Comparing a 7-day diary vs. 24 h-recall for estimating fluid consumption in overweight and obese Mexican women

Caloric beverages, particularly sugar-sweetened beverages (SSB), represent an increasing component of our diet. Mattes and many others have shown that there are differential effects of calories consumed as a liquid beverage and a food [1‐7]. In Mexico, the United States, the United Kingdom, Europe, and elsewhere increasing concern has focused on several key components of caloric beverages, those high in refined sugars, high-fructose corn syrup, or fruit juice concentrates [8‐14]. Meta-analyses have provided evidence that high intake of SSB (e.g., soft drinks, fruit drinks) are linked to increased weight, energy intake, and diabetes [15‐17]. Additional concern is raised about consumption of other beverages such as fruit juices, alcohol, and high-fat milk due to their contribution to energy intake and its association with obesity and chronic diseases such as type 2 diabetes and dyslipidemia, among others [18‐26]. Some countries have developed programs and policies to promote water intake and limit access to SSB in schools and feeding programs, mainly through banning SSB within schools, and to restrict media related to SSB through regulation of the marketing of foods (especially to children). [27‐29].

Mexico has experienced rapid increases in SSB and caloric intake from beverages in the last 20 years [12‐14]. Furthermore, the government created a beverage guidance panel [30] and recently moved to reduce caloric intake from beverages in all age groups by taxing SSB [31]. In the Mexican government recommendations and those discussed by many other countries, water intake has been viewed as an important, viable alternative to these caloric beverages; however, there is limited information on water consumption patterns in the context of a total beverage intake pattern [11, 32‐34].

There are no entirely objective measures of an individual’s food or drink consumption except in the controlled conditions of a metabolic unit. Any self-reported measurement of intake will be biased in some way by the measurement process itself. There are two main approaches to individual assessment: prospective and retrospective, each with their advantages and disadvantages [35].

Even though the increasing interest on beverages and water intake, there are few dietary tools carefully validated for U.S. adolescents [36] and adults [37‐40]. To date no specific dietary instrument for simple collection of the full array of fluid has been carefully validated for Mexicans. For epidemiological studies, retrospective methods are usually applied. A standard 24-h recall or food frequency questionnaire is the most used; both focus on obtaining data on foods and beverages containing calories and macro- and micro-nutrients. [41]. For intervention and epidemiological studies, the use of a standard 24-h recall is more common, and the 24-h urine collection could be used as a marker of fluid intake [42], but collecting 24-h urine is not practical and therefore not always used in large epidemiological studies. The purpose of this paper is to compare a prospective method: a fluid intake 7-day diary against a retrospective one: 24-h recall questionnaire for 3 days to estimate the fluid consumption in overweight and obese Mexican women participating in a clinical trial to reduce SSBs and increase water consumption.

Women included in this validation study (n = 190) did not differ in demographic characteristics, socioeconomic status, or weight status (measured as BMI) from women excluded from this analysis (n = 50) (data not shown). The subset of women in the validation study were on average 33 y old, 26 % were nulliparous, about 63 % were married or living with someone, and most of them had at least completed high school. The mean BMI was almost 31 kg/m². They were evenly distributed between the two intervention groups (51.6 % in the intervention group and 48.4 in the control group) (Table 1). The total estimated mean energy intake was 1,524 ± 468 kcal/d, of which 16.6 % (253 ± 200 kcal/d) was coming from SSB.

This study demonstrated no significant difference in the estimated total median volume of fluid intake, expressed as mL/d between a fluid 7-day diary and the 3-day 24-h recall as the reference dietary method; however, there were significant differences in mL/d and kcal/d when analyzed by beverage categories. There were highly significant crude correlation coefficients for total fluid intake, showing a stronger correlation when reporting water, alcohol, and SSB intake. The more meaningful rank correlations were for fluid intake in mL/d: water, alcohol beverages, and SSB and for fluid intake in kcal/d: alcohol beverages and SSB (rank correlation ≥ 0.6). When adjusted for age, BMI, education, socioeconomic status, and intervention group, the correlation was still high and significant for total fluid consumption in mL/d, water, and SSB (R² ≥ 0.6, p < 0.001). Hedrick and colleagues report similar results in a cross-sectional study in healthy U.S. adults aged >21 years, determining the validity and reliability of a self-administered beverage intake questionnaire, which estimates mean daily intake of beverage consumption (g, kcal) across several beverages categories [37]. The beverage intake questionnaire showed a high correlation vs. a 4-day food intake record in daily consumption (g) of water, total beverages, and SSB. To our knowledge, there is no specific validated dietary instrument to collect fluid intake in Mexican populations.

