Study design
A two-armed, pre-post test, clustered randomized control trial (CRCT) design will be used in this study. Figure 1 shows the overall design. The data will be collected at three time points: T0, baseline, T1, 3-months after the completion of an 8-week intervention, and T2, 6-months after the intervention.
Participants
Each fire station will be treated as an individual cluster. The participants will be firefighting teams’ members from the Hong Kong Fire Service Department (HKFSD), and they will be selected from more than one fire station, located in different districts in Hong Kong. Convenience and snowball sampling methods will be used for the recruitment. The participants will be assigned randomly into either intervention or control groups. All the participants will be blinded to intervention allocation.
Inclusion Criteria
Inclusion criteria will be: 1) male; 2) aged 18 years or older; 3) currently working for the HKFSD as firefighters, working on “24 hours on, 48 hours off” shift; and 4) owners of smart phones with internet access. Firefighters who are participating in any other relevant health promotion programme at the time of the study will be excluded. Written informed consent (Appendix II) will be obtained from each participant. In order to avoid interference or contamination of data, all participants will be reminded not to disclose any information which they have received from the researcher, and this reminder will be stated in the consent form.
Sample size
The sample size has been estimated on the basis of the following parameters: cluster size (number of firefighters per cluster; n = 30, standard deviation, s = 1), the probability of committing a Type I error (a = 0.05), the probability of committing a Type II error, (b = 0.2), indicating a study with a power of 80%, r = 0.05 and effect size = 0.6 (77). The estimated sample size for each group (intervention and control) is 46, which indicates 92 participants in total. By assuming an attrition rate of 17% (98, 99), the overall estimated sample size, m, was calculated to be 108 (54 for each group) in total. As the above estimation was designed for the full study, the sample size for feasibility study should be on a smaller scale. However, there is no clear definition or guidelines for estimating the sample size for a feasibility study (100). According to (101), a sample size of 20-25 in a group will probably be adequate to demonstrate intervention efficacy. For a group comparison, 10-20 participants per group would be sufficient.
Teaching materials
The contents of teaching materials have been designed already on the basis of three sources: Promoting Healthy Lifestyles: Alternative Models’ Effects (PHLAME) (102), the Centre for Food Safety (103) and the Department of Health in HK (104). In order to design an intervention on healthy eating which would be suitable for HK firefighters, existing information about healthy eating education and four out of ten sessions which focused on healthy eating from PHLAME have been modified. The teaching materials focus on (1) the rationales for healthy eating; (2) the advantages of consuming fruit and vegetable; (3) understanding methods of cooking fruit and vegetables; (4) practical tips for getting enough of fruit and vegetable when eating out and during festival seasons. As well, the TTM was designed to assess each intervention participant’s “stage of change”. The TTM includes four core constructs: stages of change, processes of change, decisional balance and self-efficacy (105). In formulating this intervention, guidelines such as goals, processes of change, strategies and health-promotion information given through WhatsApp was set with reference to those developed by Lee et al. (2017) to support the application of TTM to promote healthy eating (106). These teaching materials will be delivered to the participants in the intervention group through WhatsApp.
Development of pamphlet
The contents of the pamphlet will be the same as the teaching material, which was described above. All participants will receive all stages of intervention regardless of which stage they belong to at the baseline.
Fidelity of pamphlet
The fidelity of the teaching materials and pamphlet have been established by a panel of experts including three experienced registered nurses, one dietitian and two nutritionists. All of them have more than ten years of experience in their own areas of expertise. A self-report checklist was completed by each expert. The checklist indicates all components and aspects of the intervention that the participants will receive, from precontemplation to action stages. Self-report measures of consistency will be in binomial (yes/no) and ordinal (1 = Not relevant, 2 = Somewhat relevant, 3 = Quite relevant, 4 = Highly relevant) formats in multiple item surveys. The advantages of using this method are the low cost, ease of administration, and speed of data collection. All experts agreed that on the proposed teaching materials should include Health promotion information (HPI) items with stage-based TTM given through WhatsApp. However, one of the experts did not agree with the others on the content of one item. All the items related to goals, processes, strategies and HPI were rated either “Quite relevant” or “Highly relevant”.
