Comparison of Household Measures to Food Photographs in Portion Size Estimation of Protein Food Sources among Adults in Spintex-Accra, Ghana.

Abstract

Background: Accurate estimation of portion size remains a challenge in dietary assessment. The Photographic Food Atlas (PFA) has been identified to be a more convenient tool compared to other portion size estimation tools. While an atlas has been developed and validated for carbohydrate and protein source Ghanaian foods, the accuracy of estimations made with the photographic atlas is yet to be tested against that of household measures (HHM) commonly used by dietitians and other nutrition professionals. Aim: This study aimed to assess the accuracy of portion size estimates of commonly consumed protein-source foods using a Photographic Food Atlas in comparison with household measures. Method: This was a cross-sectional study with participants from churches selected in the Spintex area in Accra, Ghana. Data was collected in September 2022 with a structured questionnaire and analyzed with STATA 16.0 software. Simple random sampling was employed during the first two stages of the study. First, a community within Spintex Accra was selected. A list of all the churches within the selected community (Nungua) was compiled to form a sampling frame for the second stage. Following that, study participants were selected by systematic random sampling, using the calculated sampling interval specific to the selected church using a structured questionnaire. Logistic regression analysis was conducted to assess the association between participants’ characteristics and the accuracy of portion size estimations using the PFA. Also, a one sample t-test was conducted to determine differences in the portion size estimation by the assessment methods with the true weights of the food items at a statistical significance of p≤0.05. Results: A total of 70 participants (mean age = 32.66 ± 11.4 years) were recruited. Altogether, 560 portion size estimations were recorded for the boiled egg, chicken, Atlantic mackerel, and groundnut soup samples, using the Photographic Food Atlas and the Household Measure (i.e., 280 estimations per portion size estimation aid). These estimations showed an overall accuracy rate of 30% and 35.7% respectively for PFA and HHM. Portion Size Estimation Aids (PSEA) significantly influenced the weight of the food item predicted by the participants. All the mean estimates showed that for all food items, portion size estimations that participants made using either of the PSEAs differed significantly (p=0.001) from the actual weights of the foods. Use of the PFA resulted in more overestimation of portion size than that of the HHM, for example, the percentage deviation for fried chicken thigh with the use of the PFA was 107% above the true weight (45g), compared to 35.7% for HHM. Lastly, participants’ characteristics such as age, sex, BMI, and educational background were independent of how participants estimated the food samples presented to them. Conclusion: Although most food samples in the study were estimated accurately using the photographic food atlas, the smoked fish sample (Atlantic Mackerel), a major protein source in the Ghanaian diet, had a higher proportion of accurate estimation when HHM was used, which makes the household measure (matchbox) a more suitable PSEA in estimating fish in dietary surveys and assessment in Ghana. Participants’ factors such as age, sex, BMI, and educational background were not significantly associated with participants' ability to accurately estimate portion sizes of food samples presented to them.