Clinical Phenotypes of Post-tuberculosis Sequelae Among Hospitalized Adults in Kenya: A Cluster Analysis

Abstract Rationale: Survivors of tuberculosis (TB) experience an increased risk of mortality following TB treatment as compared to the general population. To better understand the wide spectrum of post-TB disease and differences in outcomes, we sought to identify clinical phenotypes among individuals with a history of TB who were admitted to a tertiary care hospital. Methods: A k-prototype cluster analysis was performed using data from a case-control study of adults admitted to Moi Teaching and Referral Hospital in Eldoret, Kenya between 2019 and 2022. The original case-control study focused on chronic hypoxemia and found that prior TB was an important risk factor for chronic hypoxemia in the population. Twenty-one variables, of a possible 78, were chosen based on data completeness, clinical relevance, and for chest x-ray data, inter-reader reliability. Silhouette index was assessed to determine the optimal number of clusters among participants with a history of prior TB. Kruskal-Wallis and chi-square tests were used to compare differences in characteristics and mortality outcomes between clusters and participants with no TB history. Results: The dataset included 417 participants, of whom 58 (12.8%) had a history of prior TB. Two distinct prior TB clusters were identified (Table 1). Cluster 1 (n=29) is characterized by younger, predominantly female individuals with no smoking history and a high rate of HIV infection. Cluster 2 (n=29) is characterized by older, predominantly male individuals with high rates of tobacco use and higher rates of pleural abnormalities and pneumothorax on chest x-ray. Cluster 2 participants had higher St. George's Respiratory Questionnaire symptom scores than Cluster 1; total, activity, and impact scores were similar between clusters. Combined inpatient and 1-month post discharge mortality was 31% for Cluster 1 and 24% for Cluster 2. Conclusions: Distinct clinical phenotypes among individuals with prior TB were identified using routinely available demographic, clinical, and imaging data. These phenotypes inform outcomes and can potentially be used to tailor interventions to improve diagnosis and treatment of post-TB complications among TB survivors in high-TB burden settings. Future research should explore clinical phenotypes in a prospectively enrolled cohort of individuals who are treated for TB and followed post treatment in order to validate findings.