Estimation accuracy of bean bags as portion size estimation aids for amorphous foods

Figures & data

Figure 1: Bean bags: description of set and uses in dietary assessment and education.

Table 1: Portion size estimation accuracy of bean bags: Comparative summary of sub-studies (N = 541)¹

Developmental evaluation sub-study
Design		Observational			Experimental
Sub-study number		1	2	3	4	5	6
Objective: ‘To determine … ’		… elderly people’s portion size estimation accuracy of a fixed volume of four food types when using bean bags as PSEAs	… female adults’ portion size estimation accuracy of 3D-printed amorphous food models when using bean bags as PSEAs (accounting for BMI^2, energy density and perceived healthfulness of the reference foods)	… middle and secondary school learners’ portion size estimation accuracy of projected photographs of amorphous foods when using bean bags as PSEAs	… the effect of four plate characteristics on portion size estimation accuracy of a fixed volume bean bag by university students: a quasi-experimental study	… the effect of bean bag colour (beige vs. black) on portion size estimation accuracy of three portion sizes of mashed butternut by university students: a randomised control, cross-over experiment	… university students’ portion size estimation accuracy of two portion sizes of three forms of maize meal porridge when using bean bags and food photographs as PSEAs: a comparative effectiveness study
Participants	Description	Elderly (mean age: 80 ± 7.4 years)	Teachers (mean age: 44.5 ± 12.3y; mean BMI 22.5 ± 4.6 kg/m^2)	13–19-year-old middle and secondary school learners	University students (mean age: 20 ± 1.7 years)	University students (mean age: 21.2 ± 3.0y).	University students (mean age: 20.9 ± 2.0 years)
	Setting	Lounge of old-age home	Staff room of 3 urban schools	School classroom: 6 sessions (Group size: 4–12)	University campus (open air)	Dining hall (university residence)	Dining hall (university residence)
	Sample (size and sex)	n = 30 (87% female)	n = 72 (100% female)	n = 51 (37% female)	n = 184 (52% female)	n = 100 (65% female)	n = 104 (56% female)
METHODS: Reference object description	Food (or alternative)	Two actual food examples of each food type: amorphous masses (AM), amorphous pieces (AP), solids (sS) liquids (L) (more details below)	3-D printed, realistically coloured, life-size food models of 12 different amorphous foods Mounds/masses (AM): Higher energy density: Stiff maize meal porridge^3 (375 ml), cooked minced meat (250 ml), scrambled eggs (190 ml) Lower energy density: Soft maize meal porridge^3 (375 ml), cooked rice (250 ml), cooked spinach (190 ml) Pieces (AP): Higher energy density: Potato salad with mayonnaise (500 ml), peanuts (125 ml), Smarties sweets (60 ml) Lower energy density: popcorn (500 ml), grapes (125 ml), mixed vegetables (60 ml)	Nine different eat-ready actual amorphous foods from different food groups, regularly (R) or less often consumed (LC): Rice (250 ml; 60 ml) (R) Spaghetti (625 ml; 375 ml) (R) Cereal (corn flakes) (500 ml; 125 ml) (LC) Chicken curry (625 ml [x2]; 250 ml [x2]) (R) Minced meat (625 ml; 375 ml) (R) Eggplant (aubergine) (250 ml; 60 ml [x2]) (LC) Mixed, diced vegetables (375 ml; 60 ml) (R) Diced kiwi fruit (125 ml [x2]) (LC) Yoghurt (500 ml; 125 ml) (R)	Bean bag, beige-coloured (generic, uniform proxy for amorphous food mass)	Actual food: mashed butternut (AM) Black bean bags as PSEA control	Three forms of actual maize meal porridge (soft, crumbly and stiff) Life-size, individual, colour photograph of each bean bag volume of each porridge form (aerial and angled 45° view) Both reference objects cooked according to same, standardised recipes^3
	Portion size	All 190 ml	Same as bean bags: 60 ml, 125 ml, 190 ml, 250 ml, 375 ml, 500 ml (each portion size 2x)	60 ml (4x), 125 ml (4x), 250 ml (4x), 375 ml (3x), 500 ml (2x), 625 ml (3x)	250 ml	Mashed butternut: 125, 250, 500 ml Black bean bags: same as beige set	For actual food (porridge): 250  and 500 ml For porridge photographs: same as bean bag set
