The Pediatric Tuberculosis (TB) Service of the Children’s
Hospital of Winnipeg assesses 500 children each year who are
identified through contact investigations to have had exposure
to adults with infectious Mycobacterium tuberculosis (MTB)
pulmonary disease. Young children are at higher risk for
developing TB disease after exposure to an infectious case.
The recognition of tuberculosis (TB) disease in children is
challenging. Children are less likely to have obvious symptoms
of disease, so the diagnosis is based mostly on epidemiological
information (contact history), immunologic evidence (tuberculin
skin test) and chest radiograph (CXR) changes. The gold standard
for TB diagnosis is the detection of MTB by culture or molecular
Older patients can submit spontaneously expectorated
sputum or liquefied airway secretions after the inhalation of hypertonic saline solution (induced sputum) , for laboratory studies.
Bacteriological confirmation of TB in pediatric patients
is difficult because respiratory secretions for microbiological
evidence are difficult to obtain (especially in young children) and
pediatric disease is often paucibacillary.
When expectorated sputum cannot be obtained for suspected
respiratory disease in a young child, aspirated stomach contents
via gastric aspiration (GA) or gastric lavage is taken for acid fast
bacilli (AFB) smear and Mycobacterium tuberculosis (MTB)
culture . The MTB culture yield from gastric aspirates ranges
from 10-30% in these asymptomatic children .
It has been shown that multiple specimens taken in one
day resulted in smear microscopy yield equivalent to specimen
obtained over consecutive days .
We present our MTB culture results obtained from GAs
done daily (2013-2015) versus multiple samples (2016-2017)
obtained in one day – either at intervals more than two hours
apart (MSx1) or hourly (Q1H).
From 2001-2008, the Pediatric TB Service relied on trained
Victorian Order of Nurses to perform GA for young children with
suspected tuberculosis. This was done early in the morning,
daily for 3 days, in their homes or in a boarding home using a
standardized method for collection. Due to the dwindling number
of experienced staff in 2008, it then became necessary to admit
children to the hospital for 3 days to obtain early morning GA
specimen. However, frequently limited in-patient bed availability
forced the service to look at other options. We collaborated
with the Pediatric Day Unit (PDU) of the Children’s Hospital for
a few years for daily out-patient morning GA. With increasing difficulty coordinating schedules with the PDU, we started
admitting young patients to hospital to obtain GAs in early 2016.
Limited bed availability, especially during respiratory syncytial
virus peak season forced the service to be creative- hence the
development of hourly GA sampling. This allowed us to admit the
child for a fasting period overnight, obtain 3 hourly specimens
in the morning, start treatment after the procedure and send the
patient home soon after.
We tabulated and described demographic, epidemiologic,
radiographic and mycobacteriology data obtained on the patients
admitted to the Children’s Hospital of Winnipeg for GA (2013-
2017). We compared the rates of positive MTB cultures from
samples taken daily for 3 days (ODx3), to the yield obtained from
MSx1 and Q1H. Statistical analysis was performed. The University
of Manitoba Health Research Ethics Board approved this study.
Fifty-seven patients who were admitted to the Children’s
Hospital of Winnipeg for GA MTB studies between January 1,
2013 to December 31, 2019. There were 31 males and 26 females,
whose ages ranged from 3-months to 16 years (mean 41 months; median 33 months). All patients had never been treated for TB in
the past and all children tested negative for HIV (Figure 1).
Of the 57 children, there were 6 patients (3M: 3F) who were
admitted to hospital with acute symptoms -– ataxia, emesis,
dehydration, fever; headache, seizure, difficulty speaking,
low energy; shortness of breath, cough; lower limb tingling
sensation; pneumonia. Ages ranged from 7 months to 16 years
(mean - 7 years; median - 5 years). None of these children had
known exposure history or TB contact information. Only 2 of
the 6 children had PPD done, both were >5mm indurated. For
imaging, 1 patient had normal CXR findings, while the other 5
had abnormal CXR descriptions – 1 with miliary pattern (and
R paratracheal lymphadenopathy) and 4 with pneumonia., The
child with the normal CXR had significantly abnormal MRI central
nervous system (CNS) imaging suggestive of tuberculosis, while
the patient with military pattern on CXR had leptomeningeal
enhancement and eye changes on MRI. Of the 4 with chest x-ray
findings of pneumonia, 1 also had paratracheal adenopathy (and
CNS imaging changes), another had mediastinal lymph nodes
(and vertebral bony changes on MRI), and one had a concomitant
pleural effusion (and abnormal CNS imaging).
