A randomized-controlled trial focusing on socio-economic status for promoting vegetable intake among adults using a web-based nutrition intervention programme: study protocol

This study describes a two-armed, matched-design, web-based RCT. The nutrition education programme that we have developed as part of a health promotion project (Fig. 1). The intervention period is 5 weeks. The RCT is designed in line with the CONSORT statement for randomised trials of non-pharmacologic treatment [28]. Participants were assessed at three points in time: at baseline (T1), at post intervention (5 weeks later; T2), and at a follow-up at 3 months (T3). The participants were randomly assigned to one of two conditions: the intervention group and the waiting list group (i.e., the control group).

The RCT was conducted according to guidelines laid out in the Declaration of Helsinki for procedures involving human subjects, and has been approved by the Ethics Review Committee on Research with Human Subjects of Waseda University, Japan (2015-167). The respondent's privacy and personal information are fully protected due to agreement between the registration monitor and the social research company used to recruit participants. All e-mails were sent to participants by the research service company, and we have received a database containing only non-specific, anonymized data.

Regarding the setting, at baseline (T1), individuals consenting to participate in the study were randomly assigned it to two groups in the order that received an answer to the intervention or control groups by the online research service company. The researchers were not involved in this allocation in any way. All participants received a notification e-mail informing them of their allocated groups. The groups were listed as P Group (intervention group) and Q Group (control group) to prevent participants from knowing which group they were assigned to. This study is signal a type of single blinding. Because it is not the intervention of face to face, there is no contact during the intervention period.

Participants assigned to the intervention group received an e-mail containing the dates of programme updates, the website URL of the intervention programme, a password for browsing the website, and the following information.

The programme updates were sent on Monday morning of each week during the intervention period. Participants received a total of five e-mails on the nutrition education programme over the 5 weeks of the intervention period. On the first occasion, participants received an e-mail containing (1) instructions on how to access the website (i.e. the URL and password), (2) an overview of the programme and how it would proceed, and (3) the programme contents for the first week. The e-mails sent in the subsequent weeks, in addition to item (1–3), included (4) a review of the contents of the previous step. The programme website was made freely accessible only for the duration of the intervention period, and after which it was closed. Participants’ passwords were effective for the duration of the study. If a participant forgot his or her password, it could be retrieved by contacting a research officer. After completion of the 5-week intervention, participants received an e-mailed request to participate in the post-intervention survey (T2). Finally, participants received another e-mail request to participate in a follow-up survey 3 months after completing the intervention (T3).

The control group surveys took place over the same period as intervention group surveys. Control group participants received an advance notification e-mail after the baseline from the survey company that another survey would take place 5 weeks later. After a 5-week-long silence, participants received an e-mail request to take part in the post-intervention survey (T2). Subsequently, participants received an e-mail request to participate in a follow-up survey 3 months later (T3).

An interactive webpage called the ‘Diet and Exercise Practices Project (http://​healthpromotionq​ol.​com)’ is designed to improve vegetable intake of visitors to help decrease the likelihood of lifestyle-related diseases (Fig. 2). It is a free website that provides information, a monitoring sheet, and advice about healthy diets, increasing vegetable intake, and preventing lifestyle-related diseases. We hypothesize that achieving an approximately 70 g (1 serving) increase in vegetable intake might help lower-income groups to catch up while simultaneously contributing to the partial resolution of the deficient vegetable intake amongst Japan’s adult population. Therefore, the action goals of intervention is a vegetable dish to eat five servings, or was a plus 1 serving (approximately 70 g) per day. The programme is divided into five steps that align with the stages of behaviour change. Table 1 shows the framework of the programme. To be updated every week, but watch the past step not proceed in the previous step. It was created to advance the based on behavioural science theory. To avoid contamination during the intervention period, access to the webpage was password-restricted so that only study participants could visit it.

In order to achieve this action goals were designed following the intervention programme. To encourage behavioural changes, the programme contents are prepared using behavioural theories and techniques tailored to the individual stages of behavioural change. In the first step, we employ the health belief model to encourage a shift from the pre-contemplative to the contemplative phase; in the second and third steps, social cognitive theory is used to encourage transition to the preparatory phase; in the fourth step, social cognitive theory and strengthening social support are used to promote progression to the execution phase; finally, in the fifth step, strengthening social capital and social support are used to promote the shift to the maintenance phase. All of the steps is composed of four items, as introducing “Today's point (Including review of the previous week from the second week onwards)”, information and skills necessary to behaviour modification as practical content “Do you know?” and “Easy in devising” are on two page, as summary “Let's try it !(to support behaviour change by using a work sheet)”.

