All participants were initially asked to record food intake and exercise using a paper-based diary and then to use a smartphone app to record food and exercise as frequently and as accurately as they found feasible for 4-5 weeks. A 2-phase study design was chosen over a parallel intervention study to avoid introducing potential confounding effects, such as social factors, familiarity with the technology, and individual academic achievement. The study design also avoided potential carry-over effects and disproportionate dropout from the app-to-paper group that may have been associated with a conventional randomized crossover design. Personal and professional contacts were used to involve schools and colleges in the design of the study and to ensure that the protocol was appropriate in an educational environment with minimal disruption to pupils. The study was not blinded or randomized, but each participant was allocated a code for data analysis purposes. The study was scientifically reviewed by the Human Research Governance Committee at the Institute of Food Research, Norwich. The study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the Oxford C Ethics Committee managed by the UK Health Research Authority (144/SC/1268).

Participants aged between 16 and 19 years, still in full- or part-time education, were recruited from sixth form colleges, schools, and other further education establishments in Norfolk and Suffolk, UK, between February and May 2015. The study was advertised through email and phone calls to education centers, followed by an initial presentation explaining the study first to the head and other interested staff and then to the students. The eligibility criteria were absence of chronic illness or disease (such as serious asthma or diabetes), not pregnant, able to give informed consent, have parental support, able or willing to use a smartphone, able to complete paperwork even with assistance, no eating disorders, no involvement in other research projects or weight management program, and not being related to any member of the study team. An inclusion criterion of BMI not below the second centile line on the BMI chart [51] was chosen as a safety measure to ensure that participants were not likely to fall below the defined healthy BMI range during the study. Signed letters of approval to run the study on each premises were obtained from the relevant school or college head (or other designated responsible person) at all participating schools (Gatekeeper approval). Written informed consent was obtained from all participants, and parental consent was also obtained for participants aged less than 18 years at recruitment.

Eighteen schools in Norfolk and Suffolk were contacted via email, phone calls, or visit by the researcher. Five schools gave consent for the study talk to be delivered to their students. Recruitment for the study took place between February and May 2015. The flow of participants through the study is described in Figure 1. A total of 157 pupils attended study talks and 38 of these attended the screening session, where participants were assessed for eligibility to take part in the study. Of those screened, 4 were excluded for not meeting the inclusion criteria. Thirty-four adolescents consented and began the paper-diary phase. Two withdrew (the first person after the first week and the second person in the second week of the study) due to academic commitments. Thirty-two participants completed the first phase of the study and were given a smartphone preloaded with the FoodWiz2 app. Two participants were lost to follow-up during this phase, with 30 completing the study. All 34 participants were, however, included in the analysis of the use of the paper diary, and the 32 that started using the app were included in its analysis.

Following consent, participants were screened based on the eligibility criteria. Weight was measured to the nearest 0.1 kg using a portable electronic scale (Salter Ultimate Accuracy, Tonbridge, Kent, UK); height was measured to the nearest millimeter using a portable stadiometer, and waist circumference was measured to the nearest millimeter using a body waist fitness caliper. Participants then completed a background information form, which provided more information about their diet, previous recording of diet or exercise, use of apps for similar purposes, and educational attainment. A score was calculated for the General Certificate of Secondary Education (GCSE) educational attainment based on a scale of 1-8 for grades G-A (including the higher level A* grade), and a cutoff of 25 (representing grade C in 5 subjects) was used to assess whether there was any difference in the interest in, or ability to, monitor diet among low and high academic achievers. The cutoff of 25 is used in the United Kingdom for assessing suitability at 16 years of age to progress to higher academic studies. Following screening, all participants received nutrition education sessions, covering the basics of a healthy diet, the importance of exercise, and different ways of achieving and maintaining a healthy nutrient intake and net energy intake based on government guidance.

