Background Tuberculosis (TB) has been the leading cause of infectious disease mortality globally, with about 1.5 million deaths annually. Limited evidence suggests higher risks of cardiovascular disease (CVD) among people diagnosed with TB disease. However, existing studies have not controlled for underlying differences in CVD risk before TB diagnosis and may also be affected by selection biases or residual confounding.

Methods Retrospective cohort analyses used 2000-2019 electronic health data from the United Kingdom (Clinical Practice Research Datalink) and United States (Veterans Medical Centers). Adults with incident TB disease (UK n=15,820; US n=2,121) were matched (health-care practice, age, sex, ethnicity (UK)/race (US)) with up to 10 patients without TB. The main outcome was incident CVD events +/- 2 years of TB diagnosis. Patients with prevalent CVD >2 years before TB diagnosis were excluded. Five analysis periods were defined: baseline (1-2 years before TB diagnosis), pre-acute (3-12 months before), acute (+/- 3 months of TB diagnosis), post-acute (3-12 months after), post-TB (1-2 years after). Poisson regression models (adjusting for socio-economic deprivation, body mass index, smoking, prescribing statins or anti-hypertensives) estimated incident rate ratios (IRR) in each period for CVD events in patients with TB compared to those without TB, and also adjusted for baseline differences.

Results Median age of patients was UK 44 years, US 69 years; most were male (51% UK, 97% US). Patients with TB were more likely to be smokers and less likely to be overweight or obese. In both UK and US cohorts, CVD incidence was consistently higher in patients with TB compared to patients without TB throughout the two years before and after TB diagnosis. Patterns were similar in both cohorts, although CVD incidence rates were higher in the US data, likely due to the different age-sex profile. The IRR was significantly higher for the acute period: UK IRR=2.7 (95%CI 2.1, 3.4), US IRR=4.0 (95%CI 2.7-5.9). After accounting for differences at baseline (1-2 years before diagnosis), the adjusted IRRs in the acute period were UK 1.6 (95%CI 1.2, 2.1), US 3.2 (95%CI 2.2, 4.4). Stratified analyses showed similar adjusted IRRs by age, sex and ethnicity/race.

Conclusion The rate of incident CVD events close to date of TB diagnosis was 3-4 times higher in patients with TB disease compared with similar patients without TB and approximately 2-3 times higher after accounting for baseline differences in CVD incidence. Expanding TB treatment to include CVD care may provide an important opportunity to improve outcomes.