The website design was settled on through consultations with registered dieticians, health movement educator, and public health experts. Regarding the size of the online text, the amount of information, and the configuration of the website, we obtained the advice of web design professionals. Figures 2 and 3 show a snapshot on programme of the website, while the website structure is shown in Fig. 4.

The primary outcome variable was assessed using a self-administered questionnaire in Japanese. Regarding the assessment, at baseline (T1), primary outcomes, transtheoretical model [31], self-efficacy [32], dietary knowledge [33], perceptions of neighborhood food environments within secondary outcomes, and all other measurements. After completion of the 5-week intervention, participants received an e-mailed request to participate in the post-intervention survey (T2). At post-intervention survey (T2), we assessed the all primary outcomes, all secondary outcomes, all other measurements excluding the subjective economic status. We evaluated about the participants’ satisfaction of process evaluation as after intervention. Finally, participants received another e-mail request to participate in a follow-up survey 3 months after completing the intervention (T3). At follow-up survey (T3), we assessed the all primary outcomes, all secondary outcomes, and all other measurements excluding the subjective economic status.

As Health Japan 21 (the secondterm), it is necessary to evaluate the intervention that applies to the improvement of diet-related quality of life. Interventions that focus on dietary lifestyle require evaluation of diet-related quality of life as primary outcomes. Two of the subscales of the Subjective Diet-Related quality of life scale includes dietary satisfaction and fun of meals. As primary outcomes, we considered Subjective Diet-Related quality of life, which comprised an assessment of the final dieting goal (2 items) [34], and self rated health, which has been reported to be associated with both socio-economic status and mortality [35, 36].

The secondary outcomes were eating behaviours, which was considered an index of behavioural change and is often held up as the goal of nutrition education. Three eating behaviours (per week) were considered: mealtime balance, or eating ‘balanced meals comprising a staple food, a main dish, and a side dish’ [37], ‘eating dark green vegetables’, and ‘eating full servings of vegetables (5 small dishes or approx. 350 g per diem)’ [38]. The mealtime balance behaviour has been shown to facilitate nutrient intake and improve nutritional status [39‐41]. To explain the staple foods, main dishes, side dishes, and vegetable servings, we posted sample photographs for reference in terms of size and amount, and stipulated that respondents should always check these photographs before responding. Because daily eating behaviours (two items) were checked using these photographs depicting single (70 g) servings of vegetables, participants were also asked to answer how many vegetable servings they consumed per diem. This item reflects the fact that, as demonstrated by Ozawa et al. [42], the behavioural goal of ‘five or six small [vegetable] dishes a day’ can potentially serve as an easily grasped indicator of consuming 350 g per diem. Additionally, the amount of vegetables eaten each day (in g) was also self-reported.

As our intention was to evaluate not only outcomes, but also the processes leading to behavioural change, we also employed secondary items assessing diet. We used a measure based on the transtheoretical model [31] (containing three items) to evaluate the intermediate factors relating to secondary-outcome eating behaviours, and a measure of self-efficacy [32] (also containing three items) as an evaluation of preparatory factors. We also used two items [33] to assess dietary knowledge concerning secondary-outcome eating behaviours. Perceptions of neighborhood food environments relating to eating behaviour was assessed with ten items [43, 44].

We also measured health literacy [45], which is defined as the knowledge, desire, and skills for acquiring, comprehending, evaluating, and making use of health information, using a scale with good validity for the Japanese population. In the context of our previous research, the relationship of health literacy with vegetable intake behaviour was shown to be unaffected by socio-economic status, whereas promotion of vegetable intake behaviour has the potential to improve health literacy [46]. In addition, we measured subjective economic status [47], household income [30], educational attainment [23, 24, 46], and other attributes (namely sex, age, marital status, living status, and employment status).

Adherence to the intervention was assessed by the number of log-ins and duration spent in the website. Furthermore, the participants’ satisfaction with the intervention was assessed by using a self-administered questionnaire at post-intervention. With reference to previous studies on programme development [48, 49], we also included an evaluation (5 items) of the nutrition education programme itself. The items were as follows: (1) was the content of the programme fun? (2) was the content of the programme easy to understand? (3) after participating, did you become aware of any problems in your own diet? (4) did you feel the programme was helpful as a health management material? and (5) as a health management material, did it make you want to participate again another time?