Participants were initially provided with paper record sheets to record the day, date, time, type, and amount of all foods and drinks consumed; they could also record any recipe in the food record sheets. These food diaries were adapted from those previously used at The Institute of Food Research [52,53]. Participants also recorded the day, date, time type, duration, and intensity of any activity undertaken in the activity sheet. The paper diaries were collected and reviewed with participants every 2 weeks during a face-to-face meeting (10-30 min) between each participant and the researcher. If participants did not attend 2 consecutive meetings with the researchers and were not contactable, they were considered as no longer wishing to take part in the study (lost to follow-up).

After a short interval in recording (1-2 weeks), coincident with a school or college holiday, a smartphone app developed by Food Angels UK Ltd, Newmarket, Suffolk, UK, was given to participants to use for the following 4 weeks (screenshots are available in Multimedia Appendix 1). FoodWiz2 was developed from an earlier version (FoodWiz), which was designed to assist people suffering from allergies. Users of the first FoodWiz app can scan the barcode of a product to discover if it contains any allergens of interest (set up via their own personal profile). The more recent FoodWiz2 app is like other food and exercise recording apps on the market and uses UK measures and databases rather than being US-based. The food composition data used in the app is taken from a commercially available source (BrandBank, Norwich, UK) combined with data for generic foods from the UK McCance and Widdowson’s The Composition of Foods Integrated Dataset [43]. The app allows users to record dietary intake by searching and selecting from a list of foodstuffs and popular meal choices. The user can search by typing in the name of the food or by scanning in barcodes from products. The app also allows users to record intake of home-prepared recipes. A small set of scales, the size of a smartphone, was also provided to allow direct measurement of portion size weights using Bluetooth technology to link the scales to the app [54]. Physical activity could also be recorded using the App, allowing the user to receive instant feedback on their overall daily calorie balance or allowance based on set targets. FoodWiz2 could also record and track weight and mood of participants. The researcher was able to see the information recorded by each participant using the app via a Web-based portal. Therefore, the researcher could monitor use and send each participant personalized feedback text messages for motivation and help. All participants were made aware that this would happen before the start of the study, both in person and in written form (Participant Information Sheet). Example feedback messages were approved during the ethical review process.

At the end of each phase of the study, participants were asked to complete an online questionnaire (Multimedia Appendices 2 and 3). The questions were about the pattern of use, what it was like to use the tools, and the perceived impact on dietary and physical activity behavior. Examples of questions asked were, “To what extent do you agree with the following statements” e.g. on the usability (‘was it time-consuming, disrupting, enjoyable, boring, convenient’), acceptability (‘comfortable to use in social setting, would use in future’) and perceived impact of each method of dietary assessment and exercise recording (‘changed portion sizes of meals, ate more fruits and vegetables, increased awareness of physical activity, increased motivation to change physical activity’). An additional questionnaire section was completed at the end of the app phase (Multimedia Appendix 3) to assess the specific features of the app (eg, search function, feedback messages), and a final section compared the 2 recording methods. Questions were scored on the Likert scale, but participants were also given the option of explaining some of their responses in more detail using a free-text approach. Free-text answers were independently reviewed by 2 researchers (FJ and PB), each identifying common topics before agreeing to the most important issues highlighted by participants. The statements included were agreed by both researchers. However, no computer-based systematic analysis of the free-text answers was undertaken due to the limited amount of data available.

For the analysis of reported food intake using paper diaries or the app, descriptive measures were calculated to describe the sample by use of percentages, means, and standard deviations. Fruits and vegetable servings per day were calculated by dividing the total fruit and vegetable servings consumed by the total number of days that eating occasions were recorded using either the paper diary or App as described by Aflague et al [55]. The number of days a participant used the paper diary or App was set at the total number of days participants recorded food intake of ≥500 kcal.

