Objective: To determine diagnostic criteria for tuberculosis among sputum s
mear acid fast bacilli negative patients: with chronic cough, based on symp
toms, signs and simple laboratory tests,
Design: A two-month prospective follow up study.
Setting: Muhimbili Medical Centre, Dar es Salaam, Tanzania,
Subjects: One hundred and seventy eight consecutive patients admitted betwe
en Ist November, 1994 and 31st March, 1995 with chronic productive cough.
Main outcome measures: Sputum smear acid fast bacilli (AAFB) negative with
TB, discriminating ability of clinical and laboratory parameters,
Results: Forty three (24.2%) were sputum smear positive for acid fast bacil
li (AAFB), In 90 (50.6%) patients, AAFB could be isolated in specimen other
than sputum and in 45 (25.3%) no AAFB could be isolated. In a univariate a
nalysis of all symptoms, signs and laboratory test results, cough of four o
r more weeks, haemoptysis, oral candidiasis, chest consolidation, pleural e
ffusion, mid zone and upper zone chest x-ray opacities were significantly d
ifferent between sputum AAFB negative TB and non TB patients. Discriminant
analysis revealed six highly significant variables: Mantoux reaction, pleur
al effusion, Kaposl's lesion, cervical lymphadenopathy, matted lymph node,
mid zone and upper zone CXR infiltrates. From these variables an equation w
as derived to calculate the probability that a sputum smear negative patien
t had tuberculosis, Then a scoring system was developed that classified cor
rectly 84% of cases of sputum AAFB negative patients into tuberculosis or n
on tuberculosis.
Conclusion: In sputum smear AAFB negative patient clinically suspected to h
ave tuberculosis, Mantoux reaction, cervical lymphadenopathy, matted Lymph
nodes, absence of mid lower zone infiltration on CXR and presence of pleura
l effusion could be used for presumptive diagnosis, but they would not make
a therapeutic trial unnecessary.