点击显示 收起
1 From the PsychoBiology Group, School of Psychology, University of Leeds, United Kingdom.
See corresponding article on page 130
2 Address reprint requests to JE Blundell, PsychoBiology Group, School of Psychology, University of Leeds, Leeds, LS2 9JT United Kingdom. E-mail: j.e.blundell{at}leeds.ac.uk.
.
If humans represent the most intelligent form of life on this planet, why is it that they find it so difficult to make the apparently small adjustments in daily behavior that we calculate would halt the continuing rise in obesity? Is a highly developed intellect useless in the presence of a permissive biological system and a provocative environment structured on consumerism? Moreover, why is it so difficult for humans to accurately record the food items they consume daily? Why have scientists found it impossible to develop a method that allows people unambiguously and accurately to reveal their habitual energy and nutrient intakes? Is there something that we do not understand or have failed to grasp about the causes of human behavior or the power of rational thought?
A dilemma for nutritional science has developed that relates to these issues. It concerns the validity of the estimates of food consumption gleaned from large-scale (and sometimes small-scale) surveys that use food diaries, food-frequency questionnaires, or other dietary recall methods to determine the amounts of energy, particularly fat, that people habitually ingest. On the basis of the available data, arguments are constructed about the relation between nutritional variables and disease endpoints. If these data themselves are unreliable, then the arguments will be questionable and the conclusions will be doubtful or actually misleading. One such argument concerns the relation between the amount of dietary fat consumed and the prevalence of obesity; the argument has received considerable exposure and is focused on American data, although the issue is far more pervasive.
The study by Goris et al (1) makes a clever contribution to the debate about the validity of food consumption estimates and the relative importance of dietary fat in obesity. It is now freely admitted that most databases do not inform us well about the actual foods that individuals habitually consume because of the massive problem of misreporting, the most common variety of which is underreporting. A key issue is whether all types of food are equally underreported. Note that different methods are subject to different forms of error. For recall methods, the problem is simply underrecording, which has been estimated to be 25% of daily energy intake (2). For food diaries, the problems are underrecording (failure to record what is eaten) and undereating (during the collection period). The study by Goris et al has advanced our thinking in this area because they measured both underrecording and undereating as they relate to underreporting.
In the study by Goris et al, a group of 30 obese men were advised by a dietitian how to keep a diary record of food consumption for 7 d. Underreporting was defined as the difference between energy expenditure (measured unambiguously by the doubly labeled water technique) and energy intake computed from the food records. Energy intake fell short of energy expenditure by 38%. Undereating (26%) was calculated from the loss of body weight measured during the observation week. Underrecording of food was calculated indirectly from water losses, assuming that the accuracy of recording water intake reflects food intake. It was deduced that there was 12% underrecording of energy intake. In other words, when asked to record their food intakes, these obese men suddenly reduced their habitual food intake by 26% and failed to record 12% of what they actually ate. In addition, the data showed an inverse relation between the reported percentage of energy from fat and the percentage of underreporting. That is, the greater the degree of underreporting, the lower was the percentage of energy from fat in foods recorded in the food diaries. This finding indicates that these obese men selectively underreported their fat intakes. If this is generally true for obese individuals participating in dietary surveys, it means that we would seriously underestimate the amount of fat being consumed by obese people. In turn, this would raise questions about the role of dietary fat in the etiology and maintenance of obesity. This issue is serious because it is known that obese individuals (ie, those with a high body mass index) are more likely to underreport their food intakes than are normal-weight individuals (3) and that women underreport more than do men (4).
What lessons can we learn from the outcomes of this study? The fact that obese people reduce the amount of food that they eat when they are asked to record it in a food diary should be no surprise. For years, self-monitoring of eating has been the cornerstone of the behavioral modification of appetite, and it has long been recognized that getting people to record the foods they eat is almost guaranteed to cause a reduction in the amount of food consumed, with a consequent weight loss.
Concerning the debate about dietary fat and obesity, the barricades have already been set up and there is evidence that scientists are digging in on each side in preparation for a drawn out war of attrition. Is it possible to take a little of the dogmatism out of the debate? For example, would it help to recognize that the relation between fat consumption and obesity is not a biological inevitability as has persistently been iterated (5)? Not everyone who eats a high-fat diet is obese; many consumers of high-fat diets are lean (6) and some will remain so for most of their lives. Can we agree that a high-fat diet is a significant nutritional risk factor for weight gain, albeit just one of many other risk factors, eg, the types of different foods consumed, an increase in portion sizes, the total amount of energy ingested, and the increasing palatability of the food supply? Can we also agree that there are different dietary "routes to obesity" (7), some of which achieve potency through combinations with particular metabolic risk factors such as a low basal metabolic rate, a high respiratory quotient, and insulin sensitivity? Even given these qualifications about the relation between fat consumption and obesity, I suspect that most scientists in the field of nutrition would advise their clients to eat as much fat as possible if these clients asked how they could most easily become obese (without the use of drugs).
