Skip to Main Content
An official website of the United States government
Epidemiology and Genomics Research Program

Pregnancy - Background & Validation Studies - NCS Dietary Assessment Literature Review

See Pregnancy and Lactation tables

Pregnancy Background

The most intensive period of human growth and development occurs during the nine months of pregnancy (11). Assessing the food and nutrient intake of pregnant women is complicated because conception triggers an array of complex and sequential physiological changes that affect maternal nutrient absorption and metabolism, energy and nutrient needs, appetite, and meal patterns (12). Individual physiologic and behavioral responses to the stress of reproduction vary widely (13), and both the type and amount of food consumed may fluctuate depending on the period of pregnancy. Nausea, which is estimated to occur in 50-80 percent of pregnancies (11), may begin as early as 4 to 6 weeks gestation, usually peaks around 8 to 12 weeks, and then declines thereafter. Heartburn and constipation are other common ailments that may trigger changes in usual food habits. Pregnant women may develop food preferences and aversions due to changes in the sense of taste and smell. Pica, an eating disorder characterized by the compulsion to eat substances that are not food, affects more than half of pregnant women in some locations in the southern part of the United States (11). Maternal concerns about optimal fetal growth and development, weight gain, and parenting may increase maternal awareness of nutrition and focus personal and health care provider attention on food habits and supplement consumption. Misrepresentation of intake, particularly with regard to alcohol and supplements, may be an issue because of respondents' fear of noncompliance.

Validation Studies in Pregnant Populations

Validation of dietary assessment methods in pregnant women is limited. Table 2.1 presents a summary of 15 validation studies; additional information on each study is included at the end of this chapter in Table 2.2. Two studies examining energy expenditure using the DLW (Doubly Labeled Water) method to measure TEE (Total Energy Expenditure) also evaluated reported energy intake. In 12 Cambridge women, reported energy intake from 7-day weighed food records underestimated TEE measured by the DLW method by 6 to 15 percent in the first trimester, 12 to 18 percent in the second trimester, and 24 percent in the third trimester (14). Although large differences between individuals in the metabolic cost of pregnancy was found, application of the Goldberg cut-off limits based on Energy Intake:Basal Metabolic Rate (EI:BMR) ratios identified under-reporters. Energy intake measured by 4-day weighed food records kept before conception by 22 Swedish women, underestimated TEE at 6 to 18 weeks gestation by 27 percent (15). In a review of six different data sets containing 228 DLW TEE measurements in pregnant and lactating women or non-pregnant, non-lactating controls in Europe and Gambia (including the DLW studies in Table 1), Prentice and colleagues (13) recommend caution in the overinterpretation of dietary intake records. Their data show food intake records can be "extremely misleading and this is only revealed when there is some external validation such as the doubly labeled water method" (13).

Three other validation studies in Table 2.1 and Table 2.4 included a biological marker to assess reported intake. Erythrocyte cell membrane fatty acid content was compared with fat intake reported on an FFQ (Food Frequency Questionnaire) in a study of 185 New England women in the first trimester (21) and on 35 third-trimester Mexico City women (18). Both studies found significant correlations between erythrocyte alpha linoleic acid content and dietary intake of the fatty acid, and that women were classified into the same intake ranges by both the FFQ and this biological marker. Serum levels of caffeine and paraxanthine were examined as a biological marker to validate caffeine intake reported on a 24HR (24-Hour Recall), with correlations between the two measures comparable to correlations between reported smoking and serum cotinine in pregnancy (17).

No other studies validating energy or nutrient intake assessed by the 24HR (24-Hour Recall) method in pregnancy populations were found. The 10 FFQ validation studies in Table 2.1 (and Table 2.4) are difficult to compare because the populations differ, the FFQ instruments differ, the studies cover various periods of pregnancy, and they differ in their reference methods and in the number of days of dietary recording. There are also statistical differences between the studies. Three FFQ studies on European pregnant women compared FFQ intake with weighed or estimated FRs (Food Records) (19;20;23). Each found the FFQ overestimated energy intake, but generally classified the women into the same or adjacent nutrient intake category as the FR. In the US, six studies have examined the validity of the Harvard FFQ (HFFQ) modified in various ways for pregnancy in each study and assessing intake usually for the past 1 to 3 months. Comparisons with multiple administrations of 24HR interviews in 4 studies found the Harvard FFQ overestimated energy intake (25-28). However, compared with 4-day weighed FRs, the modified Harvard FFQ underestimated energy intake by 10 percent during mid-pregnancy in a population of educated, white Minnesota women (22). A large study of low-income pregnant women comparing the HFFQ with three 24HR interviews concluded the HFFQ provides a reasonably accurate measure for the majority of nutrients and can appropriately rank individuals relative to one another even if absolute intakes may not be precise (27;28). However, the exclusion of 14 to 18% of the women reporting caloric intakes above 4,500 on the HFFQ suggests a significant proportion of the low-income women were unable to complete the HFFQ adequately (28). The NCI-Block HHHQ (NCI Health Habits and History Questionnaire) was found more valid for white women and black pregnant WIC participants than the HFFQ, based on correlations between each FFQ and three 24HR interviews for energy and five nutrients. Neither FFQ was valid in Hispanic women (25).

Although most of the FFQ instruments in the validation studies in Table 2.1 collected information on supplement intake, validation of reported supplement intake was not discussed.