Background
Multimorbidity, i.e. two or more non-communicable diseases (NCDs), has been associated with venous thromboembolism (VTE), but whether multimorbidity is a predictor for incident VTE is unknown.
Aims
To examine the associations between multimorbidity and its severity with risk of incident VTE, and examine the association between nine different disease clusters and incident VTE.
Methods
A cohort study using landmark analysis of 2,694,442 individuals. Swedish national registers were linked and three landmarks (L1, L2, L3), i.e. baselines, were created with 14-, nine- and four-year follow-up times, respectively. Two or more NCDs defined multimorbidity and ≥5 marked multimorbidity severity. A hazard ratio (HR) with 95 % confidence interval (CI) for VTE was calculated and adjusted for sex, education and year of birth. Death and emigration were treated as competing events.
Results
A total of 2,694,442 individuals were included. Multimorbidity was associated with incident VTE in all three analyzed landmarks: adjusted HR for VTE was 2.47 (95%CI 2.24–2.72) for L1, HR was 2.23 (95%CI 2.11–2.36) for L2, and HR was 2.16 (95%CI 2.03–2.29) for L3. HR increased with multimorbidity severity. For instance, HRs for multimorbidity with five or more NCDs was 4.29 (95%CI 2.53–7.28) in L1 analysis, 4.45 (95%CI 3.64–5.45) in L2 analysis and 4.83 (95%CI 4.20–5.55) in L3 analysis. Moreover, seven of nine different multimorbidity disease clusters were predictors for VTE.
Conclusion
This study demonstrated proof-of-concept that multimorbidity is a novel dose-graded predictor for VTE. Further studies will determine the usefulness of multimorbidity for VTE prediction in different clinical settings.