You are here
6. Assessment of Food Intake in the Distant Past
Assessment of Food Intake in the Distant Past - How Valid?
A literature search was undertaken for research papers relating to the relative validity and reliability of the retrospective measurement of dietary intake relative to dietary information provided at an earlier period. A review of the literature of such studies indicated that the level of agreement between the recall and the original diet generally showed a low correlation, indicating that assessment of past diet may lead to such misclassification as to distort the risks associated with diet. In some studies, the level of agreement was better, ranging from 0.5 to 0.8.
These studies also indicated that current diet has a strong influence on the recall diet.
A number of problems in recall have been clearly identified. These are:
- The major component of error is the inability of the respondent to provide the appropriate information.
- Correlation coefficients between the original and recalled diets decrease as The time interval between the diet assessments increased.
- Increased age was associated with decreased reliability in a number of studies.
- Subjects who had changed their diet had poorer recall, except when changes were made for health reasons
One review of the problems of memory in nutritional epidemiological research has noted that the validity of recalled data depends on the respondents' memory, and it is not clear whether the recalled data approximates true past intakes well enough to justify the time and effort involved in its collection. This review identified consistent problems encountered as:
- Individuals fail to report consumption of some foods that were actually eaten (errors of omission).
- Individuals sometimes recall foods that were never eaten (errors of commission).
- Reported estimates of amounts and kinds of foods are often inaccurate.
- Other factors including current diet, influence recall of diet in the past.
- Elaboration is an unintentional distortion with distortion of memory due to the effect of a range of factors, such as expectations and needs. For example, some individuals exaggerate the nutritional quality of their diet, reporting what they perceive should have been eaten, instead of what they actually ate.
It is concluded that recent intakes partially and gradually inhibit the recall of the diet of more distant months and years, and as the amount of time between food consumption and the attempt to recall increases, more of the original memory is likely to be inhibited.
Concerning recall by a spouse, child or other surrogate in the case of the deceased, concern has been expressed that data from the surrogate for such items as the subject's personal health habits may well be less valid than would be the data obtained from the subject himself, and that they may introduce certain biases into the data obtained.
From the very limited data available, it is concluded that the seriousness of the potential problems associated with surrogate data, are not clear. In surveys comparing the findings of dietary interviews of men with their wives' estimates of their husbands' current dietary intakes, it was found that 60 to 80 percent of respondent pairs agreed on food consumption frequencies for individual food items. However, these findings relate to the comparability of responses given by spouses for living subjects for current diet. The seriousness of the problems introduced by surrogate responses for the diet in the distant past, when the index subject is dead, remains to be documented.
The evidence from the literature clearly shows the questionable validity of recall of past diets when compared with previously recorded data. Problems identified in recalling food consumption in the past which are pertinent to the claims concerning animal fat consumption and malignant neoplasm of the prostate, include the age of the surrogate providing the information, the extensive time period for recall, errors of omission and commission, and unintentional elaboration with distortion of memory due to the effect of a range of factors, such as expectations and needs. These problems may be exacerbated when the information is given by a surrogate for a deceased person, though research has not been conducted to scientifically document this situation.