-bronchoscopy and 5
patients (8.8%) were exclusively AFB sputum smear positive on PBS microscopy. Mycobacterium
tuberculosis was cultured from the PBS of 43 patients (75.4%) and of these, 4 (7.0%) were
exclusively PBS culture positive.
Abbreviations: AFB, Acid-fast bacilli; BAL, bronchoalveolar lavage; HIV, human immunodeficiency virus; PBS, post-bronchoscopy sputum;
PTB, pulmonary tuberculosis; TB, tuberculosis.
* Corresponding author. Chest and Allergy Clinic, St Mary’s Hospital, Imperial College Healthcare NHS Trust, Praed Street, London W2 1NY,
UK. Tel.: þ44 (0) 2033121344.
E-mail addresses: email@example.com (P.M. George), firstname.lastname@example.org (M. Mehta), email@example.com.
uk (J. Dhariwal), firstname.lastname@example.org (A. Singanayagam), email@example.com (C.E. Raphael), firstname.lastname@example.org.
uk (M. Salmasi), email@example.com (D.W. Connell), firstname.lastname@example.org (P. Molyneaux), melissa.wickremasinghe@
imperial.nhs.uk (M. Wickremasinghe), email@example.com (A. Jepson), firstname.lastname@example.org (O.M. Kon).
dJoint first authors.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/rmed
Respiratory Medicine (2011) 105, 1726e1731
0954-6111/$ - see front matter ª 2011 Elsevier Ltd. All rights reserved.
Conclusion: PBS analysis can provide a simple method of rapidly diagnosing pulmonary tuber-
culosis. In this cohort, M. tuberculosis culture yield was increased by 7% through PBS sampling.
This study has important infection control implications with nearly one third of patients
becoming more infectious after bronchoscopy.
ª 2011 Elsevier Ltd. All rights reserved.
Sputum smear microscopy to detect acid-fast bacilli (AFB)
is a rapid, inexpensive, and highly specific tool for identi-
fying persons with active pulmonary tuberculosis (PTB).1
However, many patients fail to produce sputum; and even
amongst those who are productive of sputum, a significant
percentage who have active PTB on clinical and radiological
grounds are auramine smear negative despite repeated
examination. A particular challenge for clinicians concerns
the rising incidence of human immunodeficiency virus (HIV)
related TB, with an associated increase in smear negative
PTB.2,3Smear negative HIV related TB has an increased
mortality compared to smear positive disease2,4and this
may in part be related to delays in diagnosis and initiation
Patients who are not productive of sputum or are
consistently sputum AFB smear negative undergo either
fiberoptic bronchoscopy for bronchoalveolar lavage (BAL)
or sputum induction using nebulized hypertonic saline. The
choice of technique is largely dependent on local policy but
the two procedures are widely considered equivalent in
facilitating sampling of deep-seated bronchial secretions
for microbiological and cytological analysis.6e8The sensi-
tivity of BAL microscopy for the detection of AFB is variable
in PTB7,9and sensitivities for positive BAL culture of
Mycobacterium tuberculosis vary from 44 to 95%.8,10,11
Thus, a proportion of patients with PTB will remain undi-
agnosed by BAL alone.
Little is known about the clinical utility of sputum
sampling after bronchoscopy (post-bronchoscopy sputum e
PBS) and its diagnostic potential in smear negative PTB
suspects. Two previous studies have examined the yield of
various techniques including PBS for the diagnosis of smear
negative PTB12,13but numbers were small (n Z 13e25) and
as a result conclusions regarding its potential value are
difficult to establish. One larger study in Ethiopia has also
previously investigated PBS sampling in AFB smear negative
TB suspects but only included HIV infected individuals14;
post-bronchoscopy sputum culture provided comparable
results to BAL culture in this setting.
Current guidance from the American Thoracic Society
and the Centers for Disease Control and Prevention15as
well as the National Institute for Clinical Excellence6
suggests that all patients with suspected PTB be isolated
until three AFB smear negative sputum samples have been
collected. However, it is our clinical experience that
a proportion of these patients will convert to AFB sputum
smear positivity post-bronchoscopy despite being non-
productive or consistently sputa smear negative prior to
the procedure. Such patients may be inappropriately de-
isolated posing an infection risk to health workers and
other patients or contacts.
