This study recruited truck drivers from a Taiwanese freight company. Of the 840 truck drivers in this company, 450 agreed to participate in the present study. Since the participants could decide whether to join the experimental group, it was not feasible methodologically to enforce random assignment. Thus, we adopted a quasi-experimental research design (Fig. 1) with 179 participants as the experimental group and 271 participants as the control group . The participation rate was 53.6%. A total of 125 experimental group participants and 117 control group participants completed the pretest and post-test. The data of these 242 participants were included in the final statistical analysis. A goodness-of-fit test on the final and initial analysis samples indicated no significant difference in these samples according to educational levels, marital status, diagnosis with metabolic syndrome, body type, age, BMI, self-perceived health status, self-perceived susceptibility, self-perceived severity, perceived barriers, cues to action, self-efficacy, or healthy eating behaviors (p > 0.05).
The control group participated in a conventional health program, which involved a healthy diet course and the use of a nutritional education manual for intervention programs. The experimental group participated in a social-media-based health program which involved conducting health interventions through social media in addition to classroom teaching and manual use. The social media application used in this study was LINE, which is used by 91.3% of the population in Taiwan . The functions of the LINE Official Account include scheduled push notifications, one-to-one real-time feedback, a built-in loyalty e-card, and automatic question-and-answer responses. A licensed dietitian and a trained nurse administered the LINE group for enhancing, teaching, and reinforcing healthy eating behaviors. They both received standardized training at the principal investigator’s office to ensure the consistency of their responses.
In October 2019, we conducted a focus group discussion with the freight company’s manager, workplace nurses, and employee representatives to jointly develop intervention approaches and incentive programs. Subsequently, participants were recruited in January 2020. They were required to sign a consent form and were asked to fill in a pretest questionnaire before joining the social media group (experimental group) or conventional course group (control group) according to their personal preference. A 12-week intervention program was conducted from February to April 2020. After the intervention, the participants were asked to fill in a post-test questionnaire in May 2020. The research tools, research process, and informed consents were reviewed and approved by the Institutional Review Board of Cathay General Hospital (CGH-P108115).
We posted recruitment advertisements at truck driver rest areas in the freight company’s five branches in northern, central, and southern Taiwan. In January 2020, the principal investigator visited each branch to explain this research to the company’s drivers. The exclusion criteria of this study were as follows: (1) those without medical examination data, (2) those with renal insufficiency (patients who had been diagnosed with renal insufficiency by doctors, regardless of the stage), (3) those who had been diagnosed with cancer and did not receive treatment or were still under treatment, (4) those with unstable cardiovascular disease (those who had been hospitalized for related diseases or who had had their medication adjusted within 1 month), and (5) those who did not own a smart phone. Those who agreed to participate in the social media group were asked to scan a QR code with their mobile phone to join a LINE group to facilitate the intervention program. Participants who completed the pretest and post-test questionnaires received a gift worth US$13 after completing the post-test.
In this study, an intervention program was designed based on five concepts of the HBM: self-perceived severity, self-perceived susceptibility, perceived barriers, self-efficacy, and cues to action. The purpose of this program was to improve participants’ awareness of disease risks, improve their self-efficacy and perceived benefits, reduce their perceived barriers, and promote changes in their healthy eating behaviors by increasing their exposure to cues to action. The detailed content of the intervention program is shown in the Additional file 1: Appendix A.
The intervention program involved online messages, online instant responses, a picture-based food log, an audio e-book, and a loyalty e-card. These five aspects are detailed in the following text.
Provision of online messages
Twelve themes, one for each week, were planned in advance, and two to three health text or video messages were provided weekly. The message content included the five concepts (i.e., perceived susceptibility, perceived severity, cues to action, perceived barriers, and self-efficacy) of the HBM as well as simple techniques for implementing healthy eating (Additional file 1: Appendix A.). Messages were designed according to the concepts of knowledge transfer, knowledge clarification, and life application, and were disseminated directly or through a question-and-answer approach. This approach was conducted by the instructors who sent questions via the LINE group which the group members then answered.
