Serum C-reactive protein increases the risk of venous thromboembolism: a prospective study and meta-analysis of published prospective evidence

Evolving debate suggests that C-reactive protein (CRP) might be associated with the development of venous thromboembolism (VTE); however, the evidence is conflicting. We aimed to assess the prospective association of CRP with VTE risk. C-reactive protein was measured in serum samples at baseline from 2420 men aged 42–61 years, from the Kuopio Ischemic Heart Disease study. Within-person variability in CRP levels was corrected for using repeat measurements of CRP taken 11 years after baseline. Incident VTE events (n = 119) were recorded during a median follow-up of 24.7 years. The age-adjusted regression dilution ratio for log(e) CRP was 0.57 [95% confidence interval (CIs): 0.51–0.64]. In age-adjusted Cox regression analysis, the hazard ratio (95% CIs) for VTE per 1 standard deviation (SD) increase in log(e) baseline CRP was 1.17 (0.98–1.40). Further adjustment for several established and emerging risk factors did not alter the association. In a meta-analysis of nine population-based studies (including the current study) comprising 81,625 participants and 2225 VTE cases, the fully-adjusted risk estimate for VTE was 1.14 (1.08–1.19) per SD increase in log(e) baseline CRP. In a pooled dose–response analysis, a linear association between CRP and VTE risk was suggested (P for nonlinearity = 0.272). The pooled risk estimate for VTE per 5 mg/l increment in CRP levels was 1.23 (1.09–1.38). C-reactive protein was only modestly associated with VTE risk in the primary analysis. Pooled evidence, however, suggests that elevated CRP is associated with greater VTE risk, consistent with a linear dose–response relationship. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1007/s10654-017-0277-4) contains supplementary material, which is available to authorized users.