Tuesday, May 5, 2026

Time-dependent association between cancer and risk of TB [TBN 072]

Who

Adults aged 30–90 years in Taiwan, including 457,673 newly diagnosed cancer patients and 3,738,122 matched noncancer individuals. Matching was based on sex, age, payroll bracket as socioeconomic surrogate, and calendar year.

What

Cancer diagnosis was associated with higher tuberculosis risk. During follow-up, TB incidence was 358 vs 244 per 100,000 person-years in cancer vs noncancer groups, with IRR 1.47 (95% CI 1.43–1.50; P < 0.0001). TB risk peaked around cancer diagnosis, especially within 1 year before and after diagnosis. In the peridiagnostic period, cancer remained independently associated with TB: adjusted HR 2.29 (95% CI 2.22–2.35), attenuated to HR 2.20 (95% CI 2.09–2.32) after two-stage calibration for smoking and alcohol use. Highest risks were seen in younger/middle-aged cancer patients, lung cancer, mesothelioma, esophageal cancer, and hematologic malignancies.

When

Data were retrieved from January 1, 2001 to December 31, 2015. The key “peridiagnostic period” was defined as 1 year before to 1 year after cancer diagnosis.

Where

Taiwan, using nationwide population-based databases: National Health Insurance Research Database and National Health Interview Survey.

Why

The study addressed whether cancer diagnosis is associated with time-dependent TB risk, including both prediagnostic and postdiagnostic TB risk, while accounting for measured and unmeasured confounding.

How

Retrospective population-based cohort study. Incident TB was defined by ICD-9-CM TB codes plus simultaneous prescriptions of at least two anti-TB drug categories. Analyses included incidence rate ratios, Cox proportional hazards models adjusted for comorbidities, and two-stage calibration using NHIS data to adjust for unmeasured smoking and alcohol use. Main limitation: residual confounding may remain despite calibration, and claims-based definitions depend on coding accuracy.

Strength of evidence

Observational cohort evidence; large nationwide sample and time-dependent analysis strengthen inference, but causality cannot be confirmed.

Source: Shen, B. J., & Lin, H. H. (2021). Time-dependent association between cancer and risk of tuberculosis: A population-based cohort study. International Journal of Infectious Diseases, 108, 340-346. https://benangmerah.net/record/86/time-dependent-association-between

Time-dependent association between cancer and risk of TB [TBN 072]

Who Adults aged 30–90 years in Taiwan, including 457,673 newly diagnosed cancer patients and 3,738,122 matched noncancer individuals . Mat...