Exploring the impact of soil and water salinity on dietary behavior and health risk of coastal communities in Bangladesh

The purpose of this study was to evaluate the impact of soil and water salinity on dietary behavior and health risk in the coastal people of Bangladesh. This study was conducted among 240 respondents in rural coastal sub-districts in Khulna and Patuakhali of Bangladesh using a simple random sampling technique. To evaluate the association between health risk and salinity exposure categories, a multinomial logit regression analysis was conducted and statistical significance was declared at p 0.05. A significantly higher amount of salinity (NaCl) level was found in radish, potato, bean, bitter gourd, rice, shallow tube-well, and pond water from Patuakhali than Khulna. Males and those aged 36–50 years (RRR:1.89, SE:0.58) and 51–65 years (RRR:4.51, SE:1.81) were associated with hypertension compared with the females (RRR:0.57, SE:0.18) and age group 20–35 years. Consumption of shallow tube-well water (RRR:3.12, SE:1.46), salt content rice (RRR:1.36, SE:0.50), salt content vegetables (RRR:1.09, SE:0.09), salt content fish (RRR:2.77, SE:0.47), and intake of table salt (RRR:1.05, SE:0.03) were significantly associated with risk factors of hypertension (p< 0.01). A sustainable policy for salt reduction through dietary interventions along with the promotion of low saline foods and drinking water must be a priority with special emphasis on coastal areas.


INTRODUCTION
Sea level rise (SLR) is considered as one of the major reasons for salinity intrusion into soil and groundwater which results from both natural and human-induced climate change (Wassmann et al. ; Sarwar ). Due to its geographical position (low lying belt), Bangladesh experiences frequent effects of climate change. People living in coastal areas of Bangladesh are struggling with the adverse impacts of climate change (Talukder et al. ). In recent years, as a result of sea-level rise, soil and water salinity is increasing and at the same time, frequent storms and cyclones are occurring (BBS a, b; IPCC ). Increasing salinity has become a vital issue for the people living in the coastal regions of Bangladesh (Rabbani et al. ). An increasing amount of salinity in the water and soil of these regions causes scarcity of drinking water, problems in irrigation, agriculture, and severe diseases like hypertension and kidney disease (CEGIS ; Khan et al. a; Mahmuduzzaman et al. ). Several studies suggested that inhabitants of those areas have a high salt intake level as they largely depend more on rivers, groundwater sources, and ponds for drinking water (Khan et al. a, b; Rasheed et al. ) than those living more inland and this intake may increase their risk of hypertension and other diseases (Rahman & Ravenscroft ; Rasheed et al. ). In coastal areas of Bangladesh, both surface water and groundwater are contaminated by varying levels of salinity and these have the potential to affect the health of 35 million inhabitants living in these regions through direct or indirect use of water resources (Khan et al. b). The population is vulnerable to climate change and sea-level rise in low-lying settings and commonly use untreated water and food sources and are likely to have similar exposure to high sodium consumption (Nicholls et al. ; Vineis et al. ). According to the World Health Organization, a significant percentage of the population in the coastal region are exposed to a higher level of sodium than the recommended daily intake (>5 g/day) (Khan et al. a, b; Rasheed et al. ). Several studies of food consumption have In coastal areas of low lying countries like Bangladesh, food is considered as the major source of sodium which causes high blood pressure (Habiba et al. ; Talukder et al. ) and at the same time in those regions, water salinity (sodium) has been found to be associated with higher sodium consumption (Khan et al. ; Talukder et al. ). In particular, in the southwestern areas of Bangladesh, high salinity in surface and groundwater sources is affecting human health. SRTT () found that severe diseases like kidney stones and rheumatism can be caused by high saline contaminated drinking water. Rasheed et al. () claimed that coastal families have money to spend, but there are not enough fresh fruit or vegetables in the market to spend it on. The local communities are forced to survive on locally available food varieties.
Although there have been many studies on the salinity levels of soil and water and its associated effect on crop production and adaptation strategy, and on the land fertility in coastal areas of Bangladesh (Ali ; IPCC ; Khan et al. b), research on the health impact and food behavior of the coastal communities in the southern part of Bangladesh are hard to find, despite its massive significance due to climate change and increasing soil salinity in the coastal region in South Asia, particularly in Bangladesh.
The health impact is gradually increasing in the coastal communities of Bangladesh. Considering these research gaps, the main purpose of this study was to evaluate the impact of soil and water salinity on dietary behavior and the health risk of local coastal communities in Bangladesh. The problem is particularly acute in the area where soil salinity is relatively high (Figure 1). Furthermore, most of the vegetable crops are very sensitive to saline conditions and each unit increase in salinity with yield decreased from 6-19% (SDRI ; Das et al. ). Cropping intensity is reduced considerably due to high soil and water salinity.