The aim of this study was therefore to determine the
additional microbiological yield of PBS sampling and to
assess any implications on infection control, in a heteroge-
neous population within the context of a developed and low
A retrospective analysis was performed of all patients with
microbiologically confirmed PTB on the TB Register at St
Mary’s Hospital, a central London University hospital,
between 1st January 2004 and 31st December 2010. Infor-
mation was collected regarding patient demographic
profiles, HIV infection status, microbiology results and
treatment outcomes from the London TB register -
a prospectively recorded resource. Missing data was
retrieved from the case notes. At this centre, sputum
induction is not performed and all patients with suspected
PTB who are not productive of sputum or are sputum AFB
smear negative undergo fiberoptic bronchoscopy for diag-
nostic BAL. Bronchoscopy is performed by doctors special-
izing in respiratory medicine. BAL specimens are obtained
by wedging the bronchoscope into the segment of lung most
likely to yield diagnostic specimens as defined by radio-
graphic evidence of maximal pulmonary infiltrates. 30 ml
aliquots of 0.9% Saline are instilled and fluid is aspirated
until the bronchoscopist is satisfied with the lavage return.
When diffuse infiltrates are seen on imaging, the right
middle lobe is chosen as the default target. Each sputum or
BAL specimen is decontaminated using 4% NaOH and then
centrifuged at 3000 revolutions per minute for 15 min. After
decanting the supernatant, the pellet is washed and re-
suspended in phosphate buffered saline. From this,
a smear is prepared for auramine staining (with positives
being confirmed by Ziehl-Neelson) and 0.5 ml is transferred
into an MGIT tube for incubation using the BactecTMMGIT
TM960 system (BD, New Jersey USA). Sputum and BAL
samples are incubated for up to 6 weeks.
Sputum samples were collected from all inpatients
undergoing bronchoscopy in the hours after the procedure
and then on subsequent mornings whilst still in hospital.
Patients who underwent bronchoscopy as outpatients were
given sputum pots to take home and were asked to provide
samples if they became productive of sputum after the
procedure. Patients with microbiologically confirmed PTB
were eligible for inclusion into the study if they were
sputum AFB smear negative or non-productive of sputum
prior to bronchoscopy and had sputum samples sent within
one week of the procedure. Patients who were treated
empirically on clinical, radiological or biochemical grounds
were excluded from the study. Patients who were AFB
smear positive prior to bronchoscopy as well as those who
did not have a PBS sample sent within the defined
Post-bronchoscopy sputum in TB M. tuberculosis
timeframe were also excluded. In the absence of any
published work, 7 days was taken as an arbitrary but
pragmatic inclusion time point. Only the first positive PBS
sample was recorded even if it was preceded by negative
samples. In the absence of a positive PBS sample but in the
event of numerous negative samples, only the first sample
was included; ie only one PBS sample was included per
patient. Patients who were initially AFB smear positive but
whose samples subsequently cultured non-tuberculous
mycobacteria were excluded.
The study protocol was discussed with and approved by
the Chairman of the West London Research and Ethics
Committee 2 and the Joint Research Office at St Mary’s
Resultswere analyzed using Microsoft Excel for Mac 2008.
Over the study period, 419 patients had smear negative PTB
all of whom underwent bronchoscopy. 183 patients were
treated empirically with no positive microbiology but other
features consistent with a diagnosis of PTB. The remaining
236 patients had microbiologically proven smear negative
PTB and of these a total of 57 patients met the inclusion
criteria of the study by virtue of the fact that they had at
least one PBS sample sent within 7 days of the procedure.
35 were male (61.4%) and 15 were infected with HIV
(26.3%). The HIV status was unknown in 4 patients (7%). The
mean age was 41.1 years (range 16e89). 23 patients (40.4%)
were of Black African origin, 15 (26.3%) were White
Caucasianand8 (14.0%)werefrom theIndian
Subcontinent. 53 patients (93.0%) had abnormal radiology
which consisted of a chest radiograph or a computerized
tomography scan of the thorax which was consistent with
possible TB infection.
Of the 57 patients, 26 (45.6%) were sputum AFB smear
negative prior to bronchoscopy and 31 patients (54.4%)
were non-productive of sputum. At least one and up to 3
pre-bronchoscopy sputum samples were taken per produc-
tive patient (Fig. 1). All patients had PBS samples sent
within 7 days of the procedure - 39 of 57 samples (68.4%)
were sent within the first 48 h (median 2 days) (Fig. 2). PBS
samples were smear positive when taken from a range of
1e7 days post-procedure (mean 2.18 days, median 2 days)
and overall, the time to M. tuberculosis culture ranged
from 4 to 28 days (mean 13.6 days, median 13 days).
