点击显示 收起
1 From the Instituto Nacional de Perinatología, Mexico City; the Center for Studies on Sensory Impairment, Aging and Metabolism, Guatemala City; the Centre for Applied Nutrition Research, Dundee, United Kingdom; and the Facultad de Ciencias de la Salud de la Universidad Rafael Landívar, Guatemala City.
2 Supported by the small-grants program of the World Cancer Research Fund, London. 3 Address reprint requests to R Valdés-Ramos, Subdirección de Investigación en Salud Pública, Instituto Nacional de Perinatología, Montes Urales 800, 2° Piso, Torre de Investigación, Lomas Virreyes, 11000, México, DF Mexico. E-mail: rycramos{at}servidor.unam.mx.
ABSTRACT
Background: The World Cancer Research Fund convened an expert committee who analyzed the literature related to the causation of human cancers. Recommendations for preventing cancer through behavioral practices were formatted into a 14-point guideline.
Objective: We parsed the cancer prevention guidelines to determine to what extent relevant information on individual behavior could be assessed from conventional food-frequency questionnaires, which are being used in surveys conducted in developing countries.
Design: We examined a convenience sample of archival forms completed during 2 independent studies (a case-control and a field study) that used an adapted Willett food-frequency questionnaire that was translated into Spanish for use in Guatemala.
Results: All dietary related guidelines, except for salt, were evaluated by both questionnaires. Physical activity, food handling, and food preparation were not addressed by either of the questionnaires, although body mass index and dietary supplements were addressed in the case-control study and field-study questionnaires, respectively.
Conclusions: Although concordance with some of the cancer prevention goals and guidelines can be evaluated from the existing questionnaires, adjustments and additions must be made with respect to salt and supplement use, physical activity, and food handling. Actual weight and height measurements are also needed, particularly in low-income populations.
Key Words: Diet • nutritional epidemiology • food-frequency questionnaires • FFQs • cancer • Guatemala • guidelines
INTRODUCTION
Until recently, it was thought that those living in developing countries (ie, 80% of the earth's inhabitants) were protected by their poverty from degenerative chronic diseases and, thus, the public health emphasis was on undernutrition (1). Most developing countries, however, have entered a stage of nutrition transition (2–5) and more recent insights suggest that these countries may be at risk of increased rates of cancer and heart disease (6, 7). A public health community more accustomed to addressing acute infections and nutritional adversity may be both unprepared and unaware of the reality of this risk.
Although genetics plays a role in the susceptibility and resistance to cancer, there is a consensus that changes in lifestyle factors and the environment, including dietary exposures, can significantly reduce cancer incidence (8). It is estimated from a comparison of the most optimistic and the most conservative assessments that 23% of all cancers are preventable by voluntarily adopting lifestyle practices that protect against cancer incidence (9). After surveying the published results from epidemiologic, metabolic, and experimental studies, an expert panel formed a global perspective on the issue of cancer prevention and defined a set of 14 guidelines to maximize the prevention of malignant diseases in human populations (9; Table 1).
View this table:
TABLE 1 . World Cancer Research Fund/American Institute for Cancer Research 1997 guidelines for the prevention of global cancer1
To determine the current risk factors for cancer in all populations, it is necessary to establish whether existing dietary surveys can still be used or whether it is necessary to develop new questionnaires. Several authors from various countries have developed and validated dietary intake questionnaires based mainly on the food-frequency questionnaire (FFQ) (10–12). These questionnaires may be useful in identifying populations at risk of chronic diseases, but there has been little evaluation of these tools with respect to guidelines, eg, those of the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) (9).
SUBJECTS AND METHODS
Food-frequency questionnaires
We compared the abilities of 2 FFQs used in 2 different Guatemalan studies (a field study and a case-control clinical study) to assess concordance with WCRF/AICR diet-related cancer guidelines. Both FFQs were adapted from a 61-item FFQ developed by Willett (13) . The study settings and diet assessment methods are described separately.
The data for the field-study questionnaire was collected in the Republic of Guatemala in the region of the eastern highlands, ie, the Santa Rosa Province. This is a rural and agricultural setting in which coffee cultivation and subsistence farming are the primary occupations for the largely ladino (Spanish speaking of mixed Spanish-Mayan descent) peasantry that inhabits towns, townships, hamlets, and homesteads along the mountainous ridge that runs toward the Caribbean coast. The climate is temperate throughout the year, 6 mo of which make up the dry season and 6 of which make up a tropical monsoon rainy season. Illiteracy is widespread among the population, especially among women.
