Determinants of Dehydration Status and Associated Risk Factors of Cholera Outbreak in Oromia, Ethiopia

2.1. Study Area

This study was conducted in Oromia regional state of Ethiopia. The Oromia region is one of the nine regional states of Ethiopia, the homeland of the Oromo. It is bordered by the Somali Region to the east; the Amhara Region, the Afar Region and the Benishangul-Gumuz Region to the north; South Sudan, Gambela Region, and Southern Nations, Nationalities, and Peoples' Region to the west; and Kenya to the south. It is also the largest regional state covering 286,612 square kilometers.

2.2. Source Data

The data for this study was secondary data which is obtained from Ethiopian Public Health Institute (EPHI). All cholera patients reported in for a year in 2019 has been considered for this study. Thus, 965 patients in the study area and period were considered for the study.

2.3. Variables of the study

2.4. Statistical Model

Ordinal logistic regression model is a type of logistic regression model that are used to analyze ordinal dependent variables. For this study since the outcome variable is in the form of ordinal scale (ordered dehydration status as no dehydration, some dehydration and severe dehydration).

Mostly considering the method is natural for multiple categorical response having more than two possible values. And also the most well-known of these ordinal logistic regression methods is the proportional odds model. The main idea under proportional odds model is that re-expressing the categorical variable regarding a number of tertiary variables based on internal cut-points in the ordinal scale.

Therefore, the ordinal logistic regression model can be formulated as: -

$\left({ln}_{\gamma }\right)={\beta }_{i}-\left[{\beta }_{1}{X}_{1}+{\beta }_{2}{X}_{2}+\dots +{\beta }_{n}{X}_{n}\right]$ where I goes to the number categories minus one. As it is known, this is the part the researchers really want to investigate. So having seen the status of cholera diseases, it is clear to make ordinal response like no dehydrated, some dehydrated and severe dehydrated.

Consequently, if the researchers know that a category is ordinal then there are special models that tell us how the covariate relates to someone being higher or lower on the scale. A proportional odds model that deals limiting some factors to have common parameters and leaves other factors free to have separate parameters is called a partial proportional odds model. Therefore, the basic form the generalized linear written as follow: -

Where ${\gamma }_{i}$ is the cumulative probability for the${i}^{th}$category and ${\beta }_{i}$ is the threshold for${i}^{th}$ categories,${\beta }_{1}$,${\beta }_{2}\dots ..{\beta }_{k}$ are the parameters to be estimated, ${X}_{1},{X}_{2}\dots ..{X}_{K}$ are the covariates and k is the number of covariates. From this equation the numerator to the right side determine the location of the model. The denominator equation could determine the scale. The ${\alpha }_{1},{\alpha }_{2}$ ….${\alpha }_{m}$are coefficients for the scale components and ${y}_{1},{y}_{2}\dots .{y}_{m}$ are m covariates for the scale components (selected from the same set of variables as X’s).

3.1. Results

The results of the study indicated that of the total 965 cholera patients, most of them 560(58%) were severely dehydrated by cholera; whereas only 123(12.7%) was not dehydrated and the other 282(29.2%) was registered as somewhat dehydrated.

The graphical presentation in the Fig. 1 below is classification of dehydration status by age category expressed for male and female patients separately. It clearly shows that both male and female patients aged 15–44 were the most severely dehydrated compared to the other age groups. In all dehydration status, the 15–44 age groups were counted to be the largest group included as cholera infected in the study area and/or period. The graphical results supported the frequency results above that the severe dehydrated were exceeding the other stages.

The Table1 below is the summary of chi-square test of association between dehydration status and the identified factors. It indicated that there is no enough evidence to say that sex of patients their age group and whether they vomiting have significant association with dehydration status. Despite, history of travel, history of contact, other sick person in the family, IV, antibiotics and lab test has statistically significant association with dehydration status of peoples those have cholera infection. Thus, to now the magnitude and strength of those factors with the dehydration status, the researcher intended to fit appropriate statistical model.

The likelihood ration chi-square test is -780.21446 with p-value of 0.000. This indicated that the model as the whole is statistically significant as compared to the null model. It is to mean that at least one of the considered factors was significant in determining the dehydration status

The results presented under Table 2 below were the coefficient model table which is used to know the significance of individual factors. It indicated the variables like history of travel, history of contact with other patients, other sick patients in the family, Intravenous and Antibiotics drugs are significant with 5% level of significance. Thus, the researcher can conclude that those factors are potential in determining the status of dehydration. The goodness of model fit of Table 3 (Appendix) shows that the model fit the data well. The test of parallel lines of Table 4(in the appendix) is also revealed that the assumption is fulfilled.

3.2. Discussions

The descriptive results of the study indicated that of the total 965 cholera patients, most of them 560(58%) were severely dehydrated by cholera. Many other previous studies also revealed that the status of dehydration of patients was categorized under severity.

Most the factors included in study have significant association with severity of cholera and its dehydration status. However, sex of patients and their age categories was no significant to determine the status of cholera dehydration. The results of Pearson chi-square test of association indicated that history of travel of patients (p-value < 0.001), history of patients contact with other patients (p-value < 0.001), whether there is other sick patients in the family (p-value < 0.001), whether intravenous was given (p-value < 0.05), whether antibiotics was given (p-value < 0.01) and whether the patients had lab test (p-value < 0.01) have significant association with the dehydration status of cholera outbreak at 1% and 5% level of significance. Hence, the result was consistent with the previous work by [2, 4]

The ordinal logistic regression, model presented in Table 2 includes the significant factors of dehydration status of cholera outbreaks. For those have history of travel, the odds of severely dehydrated versus the combined some dehydrated and no dehydrated was exp(1.133804) = 3.11 times higher than those have no history of travel (p-value < 0.001). Thus, it is to mean that the severity of dehydration status of cholera outbreak has positive relationship with those have history of travel. It is agreed with previous work [2]

The odds of severely dehydrated versus some dehydrated and no dehydrated was exp(1.008163) = 2.74 and exp(.3112465) = 1.37 for those have history contact and those have sick patients in their family respectively (p-value < 0.01). Previous studies revealed that the variable was significant factors for cholera outbreak [5, 6]

Likewise the other factors like whether the patients had intravenous injection and whether the antibiotics treatment was given by the patients was included in the study and found significant with 1% and 5% level of significance. Other previous studies are also consistent with this result [7–9]

The results of cut1 (-1.738741) in Table 2 above refers to the estimated cut-point on latent variable to differentiate no dehydrated from some dehydrated and severe dehydrated when the value of all predictors are evaluated at their reference categories. Whereas cut2 (.186633) is the estimated cut-point used to differentiate some dehydrated from no dehydrated and severe dehydrated keeping the factors to be at their reference.

Proportional odds assumption of ordinal logistic regression was meet for this study (p-value = 0.094) which is to mean that the effects of explanatory variables over all the categories of dehydration status was expected to be similar.