The other 51 patients were identified through contact
investigation. These patients were all relatively asymptomatic. Ages ranged from 3 months to 8 years 6 months; mean of 36
months; median 32 months.
Twenty-nine children were identified as close household
contacts to an adult diagnosed with respiratory TB. Seventeen
were described as non-household contacts and another 5 children
were exposed to both household and non-household cases.
Thirty-nine children were exposed to one known adult case, 6
were being investigated for having contact with two adult cases,
and 6 had exposure to 3 index cases (from the same household).
Forty-three patients had one established contact with an
MTB index case while 6 had two separate exposures and another
2 had 3 documented occasions of contact with cases.
Forty-two source cases were described as being AFB smear
and MTB culture positive and 6 were culture positive only.
There was 1 source case who was both smear and culture
negative (referred to as a “clinical case”) and for two patients no
information was provided other than the children coming from
an “outbreak community”.
Sixteen adult contacts had cavitary findings on their CXRs and
they were all AFB smear and MTB culture positive.
Fifty patients had PPD skin testing; a 3-month old infant was
not PPD tested. Fourteen had skin test indurations measured
<5mm and 36 were ≥ 5mm.
Lymphadenopathy on CXR was the most commonly seen
radiographic finding, Of the 51 patients, lymphadenopathy was
the only finding reported in 27. In 2 patients, the only abnormal
CXR finding was paratracheal adenopathy, 21 patients had hilar
adenopathy only, while 4 children had enlarged paratracheal
and hilar lymph nodes. Densities, opacities or pneumonia were
described in 24 patients, one of whom also had a pleural effusion.
Nineteen of the patients with parenchymal changes on CXR also
had adenopathy – 14 hilar, 2 paratracheal and 3 with both. Other
imaging included MRI, CT scans, abdominal ultrasound.
Thirty-one children had daily GA done (14M: 17F), 5 MSx1
(4M: 1F) and 21 patients had QH1 (13M: 8F). A total of 169
GA procedures were performed- 93 were obtained ODx3, 13
from MSx1, and 63 by QH1 collection. Patient ages ranged from
3-months old to 14-years old, with mean of 39-months old,
median 32-months old (Figure 2).
Twenty-one males and 10 females had ODx3 GA. All the 93
GA specimens obtained ODx3 were AFB smear negative; the
MTB culture yield was 12%. The number of positive cultures
per patient was 1 of 3 samples in 2 patients, 2 of 3 samples in
3 patients, 3 of 3 samples in 1 patient. Six patients (19%) were
confirmed MTB culture positive using daily GA.
The age of the 5 patients sampled MSx1 ranged from 8 to
81-months old (mean 34 months old, median, 40 months old).
There were 4 male and 1 female patients. A total of 13 GA was
obtained (two patients only had 2 GA samples) and they were
all AFB smear negative. There were positive MTB culture results
in only 3/13 samples (23%). All the three samples that cultured
positive were from the same patient.
In total, 16/57 patients (28%) and 33/169 GA samples
cultured positive for MTB. All culture positive samples were pansensitive.
Non-tuberculous mycobacterium was not isolated in
any of the symptomatic or asymptomatic patients.
For the symptomatic patients, GA MTB culture yield was
positive in 5/6 patients (83%) and in 10/18 samples (56%). Two
of the 6 patients had ODx3 GA sampling, and in both patients the
sample that cultured positive was obtained on the 2nd day. None
of the patients underwent MSx1 GA. Three of 4 children who
had the QH1 GA procedure cultured positive. In these 3 culture
positive patients, all hourly samples isolated MTB except for the
first obtained sample in one patient. The MTB culture negative
(patient had CNS imaging suspicious of a tuberculosis process
(superior sagittal sinus and frontal cortical vein thrombosis), also
did not culture positive for MTB on non-respiratory samples, and
was MTB-PCR negative – so an alternative diagnosis was later
considered (Figure 3).
In the asymptomatic contacts, GA MTB culture yield for was
11/51 patients (22%) and in 23/151 samples (15%). Of the 23
positive cultures, 8 came from the 1st GA taken, 9 from the 2nd and
6 from the 3rd sample. Of the 12 GA taken daily from 4 patients, 9
cultured positive -3 from the 1st day, 4 from day 2 and 2 from day
3. MSx1 sampling – only one patient had confirmed MTB, with all
3 GA specimen culturing positive. Six of the 17 patients who had
QH1 GA cultured positive-4 from the 1st aspirate, 4 from the 2nd
hour GA and 3 from the last hourly specimen.