The FFQ developed by Willett (13) served as the template for the questionnaire that was used in the present study. The questionnaire used in the present study was developed in the Spanish language and used common names that had been used in another survey similar to that by Campos et al (14) in Costa Rica, which involved the same analytic laboratories and collaborators. To determine which foods to include in the questionnaire used in the present field study, preliminary 24-h dietary recalls were conducted in the county seat and the peripheral hamlets; these citizens were also asked about their consumption of seasonally available foods. These foods were then listed in the FFQs. If a food already existed in the basic questionnaire, the code number was conserved and substitutions were made for items not part of the rural Guatemalan diet. Staple foods, such as maize (tortillas and tamales), occupied vacated code numbers in the list. When the food adaptation was completed, the FFQ for the Santa Rosa population consisted of a final roster of 88 food and beverage items. For purposes of analysis, however, certain items, such as different preparations of beans (Phaseolus vulgaris), were placed into a single food category.
Willett's FFQ was designed and developed for self-administration. Given the high rates of illiteracy and our doubts about the population's ability to understand concepts such as current and historical frequency of food consumption, data were collected from individual participants by trained interviewers. In 1994 a total of 274 individuals from the county seat of Santa Cruz Naranjo and its hamlets (El Naranjo, Potreríos, and Don Gregorio) were interviewed. From these interviews, we derived usable data forms for 269 individuals (55 men and 214 women) aged 16–86 y.
The questionnaire used in the second study, a case-control study on gastric cancer, was also based on Willett's FFQ (13). The population included rural and urban Guatemalan residents of all social classes, but foods consumed by a wider range of the Guatemalan population were added to the FFQ. This modified questionnaire had a final roster of 90 items. A single nutritionist (LB) conducted all of the interviews in the series. Questionnaires completed by 10 healthy subjects (8 men and 2 women) without cancer, who served as controls in the case-control study, were randomly selected from among 147 pairs of subjects.
Validity of questionnaire to provide information on cancer-prevention behaviors
The 14 cancer prevention goals and guidelines from the WCRF/AICR expert panel (9) are summarized in Table 1. In an interactive manner, alternating between the data provided in the completed questionnaires and the language in the guidelines, we decided whether an individual assessment of concordance for each guideline could be made from the data. This was done by ranking the questionnaires on how completely cancer-prevention information could be ascertained by designating one of the following to each variable of interest: 1) evaluated (+), 2) partially (inferentially) (+/-), or 3) not at all (-). This was done in a sequential fashion for both the Santa Rosa field study and the case-control study questionnaires.
Formative research
Once the gaps in the information provided by both questionnaires had been identified, we interviewed diverse (ie, rural and urban), low-income, Spanish-speaking Guatemalans in a formative research effort to understand how the required information could be gathered in the future, as explained in the following 2 sections.
Validity of self-reported height and weight
Twenty-eight women (26 from rural and 2 from urban areas) and 22 men (4 from rural and 18 from urban areas) from a sample of low-income residents, either from a rural village in the Zacapa Province or from Guatemala City, were asked to report their individual weight and height. Immediately after the data were registered, the subjects were weighed and their height was measured by a trained nutritionist. This self-reporting was previously used in Mexico with reasonable accuracy (15).
Food handling and culinary preparation
We conducted formative research with those interviewed in a sample of 30 low-income residents of a rural Zacapan village and 20 residents of Guatemala City to identify the different food handling and food preparation methods used in Guatemala.
RESULTS
The completeness with which the cancer-prevention information could be covered differed between the 2 questionnaires because neither questionnaire included information that covered all points (Table 2). This indicated that a modified questionnaire needs to be developed that can evaluate all 14 guidelines. Nutritional adequacy, dietary variety, edible plant contribution to diet, intake of fat, fruit, vegetables, sugar, cereals, tubers and legumes, red meat, and ethanol could be evaluated with both questionnaires. The questionnaires provided no information on the discretionary use of salt or on food handling and preparation. The Santa Rosa field-study questionnaire included alcoholic beverages as part of the food listings, whereas the case-control study questionnaire included specific questions on alcohol consumption. A specific question on the use of dietary supplements was included only in the Santa Rosa questionnaire. Body mass index (BMI) could be evaluated only in the case-control study questionnaire, whereas physical activity could be derived from calculations based on BMI (see Discussion).