Two analyses were conducted on questionnaire data because of missing responses: answers to individual questions were described as percentages of all data provided with a subsequent comparison of responses for just those who included their unique identity code, allowing paired analysis of the responses given. The significance of differences in the values of responses to the questionnaire after each phase of the study was evaluated using the sign test for nonparametric paired ordinal data. This test is more conservative than the more frequently used Wilcoxon signed rank test but more appropriate for this dataset where it is difficult to prove the difference between pairs is ordinally scaled [56].

Microsoft Excel 2010 for Windows was used to enter the data, and statistical analyses were performed using SPSS, Version 22 (SPSS Inc., Chicago, IL, USA). All tests were 2-tailed and P<0.05 was taken as indicating statistical significance.

Participants (23 female, 11 male) aged 16 to 19 years were recruited in this study (Table 1). The mean participant BMI was 24 kg/m²(SD 4) with 15% (5/34) classified as either clinically obese (BMI above the 98th centile based on their sex and age) or severely obese (BMI above 99.6th centile based on their sex and age). All participants remained weight-stable throughout the study with a median weight change of −0.1 kg (interquartile range [IQR]=−0.5 to −0.3, n=33) after the paper diary phase and a median change of 0.375 kg (IQR=−0.275 to −0.925, n=28) after the app phase compared with weight at the start of each phase (overall median study weight change=−0.05 kg; IQR=−1.87 to −0.488; n=30; P=.20). Of the 34 participants, 10 females and 1 male had previously recorded their diet or exercise. Six participants (4 female, 2 male) had previously used an app for recording food intake. Average GCSE point score was 40.6 and ranged from 0 to 85, with 58.8% (20/34) of participants scoring 25 and above. All except 3 participants who scored above 25 were attending sixth form schools. Participants who scored zero (n=6) were either registered to take their GCSEs (n=5) or had emigrated from outside of Europe (n=1), where the system of education was different.

The paper-diary phase was completed by 32 out of the 34 participants who started the intervention, whereas the subsequent App phase was completed by 30 out of the remaining 32 participants. Only 12% (4/34) of participants recorded on all the possible days (28 days in both phases). The mean number of study days completed (>500 kcal) for each 28-day phase is shown in Table 2. Use of the diet recording tool was highest in the paper diary group with a mean of 24 days (SD 6) completed compared with 17 days (SD 9) for the app (P=.002). There was no significant difference in completion rate between male and female participants, and no effect of educational attainment on completion of food records was identified at either phase.

The mean percentage completion for the paper diary was 86 (SD 10), whereas that for the app was 61 (SD 7), although it should be noted that most students were taking examinations during the App-based phase. Completion rate decreased gradually throughout the study period both during the paper diary phase and the app phase (Table 2), although use showed a trend toward an increase near the end of the app phase.

Reported food intakes were analyzed by food groups, only for those days where >500 kcal were recorded to allow for comparison between results from the paper diary and the app. In general, recorded food intake was similar using either method; however, the reported consumption of chocolate snacks (P=.01) and fizzy drinks (P=.002) was significantly lower during the app phase than when paper diaries were used (Figure 2). It was noted by researchers that the quality of the data retrievable from the app was considerably better in terms of specific foods eaten and sources. For example, although a paper diary may just say “Chinese take away,” the app would prompt for a more specific description, for example, “Noodles and sweet and sour chicken.”

All participants recorded a range of exercises in the paper diaries, most frequently walking and cycling but also team sports, gym, dance, and housework. Similarly, all those still in the study used the app for this purpose on at least 1 occasion with 2 participants providing data for at least 26 days, in comparison with 8 participants using the paper diary. Interestingly, there was a significant correlation between the numbers of days completed in the paper diary and the app (P<.001), although significantly more days were completed in the paper diary phase with a median of 16 days as compared with 6 days (P<.001).

