Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 1.
Published in final edited form as: J Acad Nutr Diet. 2018 Apr 12;118(8):1482–1489. doi: 10.1016/j.jand.2018.02.002

Reliability and Validity of Digital Imagery Methodology for Measuring Starting Portions and Plate Waste from School Salad Bars

Melanie K Bean 1,✉, Hollie A Raynor 2, Laura M Thornton 3, Alexandra Sova 4, Mary Dunne Stewart 5, Suzanne E Mazzeo 6
PMCID: PMC6064651  NIHMSID: NIHMS941575  PMID: 29656934

Abstract

Background

Scientifically sound methods for investigating dietary consumption patterns from self-serve salad bars are needed to inform school policies and programs.

Objective

To examine the reliability and validity of digital imagery for determining starting portions and plate waste of self-serve salad bar vegetables (which have variable starting portions) compared with manual weights.

Design and Methods

In a laboratory setting, 30 mock salads with 73 vegetables were made and their consumption simulated. Each component (initial and removed portion) was weighed; photographs were taken of weighed reference portions and pre and post-consumption mock salads. Seven trained independent raters visually assessed images to estimate starting portions to the nearest ¼ cup and % consumed in 20% increments. These values were converted to grams for comparison with weighed values.

Statistical Analyses

Intraclass correlations (ICCs) between weighed and digital imagery-assessed portions and plate waste were used to assess interrater reliability and validity. Pearson’s correlations between weights and digital imagery were also examined. Paired samples t-tests evaluated mean differences (g) between digital imagery and measured weights.

Results

Interrater reliabilities were excellent for starting portions and plate waste with digital imagery. For accuracy, ICCs were moderate, with lower accuracy for determining starting portion of leafy greens compared with other vegetables. However, accuracy of digital imagery plate waste was excellent. Digital imagery was not significantly different from measured weights for estimating overall vegetable starting portions or waste; however, digital imagery slightly underestimated starting portions (by 3.5g) and waste (by 2.1g) for leafy greens.

Conclusions

This investigation provides preliminary support for use of digital imagery for estimating starting portions and plate waste from school salad bars. Results might inform methods used in empirical investigations of dietary intake in schools with self-serve salad bars.

Keywords: plate waste, salad bars, school lunch, nutrition policy, elementary school, food bars

There is great national support for school salad bars as a means to increase fruit and vegetable intake within the National School Lunch Program.1 Despite their increasing prevalence in schools,1 few studies have investigated the impact of salad bars on fruit and vegetable intake, and even fewer have investigated their effectiveness under the new meal standards set forth by the 2010 Healthy Hunger-Free Kids Act.2–5 To conduct such investigations, methodologically sound procedures for estimating intake from salad bars are needed to inform policies and programs designed to enhance fruit and vegetable consumption in schools.

A recent review of methods to assess children’s dietary consumption in schools reported varied outcomes, with meal observations (on-site or digital images) offering the highest interrater reliabilities.6 However, there are few reports of the accuracy of dietary assessment methods in the school context. Further, few studies included school salad bars. Obtaining weights of food waste has been considered the gold standard; however this method is time, cost, and labor-intensive and not conducive to large-scale data acquisition.7 Visual estimation on-site and digital photography overcome many of these challenges.7–11 For example, a recent study demonstrated that on-site visual estimation is a valid and reliable method for measuring plate waste in a cafeteria setting.11 However, this method can only be used when the starting portion is standard, and is not feasible with self-serve items of various portion sizes, such as salad bars. Digital imagery has the potential to address this challenge.

