| Summary: | A research report submitted to the Faculty of Health Sciences,
School of Public Health,
University of the Witwatersrand, Johannesburg
In partial fulfillment for the requirement for the degree
Master of Public Health
25 August 2014 === BACKGROUND
Tuberculosis is the leading opportunistic infection among HIV-infected adults,
including pregnant women, globally. Accurate screening tools are needed to
identify those requiring further laboratory testing and to initiate isoniazid
preventive therapy in a timely manner. This study determined the accuracy of
symptom-based screening and in particular the performance of the WHO
recommended TB symptom screening algorithm in HIV-infected pregnant
women.
METHODS
A cross-sectional study was conducted among consenting HIV-infected
pregnant women attending routine antenatal clinics in Matlosana, South Africa
recruited >1 week after first HIV diagnosis between June 2010 and February
2011. Sputum was collected from all women followed by a systematic TB
symptom screen. The performances of each symptom (cough, fever, weight loss
and night sweats) alone and in combination were assessed with TB confirmed
by sputa using microscopy and liquid culture (MGIT), as reference or gold
standard. The sensitivity, specificity, positive predictive value, negative
predictive value, positive likelihood ratio and negative likelihood ratio were
calculated for each of the four symptoms (cough, fever, weight loss and night
sweats) and their combination. Logistic regression was carried out to find
associations between patient characteristics and TB.
RESULTS
Overall, Mycobacteria Growth Indicator Tube (MGIT) confirmed prevalence of
TB was 2.4% (35/1456) in this sample group. Only 11/38 (29%) women with
confirmed TB reported any symptoms. Cough, fever, weight-loss and night
sweats, individually and in combination had sensitivities ranging from 2.7-27%
and specificities ranging from 84-97%. The positive predictive and negative
predictive values for any symptoms of cough, fever, night sweats, or weight loss
were 4.2% and 98%, respectively. TB was associated with decreasing CD4 count,
close TB contact, cough, and night sweats.
DISCUSSION
The remarkable number of asymptomatic TB in HIV-infected patients, including
in the cohort included in this study highlights the limitation of symptom-based
screening. The low sensitivity of the symptom screen would incorrectly stratify
patients who are being considered for Isoniazid Preventive Therapy (IPT).
However, one could argue that the high negative predictive value of the
symptom screen would justify its use in resource-limited settings as the initial
step in identifying patients who should receive IPT. Although household TB and
the father of the baby having TB were found not to have statistically significant
associations with active TB, they are of public health importance as they play a
role in the spread of the infection.
CONCLUSION
The WHO 4-symptom screen had low sensitive among HIV-infected pregnant
women but negative predictive value was high. Few women with TB disease
reported symptoms on direct questioning; the high rate of subclinical/
asymptomatic TB is concerning. There is an urgent need for more
sensitive screening tools for TB in HIV-infected pregnant women
|
|---|
