The analysis of factors that affecting household food security of tuberculosis patients in Surabaya

Tuberculosis (TB) cases and household food security are fundamental problem which can be concerned with each other. Household food insecurity can affect the ability of a family member in counteracting an infectious disease such as TB. This research is aimed to study the factors that affecting household food security of tuberculosis patients in Surabaya. By Binary logistic regression model, there are four significant variables; Head of Household’s Occupation, Expenditure per Month, The Household Density and Type of Roof, with the correct classification 65.6%.


Introduction
Poverty, food security and malnutrition are major concerns among international health when all people have physical, social and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life [1]. There are four food security indicators used to classify each household [2], those are: -Food Availability -Food Stability -Food Access - The Quality of Food Security where, the categories of household food security can be assigned by combining two aspects, between continuity of food availability and quality of food [2], which is presented in Table 1. Food security are fundamental problem which can be concerned with a public health. Household food insecurity has been associated with a bad ability in counteracting diseases, i.e. Tuberculosis. Tuberculosis is the type of disease that spread directly, caused by bacteria that spread through the air. TB has become the third highest cause of death in the world. According to data from the East Java Provincial Health Office, the largest TB cases in Indonesia were in East Java and Surabaya accounted for the largest number, which is 3093 out of 33999 cases in East Java in 2017 [3].
This research was conducted to study the factors that affecting variable on household food security of tuberculosis patients in Surabaya. Binary Logistic Regression was used to analyze the model in this research.

Binary logistic regression
Binary logistic regression is a method of analysis that used to find out the relationship between the response variables (Binary or dichotomous) with the predictor variables that are polychotomous [4]. The response variable (y) consists of two categories: "success" (y =1) and "failure" (y = 0). The model logistic regression given as The Equation 2.1 can be explained as a logit model: by logit transformation of π(xi) as: The test is conducted to get the best model which was built by the significant parameters. Parameters were first tested simultaneously and then tested partially to get the significant parameters.

Simultaneously test of parameters
Hypothesis:

Classification procedure
This procedure was used to evaluate the result of prediction value given by the best model to compare with the observation value, [6] give an evaluation on classification procedure to see the probability of miss classification. It is measured by apparent error rate (APER). APER value stated the proportion value of miss classification sample by the function of classification. If the subject only classified as two groups, y1 and y2, then determination of classification errors can be known through the classification table described in Table 2.

Source of data and research variables
This research used primary data which was collected by conducting survey using a random sampling technique [6] to 259 respondents taken from 3093 TB patients in Surabaya [3]. The research variables consist of three kind variables: Respond variable, Indicator variable of Food Security and Predictor variable.

Respond variable
Respond Variable (Y) in this research is Household food security Status of TB Patient in Surabaya as given in Table 3.

Indicator variable of food security
The food security status to be determined by four Indicator that gives in Table 4,

Predictors variable (X)
Predictor variable (X) gives in Table 5.  Figure 1 shows that more than 50% households are food insecure in Surabaya. 58.7% of the TB patient household are food insecure and 41.3% of them are food secure.   Table 6. , it showed that P-value = 0.000 less than α indicated that reject H0, the partial test was then conducted to find out the variables that significant in the model.

Partially test of parameter
The partial test with the hypothesis as in subsubsection 2.1.2, give the result as shown in Table 7.

Goodness of fit model
The hypothesis of goodness of fit model as in subsubsections 2.1.3, give the result as shown in Table  8. This table showed that P-value more than α = 0.05, indicated the test cannot reject H0, and so the model fit.

Classification procedure
The subject only classified as two groups, with Y1: Secure and Y2: Insecure. The determination of classification errors can be known through the classification table described in Table 9.