19 patients (33.3%) were sputum AFB smear positive on
PBS sampling (Fig. 3). Of these patients, 7 were sputum
smear negative and 12 were non-productive pre-bronchos-
copy. 5/57 patients (8.8%) were exclusively sputum AFB
smear positive on PBS sampling alone, i.e. these patients
converted to smear positivity despite being non-productive
of sputum or pre-bronchoscopy sputum AFB smear negative
as well as BAL smear negative. All 5 cultured M. tubercu-
losis by 6 weeks from BAL samples. One of these five
patients was HIV positive.
49 of 57 patients (86.0%) cultured M. tuberculosis within
6 weeks from pre-bronchoscopic sputum samples, BAL
samples, or a combination of the two (Fig. 4). In four
patients, culture samples were contaminated and so could
not be examined but in the remaining 4 patients (7.0%) M.
tuberculosis was exclusively cultured from PBS samples
within 6 weeks with negative pre-bronchoscopy sputum and
Number of pre-bronchoscopy sputum samples sent per patient.
Timeline of PBS samples collected.
1728 P.M. George et al.
BAL culture samples. Three of these four patients were
infected with HIV. A summary of these results can be found
in Fig. 5.
This study demonstrates that sampling sputum post-
bronchoscopy for AFB smear and M. tuberculosis culture
can provide an additional diagnostic tool which has to date
been underutilized in the diagnosis of PTB. 7.0% of patients
in our cohort were exclusively PBS culture positive for M.
tuberculosis and sputum sampling after bronchoscopy
provided the only means of establishing the diagnosis using
the gold standard. These data show that in this subset of
sampling for M. tuberculosis culture had a yield of 75.4%
(43 of the 57 patients enrolled into the study were PBS M.
tuberculosis culture positive). The culture yield from BAL
was 73.7%. 5 patients (8.8%) were exclusively AFB smear
positive on PBS sampling alone. Of these, 1 patient was HIV
infected and in immunosuppressed patients, where the
differential diagnosis is wide, a positive AFB smear sample
is a finding of key clinical significance. In all 5 patients, the
diagnosis was subsequently confirmed with a positive M.
tuberculosis culture from the BAL sample but an early
positive microscopy finding can allow the rapid initiation of
treatment when there is diagnostic uncertainty.
PBS sampling may not only provide an exclusive means of
achieving a rapid diagnosis of pulmonary TB through AFB
smear and microscopy and increase the yield of M. tuber-
culosis culture, but also reveal a change inpatients’ infec-
tive status. These data demonstrate that bronchoscopy
converted 33.3% of non-infectious patients (non-productive
or sputum AFB smear negative) to a higher infective status
post-procedure (productive of sputum and AFB smear
positive), requiring respiratory isolation. This finding has
significant public health and infection control implications.
There is currently no guidance as to how to manage such
patients, but where the pre-test probability of PTB is high,
and if the patient is still coughing, we advocate respiratory
isolation post-bronchoscopy until confirmation of at least
one further negative acid-fast smear sample. We are unable
to comment as to whether subsequent PBS samples could
produce even further yield after an initial negative PBS
sample and suggest that prospective studies could be
designed to specifically answer this question.
The proportion of patients with HIV co-infection in this
specific smear negative PTB cohort was 26% which is high
when considered against reported rates in London of
17e25% that include the total burden of pulmonary and
extrapulmonary TB.16This is in keeping with the growing
body of evidence that HIV associated PTB is often difficult
to confirm reflecting a lack of sputum production and AFB
smear negativity. Bronchoscopy and post-bronchoscopy
sputum sampling provided the means of culture proven
diagnoses in this group and should be considered particu-
larly when HIV infection exists. It is especially worth noting
that of the 4 patients in whom the only positive culture was
from the PBS sample, 3 were HIV positive. In the setting of
HIV the results of our study have further implications
beyond merely proving PTB. It has been estimated that 5%
of the total number of new TB cases worldwide are due to
multi-drug resistant (MDR) strains and in some settings MDR-
TB has also been shown to be almost twice as common in
HIV infected patients.17As the frequency of MDR-TB rises,
increased efforts should be made to recover a culture
isolate to determine drug susceptibilities and this data
suggests that PBS sampling can provide a simple and
The mechanism of conversion to AFB smear and M.
