Comorbidities, drug-resistance and length of hospital stay among tuberculosis inpatients in Northeastern China: a retrospective observational study from 2013 to 2021

Background: Tuberculosis (TB) remains a significant global health issue. Drug-resistant TB and comorbidities exacerbate its burden, influencing treatment outcomes and healthcare utilization. Despite the growing prevalence of TB comorbidities, research often focuses on single comorbidities rather than comorbidity patterns. This study aims to evaluate comorbidity patterns among TB inpatients in Northeastern China from 2013 to 2021 and investigate the association between comorbidities and drug-resistance with length of hospital stay (LOS). Our findings could enhance the understanding of TB comorbidity interactions and provide evidence for targeted morbidity management strategies.

Methods: Network analysis was used to evaluate comorbidity patterns, estimating centrality indices to understand the structural importance of each comorbidity. The Walktrap algorithm was used to identify clusters of highly connected comorbidities. Networks of drug-susceptible and drug-resistant TB inpatients were compared. Multivariable linear regression models were used to assess the associations between LOS with comorbidities and drug resistance.

Results: A total of 2,352 TB inpatients were included, with a median LOS of 31 (IQR: 16-51) days. Inpatients with multidrug-resistant TB (β = 12.88, 95%CI = 8.03-17.73), chronic hepatitis C (β = 31.89, 95%CI: 4.41-59.37), pneumonia (β = 37.14, 95%CI: 12.53 - 61.76), pneumoconiosis (β = 28.40, 95%CI: 11.92 - 44.87), pneumothorax (β = 19.88, 95%CI: 4.97 - 34.80), and dermatitis/eczema (β = 56.54, 95%CI: 8.18 - 104.89) were significantly associated with longer LOS. Frequent comorbidities included liver dysfunction(15.2%), hypoproteinemia (14.4%), diabetes (14.2%), pleural effusion (11.3%), and emphysema (10.2%). Hypoproteinemia showed high structural importance in the network, ranking second in strength and highest in betweenness. Nine clusters of comorbidities were detected. No significant differences were found between the networks of drug-sensitive and drug-resistant TB inpatients, except for the greater strength of cholecystitis among drug-resistant inpatients.

Conclusions: Early detection and management of drug-resistant TB and comorbidities that prolong LOS, as well as those with structural importance or within the same cluster in the comorbidity network, are crucial for improving patient outcomes.