UVM Theses and Dissertations
Format:
Print
Author:
Taylor, Jennifer C.
Dept./Program:
Nutrition and Food Sciences
Year:
2013
Degree:
MS
Abstract:
As interventions aim to increase school chidren's fruit and vegetable (F/V) consumption, feasible, reliable, and valid dietary assessment methods that are not subject to self-reporting errors are needed. The objectives of this study were to test the feasibility, reliability, and validity of methods for assessing children's F/V consumption during lunch in two elementary school cafeterias. F/V consumption was assessed by weighed plate waste (WPW), direct observation (DO) digital imaging (DI), and digital imaging with observation (DI+O) during a training simulation and over ten school days. Feasibility was tested by the ability to assess F/V consumption for a target sample of at least 30-40 lunch trays during each school day. Interobserver reliability (lOR) was tested for DO and DI based on percent agreement for individual F/V item consumption and intraclass correlations (ICC) for total portions consumed per tray. Validity was tested using Pearson correlations, paired t-tests, and Bland-Altman plots. WPW, DO, and DI assessed 340, 220, and 276 lunch trays respectively.
Thirty-one to 68 lunch trays were assessed during each school day, but no F/V items were selected on 15% of all trays. After removing trays without F/V, the net trays assessed for F/V consumption met or exceeded the target samples on all except on day; where 23% of participants did not select a F/V item. Percent agreement for DO was 90% and the ICC was 0.53. DI percent agreement was 96% and the ICC was 0.92. F/V consumption was compared to WPW over two school visits using DO (n=98) and over four school visits using DI (n=162) and DI+O (n=164). DO (r less than or equal to 0.92), DI (r less than or equal to 0.91), and DI+O (r less than or equal to 0.95) were strongly correlated with WPW for consumption of its, vegetables, and F/V combined (p<0.01). Pearson correlations for DI versus DI+O increased for lasagna, soup and whole fruit. Mean biases for total F/V consumption/tray did not differ from WPW when using DO (p=0.39), DI (p=0.56), or DI+O (p=0.38). DO underestimated mean fruit consumption/tray by 3.0 g (p<0.05), and DI+ overestimated mean vegetable consumption/tray by 2.8 g (p<0.05).
Bland-Altman limits of agreement for individual F/V consumption were within ~one ounce for DO ( -35.0 to 32.0 g), DI ( -32.9 to 31.3 g), and DI+O ( -25.0 to 26.8 g). All methods were feasible for assessing F/V consumption during school lunch. Overall, reliability for DO and DI was acceptable although reliability was fair for DO whereas it was substantial for DI. DO, DI, and DI+O were all valid for assessing mean consumption, although methods were not precise for estimating individual consumption. DI may be especially advantageous for assessing mean consumption of fruits, vegetables and F/V per tray as the method was valid in the absence of cafeteria observations. The exclusion of cafeteria observations to reduce labor and time may make DI an especially promising method for assessing children's F/V consumption in school-based interventions designed to increase consumption of these foods during school lunch.
Thirty-one to 68 lunch trays were assessed during each school day, but no F/V items were selected on 15% of all trays. After removing trays without F/V, the net trays assessed for F/V consumption met or exceeded the target samples on all except on day; where 23% of participants did not select a F/V item. Percent agreement for DO was 90% and the ICC was 0.53. DI percent agreement was 96% and the ICC was 0.92. F/V consumption was compared to WPW over two school visits using DO (n=98) and over four school visits using DI (n=162) and DI+O (n=164). DO (r less than or equal to 0.92), DI (r less than or equal to 0.91), and DI+O (r less than or equal to 0.95) were strongly correlated with WPW for consumption of its, vegetables, and F/V combined (p<0.01). Pearson correlations for DI versus DI+O increased for lasagna, soup and whole fruit. Mean biases for total F/V consumption/tray did not differ from WPW when using DO (p=0.39), DI (p=0.56), or DI+O (p=0.38). DO underestimated mean fruit consumption/tray by 3.0 g (p<0.05), and DI+ overestimated mean vegetable consumption/tray by 2.8 g (p<0.05).
Bland-Altman limits of agreement for individual F/V consumption were within ~one ounce for DO ( -35.0 to 32.0 g), DI ( -32.9 to 31.3 g), and DI+O ( -25.0 to 26.8 g). All methods were feasible for assessing F/V consumption during school lunch. Overall, reliability for DO and DI was acceptable although reliability was fair for DO whereas it was substantial for DI. DO, DI, and DI+O were all valid for assessing mean consumption, although methods were not precise for estimating individual consumption. DI may be especially advantageous for assessing mean consumption of fruits, vegetables and F/V per tray as the method was valid in the absence of cafeteria observations. The exclusion of cafeteria observations to reduce labor and time may make DI an especially promising method for assessing children's F/V consumption in school-based interventions designed to increase consumption of these foods during school lunch.