STANDARD 2.

All patients (adults, adolescents, and children who are capable of producing
sputum) suspected of having pulmonary tuberculosis should have at least two,
and preferably three, sputum specimens obtained for microscopic examination.
When possible, at least one early morning specimen should be obtained.

Rationale and Evidence Summary

To prove a diagnosis of tuberculosis, every effort must be made to identify the causative
agent of the disease. A microbiological diagnosis can only be confi rmed by culturing M.
tuberculosis complex (or, under appropriate circumstances, identifying specifi c nucleic
acid sequences in a clinical specimen) from any suspected site of disease. In practice, however, there are many resource-limited settings in which culture
is not feasible currently. Fortunately, microscopic examination
of stained sputum is feasible in nearly all settings, and the
diagnosis of tuberculosis can be strongly inferred by fi nding
acid-fast bacilli by microscopic examination. In nearly all clinical
circumstances in high-prevalence areas, fi nding acid-fast
bacilli in stained sputum is highly specifi c and, thus, is the
equivalent of a confi rmed diagnosis. In addition to being highly
specifi c for M. tuberculosis complex, identifi cation of acid-fast
bacilli by microscopic examination is particularly important for
three reasons: it is the most rapid method for determining if a
person has tuberculosis; it identifi es persons who are at greatest
risk of dying from the disease*; and it identifi es the most likely transmitters of infection.
Generally, it is the responsibility of government health systems (national tuberculosis programs
[NTPs] or others) to ensure that providers and patients have convenient access
to microscopy laboratories. Moreover, it is crucial that such laboratories undergo assessments
of quality and have programs for quality improvement. These quality assessments
are generally the responsibility of a government system (usually the NTP).

Failure to perform a proper diagnostic evaluation before initiating treatment potentially
exposes the patient to the risks of unnecessary or wrong treatment with no benefi t.
Moreover, such an approach may delay accurate diagnosis and proper treatment. This
Standard applies to adults, adolescents, and children. With proper instruction and supervision,
many children 5 years of age and older can generate a specimen. Adolescents,
although often classifi ed as children at least until the age of 15 years, can generally produce
sputum. Thus, age alone is not suffi cient justifi cation for failing to attempt to obtain
a sputum specimen from a child or adolescent.

The information summarized below describes the results of various approaches to sputum
collection, processing, and examination. The application of the information to actual
practices and policies should be guided by local considerations.
The optimum number of sputum specimens to establish a diagnosis has been examined
in a number of studies. In a recent review of data from a number of sources, it was stated
that, on average, the initial specimen was positive in about 83–87% of all patients ultimately
found to have acid-fast bacilli detected, in an additional 10–12% with the second
specimen, and in a further 3–5% on the third specimen.34 A rigorously conducted systematic
review of 41 studies on this topic found a very similar distribution of results: on
average, the second smear detected about 13% of smear-positive cases, and the third
smear detected 4% of all smear-positive cases.35 In studies that used culture as the reference
standard, the mean incremental yield in sensitivity of the second smear was 9% and
that of the third smear was 4%.35

A recent re-analysis of data from a study involving 42 laboratories in four high-burden
countries showed that the incremental yield from a third sequential smear ranged from
0.7–7.2%.36 Thus, it appears that in a diagnostic evaluation for tuberculosis, at least two
specimens should be obtained. In some settings, because of practicality and logistics, a
third specimen may be useful, but examination of more than three specimens adds minimally
to the number of positive specimens obtained.35 In addition, a third specimen is useful
as confi rmatory evidence if only one of the fi rst two smears has a positive result. Ideally,
the results of sputum microscopy should be returned to the clinician within no more than
one working day from submission of the specimen. The timing of specimen collection is
also important. The yield appears to be greatest from early morning (overnight) specimens.
35,37–39 Thus, although it is not practical to collect only early morning specimens, at
least one specimen should be obtained from an early morning collection.

A variety of methods have been used to improve the performance of sputum smear microscopy.
40–42 In general, the sensitivity of microscopy (as compared to culture) is higher
with concentration by centrifugation and/or sedimentation (usually after pretreatment with
chemicals such as bleach, NaOH, and NaLC) or both, as compared to direct (unconcentrated)
smear microscopy. A comprehensive, systematic review of 83 studies describing
the effects of various physical and/or chemical methods for concentrating and processing
sputum prior to microscopy found that concentration resulted in a higher sensitivity (15–
20% increase) and smear-positivity rate, when compared with direct smears.40 Although
there are demonstrable advantages to concentration of sputum, there are also disadvantages.
Centrifugation is more complex, requires electrical power, and may be associated
with increased infection risk to laboratory personnel. Consequently, it is not clear that the
advantages offset the disadvantages in low-resource settings.

Fluorescence microscopy, in which auramine-based staining causes the acid-fast bacilli
to fl uoresce against a dark background, is widely used in many parts of the world. A
systematic review, in which the performance of direct sputum smear microscopy using
fl uorescence staining was compared with Ziehl-Neelsen (ZN) staining using culture as the
gold standard, suggests that fl uorescence microscopy is the more sensitive method.41
The results of this review have been verifi ed in a more comprehensive, systematic review
of 43 studies. This review showed that fl uorescence microscopy is on average 10% more
sensitive than conventional light microscopy.42 The specifi city of fl uorescence microscopy
was comparable to Ziehl-Neelsen staining. The combination of increased sensitivity with
little or no loss of specifi city makes fl uorescence microscopy a more accurate test, although
the increased cost and complexity might make it less applicable in many areas.
For this reason, fl uorescence staining is probably best used in centers with specifi cally
trained and profi cient microscopists, in which a large number of specimens are processed
daily, and in which there is an appropriate quality control program.

STANDARD 3.

For all patients (adults, adolescents, and children) suspected of having extrapulmonary
tuberculosis, appropriate specimens from the suspected sites of involvement
should be obtained for microscopy and, where facilities and resources are
available, for culture and histopathological examination.

Rationale and Evidence Summary

Extrapulmonary tuberculosis (without associated lung involvement) accounts
for 15–20% of tuberculosis in populations with a low prevalence
of HIV infection. In populations with a high prevalence of HIV infection,
the proportion of cases with extrapulmonary tuberculosis is higher. Because
appropriate specimens may be diffi cult to obtain from some of
these sites, bacteriological confi rmation of extrapulmonary tuberculosis
is often more diffi cult than for pulmonary tuberculosis. In spite of the diffi
culties, however, the basic principle that bacteriological confi rmation
of the diagnosis should be sought still holds. Generally, there are fewer
M. tuberculosis organisms present in extrapulmonary sites, so identifi -
cation of acid-fast bacilli by microscopy in specimens from these sites
is less frequent and culture is more important. For example, microscopic
examination of pleural fl uid in tuberculous pleuritis detects acid-fast bacilli
in only about 5–10% of cases, and the diagnostic yield is similarly low in tuberculous
meningitis. Given the low yield of microscopy, both culture and histopathological examination
of tissue specimens, such as may be obtained by needle biopsy of lymph nodes,
are important diagnostic tests. In addition to the collection of specimens from the sites
of suspected tuberculosis, examination of sputum and a chest fi lm may also be useful,
especially in patients with HIV infection, in whom there is an appreciable frequency of
subclinical pulmonary tuberculosis.43

In patients who have an illness compatible with tuberculosis that is severe or progressing
rapidly, initiation of treatment should not be delayed pending the results of microbiological
examinations. Treatment should be started while awaiting results and then modifi ed, if
necessary, based on the microbiological findings.