Predicting smear negative pulmonary tuberculosis with classification trees and logistic regression: a cross-sectional study
2006

Predicting Smear Negative Pulmonary Tuberculosis

Sample size: 551 publication Evidence: moderate

Author Information

Author(s): Mello Fernanda Carvalho de Queiroz, Bastos Luiz Gustavo do Valle, Soares Sérgio Luiz Machado, Rezende Valéria MC, Conde Marcus Barreto, Chaisson Richard E, Kritski Afrânio Lineu, Ruffino-Netto Antonio, Werneck Guilherme Loureiro

Primary Institution: Clementino Fraga Filho Hospital, Federal University of Rio de Janeiro

Hypothesis

Can predictive models be developed for smear negative pulmonary tuberculosis (SNPT) among outpatients in resource-limited settings?

Conclusion

The study suggests that predictive models for SNPT could be useful as screening tools to optimize the use of more expensive tests and avoid unnecessary treatments.

Supporting Evidence

  • The study enrolled 551 patients with clinical suspicion of SNPT.
  • Sensitivity of the predictive models ranged from 64% to 71%.
  • Specificity ranged from 58% to 76%.
  • Models were validated using a separate sample of patients.

Takeaway

Doctors can use simple tests to guess if someone has a type of tuberculosis that doesn't show up in regular tests, which can help them decide if they need to do more expensive tests.

Methodology

The study used logistic regression and classification trees to develop predictive models based on symptoms, physical signs, and chest X-rays from 551 patients.

Limitations

The absence of the tuberculin skin test and incomplete HIV testing in some patients.

Participant Demographics

The mean age of patients was 40 years, with 91% being 60 years or younger.

Statistical Information

P-Value

<0.01

Confidence Interval

95% CI provided for various results

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2458-6-43

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication