The study design was reviewed and approved by the human research ethics committee at the Khon Kaen University (Reference No. HE582071). All participants gave written informed consent prior to the commencement of the questionnaire interview.

A cross sectional study was conducted using a structured questionnaire administered by in-person interview. Interviews were conducted between March and July, 2018. The sample comprised 749 individual residents of Phnom Penh, selected using a multi-stage random sampling technique.

Multi-stage random sampling was used to select the sample of 749 study participants. Five districts were randomly selected using a lottery method by which the researcher gives each district a number, drawing numbers from the box randomly to choose the five samples from the total of 12 districts within Phnom Penh. Then, either two or three communes were randomly selected from each selected district, to a total of 12 communes. The sample size for this study was calculated using an established formula ²¹ and the estimated sample size was 749. A systematic random sampling procedure was used to select 749 households from a list of all households in a given commune, which were provided by each commune chief, out of a total 44,436 households. All households in the 12 communes were labelled with a number and every 10 households were selected until the target was fulfilled with 749 households. Finally, one household member aged 18–59 was randomly selected from each household (unless the household had only one member).

The inclusion criteria for the study were that participants were between 18 and 59 years of age, and that they were able to understand the questionnaire. Residents with any serious health problems at time of data collection (causing them to be bedbound), diarrhea (defecating more than three times per day), pregnancy, mental illness, deformity or lower limb amputation were excluded from participating in the study.

The questionnaire was designed specifically to answer the research question, and was informed by literature review. A pilot study (n = 30) was conducted to assess content validity. One district in Phnom Penh municipality was selected for the pilot study that was different from the five districts used for the main study. Some items that were found to be difficult to understand were changed after pilot testing. Following the results of the pilot study, the questionnaire content was approved by five experts in the field of nutrition. The experts were sent the questionnaires and had three weeks to check its content. Experts recommended few changes to the questionnaire following their assessment. Internal consistency was assessed using Cronbach’s alpha, and found to have a reliability coefficient of 0.857.

The questionnaire had three parts (see Extended data) ²² : part 1 covered demographic and socioeconomic characteristics (gender, age, marital status, educational attainment, occupation, number of family members, personal monthly income); part 2 covered lifestyle and behavioral factors (tobacco use, alcohol consumption, food habits, soft drink consumption, physical activity and sedentary behavior); and part 3 covered the understanding of factors associated with fast food consumption. Prior to participant interviews, the questionnaire was translated into Khmer by an independent translator

A field team of three research assistants underwent one-day training to standardize their knowledge of the study objectives and administration of the questionnaire. Researchers used the face-to-face questionnaire to interview the participants at their home in their free time after work. Data was collected after informing participants about the objectives of the study, benefits and assuring confidentiality to those who was eligible for this study. Some participants refused to participate, the reasons being that they were tired after work, or that they did not want to provide information (due to perceptions of privacy/security). The raw data of the 749 respondents were recorded into MS-Excel for database management before in-depth analysis.

Knowledge of fast food consumption was the dependent variable used in the study, and took levels poor, fair and good. Participants were asked seven questions relating to fast food to determine their knowledge, answering true or false to each statement. The correct answer was given a score of 1, and the incorrect answer given a score of 0. The minimum total score was 0, and the maximum was 7. Responses were categorized using Bloom’s cut off with poor knowledge being assigned to scores <60%, fair knowledge for scores between 60 and 79%, and good knowledge for scores ≥80%.

Age, household income and expenditure, and the number of individuals in the household were coded as continuous variables. Gender (male or female), marital status (single, married or divorced/widowed/separated), level of education (no formal education, primary, secondary, high school, associated degree, bachelor’s degree, master’s degree, or doctoral degree and higher), occupation (farmer, unemployed, non-governmental organization employee, self-employed, student, government officer, home maker, unskilled worker, or other), people they were cohabiting with (spouse, parents, relatives, none, friend, or other), smoking (yes, no), alcohol consumption (yes, no), vegetable intake (standard portion size of rice spoons/day), fruit intake (portions/day), number of episodes of exercise per week, hours of screen time per day, and hours of sleep per night were coded as categorical variables.

Data were imported to Stata version 13 (College Station, Texas, USA) for analysis. Continuous and categorical data were inspected using descriptive statistics to determine the frequencies and percentages (categorical variables) and means, medians and standard deviations (continuous variables) of each socio-economic, demographic and lifestyle characteristic collected. Bivariate logistic regression was used to estimate the association between each socioeconomic and lifestyle factor and the outcome measure of understanding of fast food consumption. Crude odds ratios (CORs) were computed using 95% confidence intervals (CIs) and variables with statistical significance of p<0.25 were entered into the final model. In the final model, multivariate logistic regression was used to estimate the association between socioeconomic lifestyle factors and understanding of fast food consumption. Adjusted odds ratios (AORs) were then computed using 95% CIs. Significance was considered at p<0.05.

