The challenges of assessing fat intake in cancer research investigations.

Ecologic comparison of the incidence of cancer (eg, largescale differences between countries in the incidence of breast, prostate, and colon cancer) can be explained best by substantial differences in the intake of dietary fat. Additionally, there is a vast amount of animal and mechanistic data that strongly supports the hypothesis that ditary fat, independent of caloric intake, appears to have a major effect on the incidence and mortality rates for cancer. Yet, results from human case and cohort studies are inconsistent in linking carcinogenesis with fat intake. This is due to several factors. Reported intakes may not reflect previous long-term intakes and may be con-founded by several sources of error, including memory and estimates of portion size. Additionally, ongoing media reports of adverse health effects from high-fat diets may impart a social desirability bias to self-reporting of fat intake. These factors may be significant when investigating the relationship between dietary intake and cancer. Studies have shown considerable error in self-reported dietary data, with under-estimations in energy intake ranging from 3% to 18%. Such a wide range likely is due to differences in dietary assessment methodologies, which highlights the need to continue to develop improved techniques of data collection to relate nutrition better to health outcomes. The Women's Intervention Nutrition Study (WINS) is investigating the effect of dietary fat on the incidence of recurrence and survival in women with early-stage breast cancer. WINS is employing the multiple-pass 24-hour telephone recall system along with enhanced quality control measures to assess dietary intake. This dietary assessment method is particularly applicable when comparing two populations when one population is treated by an extensive dietary intervention.