Physical activity and diet pattern do not mitigate C-reactive protein increases associated with oral contraceptive use.

The purpose of this study was to examine the influence of body mass index (BMI), physical activity (PA) level, dietary inflammatory index (DII), and oral contractive (OC) use on C-reactive protein (CRP) levels, and to determine if elevated CRP values reflect systemic inflammation in OC users. Data were obtained from four cycles (1999-2006) of the U.S. National Health and Nutrition Examination Survey (NHANES) study, yielding a sample of 496 current OC users and a comparator group of 1,583 regularly menstruating women. A general linear model was used to test for interaction effects among BMI, PA level, and OC use, after adjusting for age and smoking status, with log-transformed CRP (lnCRP) identified as the outcome variable. Sequential general linear models with no interaction terms were then constructed to examine the impact of BMI, PA level, and OC use on circulating lnCRP after adjusting for age and smoking status. Follow-up analyses used general linear models to assess the relationship between lnCRP and other indices of systemic inflammation among OC users and nonusers, and to examine the predictors of lnCRP within each subgroup. The omnibus model including smoking status, age, PA level, OC use, and BMI did not identify any statistically significant two-way or three-way interaction effects (all p ≥ .259). The adjusted r2 value of the model modestly increased from 0.3789 to 0.3801 when all interaction terms were removed. After adjusting for smoking status and age, a sequentially built model indicated that PA level was inversely related to lnCRP values (p = .0019). When OC use was added to the model, it was positively associated with lnCRP values (p < .0001), with statistically and clinically significant lnCRP differences between OC users and nonusers. BMI was the last variable entered into the model, which was positively associated with lnCRP (p < .0001). Among OC nonusers, PA level (p = .0008) and BMI (p < .0001) were significantly predictive of lnCRP levels after adjusting for smoking status and age. In contrast, PA level was not significantly predictive of lnCRP values (p = .718) among OC users. All alternative indices of inflammation were positively correlated with lnCRP values (all p < .0001), but correlations were significantly stronger among OC users than nonusers (all p < .05). In a subset of OC nonusers with complete nutrition data, PA level (p = .021), BMI (p < .0001), and DII (p = .007) were significantly predictive of lnCRP after adjusting for smoking status and age. In contrast, PA level (p = .709) and DII (p = .690) were not significantly predictive of lnCRP values among OC users. In conclusion, OC-induced elevations in CRP appear to be reflective of a chronic, systemic inflammatory response. PA and low DII are associated with lower CRP among OC nonusers, but do not mitigate CRP elevations among OC users.

Duke Scholars

Author David E Eagle Duke Global Health Institute

Author Herman Pontzer Evolutionary Anthropology