This chapter provides the study aims and objectives, study design and methods of the nursery food photography feasibility substudy.
Overview
An additional feasibility study was undertaken after the end of the feasibility trial with the aim to ascertain the feasibility and acceptability of taking photographs of the foods and drinks served to 2- to 4-year-olds in nurseries to estimate portion size served and caloric intake of individual children (to replace CADET in a full trial as a method of collecting nutrition data in nurseries). This study also aimed to determine the number of children from whom one researcher could feasibly collect complete food and drink data to inform a future trial.
Sample and recruitment
A purposeful sample of nurseries within Bristol, to allow for variation in deprivation and meal provision (nursery-provided food and family-provided food), were invited to take part, with the aim of recruiting four nurseries. Nurseries were approached by e-mail and provided with an invitation letter, nursery information sheet, consent form and camera policy. Nursery managers were given 2 weeks to respond with a reminder e-mail sent after 1 week. The camera policy included details regarding storage of the camera during non-eating times and a requirement that the nursery manager check the photographs at the end of the data collection day to ensure that there were no concerns about inappropriate use of the camera. It was requested that the policy was signed along with the consent form.
Once a nursery manager had consented to taking part in the study, a date for data collection was arranged and nursery managers were asked to briefly explain mealtimes, locations and serving arrangements. Following this, information packs for all parents/carers of 2- to 4-year-olds that were going to be present on the day of data collection were posted to the nursery managers. The packs included a letter, an information sheet, an opt-out consent form and a stamped addressed envelope. Nursery managers were asked to hand these to parents/carers. Parents/carers were given 2 weeks to provide opt-out consent if they did not want their child to take part in the study.
Data collection
Two researchers visited each nursery for 1 full day of data collection. Photographs were taken of as many consented children’s food and drink as possible. Researchers followed the Fieldworkers’ Standard Operating Procedures that were developed for this study.
A set of laminated cards were prepared to be placed next to the children’s plates for each photograph. A standard credit-card-sized marker was printed on the laminated card to aid perspective adjustment and later portion size estimation. Dry-wipe marker pens were used to indicate on the laminated card the type of meal, whether the photograph was taken before or after eating and whether the photograph was of the first or second helping. The researcher ID and participant ID were written on the front of the laminated card, with the child’s name written on the back so that all meals throughout the day could be matched to that child. The laminated cards were left at the nurseries after data collection.
The type and number of drinks were recorded in one-quarter cup increments on a fieldworker data collection sheet to save time having to take ‘before’ and ‘after’ photographs of each refill of drink. A full cup of water was also weighed to determine the volume of water being drunk based on the one-quarter cup increments recorded on the data collection sheet. A researcher took one photograph before the meal and one after, at a 45-degree angle and approximately at arm’s-length distance to include all the food on the plate. The aim was for no part of the child to be in the photograph, however, it was anticipated that occasionally part of a child’s hand, arm or body would be shown in the photograph. If a child received additional portions, a photograph was taken of the ‘empty’ plate before and after an additional serving to capture a photograph of the new portion and then a final photo when the child had finished eating to capture any leftovers or confirm that the entire serving was consumed.
To optimise portion size estimation for each food in a mixed meal, foods should be separated on the plate and not layered on top of one another (e.g. pasta on one side and sauce on the other side of the plate, or sandwiches opened up to reveal the contents). To overcome the time demands of separating foods on each individual child’s plate, a portion was provided to the researchers with foods separated. For composite meals (e.g. cottage pie, curry) the researchers asked the nursery staff for a detailed description of the ingredients and quantities and for a portion. When a meal could not be weighed, standard recipes were used based on the National Diet and Nutrition Survey nutrient data bank.
In nurseries that allowed packed lunches, each consenting child’s packed lunch was photographed and each element weighed and recorded. A new pair of latex gloves was used for each packed lunch to avoid cross-contamination of allergens.
At the end of the data collection day, nursery staff who were present on the day were asked to complete a short questionnaire about their views and acceptability of the data collection method. The nursery manager was asked to check all the photograph thumbnails to confirm they had no concern about the content of the photographs.
Data analysis
The photographs were annotated as relevant and submitted for analysis to Pennington Biomedical Research Center (PBRC), an organisation based in the USA that specialises in estimating food portion and nutrient data using food photography.105–107 No information that could identify a child or nursery was sent to PBRC. Processing of photographs involved trained members of the PBRC team using the photographs and annotations to match foods pictured to food in a UK food composition database. Once a matching food was chosen, the portion size of the food was estimated by comparing the photograph to standard photographs of that food where portion size (grams) was known. The standard marker included within photographs was used to rescale photographs to the same size as the standard photographs, which aided accurate estimation of portion sizes. Portion size and food composition information was then combined to generate nutrient contents of the meal eaten. Questionnaire data from nursery staff were entered into a Microsoft Excel® (Microsoft Corporation, Redmond, WA, USA) spreadsheet and summarised using descriptive statistics. Fieldworkers were asked to complete detailed observation notes and these were analysed thematically.