The sample size was calculated using G*Power [50]. We set an effect size of 0.5, an α of 0.05, and a power of 0.95. With the expectation that two-thirds of participants would drop out, the size of the intervention group was set at 900 subjects, while that of the control group was set at 600 subjects with the expectation that a half of participants would drop out. Furthermore, we aimed to divide the intervention and control groups further according to income (low and high).

All statistical analyses will be performed with IBM SPSS Statistics 21.0. For continuous variables, independent t-tests will be used to determine inter-group differences and Man–Whitney U-tests for intra-group comparison, while the χ² or an equivalent test is to be used to determine the associations between categorical variables. Because this is a prospective RCT involving repeated measures, repeated measures analyses of variance will be applied to determine significant differences within the study groups. We also considered the possible effect of time on groups. The evaluation of the intervention is based on an intention-to-treat analysis. A p-value of 0.05 has been set as the level of significance.

This study has three main strengths. First, this is an RCT using high-quality research methods. Because the RCT participants were assigned randomly, there is very little allocation bias. Additionally, randomization took place after acquiring consent to participate, which means that the intervention group was not populated only with participants actively interested in the research topic. This helps in avoiding over-estimation of the effects of the nutrition education programme in the intervention group. Second, this is nutrition education programme was web-based, thus making it highly accessible. Furthermore, the study was carried out by a social research company with a wealth of experience in implementing academic research surveys. These points ensured high feasibility while minimizing the burden of securing an adequate sample size to increase statistical power. In addition, participants could be extracted to match the distribution of sex and age in the Basic Resident Register and household income in the Comprehensive Survey of Living Conditions. For this reason, possible confounding effects could for the most part be eliminated at the stage of allocation. Third, the programme affords numerous advantages for both participants and supporters. Participants can take part when they are free to do so (e.g. time, place, and scheduling). As this means that they can take part even if they are located far away from the researchers, it would be highly convenient for participants [26]. The programme also offers supporters the opportunity to provide a unified nutrition education programme that is not reliant on the number of participants and to provide support without being subject to spatial or geographical restrictions. While it is true that the initial cost is somewhat considerable, its operational costs are low. The economic burden on supporters is thus small. Furthermore, passwords can be set for browsing the nutrition education programme, which can prevent communication between groups that could impact the quality of research.

The RCT in this study is original in the following two regards. (1) It can verify the intervention effect for an adequate sample size, and (2) it can verify differences in the effect of a nutrition education programme by income strata. With this study, it will also be possible to verify the possible catch-up effect in relation to undesirable vegetable intake behaviour associated with low household income. Some reports have suggested that traditional classroom-based nutrition interventions might widen this disparity [6]. In contrast, with our study, it may be possible to show that a web-based nutrition education programme would inhibit the widening of this disparity.

Conversely, this study also has its limitations. It is possible that the results of the study will reflect the characteristics of individuals registered with the research company (e.g., tending to be young, high-earning, and highly educated [51‐53]). To minimize this potential limitation, we have made sure to match participants with demographic distributions in terms of sex, age, and household income. In addition, we requested research cooperation from individuals whose household income data – which was our objective – had been recorded in advance. This helps to restrict attribute bias as well as negative response rates for delicate questions pertaining to household income and other matters. Second, during participant recruitment, we excluded individuals reporting household incomes of ten million JPY or more. This means that the results do not target the entire population. Accordingly, care must be taken when drawing general conclusions. However, a limit of less than ten million JPY still applies to 88.4% of the Japanese population. Finally, participants did not have contact with one another (either in person or through indirect communication) for the period of the intervention. For this reason, it is possible that rates of adherence to nutrition education programmes will be lower than they would with enforced participation.

The promotion of vegetable intake behaviour is an important challenge for health promotion during adulthood, regardless of socio-economic status. ‘Health Japan 21 (the second term)’ policy initiative [21] is promoting the use of ICT as a health promotion strategy for the future. Because ICT frees individuals from the limitations of time or place, it may be a tool that encompasses individuals in lower socio-economic strata who have little time to spare. If we can show that web-based nutrition education programmes have a positive effect, our study would represent the outcome of a population approach as a health promotion strategy.