Observational and genetic effect estimates for CRP (mg/L) and complement C3 (g/L) by ferritin concentration (μg/L).
Observational estimatesa | Genetic effect estimatesb | |||||
---|---|---|---|---|---|---|
Adjusted logistic regression | Extreme genotype score | |||||
N | Odds ratio | 95% CI | N | Odds ratio | 95% CI | |
CRP ≥2 vs <2 mg/L | ||||||
All | 20508 | 1.12 | 1.09–1.16 | 20508 | 1.03 | 1.01–1.06 |
Men | 10415 | 1.14 | 1.10–1.19 | 10415 | 1.02 | 1.00–1.05 |
Women | 10093 | 1.16 | 1.11–1.21 | 10093 | 1.04 | 1.01–1.07 |
Complement C3 >1.04 vs ≤1.04 g/L | ||||||
All | 6674 | 1.28 | 1.21–1.35 | 6674 | 1.06 | 1.03–1.12 |
Men | 3472 | 1.26 | 1.17–1.35 | 3472 | 1.06 | 1.02–1.13 |
Women | 3202 | 1.24 | 1.14–1.36 | 3202 | 1.05 | 1.01–1.16 |
Individuals with CRP concentration ≥10 mg/L were excluded (N = 2633) to ensure that none of the participants had an ongoing infection or severe inflammation because of chronic disease. In the analysis concerning ferritin concentration, individuals with either ferritin concentration <20 μg/L or iron concentration <60 μg/dL (<10 μmol/L) were excluded (N = 8448) to ensure that none of the individuals had iron deficiency.
↵a Observational estimates are obtained from logistic regression adjusted for sex (not in sex-stratified analyses), age (<50 vs ≥50 years), alcohol intake (consumption of ≤7 units/week vs >7 units/week), BMI (<25 vs ≥25 kg/m2), smoking (current vs nonsmoker), diabetes (yes vs no), and menopausal status (yes vs no in stratified analyses with women). Ferritin was on continues scale and was logarithmically transformed [log2 (ferritin)] because of skewness of the distribution.
↵b We performed an instrumental variable analysis by constructing an extreme genotype score comparing the high-risk genotype C282Y/C282Y with wild type/wild type (reference group) using a ratio estimate to calculate the log odds ratio of increased CRP or complement C3 per unit increase in log2 (ferritin) concentration. We obtained 95% CI using Fieller's method.