Dietary Assessment Literature Review:
This page links to some files in Portable Document Format (PDF).
The most validation work in this age group has focused on assessing infant milk intake by test weighing. This method involves weighing the infant immediately before and after each feeding without change of clothing or diapers and taking the gain in weight of the infant (in grams) to be the net milk intake (in milliliters). An alternative approach in breastfed infants involves weighing the mother before and after each feeding (75). The introduction of electronic balances, which can integrate moderate movements and record these weights, has improved the accuracy and precision of measuring the weight of the infants (76;77).
Scanlon et al. published a thorough review of test weighing validation studies published through 2000 (78). Additional work in this area was not identified. Test weighing of formula-fed preterm and full-term healthy infants (Table 3.1) in the hospital by nursing staff using an electronic scale showed agreement between test weighing of the infant and the direct measurement of formula within 1 percent (79). In home settings including five to 10 mother-infant pairs, infant test weighing and formula measurements by the mother underestimated intake by 7 to 10 percent using a mechanical scale (80) and overestimated intake by 7 to 11 percent using an electronic scale (75).
Test weighing validation studies in breastfed infants have focused on modifications of procedures to reduce the maternal burden and disruptions of feeding. Results of three studies (31;81;82) examining whether breast milk intake could be estimated from the product of test weights for one or two feeds in a 24-hour period found the highest correlations between intakes estimated with 24-hour test weighing and estimates calculated from two consecutive test weights in the mid 24-hour period. Differences in mean intake estimates ranged from a 0.6% overestimation among infants 4 weeks of age to an 6% underestimation among infants 12 weeks of age (31). Meier validated the accuracy of home test weighing by mothers using the Baby Weigh electronic scale in a population of pre-term breastfeeding infants (76).
The test weighing method has several obvious limitations for a large-scale longitudinal study. Test weighing is tedious and requires careful training and supervision of mothers with some degree of technical sophistication who can operate an expensive electronic balance in the home (82). Test weighing also interrupts usual feeding routines. When milk intakes of breastfed infants are compared to those of formula-fed infants, both groups of infants should be test weighed (80). No studies have validated test weighing with combined feeding regimens (formula and breastfeeding).
DLW (Doubly Labeled Water) Method
Infant milk intake indirectly estimated from measurements of infant total energy expenditure (TEE) with the DLW method has been validated in small groups of formula-fed (83-86) and breastfed infants (83;87;88) in hospital and home settings (Table 3.1). The method involves carefully (avoiding loss from spitting up) administering a DLW dose to the infant and collecting samples of urine or saliva at baseline and over the subsequent 5 to 15 days. To increase accuracy of energy expenditure measurements, water from supplemental foods or fluids other than milk must be measured and adjusted for, as well as environmental water influx, insensible water losses, change in energy stores during the study period (change in weight), and the macronutrient content of the diet. The method has been refined over time and later studies, correcting for environmental water influx and insensible water loss, found close agreement (1 to 2% in formula studies and 2 to 5% in lactation studies) between energy intake estimated by the DLW method and direct measurement of formula or test weighing of breastfed infants.
The DLW Method has a number of advantages because it is non-invasive and requires no special equipment. The method does not interrupt infant feeding patterns, it allows for greater mobility of the mother-infant pair, it is unaffected by daily variations in intake or frequent feedings, and is practical under field conditions (87). However, the availability and cost of the isotopes, the need for sophisticated laboratory analysis, and the care required to administer the DLW dose, limit its use in large samples of infants.
Direct observation involves estimating the volume of breast or formula milk consumed by visually assessing the infant during feeding. Studies by Meier on preterm infants and/or high-risk infants found low correlations (0.48 to 0.79) and large and random errors between direct observation and test weighing when observations were performed by either mothers, nurses, or lactation consultants (89). Mothers and investigators gave comparable, yet inaccurate, estimates of infant milk intake over a single feed (r = 0.91) demonstrating that direct observation cannot be substituted for test weighing if an accurate measure of infant intake is necessary.
Only six studies examining the validity of other dietary assessment methods in older infant populations were identified (Table 3.1). A 2001 study compared energy intake measured by a 5-day estimated Food Record with a 5-day weighed Food Record and the DLW method in a cross-over study design in 6- to 12-month old infants (90). Both weighed and estimated food records overestimated DLW measurement of energy expenditure by 7%. A diet history method was compared with a weighed food record in two studies (91-93); although the diet history methods were not comparable, both overestimated intake measured with a 3- or 4-day weighed food record. The use of the Portable Electronic Tape Recording Automated (PETRA) scale in the home was found to be difficult in a British study of children from low-literacy Asian immigrant households because the equipment malfunctioned or was damaged in the home and it required intensive participant instruction and monitoring (92).
Though not validation studies, Stuff et al. (94) and Black et al. (95) each studied the day-to-day variation in energy intake of breastfed infants through rigorous tests weighing (Table 3.4). In both studies, the range of pooled within-subject coefficient of variation was wide and increased as the infant aged and more complementary foods were introduced. Black's study includes measurements through 18 months and concluded the number of days of food records needed for breastfed infants is 4 days and for toddlers is 7 days (95).
Two studies examined the validity of the 24HR (24-Hour Recall) method. In one study a 24HR collected 24 hours after collection of a duplicate diet by the parent resulted in a significant overestimation of energy and other nutrients (96). A study validating telephone 24HR interviews with face-to-face 24HR in telephone and non-telephone households in the lower Mississippi Delta Region found no significant differences in mean energy intakes, but the results for the 32 infants included in the study of 409 participants were not analyzed separately (97).
Only one FFQ validation study was found. Marshall (98) compared parental reports of beverage intake of infants at 6 and 12 months on a mailed beverage FFQ (Food Frequency Questionnaire) with a 3-day FR (Food Record) of all foods and beverages consumed. This FR was completed the week after completing the FFQ. Correlations with types of milk consumed ranged from 0.83 to 0.99 while correlations between methods for measurements of water, juice/drinks, or soft drinks were lower.
In the early 1980s, a study comparing an interview that included short questions on breastfeeding practices with the infant's medical record found mothers overestimated reporting of length of previous breastfeeding when questioned at 12 months (99).
28 Dec 2010