Seven of the 29 household contacts cultured for MTB on GA,
only 3 of the 17 children who were non-household contacts were
MTB positive, while 1 of the 5 patients with both household and
non-household exposures had GA positive culture. Both children
who were exposed in the community and the patient identified
to have been in contact with a “clinical case” did not culture on
GA for MTB.
Only 1 of the 6 children with two separate exposures at
different times (mostly in the same household) was culture
positive. One of the 2 patients with a history of 3 documented
exposures cultured positive on GA. Of the 43 children with one
identified exposure, 9/43 were positive on GA culture for MTB.
All patients who cultured on GA for MTB were exposed to
smear positive cases. None of the children exposed to the smear
negative, culture positive only cases or clinical cases yielded MTB on GA.
Three of the 11 children who cultured positive on GA were
exposed to adults with cavitary changes on CXR.
One patient who had a <5mm PPD reaction cultured positive on GA for MTB.
Two of the 11 culture positive contacts only had adenopathy
on CXR, while 8 children had both parenchymal densities and
adenopathy and one had radiographic changes of pneumonia.
Household exposure, smear status and culture positivity
of the index case, number of patient contacts and PPD were
associated with MTB culture positivity (Tables 1 and 2).
The percent of positive MTB specimen and percent of patients
who cultured positive for MTB was higher in the QH1 group, as
was the number of positive cultures taken from sample 2 but
there was no statistical significance measured (Tables 3 and 4).
It can never be overemphasized that the diagnosis of TB
disease in a child indicates recent transmission of TB disease from an adult case. And, in order to establish disease in a child
suspected to have tuberculosis, it is important to attempt
collection of respiratory secretions not only to confirm the
diagnosis with MTB culture but also to obtain culture sensitivity
to appropriately direct pharmacologic management.
The conventional procedure for GA is to insert a nasogastric
tube (NGT) in the early morning to obtain respiratory secretions
that were swallowed overnight. This is done on three consecutive
mornings - therefore entails repeated NGT insertion, which can
be and is quite traumatic for young children. Treatment with anti-
TB medications is then started only after the three specimens are
Recent studies in adults have shown non-inferior MTB culture
sensitivity and specificity of hourly sputum sampling method
compared to conventional daily sampling. A metaanalysis , reported similar sensitivity and specificity of sputum microscopy
for hourly same-day sampling and sampling obtained over two
days. They found that hourly sampling provided the additional
advantages of reducing patient drop out, and allowing faster
decision making regarding respiratory isolation and discharge.
The Pediatric TB Service adopted the hourly GA method to
obtain secretions for AFB/MTB studies in 2016 to facilitate
timely treatment and discharge. We also aimed, in line with
the Hospital’s Pain Initiative (“comfort promise”) to develop a
policy that would offer a less frequently traumatic procedure
to children, especially if the culture yield was similar or better
than obtaining GA specimen over a 3-day period. GA is used also
whenever a patient is unable to spontaneously expectorate and/
or sputum induction is unsuccessful, hence the few patients who
are older than 5 years of age.
The procedure can be used also for older children who are
unable to expectorate spontaneously or have decreased levels of
consciousness as was performed in our sicker patients with CNS
It has been reported that up to approximately 50% MTB GA
, culture yield from children with symptomatic TB , so it
was not surprising to see a higher yield of MTB cultures in our
symptomatic and sicker children.
There was an improved/better MTB culture yield from our
group of asymptomatic pediatric TB contact with the hourly GA
procedure compared to a previously reported median GA yield
of 20% , using the standardized daily GA technique . As in
previous reports , adenopathy and disseminated TB seemed to
be clinical predictors of positive yield.
GA remains the standard of care for specimen collection
prior to the initiation of anti-TB treatment , in pediatric
paucibacillary clinical TB disease. We have shown that hourly GA
sampling is a more convenient and less traumatic procedure that
not only can provide comparative culture yield/confirmation of
diagnosis but has the advantage of offering earlier treatment and
hospital discharge of patients and families. The overnight fasting
that is required may be a limitation where access to an inpatient
facility is not easily accessible. Although this paper is limited by
the retrospective and observational nature and the small number
of patients, the results provide useful and valuable information
that may lead to a change in practice for clinicians.