Development of questionnaire
A questionnaire (Appendix III) has been developed on the basis of several studies (26, 107-111). This consists of six study aspects: (1) Personal information; (2) Working characteristics; (3) Eating habits; (4) Health promotion programme for healthy eating; (5) Stage of change; and (6) Decisional balance and self-efficacy,
The questionnaire includes 11 to assess the HK firefighters’ eating habits, especially their fruit and vegetable consumption. These questions are based on a series of eating habit studies (26, 107-109). The consumption of fruit and vegetable will be measured by asking: 1) “How often did you consume fruits/vegetables in the past week?” (ranging from “Not consume” to “Seven days”), 2) “Where do you usually have fruits/vegetables?” (three choices are: “At home”, “During duty” or “No difference, at home and during duty”) and 3) “On a day you consumed the fruits/vegetables, how much did you take on average on that day?” (in bowls or serving). For eating habits, questions included “On average, how many days do you have breakfast/lunch/dinner/night snack (take away included) within a week?” (ranging from “None” to “More than five days”) and “How about the speed of having meal when you are on duty in the fire station when compared to the meals you eat at home?” (ranging from “Slower” to “Faster”) are used.
Thirty-five questions were designed to understand decisional balance. Based on Ma et al. (2003), they address the pros, cons and self-efficacy of eating fruit and vegetable. The decisional balance will be evaluated by asking the participants how important each of the pros and cons is in their decisions to consume recommended amounts of fruits or vegetables using five-point Likert scales ranging from 0 (Not at all important) to 5 (Very important). Eight pros and eight cons for fruits intake and nine pros and ten cons for vegetables intake will be used to assess the decisional balance.
Twelve questions were designed to assess self-efficacy by rating, on five-point Likert scales ranging from 0 (Very difficult) to 5 (Very easy), how difficult or easy participants find it to eat based on the recommendations in six high-risk situations, for each of the two dietary habits (110).
Four questions were designed to identify the stage of change in eating fruit and vegetable (i.e. “How many servings of fruits/vegetables the respondent usually consumed each day; intention to eat ≥ 2 servings a day of fruit or ≥ 3 or more servings a day of vegetables; whether the participant had been consuming ≥ 2 servings of fruit or ≥ 3 servings of vegetables for more than 6 months; intention to eat more). The participants will be asked to state their intentions to have servings of fruit and vegetable , by choosing one of five statements each representing a stage of change: “No, and I do not intend to in the next 6 months” [Precontemplation], “No, but I intend to in the next 6 months” [Contemplation], “No, but I intend to in the next 30 days” [Preparation], “Yes, I have been doing so for less than 6 months: [Action], Yes, I have been doing so for more than 6 months [Maintenance]. These four questions were drawn from De Vet et al. (2006).
Another three questions were developed to assess the HK firefighters’ points of view about the healthy eating promotion: 1) “Can a healthy eating promotion programme help you to change or understand your eating attitudes and habits?”, 2) “Can a pamphlet on healthy eating alone facilitate you to change or understand your eating attitudes and habits?” and 3) “Can a healthy eating promotion programme delivered through a mobile app facilitate you to change or understand your eating attitudes and habits?” They will be required to rate these above three questions on four-point Likert scales ranging from 1 “Absolutely cannot” to 4 “Absolutely can”.
Validity of questionnaire
A panel of six experts, three nursing professionals, one dietitian and two nutritionists, was invited to establish the content validity for the questionnaire. The content validity index (CVI) was 0.966, with average item-CVIs ranging from 0.667 to 1.000, and individual panel members’ CVIs ranging from 0.885 to 0.987. These results indicate that the questionnaire is valid to investigate the firefighters’ knowledge about healthy eating. The original English version of the questionnaire was translated into Chinese by a professional translator and another professional translator performed back-translation, obtaining a CVI no less than 0.8. One way to examine the reliability of this questionnaire was using test-retest reliability. Ten firefighters from the HKFSD, who fulfilled the inclusion criteria for the study, were invited to take part in this reliability test. Their knowledge about healthy eating was examined on Day 1 and Day 15 (two weeks in between) using this questionnaire. The agreement between the data collected on these two days ranged from 0.704 to 1.00, with a mean of 0.75. All of these correlation coefficients were significant at the 0.01 level.