In most cases, no significant difference in response to Likert scale questions relating to the paper diary and the app were found. The free-text responses gave some additional insight into the issues raised (Multimedia Appendix 4). Although the number and range of responses were limited, they only gave a general indication of opinions. Review of the free-text answers given in the questionnaires suggested an overall preference for the FoodWiz2 app, particularly in relation to enjoyment, convenience, recommendation to a friend, overall liking, and using again in the future, whereas the paper diary was considered time-consuming and boring. The main reasons given for preferring the app were focused on the topics of usability (eg, “Easy and fun to use”), accessibility (eg, “I can use it anywhere I want”), and the ability to track weight and calorie intake (eg, “Easy to watch calories”). Features of the app which were liked most were the smiley mood scale, bar code scanner, layout, and calorie counting.

The questions relating to patterns of use revealed that the reported level of use of the 2 approaches was similar, although the participants believed they used the paper diary for more days a week than the app (Question 1, Multimedia Appendix 2, P=.04), a result consistent with the actual data from the 2 recording methods. Participants reported that the paper diary was bulky and not easy to travel with, which often resulted in them recording what was eaten sometime after consumption. The problems associated with using a paper diary, perhaps only filling it when at home, would appear to have been more than outweighed by potential study fatigue and the issue of exam pressure during the app test phase (Multimedia Appendix 4).

Using both the paper diary and the app were considered time-consuming for different reasons. The paper diary involved having to manually write information and weigh foods, whereas the app had technical issues, for example, it worked slower than anticipated (Multimedia Appendix 4), partially explaining the lack of difference in response to the specific questions as to how time-consuming it was. Furthermore, although free-text answers suggested participants found the app more convenient, no significant difference was identified on the specific question (Section 1b-Q1, Multimedia Appendices 2 and 3). However, participants did report greater social acceptability for using the app (Figure 3). A higher proportion reported that they felt comfortable using their smartphone in social settings compared with using the paper diary (14/17 vs 3/21, P<.001). Although, when broken down into more specific social occasions, this effect was no longer significant and very few found using either approach difficult in front of their families. Furthermore, 48% (10/21) reported that they found the paper diary boring, whereas only 33% (5/15) found the app boring (P=.03, Figure 4).

Most participants reported that the use of both the paper diary and the app raised awareness of what they had been eating and how active they had been, and they felt that the app was more effective in this respect, which was again consistent with actual recorded data for chocolate snacks and fizzy drinks. However, there were no significant differences in perceived effectiveness between the 2 methods. When indicating the overall preference between the paper diary and the app, 5 out of 12 participants preferred the app, 3 out of 12 preferred the paper diary, and 4 out of 12 had no preference. Participants reported performing more aerobic and strength exercises as well as trying new activities; however, this was not significant, and indeed, the App appeared to significantly reduce confidence to do more activities. This reduction may be explained by the comments about the lack of options in the set list for exercise on the app (Multimedia Appendix 4). Specific questions addressing the quality of the app highlighted the value of being able to search for foods, but potential barriers to the use of the smartphone app were also identified. Feedback from the questionnaire showed that participants needed support with downloading and setting up FoodWiz2 as it was not available via the usual app download routes (eg, Android Store). Occasionally, participants encountered difficulty finding nongeneric food products, which may have been a limitation of the search function. Suggested improvements to the app included healthier food recommendations, recipe ideas, more physical activity options, and help with portion sizes. The problems encountered with the app were mainly the speed of loading searches and ease of finding specific food items.

Adolescence is a key point in life in establishing long-term eating patterns as young people move forward into adult life with growing independence in food choice. Results of this study demonstrated that the app is a potentially feasible method of recording diet and physical activity in adolescents. Previous similar studies using smartphone apps with adults have focused on specific applications such as weight loss or caloric balance [33,49,57], including an intervention to encourage more attentive eating [58], as well as for other clinical issues such as pain management [59,60].

