Inverse associations are consistently observed in epidemiological studies on the relations between the consumption of vegetables and fruits and different types of cancer. The strength of these associations is, however, unknown amongst others because of measurement error in data on vegetable and fruit intake. The antioxidant (pro)vitamins β-carotene, vitamin C, and vitamin E, are three of many substances in vegetables and fruits which may be responsible for the anticarcinogenic effect. This thesis is focused on the problem of intake assessment of vegetables, fruits, and antioxidant (pro)vitamins.
In the first part of the thesis, two studies on the relationships between the consumption of vegetables, fruits, and antioxidant (pro)vitamins and the occurrence of cancer are described. In the Seven Countries Study intake of vitamin C was inversely related to stomach cancer mortality at ecological level. Subjects with low intakes of vegetables, fruits, and β-carotene that were stable over time experienced more than two-fold increased risks of lung cancer in the Zutphen Study than subjects with high stable intakes. A lack of information on the extent of measurement error in the dietary data in both studies hampered the correct interpretation of the results.
The second part of the thesis includes several studies on the estimation of measurement error in data on vegetable, fruit, and antioxidant (pro)vitamin intake and biochemical markers. In a study on the effects of frozen storage at -20°C it was shown that vitamin E concentrations in EDTA-plasma decreased dramatically between 6 and 12 months, whereas for β-carotene this took place after 1 year. The use of such plasma in nested case-control or case-cohort studies would result in highly attenuated odds ratios for β-carotene and vitamin E.
Reproducibility and relative validity for food group and nutrient intake assessed with an extensive semi-quantitative food frequency questionnaire was also investigated. The questionnaire seemed adequate for ranking subjects according to intake of most nutrients and food groups including fruits, but it did not yield such good results for vegetables, β-carotene, vitamin C for men, and vitamin E for women. The observed correlation coefficients between questionnaire and repeated 24-h recall data may be either over- or underestimates of the true validity coefficients, because of unknown error structure in both types of data. Validity coefficients estimated by a triangular comparison between questionnaire, 24-h recall, and biomarker measurements will probably be overestimates of true validity coefficients.
From these studies it is concluded that measurement error in assessing vegetable, fruit, and antioxidant (pro)vitamin intake may be large, which is a handicap for epidemiological studies. Further progress lies in improvement of dietary assessment methods, and probably even more in understanding error structures and the development of analytical methods to recognize and cope with those structures.
|Qualification||Doctor of Philosophy|
|Award date||6 Mar 1996|
|Place of Publication||S.l.|
|Publication status||Published - 1996|
- fruit crops