Knowledge transfer and life application involved providing information on various meal combinations recommended by the Health Promotion Administration (HPA) of Taiwan, including Taiwanese breakfast, Taiwanese snacks, cafeteria food, and convenience store meals. Examples of a direct message are as follows: (1) “A healthy day starts with breakfast. Do you know what type of breakfast is the healthiest? The figure below provides an example.” and (2) “Let’s see whether you still remember ‘My Healthy Eating Plate.’ Click on the link below to refresh your memory.” Examples of question-and-answer content are as follows: (1) “What type of food will make you sleepy while driving?” (2) “Do you know that your eyes need protection while driving under the sun for a long time? Do you want to know which vegetables and fruits can protect your eyes?” and (3) “Do you know what a ‘sweet burden’ is? Guess how many cubes of sugar someone who consumes 2,000 cal a day can eat?” The video messages involved precautions for maintaining healthy eating habits and continuous exercise.
A licensed dietitian and a trained nurse served as online coaches to provide instant responses and cues to action and reduce perceived barriers. This study provided online coaching services from 8 am to 8 pm daily, including weekend. When participants sent messages or asked questions individually, coaches would provide an instant response for immediately solving the participants’ diet-related problems.
Because it is inconvenient and dangerous for truck drivers to read while driving, we converted the health manual into an audio e-book. This step increased the possibility of exposing the truck drivers to health information and improved accessibility to cues to action.
Picture-based food log
Participants captured photos of their meals and snacks and uploaded them to the LINE group. The online coaches provided suggestions for improvement or encouragement according to My Plate established by the Ministry of Health and Welfare of Taiwan in 2019.
We designed a point accumulation mechanism using a loyalty e-card to encourage drivers in the social media group to participate in online activities. Under the incentive mechanism, the drivers were awarded one point for uploading a photo of one meal in the first 4 weeks. From weeks 5 to 12, the participants were required to upload two photos of their meals to receive one point. In week 12, one additional point was awarded if the drivers’ meals met the criteria (types and servings of food) of My Plate. One point was awarded each time a participant answered an online question correctly. Finally, at the end of this research, the drivers who finished first, second, and third in terms of points at each branch of the company were given a backpack worth US$60, a thermos mug worth US$30, and a US$15 gift voucher, respectively.
The questionnaire was developed based on previous HBM related studies [41,42,43]. Six experts from nutrition and public health specialties evaluated the content of the questionnaire regarding its applicability, correctness, and completeness to ensure its conformity with the research purpose. The content validity index of the overall questionnaire was 0.79. The questionnaire consisted of the following sections: sociodemographic factors, health examination data, healthy eating behavior, self-perceived health status, self-perceived susceptibility, self-perceived severity of chronic diseases, barriers to healthy eating, self-efficacy of healthy eating, and cues to action. We conducted principal component factor analyses for the sections mentioned above, besides sociodemographic factors, health examination data, as well as healthy eating behavior, and retrieved one factor from each section. The cumulative explained variances of the sections were 60.7% for self-perceived health status, 83.7% for self-perceived susceptibility of chronic diseases, 87.3% for self-perceived severity of chronic diseases, 72.4% for barriers to healthy eating, 83.5% for self-efficacy of healthy eating, and 81.8% for cues to action. Thus, we next calculated the total score by averaging the item scores since the items of each section reflect one domain.
Self-perceived health status
The measurement of self-perceived health status was developed based on the questions of the Taiwan National Health Interview Survey and consisted of five items . The items included three forward-scored items: “In general, what do you think of your current health condition?” “I think that my current health condition is very good,” and, “I think that I am healthier than other people the same age as me,” and two reverse-scored items: “I think that I get sick more easily than others do” and “I think that my health is becoming worse.” The responses were measured using a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). The average of the five items created the total score (ranging from 1 to 5) with a higher score indicating a more positive self-perceived health status. The Cronbach’s α value was 0.83.
Healthy eating behavior
The drivers’ healthy eating behavior was measured using the food frequency scale , which consisted of the intake frequency of nine food items: unrefined grains, vegetables, fruits, beans or bean products, seafood, eggs, meat, nuts, and water. The options included “0–1 days per week,” “2–3 days per week,” “4–5 days per week,” and “6–7 days per week.” Each food frequency was evaluated by the criteria of My Plate which indicated the frequency of meat, water, nuts, vegetables and fruit is almost every day, while the frequency of seafood, eggs, beans, and bean products is at least two to three times a week. For each food, a point was awarded if a participant’s consumption was consistent with the suggestion of My Plate. Otherwise, no point was awarded. The total score for healthy eating behavior ranges from 0 to 9, with a higher average score indicating a more favorable healthy eating behavior.
The self-perceived susceptibility scale was revised from Huang et al.’s study . The respondents were asked how likely the following six diseases will happen within the next six months: heart attack, stroke, kidney disease, diabetes, cancer, and other chronic diseases. The scale was measured by a 4-point scale ranging from 1 for “strongly disagree” to 4 for “strongly agree.” The average of the item scores represented the total score with a higher score indicating a higher self-perceived susceptibility. The Cronbach’s α value was 0.96.
The self-perceived severity scale was revised from Huang et al.’s study . The respondents were asked how severe they thought the six aforementioned diseases would be if they were diagnosed in the next six months. A 4-point scale ranging from 1 for “not very serious” to 4 for “very serious” was used for assessment. The average of the six item scores was calculated as the total score, with a higher score indicating a higher self-perceived severity. The Cronbach’s α value was 0.97.
The measurement of perceived barriers was developed according to the previous literature [43, 45] and consisted of two questions: “I don’t like the taste of most high-nutrient foods” and “I think it is difficult to change my eating behaviors to consume more high-nutrient foods within the next two weeks.” These items were scored on a 4-point scale ranging from 1 for “strongly disagree” to 4 for “strongly agree.” The average of the two item scores was used as the total score with a higher score indicating higher perceived barriers to implementing healthy eating behaviors. The Cronbach’s α value was 0.62.
The measurement of self-efficacy was developed according to the previous research [43, 46, 47]. The scale measured how confident the respondents felt in their ability to manifest healthy eating behaviors in the next two weeks by two items, scored on a 4-point scale. The average of the two item-scores was calculated as the total score for self-efficacy with a higher average score indicating a higher self-efficacy. The Cronbach’s α value was 0.80.
Cues to action
Cues to action was measured by a scale which was developed according to previous studies [43, 48]. The scale measured how much the respondents would agree they would follow the information from mass media (e.g., news stories, ads, and other programs), health professionals (e.g., doctors, nurses, and other medical professionals), and laypersons (e.g., family members and friends) about the food choices. The three items were scored on a 4-point scale ranging from 1 for “strongly disagree” to 4 for “strongly agree.” The average of the three item scores was used as the total score of cues to action with a higher average score indicating a higher likelihood to follow cues to action. The Cronbach’s α value was 0.89.
The sociodemographic variables considered in this study included birth year, marital status, educational level, and health examination data, including height, weight, waist circumference, high-density lipoprotein, triglycerides, fasting blood glucose, and blood pressure. Information about the variables was provided by the participants on the basis of their annual employee health examination reports. The BMI was calculated from the participants’ height and weight. The metabolic syndrome was determined according to the biochemical value .
Statistical analysis was conducted using IBM SPSS Statistics version 19. Means, standard deviations, and percentages were calculated to demonstrate the distributions of sample characteristics. To ensure comparability, we examined the distributions of pretest variables between the experimental and control groups by using the t-test and chi-square test. The mean differences between the pretest and post-test results of the two groups were examined by paired t-test. Generalized estimating equation (GEE) regression models were adopted to investigate the intervention effects.