To explore the association between C-reactive protein(CRP) level at baseline and the risk of diabetes mellitus(DM) during follow-up in a community cohort.

Subjects were the residents (35-74 years old) derived from a community-based diabetes prospective cohort study in Pudong New Area in Shanghai, which began in 2009 and followed up in 2012. Baseline non-diabetic population was divided into four groups according to the level of baseline CRP (CRP<1.00 mg/L, 1.00-1.99 mg/L, 2.00-2.99 mg/L or ≥3.00 mg/L). Analysis of variance, Kruskal Wallis H nonparametric test and Chi-square test were used to compare the baseline characteristics, and COX regression model was used to analyze the association between baseline CRP level and DM risk during follow-up.

The mean follow-up time was (2.7±0.5) years. For those with CRP<1.00, 1.00-1.99, 2.00-2.99 and ≥3.00 mg/L, the incidence rate of DM was 14.68/1 000, 21.30/1 000, 28.92/1 000 or 46.38/1 000 person-years, respectively; the incidence rate of DM increased with baseline CRP levels (χ_linear²=13.63, P<0.001). After adjusting for family history of diabetes, history of hypertension, age, body mass index, triglycerides, and the baseline blood glucose status, the risk of DM in group with CRP level ≥3.00 mg/L was 3.26 times (95% CI: 1.39-7.64) that of group with CRP level<1.00 mg/L. The areas under the receiver operating characteristic curves of baseline CRP level in predicting the incidence of DM was 0.604 (95% CI:0.522-0.686).

Baseline CRP level is an independent risk factor of diabetes; however, the predict power of CRP is poor. Therefore, combiding CRP with other factors may be a good choice to predict diabetes.

Diabetes mellitus; C-reactive protein; Prospective cohort study