Study design, setting, and population
These sub-districts are also prone to salinity intrusion as it is part of the exposed coastal classification (PDO-ICZMP 2003;SDRI ). In short, we selected three unions (Dacope Sadar, Paikgacha Sadar, and Kalapara Sadar), based on the diversity of foods and potable water uses. In this study, four villages were randomly selected from each union using a simple random sampling technique. Trained research staff made household visits for data collection from the selected villages and, of those, 240 participants were available for interview and health assessments.

Questionnaire regarding dietary measurements
The daily dietary intake and the previous 7 days diet of the respondents was assessed by a 24-hour recall method and Moreover, the amount of water a person drinks per day was collected through a questionnaire survey, and salinity was measured by testing the samples from that specific source. The amount of raw salt (table salt-NaCl) consumed by the respondent during the meal was also collected through a questionnaire survey.  grams per liter (g/1,000 g) for drinking water and food sample analysis where 1 ppt ¼ 1,000 mg/L).

Covariates
Information on socio-demographic conditions, the sources of drinking and cooking water, years of using the water sources, health status, and diet of the past 7 days including consumption of rice, vegetables, fish, and table salt intake of each participant were collected. Using a graphical representation of these methods, the data were described by compiling it into graphs and tables.

Empirical analysis
The potential uncertainty and bias of univariate analysis have emphasized the need for multinomial logit model analysis which considers the effects of all risk factors influencing disease in coastal areas. In contrast with gender, age category, region, drinking water source, cooking water source, bathing water source, amount of salt intake from rice consumption, amount of vegetable intake, amount of table salt intake and amount of water intake data are in a discrete format. The database used in the current study has five discrete severity of disease categories: disease-free, hypertension, kidney disease, skin disease, and other diseases. The disease classification is based on only the individual disease. An ordered probability model may be most convenient in this study. Following the common use of unordered discrete outcome models, a multinomial logit model for health impact levels was used to explain the severity of health impact (salt intake from saline foods in coastal people) on dietary behavior among the five categories of disease. This model permits the use of a categorical dependent variable (Theil ).
In the general case of a random effect context of the health impact outcomes, the probability of disease by consuming salinity containing foods and water of a coastal community can be specified as: where Y ki is an outcome variable such as the different types n is the total number of observations), α k is a constant parameter for health impact outcome category k; β k is a vector of the estimable parameters of each category; X ki represents a list of explanatory variables which are responsible for determining health impact outcomes and ϵ ki is a random error term that is independently and identically distributed. As the samples are categorized into different disease types, therefore, we assume the error term (ϵ ki ) is following the generalized extreme value (i.e. Gumbel) distribution, where the probability of falling disease types k of sample i, conditioning on the explanatory variables X ki and a constant α k . Hence, the multinomial logit model forms as follows: In our empirical framework, the estimated coefficients were used to evaluate the probabilities of the disease falling into one of the five categories. Our model consists of five probabilities, P k (k ¼ 1, …, 5), related to the five categories of disease (i.e. disease-free, hypertension, kidney disease, skin disease, and other diseases). The probability of being no disease, hypertension, kidney, skin disease, and other diseases are denoted as P 1 , P 2 , P 3 , P 4 , and P 5 respectively. As a result, the sample likelihood of the multinomial logit model (Equation (2)) follows the standard maximum likelihood method. Since the explanatory variables are continuous and discrete determining factors, log-odd ratios of the outcomes become: Thus, the coefficients are distinguishable only up to an additive constant so only the difference in coefficients is identifiable. One outcome (the base category) of the coefficient is fixed to zero to resolve this interdeterminacy. Due to the non-linear characteristics of the multinomial logit model, the estimated coefficients of the independent variables do not represent their effects on the dependent variable. In that manner, the relative risk ratio (RRR) represents the effect of a relevant risk factor. In our analysis, the RRR of risk factors is computed relative to the base category (i.e. no disease). For instance, the relative probability of health impact outcome (k ¼ 2) to the base category (k ¼ 1) is given as: Therefore, the RRR is written as: Equations (4) and (5) imply the RRR of hypertension larly, if we consider k ¼ 3, 4, 5 it will indicate the RRR of kidney, skin, and other diseases relative to disease-free (k ¼ 1) types. Moreover, the intuition of RRR of an independent variable indicates the increase (RRR >1) or decrease (RRR < 1). In this study, the multinomial logit model and the associated RRR were estimated using Stata (version 13.0). All dietary patterns were entered into the same model. The model was adjusted for potential confounding by gender, age, region, source of drinking water, source of the cooking water, and source of bathing water, which influenced the risk of health status and is also associated with the dietary pattern scores. Statistical tests were twosided, and p < 0.05 was considered a statistically significant level.

