Influence of individually estimated portion size on the assessment of nutritional risk in colorectal cancer in Portugal.

We evaluated the influence of individually estimated portion sizes on the estimate of nutrient related risk of colorectal cancer, using data from a Portuguese hospital based case-control study on diet and colorectal cancer. A total of 100 patients with colorectal adenocarcinoma (aged 15-92 years) and 211 controls (aged 36-89 years) were included. Two data sets were created for nutrient analysis, the first one allowed estimates of food intake using data on portion size as collected with visual aids during the interview. The second estimate substituted respondents' estimate with a standard portion size, as used in the semi-quantitative (SQ) food frequency approach. The two analytic approaches yielded similar energy and nutrient intakes in cases and controls. The percent range of concordance is acceptable, in the same quartile varying from 44 to 82% (mean: 56%) and very good in the same or adjacent (+/-1) quartile (mean: 91%, range: 85-97%). The two estimates lead to a similar pattern of multivariate odd's ratio, however the SQ estimates resulted in more significant findings. We conclude that little extra information is gained by including individual portion size information when assessing diet-related risk of colorectal cancer.