Body measurements
Anthropometric data will be the secondary outcome of this study. The participants will be weighed wearing light clothing and no shoes. Body weight will be measured by the researcher using the Innocare weighting scale. When measuring body height, a Butterfly Brand measuring tape and a 10-inch triangular ruler will be placed perpendicular the wall to mark the top of the participant’s head and a Butterfly Brand measuring tape will then be used to measure the distance along the wall from the floor to the ruler. The BMI (kg/m2) will be calculated from height and weight measurements. Waist circumference and hip circumference will be measured by the researcher, using the Butterfly Brand measuring tape.
In order to ensure the reliability of the data collected by the researcher, a professional nurse will be invited to perform the data collection alongside the researcher to ensure inter-rater and intra-rater reliability. The measurement will be performed by the researcher the professional nurse. The agreement between the professional nurse and the researcher as well as within the researcher should be kept the intra-rater correlation coefficients (ICC) no less than 0.8 (112-114) for each measurement.
Randomization
The participating firefighters who fulfill the inclusion criteria will be assigned randomly to either the intervention or the control groups (based on their clusters) by using computer generated random numbers. This will be done by an information technology (IT) expert with more than fifteen years’ experience in this field and he will be blinded.
Data collection
Ethical approval has been obtained from the Human Subjects Ethics Sub-committee (HSESC) [HSEARS20180527001] of the Hong Kong Polytechnic University. An information (study aims, procedure and duration) sheet will be provided and a signed consent form (Appendix II) will be obtained from each participant prior to the data collection. The baseline questionnaire will then be completed by the participants and anthropometric data including body height, body weight, waist circumference, hip circumference, BMI and WHR will be measured. The data will be collected by the researcher in either a University laboratory or the participants’ fire stations at three different time points: T0 (Baseline), T1 (three months after the completion of the 8-week intervention), and T2 (six months after the completion of the intervention). The participants will be reminded about the follow-up data collection one month and two days in advance through WhatsApp. All the data will be accessed by the research team only. The schedule for enrolment, interventions, and assessments is showed in Figure 2.
Intervention group
The participants in the intervention group will be required to complete the questionnaire and their anthropometric data will be recorded. Then they will receive the pamphlet from the researcher at T0 and the baseline stage-matched teaching material every two weeks through WhatsApp.
Control group
Similarly, the participants in the control group will also be required to complete the questionnaire and their anthropometric data will be recorded. The only difference between the intervention and control groups is that the participants in the control group will receive the pamphlet (the same as the intervention group) from the researcher at T0 only.
Data analysis
All of the collected data will be cleaned prior to the analysis process. The data analysis will be conducted by a statistician who is blind to the participants’ group allocation. The participants’ demographic characteristics of will be presented using descriptive statistics (mean and standard deviation) for continuous data, including age, body height and weight, BMI and percentage frequencies for categorical data, including gender, marital status, and educational level. Inter-relationships between the variables will be assessed using Pearson’s correlation coefficients. Differences between the intervention and control groups on outcome indicators will be compared using t-tests and chi-square tests for the continuous and categorical data respectively. The effect of time (baseline; three and six months after the completion of the programme) on the outcome measurements will be investigated using repeated measures ANOVA. The impact of “baseline stage-matching” on “stage of change” and other outcome measures within the intervention group will be compared using the Kruskal-Wallis H test and one-way ANOVA. Normality tests will be conducted. If the data are normally distributed, it will be appropriate to use parametric tests such as one-way ANOVA. But if they are not normally distributed, non-parametric tests such as Kruskal-Wallis H test will be used. P-values of less than 0.01 will be considered as statistically significant for all comparisons. All statistical analyses will be conducted using the Statistical Packages for Social Sciences (SPSS Inc., Chicago) version 22.0. Intention to treat analyses will be applied. Up to now, there is no clear cut-off for the acceptable proportion of missing data in a dataset for statistical data analysis. Questionnaires with missing rates of more than 10% will be regarded as “disqualified” and excluded from the analysis. Various statistical methods will be used to treat missing data, including replacement by means or values from the regression analyses, depending on the amount and type of missing data.