We used primary data obtained through a survey in Tamansari sub-district, Tasikmalaya City with a population of 79,392 people. Tamansari sub-district has the most number poor households which is 13.48 percent of the total population among ten sub-districts in Tasikmalaya City. Using Slovin and purposive random sampling methods, a total sample of 358 respondents were interviewed, encompassing parents with children aged 2–14 years. After the data cleaning process, 329 respondents were included in our analysis. We identified health complaints from children with low food diversity. This study was approved for ethical consideration by the University of Padjadjaran Bandung with ethical number of 1081/UN6.KEP/EC/2022.
This study uses food diversity as a research object by collecting data on food groups consumed by children within the last 24 hours including cereal, milk and cheese, meat, nuts, vegetables, and fruits. In addition, children’s health status, children’s characteristics, parents’ characteristics, household characteristics, and children’s medical costs are also indicators of the respondents’ assessment. The parameter used for children’s health is to identify children with low food diversity that has experienced health complaints. Several types of health complaints that were asked of respondents included typhus, stomach ulcers, coughs, flu, and fever. We make sure that the health complaints come from children with low food diversity, however it also possible that the health complaints due to other factors.
We used two probit model regressions to identify the factors that cause children's health complaints. The first model used DDS an independent variable, while the second model used DSS. Variables used in the estimation method can be seen in Table 1. In general, the probit model equation can be represented by the following equation:
$${P}_{i}=f\left({Y}_{i}\right)$$
$$=f\left({\beta }_{0}+{\beta }_{1}{X}_{1i}+\dots +{\beta }_{p}{X}_{pi}\right)$$
1
From the above equation, the function can be transformed into a linear form as follows:
$${f}^{-1}\left({P}_{i}\right)={\beta }_{0}+{\beta }_{k}{X}_{ki}+\dots +{\beta }_{p}{X}_{pi}$$
2
Where \({P}_{i}\) is the probability of success in the i-th model, while \({f}^{-1}\left({P}_{i}\right)\) is a normal cumulative distribution function (CDF), \({\beta }_{k}\) and \({X}_{ki}\) respectively are the coefficients of the regression model parameters and independent variables with k = 1,2, 3…, p.
Table 1
Variable | Description | Unit |
Health Complaints | Health complaints experienced by children with low food diversity (typhus, stomach ulcers, coughs, flu, and fever). | 1 = Yes 0 = Otherwise |
DDS (Dietary Diversity Score) | The food diversity score is divided into two parts, namely below average and above average | 1 = Above the average (> 3.863222) 0 = Below the average (< 3.863222) |
DSS (Dietary Serving Score) | The dietary serving score is divided into two parts, namely below and above average | 1 = Above the average (> 16.74468) 0 = Below the average (< 16.74468) |
Food intake | The frequency of meals the child eats in one day | Intervals ranging from 1 to 5 |
Age child | Age of children | Year |
Gender child | Child gender | 1 = Boy 0 = Girl |
Vaccine | Completed COVID-19 vaccine | 1 = Yes 0 = Otherwise |
Mother age | Age of mother | year |
Housewife | Mother's employment status | 1 = Work 0 = Housewife |
HH income | Household income is divided into two parts, namely above and below average of all sample | 1 = Above the average (> Rp. 2.290.237) 0 = Below the average (< Rp. 2.290.237) |
HH member | Number of family members | Person |
Distance | Distance from home to health care facility | 1 = More than 1 km 0 = Otherwise |
Measuring food diversity in this study refers to the calculation of the Dietary Diversity Score (DDS) (Rathnayake et al., 2012). The DDS is calculated by adding up the values of the various food groups consumed in the last 24 hours (Krebs-Smith et al., 1987). We used a food group consisting of cereals, vegetables, fruits, whole grains, meat/fish/eggs and milk/dairy products. To calculate the DDS, we carried out several steps. First, we asked the child's consumption of those food group. Second, we scored the children who consumed any of the food groups (see Table 2). Third, we accumulate the DDS by enumerating the total food groups consumed with a score ranging from 1 to 6.
Table 2. Scores for Each Food Group in DDS Measurement
Food Group | Dietary Diversity Score |
Cereal | 1 |
Diary and Cheese | 1 |
Meat | 1 |
Nuts | 1 |
Vegetable | 1 |
Fruit | 1 |
Total | 6 |
Source: B2SA |
In addition, we use the Dietary Serving Score (DSS) to measure food diversity, where this calculation refers to the portion recommendations for each food group that is given a certain score. The recommendations for meal portions refer to the Pedoman Gizi Seimbang (PGS) or Balanced Nutrition Guidelines through the principles of Beragam, Bergizi, Seimbang dan Aman (B2SA) or Diverse, Nutritious, Balanced, and Safe menu [13]. The scoring method takes into account the same six major food categories as in the DDS calculation, with each food group receiving a maximum score of 20 (see Table 3). Cereal and fruit groups were received maximum of 5 point for each five recommended servings. Dairy & cheese and nuts groups were received maximum 1 point for each one recommended serving, whereas for one recommended serving of meat have 2 points of maximum. For three recommended serving of vegetable groups have 6 points of maximum.
Table 3. Scores and Serving Recommendations from Each Food Group
Food Group | Serving (B2SA) | Assigned Score |
Cereals | 5 | 5 |
Diary and Cheese | 1 | 1 |
Meat | 1 | 2 |
Nuts | 1 | 1 |
Vegetable | 3 | 6 |
Fruit | 5 | 5 |
Total | 20 |
Source: B2SA
The cost of child illness in this study was calculated using medical direct costs and government cost. According to Nørgaard et al., direct costs can be calculated using medical expenses such as medical registration fees, medical action costs, transportation costs, and other costs [23]. We included these costs into our survey to obtain costs from patient perspective. The calculation of direct costs in this study refers to the research of Rein et al. which represents medical costs and transfer fees from the government. The formula used in calculating direct costs is as follows [24]:
$$Medical cost=Prevalence x Population x Direct Cost \text{x} 12$$
3
Prevalence is the proportion of children experiencing health complaints, population is the total of respondents; and medical cost is the amount spent by parents when children experience health complaints. Meanwhile, the formula used to calculate the economic burden borne by the government due to low food diversity is as follows:
$$Gov\_Cost=Prevalence x Population x Gov\_Assistance \text{x} 12$$
4
The economic costs borne by the government are calculated based on the prevalence of children who have low food diversity multiplied by the total number of respondents and assistance from the government. Bantuan Sosial Non Tunai (BSNT) or Non-Cash Social Assistance is used as an indicator of government assistance provided to the lower middle class, where the amount of assistance given if it is monetized is as much as $23.32 (Rp. 200,000), one of which is aimed at improving child nutrition. We are using BSNT as a proxy for government costs due to the aims of this program to improve coverage of food and nutritional needs in the community. In addition, this program is also one of the government efforts to alleviate stunting.