Validity of electronic diet recording nutrient estimates compared to dietitian analysis of diet records: randomized controlled trial

Abstract

Background: Dietary intake assessment with diet records (DR) is a standard research and practice tool in nutrition. Manual entry and analysis of DR is time-consuming and expensive. New electronic tools for diet entry by clients and research participants may reduce the cost and effort of nutrient intake estimation.

Objective: To determine the validity of electronic diet recording, we compared responses to 3-day DR kept by Tap & Track software for the Apple iPod Touch and records kept on the Nutrihand website to DR coded and analyzed by a research dietitian into a customized US Department of Agriculture (USDA) nutrient analysis program, entitled GRAND (Grand Forks Research Analysis of Nutrient Data).

Methods: Adult participants (n=19) enrolled in a crossover-designed clinical trial. During each of two washout periods, participants kept a written 3-day DR. In addition, they were randomly assigned to enter their DR in a Web-based dietary analysis program (Nutrihand) or a handheld electronic device (Tap & Track). They completed an additional 3-day DR and the alternate electronic diet recording methods during the second washout. Entries resulted in 228 daily diet records or 12 for each of 19 participants. Means of nutrient intake were calculated for each method. Concordance of the intake estimates were determined by Bland-Altman plots. Coefficients of determination (R(2)) were calculated for each comparison to assess the strength of the linear relationship between methods.

Results: No significant differences were observed between the mean nutrient values for energy, carbohydrate, protein, fat, saturated fatty acids, total fiber, or sodium between the recorded DR analyzed in GRAND and either Nutrihand or Tap & Track, or for total sugars comparing GRAND and Tap & Track. Reported values for total sugars were significantly reduced (P<.05) comparing Nutrihand to GRAND. Coefficients of determination (R(2)) for Nutrihand and Tap & Track compared to DR entries into GRAND, respectively, were energy .56, .01; carbohydrate .58, .08; total fiber .65, .37; sugar .78, .41; protein .44, .03; fat .36, .03; saturated fatty acids .23, .03; sodium .20, .00; and for Nutrihand only for cholesterol .88; vitamin A .02; vitamin C .37; calcium .05; and iron .77. Bland-Altman analysis demonstrates high variability in individual responses for both electronic capture programs with higher 95% limits of agreement for dietary intake recorded on Tap & Track.

Conclusions: In comparison to dietitian-entered 3-day DR, electronic methods resulted in no significant difference in mean nutrient estimates but exhibited larger variability, particularly the Tap & Track program. However, electronic DR provided mean estimates of energy, macronutrients, and some micronutrients, which approximated those of the dietitian-analyzed DR and may be appropriate for dietary monitoring of groups. Electronic diet assessment methods have the potential to reduce the cost and burden of DR analysis for nutrition research and clinical practice.

Trial registration: Clinicaltrials.gov NCT01183520; https://ichgcp.net/clinical-trials-registry/NCT01183520 (Archived by WebCite at http://www.webcitation.org/6VSdYznKX).

Keywords: diet records; electronic data; nutrition assessment.

Conflict of interest statement

Conflicts of Interest: None declared.

Figures

Figure 1

Percentage agreement between electronic methods of diet recording and dietitian-entered 3-day diet record (DR) comparing Nutrihand and Tap & Track to values obtained from Grand Forks Research Analysis of Nutrient Data (GRAND).

Figure 2

Percentage agreement between electronic methods of diet recording and dietitian-entered 3-day diet record (DR) comparing Nutrihand to values obtained from Grand Forks Research Analysis of Nutrient Data (GRAND).

Figure 3

Bland-Altman plots comparing electronic diet entry by participants to dietitian entry of 3-day diet record (DR) into Grand Forks Research Analysis of Nutrient Data (GRAND). Plots for energy and macronutrients comparing Nutrihand and Tap & Track to GRAND. Solid horizontal line indicates mean of differences between Nutrihand or Tap & Track and GRAND. Upper and lower limits of agreement (dashed lines) define range within which most differences between methods are expected to occur. Dotted line at y=0 is given for reference.

Figure 4

Plots comparing nutrient intakes estimated from 3-day diet records coded by investigator to same 3-day diet records entered electronically: Comparing Nutrihand (black circle) or Tap & Track (grey square) to Grand Forks Research Analysis of Nutrient Data (GRAND). Each point represents mean of food records for 3 days for each individual (n=19). Regressions comparing intake estimates from Nutrihand (solid line) and Tap & Track (dashed line) to estimates obtained from investigator-coded records were performed and R2 values are reported. *Statistical significance at P<.05.>

Figure 5

Plots comparing nutrient intakes estimated from 3-day diet records coded by investigator to the same 3-day diet records entered electronically: Comparing Nutrihand (black circle) to Grand Forks Research Analysis of Nutrient Data (GRAND). Each point represents mean of food records for 3 days for each individual (n=19). Regressions comparing intake estimates from Nutrihand (solid line) and Tap & Track (dashed line) to estimates obtained from investigator-coded records were performed and R2 values are reported. *Statistical significance at P<.05.>