Ethics statement
The research method was approved by the Institutional Review Board (IRB) of the Human Research Ethics Committee at Jashore University of Science and Technology, Bangladesh. All the study participants were formally consented before their participation and signed consent was obtained for each participant.

Sociodemographic status
A total of 54.16% of the survey respondents were male and 45.84% were female (Table 1). About 36% were 20-35 years old, 40% were 36-50 years old, 19.17% were 51-65 years old and 5% were >65 years old. Among them, 36.25% of the respondents were farmers and 13.75% of the respondents were housewives. Salt content test in some local food items grown in the coastal region

Water salinity (NaCl) levels
Location-wise variations in salinity (NaCl) levels of drinking water are shown in Figure 3. The highest proportion of respondents exposed to more than 600 mg/L sodium chlor-

Impact of salinity on dietary behavior and health risk
The estimated severity of the disease model is presented in    shows that both gender and age were important factors for increasing the risk of hypertension (Figure 4(a)). It is indicated that the 36-50 year age range had moderate risk and those aged 51-65 years had a higher risk of hypertension. Above this, it also explains that males (51-65 years and 36-50 years old) had more risk than females. Figure 4(b) shows that females had a higher risk of kidney disease than males and both gender and age were significant factors to increase the risk of kidney disease. The age 36-50 year age range had a lower risk than those aged 20-35 years.  Note: The estimated model is interpreted as the relative risk ratio (RRR). The relative risk ratio represents whether the probability is increased or decreased by that variable for each disease severity outcome. Significance Level: ***p < 0.01, **p < 0.05, *p < 0.1.
a comparatively higher risk of skin diseases than females. It also explains that both age and gender were important factors for increasing the risk of skin disease. Figure 5(b) shows the varying probabilities of other diseases (diarrhea, fever, gastric, etc.) in both males and females and at the same time indicates that not only age but also gender was another influencing factor for an increase or decrease in these diseases.

DISCUSSION
This study revealed that the consumption of high salt containing foods and potable water was positively associated with hypertension, kidney disease, and skin disease in the studied population aged above 20 years in the rural coastal  year-olds. Our findings were consistent with previous studies

CONCLUSIONS
In summary, salinity levels of different food and water samples were significantly higher in Patuakhali and Khulna. This research indicates that the increased risk of hypertension, kidney, and skin disease among the coastal population in Bangladesh was due to the high salinity concentration in daily food intake, drinking, and cooking water sources. Males and those aged 36-50 and 51-65 years had a significantly higher probability of hypertension and skin disease compared to females. From this research work, it has been shown that most of the people in the concerned study consumed more than 5 g of salt daily compared to their biological needs and this was significantly associated with hypertension. In the interim, safe water options having low saline content alone or together with dietary and lifestyle interventions need to be investigated. So, at the community level, certain strategies should be developed to reduce salt intake and at the same time awareness should be created about the effect of excessive salt intake on health.

DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.