Severe Tuberculosis Subtypes Stratify Short-Term Ventilation and ICU Risk

A retrospective cluster analysis of 500 patients with severe pulmonary tuberculosis has identified two distinct clinical subtypes characterized by differing inflammatory profiles and acute clinical outcomes. The study, conducted by researchers Jiaqi Pu and Jian-Qing He and published in the European Journal of Medical Research, demonstrated that six routine clinical variables can separate severe tuberculosis into hyperinflammatory and hypoinflammatory groups with markedly different requirements for respiratory support and critical care.

Prior to this investigation, it remained unknown whether distinct clinical subtypes with differing short-term outcomes existed within this patient population.

To identify potential subtypes, the investigators examined a retrospective cohort of 500 severe tuberculosis patients. The team applied k-means clustering based on six routine admission variables: age, body mass index (BMI), serum albumin, lymphocyte count, neutrophil count, and C-reactive protein (CRP). Clinical outcomes, including respiratory interventions and hospital stays, were then systematically compared between the resulting clusters.

The clustering algorithm partitioned the cohort into two distinct subgroups designated as Cluster 1 and Cluster 2. Cluster 1, comprising 228 patients, presented as a hyperinflammatory phenotype marked by older age, lower BMI, reduced albumin concentrations, lymphopenia, and elevated neutrophils and CRP. In contrast, Cluster 2 consisted of 272 patients displaying a hypoinflammatory profile characterized by younger age, higher BMI, preserved albumin, and lower inflammatory measures.

Marked differences in clinical interventions and resource utilization were observed between the two patient groups. Compared with Cluster 2, patients in the hyperinflammatory Cluster 1 experienced significantly higher rates of non-invasive ventilation (11.8 percent versus 1.8 percent), invasive mechanical ventilation (8.8 percent versus 2.2 percent), and intensive care unit admission (6.1 percent versus 1.1 percent). Furthermore, Cluster 1 patients required significantly longer hospitalizations, with a mean duration of 14.0 days compared to 11.0 days in Cluster 2, with all differences reaching statistical significance at P less than 0.01.

Multivariate logistic regression adjusting for age, BMI, autoimmune disease, and chronic obstructive pulmonary disease confirmed that the risk differences remained independent. With Cluster 1 serving as the reference group, Cluster 2 exhibited a significantly lower odds of requiring non-invasive ventilation (odds ratio 0.138, 95% confidence interval 0.037 to 0.421, P = 0.001), invasive mechanical ventilation (odds ratio 0.149, 95% confidence interval 0.050 to 0.391, P < 0.001), and intensive care admission (odds ratio 0.078, 95% confidence interval 0.017 to 0.267, P < 0.001).

In-hospital mortality comparisons showed no statistically significant difference between the two groups, with two deaths recorded in Cluster 1 (0.9 percent) and one death in Cluster 2 (0.4 percent), corresponding to a P-value of 1.000. However, the researchers emphasized that the very low absolute number of deaths substantially limits the interpretability and statistical power of this mortality comparison.

The authors concluded that six routine clinical variables effectively stratify severe pulmonary tuberculosis into hyperinflammatory and hypoinflammatory clusters. Because patients in the hyperinflammatory cluster require significantly more respiratory support and intensive care management, this clustering approach provides a pragmatic risk-stratification tool suitable for clinical implementation in resource-limited healthcare settings.

Subject of Research: Clinical subtyping and short-term risk stratification in severe pulmonary tuberculosis

Article Title: Two distinct subtypes of severe pulmonary tuberculosis for short-term risk stratification: a cluster analysis

Article References: Pu, J., & He, J.-Q. (2026). Two distinct subtypes of severe pulmonary tuberculosis for short-term risk stratification: a cluster analysis. European Journal of Medical Research. https://doi.org/10.1186/s40001-026-05380-y

Image Credits: AI Generated

DOI: 10.1186/s40001-026-05380-y

Keywords: severe pulmonary tuberculosis, severe tuberculosis, k-means clustering, cluster analysis, risk stratification, hyperinflammatory subtype, hypoinflammatory subtype, non-invasive ventilation, invasive mechanical ventilation, intensive care unit admission, C-reactive protein, serum albumin, lymphopenia, neutrophils, body mass index, hospital length of stay, chronic obstructive pulmonary disease, autoimmune disease