Correlation coefficients between dietary fat and indicators of body weight from epidemiologic data appear low; however, this is not surprising given the number of other variables that affect body weight. However, to what extent are these correlations influenced by the selective underreporting of high-fat foods by overweight and obese subjects? The observation that obese men in the Netherlands selectively underreport the amount of fat they eat should not surprise us (1). There is also good evidence from a UK database that obese women selectively underreport foods that are both sweet and high in fat (8). In some countries, the degree of underreporting increased with successive surveys and probably reflects both an increase in the participants' diet consciousness and in their ability to readily identify taboo foods (9). These findings are as expected because food has symbolic, moral, and emotional qualities for many individuals. Few individuals in the United States and Western Europe have not been subjected to messages about the important role of dietary fat in obesity. Individuals are aware that they are judged on the basis of the foods they eat.
The more we tell people that they should reduce the amount of fat in their diets, the more they tell us how little of it they eat; however, actual intakes do not appear to change or even increase. Individuals may not wish to admit (via food diaries, food-frequency questionnaires, or dietary recalls), even to themselves, exactly what they put in their mouths. Consequently, the fact that Goris et al showed that the obese men in their study selectively underreported their fat intakes confirms many suspicions. Thus, we should be cautious about making too many doctrinaire statements on nutrition-disease relations on the basis of data that may be seriously flawed and that may generate spurious correlation coefficients.
The task of finding out exactly what foods people consume seems a simple task in an increasingly technological scientific environment. Why is it so difficult to achieve? First, what do subjects in nutrition studies or dietary surveys have to gain by reporting accurately to us highly personal information? How many nutritionists would report accurate information if asked to recall the number of their extramarital sexual encounters or the amount of their donations to charity? I suspect that there would be considerable underreporting and overreporting, respectively, in response to these latter questions for the same reason that subjects underreport food intakes: the high degree of sensitivity of these issues. In one study, subjects identified as underreporters were interviewed and given the opportunity to indicate whether their food records reflected what they habitually ate (10). The subjects admitted volitionally selecting foods that were easier to record (ie, they changed their eating habits) or deliberately omitting to record foods that they had eaten because of embarrassment (underrecording). Interestingly, some subjects not identified as underreporters also admitted that their records had also been adjusted. Consequently, we can identify with some confidence the underreporters whose food records are probably not a true reflection of what they usually eat; however, even the records of those subjects who appear to be "above suspicion" cannot be regarded as safe (1112). A recent study in the Journal selected subjects whose reported energy intakes were regarded as accurate because these intakes fell within 30% of predicted requirements. This margin of error would be completely unacceptable in other fields of nutritional science.
The questionable reliability of food collection data is actually part of a much wider problem that concerns the capacity and willingness of people to disclose details of their personal lives to total strangers (albeit protected by anonymity). Scientists in the field of nutrition are fortunate because there exists a formula (the ratio of energy intake to basal metabolic rate ratio) that provides a rough and ready means of detecting those records that are false reflections of true habitual consumption. Researchers thus have the option of dealing in particular ways with these flawed data. The omission of data from subjects whose intakes are implausible (as indicators of habitual intake) results in exclusion from the analysis of a high percentage of those very subjects (obese individuals) who are of special interest. The inclusion of data from all subjects in the analysis means that the database is inaccurate.
Perhaps it should be recognized that although the nutritional issue under discussion is a dilemma for nutrition, it is not a problem in nutrition but in psychology. The processes of recalling and recording food intakes require attention and perception; the registration, storage, and retrieval of daily events (some of which are detailed and complex); and sometimes the computation of the frequency of occurrence of these intakes and events. Other mediating factors include motivational variables that modify the capacity to comply with instructions and the emotional and moral feelings that influence the perceived value of making a true record of events.
Can the problem ever be solved? Perhaps some future miracle of technology will allow the automatic monitoring and nutrient analysis of every morsel of food as it is eaten. In the meantime, we have to live with inadequate methodology and the questionable data it produces. Apparently, our subjects are unwilling to yield to us the rather modest information we seek. Because food provides one of life's most accessible (and potent) forms of pleasure, perhaps we should accept that we are dealing with behavior governed by hedonism rather than by rationality.
REFERENCES
Related articles in AJCN: