Review
Sputum monitoring during tuberculosis treatment for predicting outcome: systematic review and meta-analysis

https://doi.org/10.1016/S1473-3099(10)70071-2Get rights and content

Summary

WHO has previously recommended sputum-smear examination at the end of the second month of treatment in patients with recently diagnosed pulmonary tuberculosis, and, if positive, extension of the intensive therapy phase. We did a systematic review and meta-analysis to assess the accuracy of a positive sputum smear or culture during treatment for predicting failure or relapse in pulmonary tuberculosis. We searched PubMed, Embase, and the Cochrane Library for studies published in English through December, 2009. We included randomised controlled trials, cohort, and case-control studies of previously untreated pulmonary tuberculosis patients who had received a standardised regimen with rifampicin in the initial phase. Accuracy results were summarised in forest plots and pooled by use of a hierarchical regression approach. 15 papers (28 studies) met the inclusion criteria. The pooled sensitivities for both 2-month smear (24% [95% CI 12–42%], six studies) and culture (40% [95% CI 25–56%], four studies) to predict relapse were low. Corresponding specificities (85% [95% CI 72–90%] and 85% [95% CI 77–91%]) were higher, but modest. For failure, 2-month smear (seven studies) had low sensitivity (57% [95% CI 41–73%]) and higher, although modest, specificity (81% [95% CI 72–87%]). Both sputum-smear microscopy and mycobacterial culture during tuberculosis treatment have low sensitivity and modest specificity for predicting failure and relapse. Although we pooled a diverse group of patients, the individual studies had similar performance characteristics. Better predictive markers are needed.

Introduction

Tuberculosis is a major global health problem with 9 million new cases and almost 2 million deaths per year.1 WHO recommends that patients with previously untreated pulmonary tuberculosis receive a four-drug regimen during the 2-month initial phase of treatment that includes rifampicin.2 The overall rate of failure or relapse (poor outcome) in patients receiving short-course directly observed treatment with a rifampicin-containing regimen is low.3, 4, 5 In patients receiving 6-month regimens who have drug-susceptible organisms, the estimated failure rate is 1–4% and the relapse rate 7% or less.3 Relapse of tuberculosis continues to place a substantial burden on patients and tuberculosis control programmes; in 2007, at least 270 000 patients worldwide returned after relapse (5% of tuberculosis notifications).1 Early identification of patients who have an increased risk of a poor outcome, coupled with an intervention such as treatment modification, could potentially reduce this burden.

Previous treatment guidelines have recommended the examination of a sputum smear at the end of the 2-month initial phase of treatment and, if the smear is positive, to continue the initial phase for an extra month before proceeding with the standard 4-month or 6-month continuation phase.6 These recommendations, designed for low-resource settings that lack the ability to do culture or drug-susceptibility testing, have been questioned due to lack of evidence, and the ability of a positive sputum examination result to predict poor outcome has not been fully assessed.7, 8, 9, 10, 11, 12 A recently published review on tuberculosis biomarkers identified an inverse relation between 2-month sputum-culture conversion and relapse, although the use of 2-month culture conversion as a guide to treatment of individual patients was limited by poor positive predictive value (PPV).13 However, the review was not a systematic review of all available evidence.

To estimate the accuracy of a positive sputum smear or culture for predicting poor outcome in patients with pulmonary tuberculosis who received a standardised regimen that included rifampicin in the initial phase, we did a systematic review and meta-analysis. In addition to estimating the accuracy of the sputum examination result, the quality of studies was appraised. Results from this systematic review were used in the development of the most recent WHO tuberculosis treatment guidelines.2

Section snippets

Methods

We used standard methods for systematic reviews of diagnostic accuracy studies.14, 15, 16, 17

Results

The searches of published work identified more than 12 000 citations, from which 15 papers (28 studies) were selected (figure 1).7, 9, 12, 20, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38 Table 1 summarises study quality, and Table 2, Table 3 show characteristics of individual studies by relapse (18 studies) or treatment failure (ten studies), respectively. 23 (82%) studies were done in low-income countries. Most studies assessed the examination of a sputum sample at month 2, consistent with

Discussion

We did a systematic review and meta-analysis of the accuracy of sputum examination during treatment to identify pulmonary tuberculosis patients who will fail treatment or experience relapse. We found low sensitivity and moderate specificity for prediction of relapse or failure in all studies, irrespective of sputum examination or time of assessment. These results were similar for individual studies and pooled analyses.

If an individual's status is unknown, predictive values are used to estimate

Search strategy and selection criteria

These are detailed in the Methods section.

References (50)

  • Treatment of tuberculosis: guidelines for national programmes (3rd edn) [WHO/CDS/TB/2003.313]

  • D Wilkinson et al.

    Should we take a history of prior treatment, and check sputum status at 2–3 months when treating patients for tuberculosis?

    Int J Tuberc Lung Dis

    (1998)
  • DA Enarson et al.

    Appropriateness of extending the intensive phase of treatment based on smear results

    Int J Tuberc Lung Dis

    (2004)
  • D Benator et al.

    Rifapentine and isoniazid once a week versus rifampicin and isoniazid twice a week for treatment of drug-susceptible pulmonary tuberculosis in HIV-negative patients: a randomised clinical trial

    Lancet

    (2002)
  • DA Mitchison

    Shortening the treatment of tuberculosis

    Nat Biotechnol

    (2005)
  • FZ Zhao et al.

    Sputum microscopy results at two and three months predict outcome of tuberculosis treatment

    Int J Tuberc Lung Dis

    (1997)
  • WL Deville et al.

    Conducting systematic reviews of diagnostic studies: didactic guidelines

    BMC Med Res Methodol

    (2002)
  • C Gatsonis et al.

    Meta-analysis of diagnostic and screening test accuracy evaluations: methodologic primer

    AJR Am J Roentgenol

    (2006)
  • M Pai et al.

    Systematic reviews of diagnostic test evaluations: what's behind the scenes?

    ACP J Club

    (2004)
  • MM Leeflang et al.

    Systematic reviews of diagnostic test accuracy

    Ann Intern Med

    (2008)
  • WN Rom et al.

    Tuberculosis

    (2004)
  • JP Cao et al.

    Two-year follow-up of directly-observed intermittent regimens for smear-positive pulmonary tuberculosis in China

    Int J Tuberc Lung Dis

    (1998)
  • P Whiting et al.

    The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews

    BMC Med Res Methodol

    (2003)
  • GH Guyatt et al.

    GRADE: an emerging consensus on rating quality of evidence and strength of recommendations

    BMJ

    (2008)
  • CM Rutter et al.

    A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations

    Stat Med

    (2001)
  • Cited by (207)

    • Molecular Imaging of Tuberculosis

      2023, Seminars in Nuclear Medicine
    • Transcriptional profiles predict treatment outcome in patients with tuberculosis and diabetes at diagnosis and at two weeks after initiation of anti-tuberculosis treatment

      2022, eBioMedicine
      Citation Excerpt :

      The mechanisms underlying DM-induced TB treatment failure remain, however, poorly understood. Prediction of TB treatment failure based on sputum-smear microscopy and mycobacterial culture lacks sensitivity10 and depends on the quality of sputum samples, which are difficult to collect and are frequently inconsistent in quality.11–13 As well as more advanced sputum-based diagnostics, monitoring of whole blood transcriptomics may be an additional, complementary but independent, method to monitor treatment responses, possibly with increased sensitivity.14

    View all citing articles on Scopus
    View full text