This study had several strengths including the fact that it focused on encouraging overall dietary improvement rather than calorie control in adolescents. This was considered a key issue, as some young people are very sensitive about their weight and may be particularly at risk of developing eating disorders such as anorexia or bulimia [61]. FoodWiz2 included information from the UK food composition tables, so it was more comprehensive than those using solely commercial brand data sources that do not include products that are not prepacked, for example, fresh fruit, vegetables, and meat. The study also compared the new app technology with the standard paper diary, and participants were able to give feedback on their experience of using either recording method by answering both closed and open questions. The data were considered in the light of recorded food intake data as well as comparing intake data from the app and the paper-based diary. The trial retention in this study was higher (94% [30/32] of participants in the app phase) than that reported by Carter et al [57] testing a different app with women aged 35 years (SD 9). An equal number of participants were lost from the paper diary phase and the FoodWiz2 app phase and were all linked with academic commitments. Participants were either in full-time or part-time education and so this was not unexpected. The study was subject to several challenges. Recruitment was lower than planned, such that only 57% (34/60) of the original recruitment target was reached. Direct contact with schools was the most successful recruitment strategy, but delays in starting arose as a result of one large potential source of participants withdrawing their support just as recruitment was about to start. This delay meant that the study ran over the normal school examination period in the United Kingdom from May to June. The frequency of use of the interventions was significantly higher in the paper diary compared with the app group (P=.002). This was unexpected and is not consistent with what has been previously reported [33,57,62]. This is likely to be due to two factors. First, the participants completed 4 weeks of recording using the paper diary before using the app for the same period and completion consistently dropped over this period, suggesting a fatigue effect. The use of the paper diary declined from the first week, but the use of the app increased slightly in the fourth and final week after exams had finished. The decline in the frequency of use of self-monitoring devices has been previously reported [57,63] in 2 studies comparing adherence to completion of dietary records using smartphone, website, or paper diary in randomized controlled trials. Second, review of the free-text answers highlighted that many participants were under pressure from academic commitments, especially those that had to prepare and appear for their advanced level examinations during the study, which impacted their ability to use the app fully. The comments by the participants that the paper diary was bulky and not easy to travel with resulting in them not recording what was eaten until sometime after consumption is likely to introduce inaccuracy in recording [49,64], although this was not detectable in this study. It is interesting to note that there was no significant difference between males and females or effect of educational attainment on the numbers of days on which food and exercise was recorded, although it should be noted this was a small self-selected subpopulation of all potential participants. The observed similarity in responses between male and female participants contrasts with the results reported in a recent study from the United States with children aged 3-10 years [55].

Free-text responses in the questionnaires indicated that using either the paper diary or the FoodWiz2 app raised participants’ awareness of what they were eating and their level of activity. This also seemed to impact participants’ view of their choice of food and the level of activity, particularly when participants used the app. This result is consistent with previous studies, where the recording of diet or activity has led to increased awareness and, in some instances, change in behavior [65-67]. Furthermore, the questionnaire data in this study suggesting that participants believed they had modified their intake of unhealthy foods, albeit not statistically significant, was reflected in the results of the quantitative analysis of food entries in both the paper diary and the app, where significant reductions in chocolate snacks and fizzy drinks were reported using the app. In terms of acceptability and ease of use, participants were more comfortable using the app in different settings, especially in school and in social settings. These results are similar to those of previous studies [57,63]. In these previous studies, a significantly higher proportion of participants reported that the smartphone and website records were convenient to use compared with the paper-based record.

Our results indicate that, in general, participants preferred the use of a smartphone app to the more traditional paper diary, although some technical issues need to be addressed. In particular, participants found it more comfortable to use the app in social settings. They perceived that the use of the app had more impact on their dietary intake as well as physical activity, compared with the paper diary. Analysis of data from the recorded food intake also showed significantly reduced consumption of chocolate snacks and fizzy drinks among participants when they used the app to record their food intake compared with using the paper diary. The use of mobile technology shows great promise for reducing the burden of self-monitoring lifestyle in this age group, but future apps need to be more sophisticated than the one used for this study. Finally, the relative ease of data extraction for the app compared with coding food diaries and the quality of detail provided mean similar tools show great promise for research purposes.