Digital imagery plate waste assessment methods9,12,13 include labeling trays and taking a photograph from a standard angle prior to and after consumption; each pair of images is then viewed in the laboratory and plate waste visually estimated. The primary benefit of digital imagery is the ability to obtain a large amount of data in a faster and less disruptive way than manually weighing; it also has the benefit of conducting the rating via viewing the photographs in the unhurried laboratory setting, a sharp contrast with the noisy, hectic cafeteria environment. The variety of foods selected can also be easily recorded using this method. Although studies have supported the reliability and validity of digital imagery for determining plate waste with school lunches,9,12,13 its use to estimate variable starting portions, in addition to consumption, from self-serve salad bars, has not been empirically supported. Indeed, Taylor et al9 reported that foods served in varying portions (e.g., leafy greens) were more challenging to assess via digital imagery. Given the variable starting portions of salad bar items (as students self-serve their plate), determining the starting portion is essential, particularly if assessment of energy intake (or other nutrition information) is desired. Prior investigations using digital imagery plate waste7,14 might not translate to salad bars, given the potential for self-serve salad to contain overlapping vegetables, be obscured by salad dressing, or vary widely in composition and portion sizes. Thus, the current study compared digital imagery (i.e., laboratory-based coding of digital images before and after consumption) to measured weights and investigated if digital imagery can accurately and reliably determine 1) starting portions, and 2) plate waste from self-serve salad bars.

METHODS

This study was conducted by trained research staff at Virginia Commonwealth University in preparation for a larger investigation of the impact of school salad bars on dietary consumption patterns among Title I elementary school students. The current validation study was conducted in the laboratory setting and did not involve human subjects, thus was not subject to Institutional Review Board approval.

Preparation of Weighed Reference Portions and Test Salads

Salad bar menus were obtained from School Nutrition Services in the school district where the trial was going to occur, to identify all vegetables offered on the salad bars, including recipes for mixed items (e.g., corn and black bean salsa). Only fruits and vegetables are served on the salad bar (e.g., no meats or grains). Fruit on the salad bars is served cupped in a standard 4oz serving; thus only vegetables were used in this investigation. Study staff recreated salad bar vegetables using identical preparation methods as those used in the target schools (e.g., diced, sliced, or whole). In the schools, students use a ¼ cup spoodle (serving spoon) and are allowed up to three spoodles (6oz) of vegetables. For this study, two trained research staff independently served three portions (¼ cup, ½ cup, and ¾ cup) of each vegetable into paper boats (identical to those used in the schools), for consistency with school presentation and portion parameters. Each portion was labeled and photographed to serve as a reference (Figure 1). Portions were weighed in triplicate to the nearest gram, and the average used as the measured weight of each vegetable.

Figure 1.

Figure 1

Open in a new tab

Column 1 presents samples of three reference portions of spinach in: a) ¼ cup; b) ½ cup; c) ¾ cup portions. Column 2 provides a sample mock salad at: d) pre-consumption; e) post-consumption without dressing; and f) post-consumption with salad dressing. Mock salads and reference portions were created for use in a simulated laboratory-based study to establish the reliability and validity of digital imagery for determining starting portion size and waste from elementary school self-serve salad bars.

Study staff viewed 44 pairs of images (“pre” and “post-consumption”) of students’ salads from a prior school lunch investigation to observe typical portion and consumption patterns. Investigators trained study staff to generally adhere to student portion guidelines (1–3 spoodles) and then staff then made 30 mock salads in paper boats using various portions of the 16 vegetables (up to three vegetables per salad) that would be served on the salad bar. Each ingredient (n=73 across the 30 salads) was weighed in triplicate to the nearest gram and an average calculated to determine the reference weight. This weight was then converted to cups. Researchers then simulated consumption by removing variable amounts of vegetables from the salads, weighed these “consumed” amounts in triplicate and calculated waste (average grams remaining for each ingredient). Weighed % waste was determined for each salad ingredient: (starting portion [g]–portion consumed [g])/starting portion [g])*100).

Photographs were taken of each salad prior to and after “consumption.” (Figure 1). To simulate presentation at school which might include salad dressing (each student is permitted up to two 1.5oz low-fat ranch salad dressing packets, which would only be applied in the “post-consumption” image), dressing was then added, and a second image taken. All photographs were taken from ~45° angle using iPads, consistent with methods from prior studies.15 Photographers were instructed to ensure all four corners of the salad boat were in the image prior to taking the photograph, for size consistency. All weights were taken in triplicate to the nearest gram and averaged using the MY Weight Food Scale 3001P (Phoenix, AZ). Scales were tared before each use to remove the weight of the salad boat.