	METHODS: Data collection	Three sets of plated foods typically served for breakfast; each set presented on a tray, i.e. tea (L), soft porridge^3 (AM), diced apple (AP) whole apple (S), fruit juice (L) scrambled eggs (AM), bread (S), grapes (AP) Individual administration; researcher-assisted; one food set shown at a time for 30 seconds and then removed Bean bags presentation: random arrangement on neutrally coloured surface Reproducibility: Not performed	Individual (separated) booths, with reference objects individually presented on identical plates Each booth own set of test objects, randomly laid out on neutral-coloured surface Reproducibility: Three food models presented twice; completed by all participants	PowerPoint presentation of food photographs (aerial and 45° angled view per slide, food dished up on black plates or bowls on white background using standardised photographic technique) displayed for 30 seconds Each photograph (slide) included a spoon as reference size Bean bag sets, randomly ordered, and the reference spoon were physically on display next to participants (not plated) Total: 20 slides Reproducibility: Three photographs (different foods and different portion sizes) were shown twice (randomly placed in the slide show)	Response to ‘What is the size of the beanbag on the plate?’ given on 4 response options provided Individual presentation at 13 booths: Size: 6 plate pairs: 20 vs. 25 cm diameter Colour: 4 plate pairs: white vs navy Pattern: 2 plate pairs: navy ornaments on white vs white Rim: 4 plate pairs: presence or absence of rim Five distractor bean bag volumes (125–500 ml) randomly presented in between the 250 ml fixed-size bean bag Reproducibility: Two randomly placed duplicate stations completed by all participants	Separate stations for beige and black (both brown background) Bean bags individually placed on identical, white plates, randomly arranged and spread out on each station (table) Individual portions of butternut on same plates as bean bags presented one after the other Random sequence of crossover (beige vs. black station) Reproducibility: Ten participants repeated whole assessment after 2 days	On cream-coloured, commonly used, enamel plates (24 cm diameter; identical for actual food and photographs) 12 + 3 separate stations (tables, i.e. 3 foods x 2 portion sizes x 2 PSEA + 3 repeats) Random presentation of food type, portion size and PSEA Reproducibility: Three test-retest repeats (different forms of porridge, portion sizes and PSEA)
RESULTS:	Portion size estimation accuracy of bean bags as test objects	Total: 240 estimations Perfect: 22% (AM & S); 23% (L); 37% (AP) Acceptable: > 73% for each of the food types Trend: Overestimation of AM Underestimation of S and L	Total: 864 estimations Perfect: 29% Acceptable: 73% Slight overestimation predominated, especially of AP No obvious association with BMI category^2, energy density, or perceived healthfulness Reproducibility: about 50% of participants gave identical answers in the two administrations If slight misestimation is included: > 85%	Total: 867 estimations The mode of learners estimated 19/20 (95%) of food photographs perfectly or acceptably The secondary school learners performed slightly better The largest portion size (625 ml) had lowest % (range: 2–25%) of perfect estimations No obvious relation to sex, food group or frequency of consumption Reproducibility: Perfect for 2/3 (67%) of estimations	Total: 2392 estimations Only the size of the presentation plate influenced estimation accuracy Large (25 cm diameter) plates yielded a larger proportion of perfect estimations Reproducibility: 67% of participants gave identical answers on two administrations (no gross estimation changes)	Total: 600 estimations Beige: 35% Perfect Black: 13% Perfect Overestimation: Beige: 58% Black: 78% Largest portion (500 ml) most over-estimation Reproducibility: Almost 50% of participants changed response in second administration	From all 1243 estimations: Perfect: 37% Total underestimation: 27% Total overestimation: 36% Bean bags only: Perfect: 28% Acceptable: 83% Total underestimation: 20% Total overestimation: 52% Photographs only: Perfect: 47% Acceptable: 91% Total underestimation: 34% Total overestimation: 19% Reproducibility: Perfect agreement between administrations: 28% and 51% for bean bags and photographs respectively
REC Approval number		22/2016	4/2017	14/2016	631/2018	800/2019	567/2017

¹ More details per sub-study available in Supplementary file 1.

² BMI: body mass index (body weight in kg divided by height in metres squared); body weight and height measured by researchers; BMI categories: < 18.5 kg/m² (underweight), 18.5–24.9 kg/m² (normal weight), 25.0–29.9 kg/m² (overweight) and ≥30 kg/m² (obese).

³ Wolmarans P, Danster N, Dalton A, Rossouw K, Schönfeldt H (eds). Condensed Food Composition Tables for South Africa. Medical Research Council, Parrow Valley, Cape Town, 2010. (http://safoods.mrc.ac.za). Ratios (w/w) of water:maize meal: Stiff porridge: 73:27; Soft porridge: 88:12; Crumbly porridge: 53:47.

Abbreviations used: PSEA: portion size estimation aid; 3-D: three-dimensional; AM: amorphous masses; AP: amorphous pieces; L: liquids; LC: less consumed; S: solids; y: years; R: regularly consumed; REC: Research Ethics Committee of the Faculty of Health Sciences, University of Pretoria, South Africa.