Methodological issues in etiologic studies of diet and colon cancer

The relative validity of retrospectively assessed energy intake data in cases with myocardial infarction and controls (the Zutphen Study)

The relative validity of energy intake data, retrospectively collected after a 15-year period, was investigated. In 1970 dietary data were obtained by the cross-check dietary history method among 615 men aged 50-69. Between 1970 and 1985, 43 of these men had a non-fatal myocardial infarction. As controls 86 healthy men were selected. Of these 129 men, 112 were interviewed about their usual food intake in 1985 and in 1970. The retrospectively collected food intake data were compared with the data obtained in 1970. Retrospectively the average energy intake was overestimated by approximately 300 kcal/day. The amount of overestimation was statistically associated with change in energy intake, education and interviewer. The data collected in 1970 showed a significantly lower energy intake in cases compared with controls. However, the retrospectively assessed energy intake data showed no significant difference. A difference in energy intake between cases and controls may not be found when dietary intake data are assessed retrospectively.

Nutrient intakes of whites, blacks, and Mexican Americans in southeast Texas

Ethnic groups in the United States exhibit different patterns of cardiovascular disease and cancer morbidity and mortality. This has, in part, been attributed to differences in dietary intake. However, there is limited comparative information available regarding the dietary patterns of whites, blacks, and Hispanics residing in the same geographic area. Selected nutrient intakes were obtained by an interviewer-administered 24-hr dietary recall from 231 white, 102 black, and 98 Mexican-American persons residing in the same communities in Southeast Texas. Mean caloric intakes were highest for whites, followed by Mexican Americans and blacks. Mexican Americans had carbohydrate intakes that were significantly higher, but total fat intakes that were significantly lower, than those of whites. Blacks of both sexes had the highest cholesterol intakes and black males had the highest saturated fat intakes. Neither was significantly higher than that of whites or Mexican Americans. Overall, the mean vitamin A and C values were highest for blacks and lowest for whites, although the differences were not statistically significant. Mean calcium and phosphorus intakes were significantly higher for whites compared with those for blacks and Mexican Americans. Blacks had significantly lower mean fiber values than whites or Mexican Americans. International ethnic differences in disease distribution have long been used to provide clues to etiologic factors. National ethnic differences in disease distribution related to dietary intake can further elucidate these causative and/or preventive factors. However, to do so will require additional attention to dietary methodology of the type presented here.

The relative validity of a retrospective estimate of food consumption based on a current dietary history and a food frequency list

The relative validity of information and food consumption in the distant past was assessed by combining a dietary history (referring to the recent past) with a food frequency list (monitoring major changes over the past 12-14 years). This approach was evaluated in a study of two groups of apparently healthy elderly people (mean age 80 years) who had participated in a food consumption study 12-14 years before the start of the present study. One group consisted of 18 harbor employees who retired subsequent to the initial assessment of food intake. On the average, each member of this group had reduced his food consumption by about 1,000 kcal. The other group consisted of 46 elderly men and women who had retired before their food consumption was measured initially. This group had not markedly changed their food intake. The results showed that both groups overestimated changes in their food intake and that the systematic overestimation and random error were similar for both groups. If the men in both groups were combined to form one group, a valid ranking of subjects in small and large consumers of energy and most of the selected nutrients was possible. However, current food intake influenced the accuracy of the measurement of past food intake.

Epidemiologic studies on vitamin A and cancer

Data derived from epidemiologic studies on human populations are consistent with the protection from cancer afforded by vitamin A seen in animal studies. The populations studied are diverse, including groups living in India, Singapore, Norway, the United Kingdom, and the United States. The methodologies brought to bear on the question have been equally varied. Although there are inconsistencies in findings, and instances in which an association has not been observed, the weight of evidence suggests that the intake of vitamin A from dietary or other sources may inhibit the onset of lung cancer and possibly other cancers. However, the evidence from human populations is not experimental and it is conceivable that the associations observed are not causal. Additional epidemiologic research is needed to determine what sites of cancer may be inhibited by vitamin A and whether cancer growth at any other site is enhanced by high vitamin A intakes. It is also important that controlled trials using vitamin A as a chemopreventive agent be considered as a means of determining whether the epidemiologic findings are of clinical significance.

Dietary fibre and 10-year mortality from coronary heart disease, cancer, and all causes. The Zutphen study

In 1960, 871 middle-aged men in the town of Zutphen, The Netherlands, participated in a survey of risk indicators (including diet) for coronary heart disease (CHD). Information was collected about the usual food intake for the 6--12 months before the interview by the cross-check dietary history method. During 10 years of follow-up, 107 men died from all causes, 37 from CHD, and 44 from cancer. Mortality from CHD was about four times higher for men in the lowest quintile of dietary-fibre intake than for those in the highest quintile, but this inverse relation disappeared after multivariate analyses. Rates of death from cancer and from all causes were about three times higher for men in the lowest quintile of dietary-fibre intake than for those in the highest quintile, and these relations persisted after multivariate analyses. A diet containing at least 37 g dietary fibre per day may be protective against chronic diseases in Western societies.

Summarizing dietary patterns using multivariate analysis

Data on the frequency of consumption of 76 foods were collected from 142 native-born Australians of Australian or British parentage, 26 of Italian parentage and 140 Italian migrants to Australia. Clear separation of the three groups was achieved, using a few as 27 food items in a discriminant function analysis. Potatoes, cauliflower, pumpkin and salt were consumed more by Australians, while salad dressings and grapes were eaten more by Italians. Australian-Italians fell between the two groups, although they tended to be more like Italians. The discriminant score derived for each person was used as an overall food score and this was then related to the serum cholesterol level. There was no consistent relationship between food score and serum cholesterol level.

Diet and cancer of the esophagus

The reported dietary, alcohol consumption and smoking habits of 147 Roswell Park Memorial Institute white male patients diagnosed with cancer of the esophagus were compared with the reports of 264 white males of comparable ages with diagnoses other than cancer. Overall frequency of vegetable an fruit consumption was associated with lower risk; persons reporting fruit and/or vegetable consumption 31-40 times a month had significantly greater risk than those who reported consumption 81 times a month or more. Calculated indexes of vitamin A and vitamin C intake were similarly related to reduction in risk. Dose-response gradients were observed for frequency of vegetable and/or fruit consumption, as well as for vitamin A and C intake. The putative protective effect of vegetable and fruit intake remained evident after controlling for its possible association with smoking and drinking. Previously reported associations of smoking, alcohol use and social class, as measured by type of occupation, were replicated in these data. The findings of this investigation in instances of colon, lung, bladder, oral, and laryngeal cancers, and with evidence of tumor inhibition by vegetable properties in animals. Interpretation of the findings is limited by the difficulties of retrospective assessment of dietary intake and by possible confounding by other factors known to be related to esophageal cancer.