Digital Imagery Assessment of Portion Sizes and Waste

Seven undergraduate students, who received research credit for participating, were trained on digital imagery methods by the lead investigator. Training included an overview of rating methods, presentation of reference portion photographs, and group viewing of sample images and determination of waste in 20% increments. Raters then independently viewed 114 pairs of images (“pre” and “post-consumption”) from a prior school-based investigation (44 of which included salads) and assessed the starting portions (salad bar vegetables only) to the nearest ¼ cup (¼ cup, ½ cup, ¾ cup), and % waste (all vegetables) in 20% increments. Visual stimuli (pie charts) assisted raters in making judgments.16 Two study investigators had previously rated each image (independently) to determine gold standard ratings for each item. Any disagreements among investigators were discussed and resolved, with a third investigator available if discrepancies could not be resolved (although she was not needed). Intraclass correlations (ICCs) were calculated, with a-priori ICC>.80 established as the criterion indicating readiness to rate in the current study. This level was achieved after one pass, with no re-training needed.

Trained raters then independently viewed photographs of “pre” and “post-consumption” salads simultaneously to estimate starting portions and % waste of each vegetable. Salads with and without dressing were rated on separate occasions to avoid bias. Photographs of reference portions (¼ cup, ½ cup, ¾ cup) were used to assist in determining starting portions of each vegetable to the nearest ¼ cup. Raters then assessed % left uneaten (% waste) of each ingredient in 20% increments. Raters could also select “cannot determine” if the starting portion or waste estimate could not be made (e.g., item was obscured). To facilitate comparison with weighed starting portions, digital imagery estimates were converted to grams based on average weight per cup. Agreement among raters (interrater reliability) and comparison between digital imagery and weighed assessments of portions and waste (validity) were calculated.

Statistical Analysis

Statistical Package for the Social Sciences (v.23.0, SPSS Inc)17 was used for analyses, with P<0.05 required for significance. Based on different visual presentations, vegetables were categorized as “Leafy Greens” (spinach and romaine) and “Other Vegetables” (cucumbers, bell peppers, broccoli, petite carrots, cauliflower, zucchini, squash, grape tomatoes, sweet potatoes, kidney beans, peas, corn, garbanzo beans, corn and black bean salsa). ICCs were used to determine both interrater reliability (IRR) and validity of starting portion and plate waste estimates (both assessed as interval data).18 ICCs were considered fair if 0.41–0.60; moderate if 0.61–0.80; and excellent if 0.81–1.0.19 IRR of digital imagery portions (to the nearest ¼ cup) and % waste (in 20% increments; with and without salad dressing) was determined by calculating ICCs among the seven raters, using a fully crossed design.18 Two-way models were applied: all raters assessed each plate, and absolute agreement was examined.18 IRRs were examined overall as well as for vegetable subgroups (Leafy Greens and Other Vegetables).

Validity was investigated with ICCs comparing digital imagery to the weighed reference value. Specifically, ICCs assessed agreement between 1) weighed and digital imagery-estimated starting portions; and 2) weighed and digital imagery-estimated % waste. Validity was examined overall and for vegetable subgroups. Pearson’s correlations between digital imagery and weights were also examined, consistent with prior studies, to facilitate comparison.9 ICCs are preferable to correlations, because they provide a measure of agreement with the reference value (weights) and do not merely reflect linear association.20 Paired samples t-tests evaluated differences between measured weights (g) and digital imagery-assessed portions and waste (converted to g) to examine potential implications of biased estimates on consumption estimates. Bland-Altman plots were used to quantify bias and the range of agreement between measured weights and digital imagery estimates. Specifically the mean difference of digital imagery compared with measured weights for both starting portions and waste was plotted. Bland-Altman regressions were then used to compare the results from the two estimation methods.

RESULTS

IRRs were 0.91 for determining starting portions of vegetables and 0.99 for determining % plate waste among the seven raters, indicating excellent agreement. When examined by category, ICCs for determining starting portion estimates were lower for Other Vegetables (0.89) compared with Leafy Greens (0.92), yet both demonstrated excellent rater agreement. IRRs for determining plate waste were excellent for both Leafy Greens (0.99) and Other Vegetables (0.99; Table 1). IRRs for plate waste estimates with dressing (not shown in Table 1) remained excellent (ICC=0.98–0.99; P<0.001).

Table 1.

Interrater reliabilities and validity of weights compared with digital imagery for determining starting portions and plate waste for salad bar vegetables in a laboratory-based study (n = 30 mock salads with n = 73 vegetables).

Food item Interrater Reliabilities Validity
Portion Estimates % Plate Waste Portion Estimate (Digital Imagery vs Measured Weight) % Plate Waste (Digital Imagery vs Measured Weights)
n ICCb ICC ICC rc ICC r
Overall Salad Bar Vegetables 73 0.909*** 0.986*** 0.741*** 0.753*** 0.979*** 0.962***
Other Vegetablesa 52 0.868*** 0.985*** 0.767*** 0.713*** 0.986*** 0.975***
Leafy Greens 21 0.920*** 0.987*** 0.671** 0.732*** 0.959*** 0.931***
Open in a new tab

Note: Estimates are based off of weighed and visually estimated values in a simulated laboratory based validation study, conducted in preparation for an investigation related to the role of salad bars on elementary students’ dietary intake.

a

Other Vegetables are all salad bar vegetables other than leafy greens

b

ICC = intraclass correlation; interrater reliabilities are based on 7 raters

c

r = Pearson’s correlation

**

P<.01

***

P <.001

With respect to accuracy of digital imagery compared with weighed reference portions, ICCs were moderate (0.74), with lower accuracy of determining starting portions of Leafy Greens (0.67) compared with Other Vegetables (0.77). However, accuracy of digital imagery plate waste assessments was excellent compared with measured weights for Vegetables overall (0.98) and for Leafy Greens (0.96) and Other Vegetables (0.99). Pearson’s correlations between weights and digital imagery were lower (r=0.75) for estimating portions compared with % plate waste (r=0.96), although both were significantly and strongly related (P<0.001).

T-tests examined mean differences in portion estimates (g) and plate waste (% and g) between digital imagery and weights. (Table 2). Digital imagery was not significantly different from measured weights for estimating overall vegetable portions (0.61g difference). However, digital imagery underestimated starting portions of Leafy Greens by 3.5g (P=0.012). Similarly, digital imagery underestimated plate waste by 3.1% for vegetables overall, (2.2% for Other Vegetables and 5.2% for Leafy Greens. When converted to grams, this difference was only significant for Leafy Greens (digital imagery underestimated waste by a mean of 2.13g [SE=0.53]; P=0.012).

Table 2.

Mean difference between weighed and digital imagery-estimated starting portions and plate waste for salad bar vegetables in a laboratory-based study (n = 30 mock salads with n = 73 vegetables).

Food item n Mean grams starting portion ±SE Mean grams wasted ±SE
Weighed Digital Imagery P Weighed Digital Imagery P
Overall Salad Bar Vegetables 73 27.49±1.8 28.09±2.0 0.605 9.58±1.3 8.3±1.2 0.144
Other Vegetablesa 52 33.32±1.9 35.60±2.0 0.139 11.75±1.6 10.86±11.4 0.438
Leafy Greens 21 13.03±1.7 9.52±0.8 0.012 4.19±1.5 2.06±4.1 0.012
Open in a new tab

Note: Estimates are based off of weighed and digital imagery-estimated values in a simulated laboratory-based validation study, conducted in preparation for an investigation related to the role of salad bars on elementary students’ dietary intake. P values are results of paired t-tests; SE = standard error

a

Other Vegetables are all salad bar vegetables other than leafy greens

Visual examination of the Bland-Altman plots (Figure 2) does not suggest systematic over or under estimation bias using digital imagery for either starting portions or waste. The 95% limits of agreement for starting portions were −20.27g to 19.05g. Bland-Altman regressions suggested that these means were not significantly related, accounting for 1.2% of the variance of the mean measurement (F(1,71)=1.88; P=0.175). The 95% limits of agreement for plate waste were −12.9g to 15.4g. Bland-Altman regressions suggested that these mean values were not significantly related, accounting for 1.1% of the variance in the mean measurement difference (F(1, 71)=.22; P=0.639).

Figure 2.

Figure 2

Open in a new tab

Bland-Altman plots comparing digital imagery-estimated measurements with measured weights of A) starting portions and B) plate waste of vegetables from salad bars in a simulated laboratory-based study. The difference between methods was plotted as weighed method minus digital imagery estimation method. The center line represents the mean difference between weighed and digital imagery methods. The upper and lower boundaries represent the 95% limits of agreement between methods.

DISCUSSION

Main findings from this study were that digital imagery was both reliable and valid for estimating consumption (starting portion and waste) of salad bar vegetables and did not differ significantly from weighed methods overall. IRRs using digital imagery were excellent for determining consumption, and comparable to those reported previously with other foods.7,9 Although a handful of prior investigations have used digital imagery with salad bars,9,21 this study is the first to demonstrate that digital imagery can reliably and accurately estimate starting portions of salad bar vegetables, using reference weights corresponding to variable portion sizes. Findings also strongly support the accuracy of digital imagery to assess plate waste from salad bars. Accuracy of determining starting portions was somewhat more challenging (although “moderate” agreement was achieved), particularly when estimating starting portions of leafy greens. Nonetheless, accuracy of plate waste estimation via digital imagery for leafy greens was excellent. Taylor et al9 previously validated digital imagery (using correlations) for determining plate waste of fruits and vegetables, including vegetables from a salad bar. Consistent with the current report, correlations between digital imagery and weighed measurements were lower for leafy greens. However, in the prior investigation, Taylor et al9 used a “standard serving” (which was not defined) as a reference weight for leafy greens, or “counts” (e.g., slices, baby carrots) for other variable portion items. There are limitations to this method, as it relies on the assumption that all slices/items are equal in size; further, use of a standard serving does not account for the variability of children’s serving sizes when they are permitted to self-serve. Current findings build on prior investigations by validating digital imagery against reference weights, and provide preliminary support for use of digital imagery for determining the starting portions and waste from self-serve salad bar vegetables.

Examination of mean differences between digital imagery and weighed values can help understand potential implications of digital imagery on consumption estimates. Across vegetables, weighed and digital imagery-estimated portions and waste (all in grams) were not significantly different. However, digital imagery underestimated portions (by 3.5g) and waste (by 2.1g) for leafy greens. This difference corresponds to <1kilocalorie, suggesting negligible influence on nutritional estimates. In contrast, a similar laboratory-based validation study (that did not include variable portions of salad bar vegetables) reported inconsistent bias across food groups using digital imagery (e.g., entrees, beverages, fruits and vegetables), with ~5g overestimation of portions for fruits and vegetables.7 Another investigation reported ~3g overestimation of consumption by digital imagery for salad greens.9 Given sample limitations, this study could not assess if a consistent pattern of over- or underestimation existed for each vegetable. Nonetheless, the potential for digital imagery to underestimate selection and waste of leafy greens should be considered when using these methods.

The limits of agreement indicate that, among 95% of all vegetables, starting portion estimates could be mis-estimated by up to 20g (for starting portions) or 13g (for waste). Although no a priori acceptable limit was set in the current investigation, prior school plate waste investigations suggested that acceptable limits of agreement are within 20g, as this volume is equal to ¼ of a standard serving of vegetables (½ cup = 80g).22,23 According to this guideline, starting portion limits of agreement are just at the “acceptable” range, although nonetheless suggest that digital imagery could be less sensitive to small changes in individual starting portions. Waste estimates were well within this acceptable range and were smaller than those reported in a prior school-based validation study (−32.9g to 31.3g).22 These differences could in part be due to the simulated design of the current investigation; however, the enhanced estimates could also reflect use of reference portions for variable portion salad bar vegetables (versus standard portions, in the prior study), leading to more accurate estimates.

