C-reactive protein (CRP) has been studied extensively for association with a large number of non-infectious diseases and outcomes. We aimed to evaluate the breadth and validity of associations between CRP and non-infectious, chronic health outcomes and biomarkers. We conducted an umbrella review of systematic reviews and meta-analyses and a systematic review of Mendelian randomization (MR) studies. PubMed, Scopus, and Cochrane Database of Systematic Reviews were systematically searched from inception up to March 2019. Meta-analyses of observational studies and MR studies examining associations between CRP and health outcomes were identified, excluding studies on the diagnostic value of CRP for infections. We found 113 meta-analytic comparisons of observational studies and 196 MR analyses, covering a wide range of outcomes. The overwhelming majority of the meta-analyses of observational studies reported a nominally statistically significant result (95/113, 84.1%); however, the majority of the meta-analyses displayed substantial heterogeneity (47.8%), small study effects (39.8%) or excess significance (41.6%). Only two outcomes, cardiovascular mortality and venous thromboembolism, showed convincing evidence of association with CRP levels. When examining the MR literature, we found MR studies for 53/113 outcomes examined in the observational study meta-analyses but substantial support for a causal association with CRP was not observed for any phenotype. Despite the striking amount of research on CRP, convincing evidence for associations and causal effects is remarkably limited.
Keywords: Bias; C-reactive protein; CRP; Mendelian randomization; Meta-analysis; Systematic review; Umbrella review.