BackgroundBody mass index (BMI) based on self-reported height and weight has been criticized as being biased because of an observed tendency for overweight and obese people to overestimate height and underestimate weight, resulting in higher misclassification for these groups. We examined the validity of BMI based on self-reported values in a sample of Norwegian women aged 44–64 years.MethodsThe study sample of 1,837 participants in the Norwegian Women and Cancer study self-reported height and weight, and then, within 1 year, either self-reported anthropometric again, or were measured by medical staff. Demographic and anthropometric were compared using t-tests and chi-square tests of independence. Misclassification of BMI categories was assessed by weighted Cohen’s kappa and Bland–Altman plot.ResultsOn average, the two measurements were taken 8 months apart, and self-reported weight increased by 0.6 kg (P<0.05), and BMI by 0.2 kg/m2 (P<0.05). The distribution of BMI categories did not differ between self-reported and measured values. There was substantial agreement between self-reported values and those measured by medical staff (weighted kappa 0.73). Under-reporting resulting in misclassification of BMI category was most common among overweight women (36%), but the highest proportion of extreme under-reporting was found in obese women (18% outside the 95% limits of agreement). The cumulative distribution curves for the measured and self-reported values closely followed each other, but measurements by medical staff were shifted slightly toward higher BMI values.ConclusionWhile there was substantial agreement between self-reported and measured BMI values, there was small but statistically significant under-reporting of weight and thus self-reported BMI. The tendency to under-report was largest among overweight women, while the largest degree of under-reporting was found in the obese group. Self-reported weight and height provide a valid ranking of BMI for middle-aged Norwegian women.

Several studies support a protective effect of dietary magnesium against type 2 diabetes, but a harmful effect for iron. As diabetes has been linked to pancreatic cancer, intake of these nutrients may be also associated with this cancer. We examined the association between dietary intake of magnesium, total iron and heme‐iron and pancreatic cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort. In total, 142,203 men and 334,999 women, recruited between 1992 and 2000, were included. After an average follow‐up of 11.3 years, 396 men and 469 women developed exocrine pancreatic cancer. Hazard ratios and 95% confidence intervals (CIs) were obtained using Cox regression stratified by age and center, and adjusted for energy intake, smoking status, height, weight, and self‐reported diabetes status. Neither intake of magnesium, total iron nor heme‐iron was associated with pancreatic cancer risk. In stratified analyses, a borderline inverse association was observed among overweight men (body mass index, ≥25 kg/m2) with magnesium (HRper 100 mg/day increase = 0.79, 95% CI = 0.63–1.01) although this was less apparent using calibrated intake. In female smokers, a higher intake of heme‐iron was associated with a higher pancreatic cancer risk (HR per 1 mg/day increase = 1.38, 95% CI = 1.10–1.74). After calibration, this risk increased significantly to 2.5‐fold (95% CI = 1.22–5.28). Overall, dietary magnesium, total iron and heme‐iron were not associated with pancreatic cancer risk during the follow‐up period. Our observation that heme‐iron was associated with increased pancreatic cancer risk in female smokers warrants replication in additional study populations.