Although analyses of agreement (weighted kappas or ICCs) are more appropriate for measuring agreement compared with correlations,18 we conducted Pearson’s correlations between weighed and digital imagery estimates to facilitate comparison with prior work. Correlations between digital imagery and measured weights were higher in the current study compared with those found in a prior study9 for estimating consumption of leafy greens, which might reflect differences in use of a variable reference portion in the current trial, compared with a standard reference portion in the prior trial. Williamson et al.7 correlated digital imagery with measured weights of pre-portioned fruits and vegetables (without variable starting portions) and reported r=0.93 (compared to r=0.75 in the current study), further suggesting that estimating portions from variable serving sizes is more challenging than if vegetables are a standard portion. However, correlations between digital imagery and weighed plate waste estimates were comparable (r=0.849 and r=0.897 compared with 0.96 in the current study). As noted earlier, correlations suggest linear association, but not necessarily agreement, in contrast with the ICCs used in the current investigation.

Limitations of this study include the relatively small number of salads created and corresponding inability to conduct reliability and validity estimates for each vegetable separately. Although grouping (Leafy Greens vs. Other Vegetable) was similar to that used in prior investigations,9 future research should include larger samples of individual vegetables to permit examination of psychometric properties with greater specificity. Salad creation and consumption simulation was based on school serving parameters but did not follow a systematic protocol to ensure equal representation of each portion and waste per vegetable.

Importantly, the reference portions were created in ¼ cup increments, based on the ¼ cup serving utensil used in the target school district, thus might not be generalizable to all other districts. While there is variability in salad bar implementation practices between schools,24,25 all schools participating in the National School Lunch Program must ensure students take at least ½ cup of fruit or vegetables with each meal. Thus use of ¼ cup reference portions would likely be appropriate to capture the range of starting portions. Future studies using digital imagery methods with salad bars should consider the specific schools’ implementation practices and adjust reference portions accordingly.

It is unclear if this simulated design translates to the school setting; however, careful attention was paid to match consistency with school preparation methods (menus, preparation, and presentation). Although ideal, validating these methods in the school setting might not be feasible. Specifically, it would be very challenging and overly intrusive to weigh components from students’ already-assembled salads to determine the starting portion prior to consumption. However, the simulated methods might have resulted in overestimation of estimates. Further, raters were able to rate >99% of vegetables, which is likely an overestimation of what would occur in a real-world school setting (e.g., due to vegetables being obstructed or missing from trays at post). For example, a school-based investigation reported that 2–9% of fruits and vegetables could not be assessed for these reasons.21

This study did not include estimation of portions or waste from the salad dressing, although the dressing was used to mimic the visual presentation in the schools. Only IRRs could be calculated for salads with dressing, as the weight of the dressing could not be calculated once applied. Future studies should investigate ways to reliably and validly capture intake from dressing and other condiments, as they often add considerable calories and have frequently been excluded from plate waste investigations.10

Conclusions

This investigation builds on prior digital imagery validation studies by specifically investigating use of digital imagery for assessing variable starting portions of vegetables from self-serve salad bars, and use of a weighed reference portion and statistical approach using a measure of agreement (ICC).20 Results suggest that use of digital imagery with photographs of measured reference portions to aid in assessment of starting portions might be a reliable and valid method to estimate dietary intake from salad bars. This approach should be further validated in the school setting as is feasible; however, this study provides preliminary support for its use within empirical investigations of dietary intake in schools with salad bars, informing school nutrition policies and programs.

Research Snapshot.

Research Question

There is a need to establish psychometrically sound methods for assessing dietary intake from salad bars, given their increasing prevalence in schools. In a laboratory setting, this research investigated if digital imagery can reliably and accurately determine starting portions and plate waste from self-serve salad bars.

Key Findings

Results supported the interrater reliability and accuracy of digital imagery for estimating starting portions and waste from salad bars. Although the nutritional impact is negligible, digital imagery underestimated portions and waste for leafy greens. Results provide preliminary support for use in investigations of dietary consumption in schools with salad bars.

Acknowledgments

Funding: This work was supported in part by Impact 100 to Greater Richmond Fit4Kids (PI: Mary Dunne Stewart) and NICHD 1R03HD088985-01 to Melanie K Bean.

The authors would like to thank April Williams for her assistance with this study.