View this table:
TABLE 2 . Differences in the completeness with which cancer-prevention information could be covered by both the field-study and case-control study questionnaires1
Validity of self-reported height and weight
Only 21 of the 28 women interviewed were willing to report their weight, and only 7 of the 28 reported their height; all but 1 of the 22 men reported their weight and all 22 reported their height (Table 3). The differences between self-reported and measured height were minimal between those from rural and urban areas (-0.08 ± 0.08 and -0.02 ± 0.02 cm, respectively), whereas the differences in weight were greater (2.2 ± 4.8 and 0.04 ± 5.8 kg, respectively). When BMI was calculated from these self-reported data, the relative SDs were reduced for those from rural and urban areas (-3.3 ± 1.6 and –0.7 ± 2.4 kg/m2, respectively).
View this table:
TABLE 3 . Data obtained for the validity of self-reported height and weight1
Food handling and preparation
When those interviewed were asked about their cooking methods and the types of foods they cooked, the most common responses were gas stoves for boiling food and open pit fires for cooking tortillas. When subjects were asked whether they had electricity and owned a refrigerator and, if so, what food items were usually kept cold, most households answered that they did have electricity, but that they did not have a refrigerator. Finally, when asked where and when they obtain maize and beans, subjects responded that they frequently bought these staple items in the local market. Maize was either purchased every 2 or 3 d or harvested and kept in either plastic bags or in a granary throughout the year. Beans were bought once or twice a week, soaked and cooked, and kept in cooking pots on the stove until consumed.
DISCUSSION
Cancer is no longer a novelty in countries characterized by epidemics of infection and food shortages. Control of infection and famine since World War II has led to a rapid extension of longevity worldwide (16). More persons in developing countries are living longer, placing them at a higher risk of cancer. Such individuals are thus confronted with environmental risk factors that may have always existed and those that may have arisen in the wake of expanding technology. Wherever cancer is a concern, however, diet and lifestyle should play a role in providing some form of protection from risk of this disease.
Although statistical methods deal with group data, cancer prevention operates on the level of the individual person. Hence, any strategy for applying a scoring of concordance with the preventive practices of dietary guidelines must find a way to make a categorical classification of the individual in terms of concordance behaviors. We evaluated whether we could assess cancer prevention at the level of the individual and apply the standard FFQ to the setting of a low-income country like Guatemala.
The 2 FFQs did not address physical activity. However, it is theoretically possible to estimate the physical activity level from average daily energy expenditure, age, sex, weight, and height. The basal energy expenditure can be estimated from published equations, eg, Harris-Benedict (17) and Schofield (18). Physical activity level is the difference between estimated energy consumption (from dietary data) and basal energy expenditure (from anthropometry). Weight and body composition would have to be assumed to be stable and any over- or underestimation of true energy intake would produce reciprocal errors in physical activity levels. This was likely the case in the present study because FFQs tend to overestimate energy intake (13).
One of the shortcomings of the diet-related cancer guidelines, as applied to developing countries, comes with respect to dietary supplement use. The apparent proscription of the guidelines is clarified as being a relative one, suggesting that "dietary supplements are probably unnecessary and possibly unhelpful in the context of cancer prevention" (9). Meanwhile, the guidelines recognize that developing countries have underlying endemic conditions of micronutrient deficiency that might need to be addressed by the consumption of additional micronutrients. Knowledge of the different classes of foods provided in the FFQs makes it possible to use regional food-composition tables, such as those developed at the Institute of Nutrition of Central America and Panama (19), to generate quantitative and proportional data required to estimate nutrient adequacy.
It is clear that the FFQs already used in Guatemala, such as the ones evaluated in the present study, can be used to assess whether there is concordance with WCRF/AICR guidelines (9); however, some modifications or additions may be pertinent to obtain more appropriate and complete data. For example, the data we obtained with our formative research approach indicated that self-reported weight and height cannot be used as a basis for establishing concordance with WCRF/AICR guidelines, particularly in women in Guatemala. This differs from the findings of a study by Tavano-Colaizzi (15) in Mexican women in which the reported and measured weights and heights did not differ significantly. Finally, to maximize the data capture for evaluating concordance with the WCRF/AICR guidelines in populations, some new questions on salt intake and food handling and preparation should be developed and added to the existing FFQ.
ACKNOWLEDGMENTS
We appreciate the wise counsel and encouragement of Geoffrey Cannon. We thank Elena Kourchenko, Ana Garcés, and Manolo Mazariegos for their collaboration in the formative research issues.
REFERENCES