Validation of a severity-of-illness score in patients with tuberculosis requiring intensive care unit admission

Abstract

Background. There is a paucity of data on the determinants of mortality due to tuberculosis (TB) in the intensive care unit (ICU). Objective. To develop a simple severity-of-illness score for use in patients with TB admitted to an ICU.

Methods. A scoring system was generated by retrospectively identifying the four most significant and clinically unrelated predictors of mortality from an existing prospectively collected dataset (January 2012 - May 2013), and combining these with known predictors of poor outcome.

Results. Of 83 patients admitted with TB, 38 (45.8%) died in the ICU. The four parameters identified from the retrospective analysis were: (i) HIV co-infection with a CD4 cell count <200/μL; (ii) a raised creatinine level: (iii) a chest radiograph showing diffuse parenchymal infiltrates/miliary pattern; and (iv) absence of TB treatment on admission. These were combined with septic shock and a low arterial partial pressure of oxygen/fractional inspired oxygen (P:F) ratio to generate a six-point severity-of-illness score (one point for each parameter). The scores for survivors were significantly lower than those for non-survivors (mean (standard deviation) 2.27 (1.47) v. 3.58 (1.08); p<0.01). A score of ≥2 was associated with significantly higher mortality than a score of <2 (7.1% v. 46.4%; odds ratio (OR) 15.03; 95% confidence interval (CI) 1.86 - 121.32; p<0.01), whereas a score of ≥3 was associated with a significantly higher mortality than a score of <3 (64.6% v. 20.0%; OR 7.29; 95% CI 2.64 - 20.18; p<0.01).

Conclusion. The proposed scoring system identified patients at increased risk of dying from TB in the ICU. Further prospective studies are indicated to validate its use.