Footnotes

Author Contributions: MKB designed the study, conducted data collection and analyses, and drafted the manuscript; HR contributed to study design and methods; LT informed study analyses and interpretation of results; AS collected the data and oversaw data management; MDS coordinated school salad bar information with Eat Fresh RPS; SEM contributed to study design and implementation; All authors reviewed and contributed to subsequent drafts of the manuscript.

Conflicts of interest: none to report

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Contributor Information

Melanie K Bean, Associate Professor, Department of Pediatrics, Children’s Hospital of Richmond at Virginia Commonwealth University, PO Box 980410, Richmond, VA 23298-0410, 804-527-4765.

Hollie A. Raynor, Professor, Department of Nutrition, University of Tennessee, 1215 Cumberland Ave, Knoxville, TN 37996, 865-974-6259.

Laura M. Thornton, Associate Research Professor, Department of Psychiatry, University of North Carolina, Chapel Hill, 101 Manning Drive, Chapel Hill, NC 27599-7160, 804-690-3079.

Alexandra Sova, Research Assistant, Department of Pediatrics, Children’s Hospital of Richmond at Virginia Commonwealth University, PO Box 980410, Richmond, VA 23298-0410, 804-527-4765.

Mary Dunne Stewart, CEO, Greater Richmond Fit4Kids, PO Box 1092, Richmond, VA 23218, 804-307-9161.

Suzanne E. Mazzeo, Professor, Department of Psychology, PO Box 842018, Virginia Commonwealth University, Richmond, VA 23284-2018, 804-827-1708.