A total of 749 participants from five districts and 12 communes were recruited for this study. The socio-demographic characteristics of respondents are summarized in Table 1. Participants were 50.20 % female with an average age of 32 ± 11 years ²³ . The relationship status of participants was 53.94% married, 43.52% single, and 2.54% divorced/widowed/separated. Regarding educational attainment, 31.91% of participants had completed high school, 26.44% had completed Bachelor’s degrees, whereas 15.22% had only completed primary school. The most common occupations were private company workers (28.97%), self-employed (21.36 %) and students (20.16%), with only 0.67% reporting to be farmers and 0.93% unemployed. Nearly half of the sample lived with their spouse (47.93%), a quarter lived with parents (25.37%), 12.95% lived with other relatives, and only 13.89% of participants lived with fewer than three family members. Monthly earnings ranged from $40 to $5,100 US with a median of $300 ± 687 USD. Monthly expenditure ranged from $20 to $3,750 USD with median $200 ± 394 USD.

Only 10.41% of participants smoked, and 54.34% had consumed alcohol in the past 12 months. Most participants (65.95%) ate at least 6 portions of vegetables per day, 21.76 % ate between 4 and 6 spoonfuls of vegetables per day, and only 12.28% ate less than 4 spoonfuls. Similarly with fruit, only 44.33% of respondents ate fewer than three portions of fruit portions per day, 20.03% ate between three and five portions per day, and 35.65% ate at least five portions. 54.21% of respondents consumed fewer than six spoonfuls of meat per day, and 45.79% consumed at least six spoonfuls of meat per day. With respect to physical exercise and sedentary behavior, 50.07% of the respondents exercised at least weekly. In addition, 73.56% of respondents had at least two hours of screen time a day and 56.21% of the sample slept more than eight hours each night ( Table 2).

In their responses to seven questions related to knowledge about fast food consumption, there were four questions which respondents answered correctly at least half the time, and the remaining three questions were answered correctly less than half the time. For example, the question “Milk tea such as pearl tea is good for health because it contains both carbohydrate and dairy milk” was answered correctly as false 38.72% of the time; “Cola drinks containing high carbohydrates could help digestion” was answered correctly as false 45.66% of the time, and “Fast food such as hamburgers and pizza contain fiber which is good for your digestive system” was answered correctly as false 49.93% of the time. Descriptive statistics for all questions are given in Table 3.

Based on respondents’ answers to the questions in Table 3, a level of knowledge was assigned as one of three levels; good, fair, and poor, according to Bloom’s cut-off point. The range of possible scores was from 0 to 7, and the mean sample score was 4.20 (±1.70). Boundary criteria used were a score of less than 60% signified poor knowledge, a score of between 60 and 79% signified fair knowledge, and a score of 80% or higher signified good knowledge. In the sample of respondents, 52.07% had poor knowledge, 22.70% had fair knowledge, and 25.23% had good knowledge ( Table 4).

Knowledge levels poor and fair were deemed unsatisfactory for the purpose of performing logistic regression to determine the predictive factors of knowledge level. The overall prevalence of poor and fair knowledge of fast food consumption was 74.76% (95%CI: 71.64-77.88%). In a simple bivariate logistic regression, the factors gender, marital status, weekly exercise and nightly hours of sleep were found to be associated with having poor or fair knowledge of fast food consumption (p<0.25) and were taken forward as factors in the final model ( Table 5).

The final model using a multiple logistic regression, found only two factors to be associated with poor and fair knowledge of fast food consumption. These were weekly exercise and nightly hours of sleep. Those who did not exercise at least once a week (AOR: 1.53, 95%CI: 1.15-2.25; p <0.001) and those who slept for less than eight hours per night (AOR: 1.45, 95%CI: 1.09-2.12; p =0.014) were more likely to have poor or fair (“unsatisfactory”) knowledge of fast food consumption ( Table 6).

Firstly, the participants in this investigation were from a capital city and these results may not be representative of the whole population of Cambodia; if other provinces or cities were included, the results might be different. Secondly, as the current study was a cross-sectional analytical study, we could not infer causality. Lastly, all information collected in the study was based on self-reporting; there might be methodological bias due to under- and over-reporting of certain behaviors.

Figshare: Socio-demographic and Lifestyle factors associated with Understanding Fast Food Consumption among Adults in Cambodia. https://doi.org/10.6084/m9.figshare.12894740.v1 ²³

This project contains the following underlying data:

Figshare: Socio-demographic and Lifestyle factors associated with Understanding Fast Food Consumption among Adults in Cambodia. https://doi.org/10.6084/m9.figshare.12894644.v1 ²²

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).