One potential explanation of the energy intake differences in our study is the methodology of each dietary tool in the registration of detailed information about added sugar. For the 3-day 24-h recall, well-trained interviewers asked the amount of sugar added to the beverages. On the other hand the 7-day diary was a self-administered questionnaire, and in spite of the meticulous review of the questionnaires once women handed them back, when it was not specified whether beverage had added sugar or there was not an exact amount of it, it was assumed that they were SSB and a standard sugar portion was assumed for energy estimations. This hypothesis is supported by the results of beverages without added sugars. When comparing the reported energy between the two instruments, these two categories were the only ones with no statistically significant differences. The correct measurement of added sugar in beverages is particularly important in Mexico where 100 % fruit juices with water and added sugars, often called aguas frescas, represent a major component of caloric beverage intake for all age groups and added sugar is important in coffee and 100 % juices also [12‐14].

The difference on beverage intake in mL/d by type of beverage might be explained by the reported portion size. As already stated, the main difference between instruments, besides the temporality, was that the dietary information with 3-day 24-h recall was collected by well-trained interviewers with special care on portion sizes, versus the 7-day diary self-administered questionnaire.

The notable benefit of the 7-day diary was in the one kind of beverage that has received great attention in Mexico, SSB. The schools ban or limit them since 2010 [48], and recently the government approved a 10 % tax on SSBs, with the purpose of discouraging and decrease SSB consumption among the population [31]. As a less healthful beverage, research from other countries might suggest recall of SSB and other food with high energy, sugar, and fat content would be underestimated systematically [49‐52]. The highly significant crude and adjusted correlation coefficients of the 7-day diary to the reference method on this study (24-h recall) is indicative of one potential strength of the 7-day diary to measure SSB intake.

The major weakness of this study is the lack of an ideal gold standard method to record fluid intake of free-living individuals. So far, 24-h urine is one of the best markers of fluid intake in subjects with normal renal function [42]; however, it has some limitations. For individuals who perspire in abundance (i.e., active persons and athletes), 24-h urine is not an accurate marker because it does not take into account water losses via perspiration and feces [53], and 24-h urine sample collection represents a burden for participants. In this study we used as a reference measure repeated 24-h recalls. Even though this is considered an adequate reference [54], it is important to mention that it also has some limitations. The 24-h recall has inherent errors from memory, portion size estimations, and distortions of reported dietary recall, resulting in potential loss of validity as a reference method.

The fact that we used the reported and registered beverage intake from the 24-h recall and 7-day diary, respectively, after 3 months of the beginning of the intervention of the original trial might affect both internal and external validity of the study. However, there were no differences between the intervention and control groups for either total beverage intake or any beverage category (data not shown). We adjusted the regression analysis for the intervention group. For external validity, women in the original trial, even those in the control group, were receiving general information of healthy life styles and were contacted every month. The fact that women received this information and the continuous contact with the personnel from the clinical trial may have resulted in a group of women more conscious about their eating and drinking habits than the general population, thus having a better report and register of beverages and diet in general. In addition, for external validity, the fact that our study included only overweight and obese women with a high intake of SSB (>250 kcal/day) limits the generalization of the results to women with same characteristics. Furthermore, it is well recognized that the prevalence of underreporting of energy consumption and certain foods and beverages considered as “unhealthy” is higher in women and even more among overweight and obese women compared to men and normal weight women, respectively [55, 56]. One other limitation to this study was the fact that the 7-day diary was provided to women during the same week as the 24-h recall. The sequence of measurement in this study might have resulted in an increased correlation between both instruments [57]. To confirm our results, a study evaluating the methodologies in a design where subjects will alternatively filled a 3-day 24-h recall or a 7-day diary, in a randomized manner, with a “wash-out” period between the two fillings would be more appropriate.

The study has important strengths. Our sample size (n = 190 women) and number of repeated measures were appropriate, considering that a sample size of 110 subjects with at least two replicates of 24-h recall and 2 to 4 days of diet recording will provide adequate statistical power and remove the within-person variation. Overall, studies using 3 days of repeated measured per subject, 200 subjects is adequate for validation studies [54]. We evaluated potential selection bias by comparing women included in this analysis with those recruited for the original trial. There were no differences in age, marital status, or other demographic characteristics (data not shown).

Overall, the 7-day diary showed high and strong rank correlations with that reported in the 24-h recall, suggesting that the diary method is a valid tool to evaluate total fluid, water, and SSB intake in this population, but not all individual beverage categories. The gaps between mL/d and kcal/d for the key sources of caloric beverages suggest that further refinements may be useful. This could include improved participant training and more conscientious review by the field workers. Diet recording methods are appropriate instruments that do not depend on memory, so the measurement error, when used in literate and cooperative populations, might be reduced. However, there is always the problem that this instrument may influence the actual fluid intake [55]. As caloric beverages intake increase across the globe, as well as the strategies to discourage the intake of these beverages along with the water intake promotion, accurate methods to assess beverage intake are needed. Dietary tools such as the 7-day diary might be one approach for data collection for large-scale samples with similar characteristics to our study population (overweight and obese women with high intake of sugar-sweetened beverages).