References

  • 1.Harris DM, Seymour J, Grummer-Strawn L, et al. Let’s Move Salad Bars to Schools: a public-private partnership to increase student fruit and vegetable consumption. Child Obes. 2012;8(4):294–297. doi: 10.1089/chi.2012.0094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Adams MA, Bruening M, Ohri-Vachaspati P. Use of salad bars in schools to increase fruit and vegetable consumption: where’s the evidence? J Acad Nutr Diet. 2015;115(8):1233–1236. doi: 10.1016/j.jand.2015.02.010. [DOI] [PubMed] [Google Scholar]
  • 3.Johnson CC, Myers L, Mundorf AR, O’Malley K, Spruance LA, Harris DM. Lunch salad bars in New Orleans’ middle and high schools: student intake of fruit and vegetables. Int J Environ Res Public Health. 2017;14(4) doi: 10.3390/ijerph14040415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Adams MA, Bruening M, Ohri-Vachaspati P, Hurley JC. Location of school lunch salad bars and fruit and vegetable consumption in middle schools: a cross-sectional plate waste study. J Acad Nutr Diet. 2016;116(3):407–416. doi: 10.1016/j.jand.2015.10.011. [DOI] [PubMed] [Google Scholar]
  • 5.Healthy Hunger-Free Kids Act of 2010. One Hundred Eleventh Congress of the United States of America; 2nd Session. S. 3307; 2010. [Accessed June 30, 2015]. https://www.govtrack.us/congress/bills/111/s3307/text. [Google Scholar]
  • 6.Tugault-Lafleur CN, Black JL, Barr SI. A systematic review of methods to assess children’s diets in the school context. Adv Nutr. 2017;8(1):63–79. doi: 10.3945/an.116.013144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Williamson DA, Allen HR, Martin PD, Alfonso AJ, Gerald B, Hunt A. Comparison of digital photography to weighed and visual estimation of portion sizes. J Am Diet Assoc. 2003;103(9):1139–1145. doi: 10.1016/s0002-8223(03)00974-x. [DOI] [PubMed] [Google Scholar]
  • 8.Hanks AS, Wansink B, Just DR. Reliability and accuracy of real-time visualization techniques for measuring school cafeteria tray waste: validating the quarter-waste method. J Acad Nutr Diet. 2014;114(3):470–474. doi: 10.1016/j.jand.2013.08.013. [DOI] [PubMed] [Google Scholar]
  • 9.Taylor JC, Yon BA, Johnson RK. Reliability and validity of digital imaging as a measure of schoolchildren’s fruit and vegetable consumption. J Acad Nutr Diet. 2014;114(9):1359–1366. doi: 10.1016/j.jand.2014.02.029. [DOI] [PubMed] [Google Scholar]
  • 10.Swanson M. Digital photography as a tool to measure school cafeteria consumption. J School Health. 2008;78(8):432–437. doi: 10.1111/j.1746-1561.2008.00326.x. [DOI] [PubMed] [Google Scholar]
  • 11.Getts KM, Quinn EL, Johnson DB, Otten JJ. Validity and interrater reliability of the visual quarter-waste method for assessing food waste in middle school and high school cafeteria settings. J Acad Nutr Diet. 2017;117(11):1816–1821. doi: 10.1016/j.jand.2017.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Smith SL, Cunningham-Sabo L. Food choice, plate waste and nutrient intake of elementary- and middle-school students participating in the US National School Lunch Program. Public Health Nutr. 2014;17(6):1255–1263. doi: 10.1017/S1368980013001894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Nicklas TA, Liu Y, Stuff JE, Fisher JO, Mendoza JA, O’Neil CE. Characterizing lunch meals served and consumed by pre-school children in Head Start. Public Health Nutr. 2013;16(12):2169–2177. doi: 10.1017/S1368980013001377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Martin CK, Newton RL, Jr, Anton SD, et al. Measurement of children’s food intake with digital photography and the effects of second servings upon food intake. Eat Behaviors. 2007;8(2):148–156. doi: 10.1016/j.eatbeh.2006.03.003. [DOI] [PubMed] [Google Scholar]
  • 15.Gemming L, Utter J, Ni Mhurchu C. Image-assisted dietary assessment: a systematic review of the evidence. J Acad Nutr Diet. 2015;115(1):64–77. doi: 10.1016/j.jand.2014.09.015. [DOI] [PubMed] [Google Scholar]
  • 16.Comstock EM, St Pierre RG, Mackiernan YD. Measuring individual plate waste in school lunches. Visual estimation and children’s ratings vs. actual weighing of plate waste. J Am Diet Assoc. 1981;79(3):290–296. [PubMed] [Google Scholar]
  • 17.IBM SPSS Statistics for Windows. Version 23.0. Armong, NY: IBM Corp; [Google Scholar]
  • 18.Hallgren KA. Computing inter-rater reliability for observational data: an overview and tutorial. Tutor Quant Methods Psychol. 2012;8(1):23–34. doi: 10.20982/tqmp.08.1.p023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159–174. [PubMed] [Google Scholar]
  • 20.Dubois S. Accuracy of visual estimates of plate waste in the determination of food consumption. J Am Diet Assoc. 1990;90(3):382–387. [PubMed] [Google Scholar]
  • 21.Amin SA, Yon BA, Taylor JC, Johnson RK. Impact of the National School Lunch Program on fruit and vegetable selection in northeastern elementary schoolchildren, 2012–2013. Public Health Rep. 2015;130(5):453–457. doi: 10.1177/003335491513000508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Taylor JC, Yon BA, Johnson RK. Reliability and validity of digital imaging as a measure of schoolchildren’s fruit and vegetable consumption. J Acad Nutr Diet. 2014;114(9):1359–1366. doi: 10.1016/j.jand.2014.02.029. [DOI] [PubMed] [Google Scholar]
  • 23. [Accessed June 30, 2015];New Meal Pattern Requirements and Nutrition Standards, USDA’s National School Lunch and School Breakfast Programs. http://www.fns.usda.gov/sites/default/files/LAC_03-06-12_0.pdf.
  • 24.Blumenschine M, Adams M, Bruening M. Prevalence of and differences in salad bar implementation in rural versus urban Arizona schools. [E-pub ahead of print] J Acad Nutr Diet. 2017 doi: 10.1016/j.jand.2017.09.004. [DOI] [PubMed] [Google Scholar]
  • 25.Bruening M, Adams MA, Ohri-Vachaspati P, Hurley J. Prevalence and implementation practices of school salad bars across grade levels. [E-pub ahead of print] Am J Health Promot. 2017 doi: 10.1177/0890117116689159. [DOI] [PubMed] [Google Scholar]

RESOURCES