| dc.description |
ABSTRACT
INTRODUCTION: To facilitate the detection of sputum smear-negative tuberculosis (TB) in resource-limited settings, the Tuberculosis Probability Scoring (TPS) chart has been developed recently, as part of a manual entitled “Handbook for quality improvement of Chest X-ray reading in adult TB suspects” [Magis, et al., under review for publication]. The scoring system of the chart is based on clinical and chest X-ray (CXR) features that were identified by several valid studies to be predictive for TB. Although its validity needs to be confirmed, the TPS chart might be a promising tool for improving TB case-detection rates, a key strategy in global TB control.
Study objective: To assess the validity and feasibility of the TPS chart for diagnosing sputum smear-negative TB.
Methods: A cross-sectional study conducted between November, 2009 and May, 2010 in healthcare institutions of Kilimanjaro Region, Tanzania. Clinical data from smear-negative TB suspects were collected and their CXRs were evaluated in a structured way in accordance with the TPS chart. Each suspect was classified as ‘probable TB’, ‘possible TB’ or ‘TB unlikely’ based on the TPS chart score. The TPS chart classifications were compared with sputum culture results (the gold standard test for diagnosing TB).
A standardized questionnaire, filled by Regional/District TB and leprosy co-ordinators (R/DTLCs) dealing with TB patients, was used to assess the feasibility of using the TPS chart in clinical settings of the Tanzania’s Kilimanjaro Region. The findings from the questionnaire were evaluated in a focus group discussion that was carried out among the same respondents.
Results: A total of 153 smear-negative TB suspects were enrolled and from 130 of them complete data were available for analysis. Using the pre-defined cut-off values (6 and 12) of the TPS chart, 68.5% (89/130) of study participants were classified as “probable TB”, 28.5% (37/130) as “possible TB” and 3.0% (4/130) as “TB unlikely”. Among those categorized as possible TB (a group requiring follow-up evaluation) 37.8% (14/37) had a positive sputum culture result. The chart is highly sensitive (98.3%) to predict TB among smear-negative TB suspects at cut-off value of 6 with very low specificity (4.2%). The sensitivity and specificity at a cut-off value of 12 are 74.6 and 36.6%, respectively.
The clinical information required to fill the TPS chart is readily available from patients. However, it is difficult for R/DTLCs to interpret CXRs of smear-negative TB suspects in accordance with TPS chart. The cost burden, especially CXR services and blood sugar tests cost, of TPS-based smear-negative TB diagnosis is a concern as discussed by R/DTLCs.
Discussion: according to TPS chart validity study findings, the chart is sensitive enough (98.3%) at a cut-off value of 6 to identify those unlikely to have TB. However, the specificity of the chart is very low at both cut-off values (4.2% and 36.6% at 6 and 12 cut-off values, respectively) which may result in massive misclassification of those without the disease as having smear-negative TB and unnecessary initiation of TB treatment.
The chart appears to be suitable to implement in Tanzania’s Kilimanjaro Region and similar other settings if it is revised and re-evaluated to make it valid. The major concern in this setting is the difficulty to interpret CXRs of sputum smear-negative TB suspects, especially in accordance with the TPS chart.
Conclusion: The TPS chart without further revision and adjustments of the scoring system and the cut-off values is not valid to implement in the routine clinical practice. Provided that the chart will be valid after revision and re-evaluation, training on how to interpret CXRs of sputum smear-negative TB suspects in accordance with the TPS chart and developing a handbook describing common CXR features of TB is necessary to implement the chart in resource-limited settings. In addition, the cost (esp. CXR services and blood sugar test) of TPS-based smear-negative TB diagnosis is a concern that may limit the availability of the required information to score the chart.
Recommendation: Further revision and adjustments of the scoring system and the cut-off values is needed to make it valid in resource-limited TB high burden settings like Tanzania’s Kilimanjaro Region. Inputs from TB experts having experience in resource-limited TB/HIV high burden settings could be of help in revising the chart.
Provided that the chart will be valid after the above efforts, training on how to interpret CXRs of TB suspects is mandatory to make the chart helpful in the routine clinical practice. TB control program policy makers should consider reducing the indirect cost of TB diagnosis to the poor. |
|