- Mycobacterium tuberculosis bloodstream infection (BSI) in HIV-associated tuberculosis (TB) is independently associated with mortality (4 days.
Why this matters
- Consider adding urine lipoarabinomannan (LAM) and first-line sputum testing in patients with TB and HIV who are hospitalized with >1 WHO danger signs.
- Editorial : TB prevalence may help define appropriate strategy; consider risk/benefits of initiating treatment in all patients, including those with negative rapid TB tests.
- 20 data sets, 5751 individualized patient data (IPD) included.
- Mean predicted M tuberculosis BSI probability (>1 WHO danger signs, CD4 76 cells/µL): 0.45 (95% CI, 0.38-0.52).
- Pooled summary for urine LAM:
- Diagnostic yield: 0.52 (95% CI, 0.35-0.69).
- Sputum: 0.77 (95% CI, 0.63-0.87).
- Composite: 0.89 (95% CI, 0.80-0.94).
- Significant heterogeneity across studies (urine, sputum, P<.0001 both>
- M tuberculosis predicted mortality risk before 30 days (adjusted HR, 2.48; 95% CI, 2.05-3.08).
- Association observed between delayed therapy and early mortality (OR, 3.2; 95% CI, 1.2-8.8).
- IPD metareview exploring prevalence, diagnostic yield, mortality risk, and related treatment effect of M tuberculosis in BSI in adults with HIV-associated TB.
- Funding: None disclosed.
- Inclusion, patient selection bias.
- Study design, conduct variations.
- Overestimated diagnostic yield.
- Missing data bias.
- Unmeasured confounding.