tuberculosis culture positivity post-bronchoscopy is yet to
be clearly elucidated. Targeted bronchoalveolar lavage
directed by radiological appearances could cause mobili-
zation of deep-seated bacillary-laden secretions, which are
then expectorated in the following days. The longest time
to a positive PBS sample was seven days post-procedure and
so it would appear unlikely that airway irritation and
increased cough frequency caused by the bronchoscope
itself, as is the case with induced sputum sampling, could
solely explain the phenomenon. We made an estimate of
sputum bacillary load using days to culture as a surrogate in
an attempt to identify subgroups of patients in whom there
was a greater chance of conversion to smear positivity post-
procedure. This did not differ significantly in patients
infected with HIV nor those that were smear negative
There are several recognized limitations to our observa-
any case note analysis. Stratification for confounding vari-
ables would be difficult with a cohort of this size and so any
conclusions drawn would need to be handled with care. The
HIV status of a small number of patients in this study is
unknown and although it is our practice to establish the
retroviral status of every patient attending our services, the
Summary of AFB smear results.
Summary of M. tuberculosis culture results.
Post-bronchoscopy sputum in TB M. tuberculosis
uptake of an offer of a test was variable. As a result, we may
have underestimated the burden of HIV infection in this
group of patients but to our knowledge, no patient was
subsequently diagnosed and our data is in keeping with prior
prevalence reports of co-infection.16PTB is commonly
factor may be that individuals most likely to have undergone
who may have more severe or complex disease. Some of the
patients attending our respiratory unit proceeded to bron-
choscopy before three negative sputa were obtained. When
tuberculous disease is suspected, we routinely advocate
obtaining three negative sputum samples prior to invasive
testingifthe patientisclinicallystable andintheoutpatient
setting. However given the diverse nature of the patient mix
in our unit where TB is only one of a number of differential
diagnoses, where HIV co-infection is common and where
acutely unwell inpatients are a significant clinical issue, it is
at times inappropriate to delay proceeding to the most
sensitive diagnostic test and hence delay treatment initia-
tion. We recognize that this practice may limit the gener-
alized applicability of the yield of PBS when 3 samples are
taken. In addition the results of the study can only be
generalized to centers that utilize bronchoscopy as a means
of increasing the microbiological yield in smear negative
PTB. We acknowledge that bronchoscopy is not a risk free
procedure requiring expertise and technical equipment that
may not be readily available in all settings. Many centers
perform sputum induction in preference and in this context
the results of this study may not be applicable. Not with-
standing this, our findings demonstrate the additional chal-
lenges that are presented in smear negative PTB particularly
when HIV co-infection exists and illustrate the additional
diagnostic potential that bronchoscopy may yield post-
procedure in a simple and clinically applicable manner.
In conclusion, sampling sputum post-bronchoscopy can
provide a previously underutilized method of making
a rapid diagnosis of PTB and reduce the number of patients
who are treated on an empiric basis, particularly in the
context of sputum smear negative or non-productive
disease. Importantly it can increase culture yield by up to
7% hence allowing a greater proportion to have full drug
sensitivity testing and therefore appropriate management
of potential drug resistant strains. In our study this added
culture sensitivity was in fact found largely in HIV positive
individuals, a group already recognized to have significantly
higher drug resistant rates. PBS sampling is also a poten-
tially key infection control issue that should be considered
following bronchoscopy. Further studies are now required
to establishthe duration
bronchoscopy inpatients who were previously considered
non-infectious but in the light of this data, we consider it
best practice to only de-isolate such patients when their
infective status can be ascertained with at least one post-
bronchoscopy sputum sample.
of smear positivitypost-
Conflict of interest
We declare that we have no conflict of interest.
OMK designed the study, advised regarding data collection
and edited the manuscript.
PMG helped design the study, coordinated data collec-
tion, analyzed the data and wrote the manuscript
MM and JD collected the data, analyzed the data and
wrote the manuscript
AS analyzed the data and edited the manuscript
CR and YS collected and analyzed the data
PM, DC, and MW were involved in data collection and
edited the manuscript
AJ provided microbiology records for all the patients and
critically read the manuscript
Summary of PBS results.
1730 P.M. George et al.