Spatial distribution of tuberculosis disease among men and women in Bandung city, Indonesia

: Tuberculosis (TB) is still a serious health problem particularly in developing countries caused by bacteria (Mycobacterium tuberculosis) that most often affect the lungs. Tuberculosis is curable and preventable. In general, tuberculosis (TB) incidences rates are usually higher in men than in women, but it has similar spatial distribution with a very high positive correlation. This study evaluated the spatial distribution of Tuberculosis Disease among men and women in Bandung City, Indonesia by means Bayesian shared component model. We found the joint effects dominated the spatial variation of TB for men and women which indicates there are similar risk factors that influence the spatial distribution of TB for men and women. The factors could be a health facility, health behaviours, and environmental factors. By knowing, the spatial pattern of TB for men and women, the government may use this information to develop an effective and efficient strategy to control TB transmission.


INTRODUCTION
Tuberculosis (TB) is caused by bacteria (Mycobacterium tuberculosis) that most often affect the lungs.To date, it is still a serious health problem particularly in developing countries with low health facility and health behaviors [1].Tuberculosis is curable and preventable.However, it is among the leading causes of death from infectious disease [2].In 2018, a total of 1.5 million people died from TB and recorded as the one of the top 10 causes of death and the leading cause from a single infectious agent (above HIV/AIDS) [1].The previous studies reported tuberculosis (TB) incidences rates are usually higher in men than in women [3] [4] [5] [6] [7].Globally, men are more likely to be diagnosed with TB than women, with a male-to-female ratio of 1.6:1 [4].
TB transmission are commonly found occurs within a household or community, leading to heterogeneous spatial patterns [8].A clear spatial clustering of TB could reflect ongoing transmission and causative risk factors.In order to develop efficiency and effective strategy to control and prevent TB transmission, identify the hotspot of TB and its spatial distribution by gender are required.It is important to evaluate the disease transmission considering gender to find more accurately the risk factors that have a significant effect in explaining the disease variation over space [9].
The ratio in Bandung was recorded 1.3:1.Several factors are suspected to contribute to the different number of cases based on gender such as healthy life behaviors.This study was conducted to evaluate gender differentials in TB for sub-districts in Bandung city.Data exploration shows in several sub-district the ratio higher than 1.3:1.However, there is a high positive correlation between the number of TB between men and women which indicates there are shared effects that influence by similar risk factors.The Bayesian shared component model reported the shared effect dominated the spatial variation of TB for men and women which is indicated by the standard deviation of the hyperparameter.The standard deviation estimate of the joint spatial effects is much higher than the standard deviation estimate of the individual spatial effects.In this study, we propose to use spatial joint modelling by mean Bayesian approach to explore the geographical variation of TB risk for men and women in Bandung, Indonesia.To data, the TB positive cases in SPATIAL DISTRIBUTION OF TUBERCULOSIS DISEASE AMONG MEN AND WOMEN Bandung are still high with the ratio between men and women is 1.2:1.Bayesian shared component model is better to be used to strengthen inference and correct for any spatially structured sources of bias, when distinct data sources on one or more related diseases are available [10].

Material
Bandung is a capital city of West Java, Indonesia with very dense population.In 2018, the population of Bandung are 2,452,179 inhabitants.TB is still becomes health issue because the number of incidences were still high.TB and population at risk data were collected from Bandung city health office.It can be accessed for free from http://data.bandung.go.id/.

Method
Based on the model that have been developed by Knorr-Held and Besag (2001) [11], we propose the following model: where   denotes the expected count for group  at location  which is defined as: and   is the relative risk for group  at location .The relative risk   is commonly modeled using log linear model by defining the expected count   as the offset variable.
For the spatially structured component is assumed follow iCAR prior [12] and for unstructured is exchangeable prior (iid).The model is estimated using R-INLA [13]

Data exploration
Tuberculosis (TB) was found in all sub-district in Bandung city.In 2018, the official health reported of 11,000 number of TB cases with the ratio between men and women is 1.2:1.The descriptive of statistics are presented in Table 1.The high incidences were found in the southwest, northern, and center regions of Bandung (see Fig. 1).The correlation between TB men and women based on sub-district observations of 0.856 which indicates there is a strong correlation.Figure 4 shows that there is evidence of a common spatial pattern in both gender.

Bayesian shared component model
In this section we present an advanced spatial analysis, in which both genders are analyzed simultaneously.In this case, the statistical model assumes a joint spatial effect that is shared between the two gender categories, men and women.represented the overall risk of TB for each gender.For men the overall risk is (−0.078)= 0.925 and for women (−0.099)= 0.906.The overall risk TB of men slightly higher than women.The density of both intercepts are presented in Figure 5. Table 3 shows the standard deviation of the hyperprior parameter.The higher value of the standard deviation explains that the random effect component has higher effects in explaining the spatial variation of TB for each gender.Here we found the joint spatial effect is the most important component.The distribution of each component is presented in Figure 5.We found the spatial pattern of TB for both genders is similar.The high risks of TB were found in the southwest and northern region of Bandung while for the eastern of Bandung, the risk of TB is relatively low.

DISCUSSION
The differential issue of gender in tuberculosis (TB) have been discussed worldwide [4].Globally, men are more likely to be diagnosed with TB than women, with a male-to-female ratio of 1.6:1 [4].The ratio in Bandung was recorded 1.3:1.Several factors are suspected to contribute to the different number of cases based on gender such as healthy life behaviors.This study was conducted to evaluate gender differentials in TB for sub-districts in Bandung city.Data exploration shows in several sub-district the ratio higher than 1.3:1.However, there is a high positive correlation between the number of TB between men and women which indicates there are shared effects that influence by similar risk factors.The Bayesian shared component model reported the shared effect dominated the spatial variation of TB for men and women which is indicated by the standard deviation of the hyperparameter.The standard deviation estimate of the joint spatial effects is much higher than the standard deviation estimate of the individual spatial effects.Therefore, it can be SPATIAL DISTRIBUTION OF TUBERCULOSIS DISEASE AMONG MEN AND WOMEN explained that the joint spatial effects have captured the spatial variability of the data.Moreover, the standard deviation estimate of the random effects for both men and women groups is small, which indicates that there is no other important source of variability in the data.The basic point of the proposed model is to observe the joint spatial pattern and to understand the effect of the individual spatial pattern of each gender.Based on the estimation of a statistic that measures the differences patterns between men and women we found there is a slight difference of 1.208.Based on the shared component model, we found the high-risk of TB for men and women are generally in southwest and northern of Bandung with the relative risk higher than its expected count.By knowing, the spatial pattern of TB for men and women is similar which indicates they have similar risk factors, the government may use this information to develop an effective and efficient strategy to control TB transmission.

CONCLUSION
In general, tuberculosis (TB) incidences rates are more likely higher in men than in women, but it has similar spatial distribution with a very high positive correlation.The spatial distribution of TB for men and women in Bandung city, Indonesia are dominated by the joint effects which indicates there are similar risk factors that influence the spatial distribution of TB for men and women.
Although men are more likely to be infected with TB than women, risk factors related to space are found to be the same.

Figure 1 .
Figure 1.Number of Tuberculosis by Gender and sub-district

Figure 2 .
Figure 2. Number of Population at risk by gender and sub-district

Figure 3 .Figure 3
Figure 3. Number of Tuberculosis (TB) and Population at Risk per sub-district

Figure 4 .
Figure 4. Number of Tuberculosis (TB) and Population at Risk per sub-district

Figure 4 .
Figure 4. Intercept of (a) Man and (b) Woman Models Table 3. Posterior mean of standard deviation (SD) of the random effects

Figure 8 .
Figure 8. Relative Risk Estimate of TB (BTA+) for Men and Women

Table 1 .
Descriptive of statistics