A Comparison on Impact of HIV / AIDS Patients Characteristics on their Blood Pressure in Nigeria

DOI: http://dx.doi.org/10.24018/ejers.2020.5.8.1807 Vol 5 | Issue 8 | August 2020 884  Abstract—The study focused on comparison on impact of HIV/AIDS patient’s characteristics on their blood pressure in Nigeria: a case of NAUTH, COOUTH and Onitsha general hospital in Anambra State. The blood pressure being the response variables are systolic blood pressure & diastolic blood pressure, while the predictor variables being the HIV/AIDS patient’s characteristics are age, baseline count, initial weight, present weight and CD4 count of HIV/AIDS patients. The R software package was employed to facilitate the data analysis. The Multivariate Regression Model of the two response variables (Systolic PB and Diastolic PB) was first fitted with the coefficient of determination of 31.88% and 46.80% respectively for NAUTH data, 27.9% and 37.98% respectively for COOUTH data and 97.35% and 57.15% respectively for general hospital, Onitsha data. The test on the significance of the parameters for the multivariate regression for NAUTH data revealed that age and baseline count of HIV/AIDS patients have significant relationship with systolic BP at 5% level of significance, whereas other predictor variables (initial weight, present weight and CD4 count of HIV/AIDS patients) are not significant, while in the second model, only age has a significant relationship with diastolic BP, whereas initial weight, present weight, baseline count and CD4 count of HIV/AIDS patients do not have significant relationship with diastolic BP at 5% level of significance. The test on the significance of the parameters for the multivariate regression also revealed that only age has significant relationship with systolic and diastolic BP at 5% level of significance, whereas other predictor variables are not significant for both COOUTH and general hospital Onitsha data. It was further revealed that the data collected from the general hospital Onitsha has the highest coefficient of determination (0.9735) with the lowest AIC (1348.944), BIC (1374.462) and residual standard error (2.587) for systolic blood pressure model which makes the data used in this study the most suitable for the model employed under the stipulated year of study. Also observed that the same data collected from the general hospital Onitsha has the highest coefficient of determination (0.5715) with the lowest AIC (1825.917), BIC (1851.435) and residual standard error (6.008) for diastolic blood pressure model which equally makes the data used in this study the most suitable. It is clear from the result obtained in this study that the data set collected from general hospital, Onitsha from 2003 to 2017 is most appropriate for the multivariate multiple linear regression models.



Abstract-The study focused on comparison on impact of HIV/AIDS patient's characteristics on their blood pressure in Nigeria: a case of NAUTH, COOUTH and Onitsha general hospital in Anambra State. The blood pressure being the response variables are systolic blood pressure & diastolic blood pressure, while the predictor variables being the HIV/AIDS patient's characteristics are age, baseline count, initial weight, present weight and CD4 count of HIV/AIDS patients. The R software package was employed to facilitate the data analysis. The Multivariate Regression Model of the two response variables (Systolic PB and Diastolic PB) was first fitted with the coefficient of determination of 31.88% and 46.80% respectively for NAUTH data, 27.9% and 37.98% respectively for COOUTH data and 97.35% and 57.15% respectively for general hospital, Onitsha data. The test on the significance of the parameters for the multivariate regression for NAUTH data revealed that age and baseline count of HIV/AIDS patients have significant relationship with systolic BP at 5% level of significance, whereas other predictor variables (initial weight, present weight and CD4 count of HIV/AIDS patients) are not significant, while in the second model, only age has a significant relationship with diastolic BP, whereas initial weight, present weight, baseline count and CD4 count of HIV/AIDS patients do not have significant relationship with diastolic BP at 5% level of significance. The test on the significance of the parameters for the multivariate regression also revealed that only age has significant relationship with systolic and diastolic BP at 5% level of significance, whereas other predictor variables are not significant for both COOUTH and general hospital Onitsha data. It was further revealed that the data collected from the general hospital Onitsha has the highest coefficient of determination (0.9735) with the lowest AIC (1348.944), BIC (1374.462) and residual standard error (2.587) for systolic blood pressure model which makes the data used in this study the most suitable for the model employed under the stipulated year of study. Also observed that the same data collected from the general hospital Onitsha has the highest coefficient of determination (0.5715) with the lowest AIC (1825.917), BIC (1851.435) and residual standard error (6.008) for diastolic blood pressure model which equally makes the data used in this study the most suitable. It is clear from the result obtained in this study that the data set collected from general hospital, Onitsha from 2003 to 2017 is most appropriate for the multivariate multiple linear regression models.

I. INTRODUCTION
The Human Immune Virus and Acquired Immune Deficiency Syndrome epidemics are both global phenomena threatening the health of various peoples, culture and population in the world. The Sub-Saharan Africa (SSA) with about 10% of the world's population has over two third of the people living with HIV [1]. HIV means Human Immune Virus. It is a virus that attacks, destroys and continues to deplete human immune system. The acronym AIDs means Acquired Immune Deficiency Syndrome. This suggests that the condition or illness is not inherited but acquired from possible environment factors such as virus infections. Similarly, immune deficiency means that the viruses have gradually caused deficient immunity as clearly manifested in poor nutrition and low resistance to opportunistic infections [2].
The threat of HIV has continued to be one of the most dreaded health challenges in the world since 1980s. The global AIDS response revealed that the national meridian HIV prevalence infection in Nigeria as 4.1% [3]. HIV/AIDS affects both the old, young, men and women in the society and in fact affect the productivity of every nation. From its inception this disease has destroyed lives, families and societies. HIV and AIDS deplete human immune system which kills the white blood cells resulting to death of its victims. The epidemic has become a serious issue globally. It is no longer only a health issue but a substantial threat to blood pressure, imposing a heavy burden, first on families, communities and eventually on the economy.
It is obvious that some characteristics of HIV/AIDS patients may be influenced by their blood pressure. Blood pressure measures cardiovascular function by measuring the force of blood exerted on peripheral arteries during the cardiac cycle or heartbeat. The measurement consists of two components [4]. The first is the force exerted on the arterial walls during cardiac contraction and is called systole. The second is the force exerted during cardiac relaxation and is called diastole. They represent the highest (systole) and lowest (diastole) amount of pressure exerted during the cardiac cycle. Blood pressure is recorded as fraction, with the systolic measurement written, followed by a slash and then the diastolic measurement [5]. Blood pressure may be affected by many factors, including blood volume, peripheral resistance, age, condition of the muscle of the heart genetics, diet and weight, activity, and emotional state [6]. In this paper, we used the R software for estimating the parameters of Multiple Linear Regression Model for blood pressure; examine whether there is any significant relationship between HIV/AIDS patients' characteristics and their blood pressure; and to ascertain among the hospitals the best that adequately fit the multivariate multiple linear regression model.

II. REVIEW OF RELATED LITERATURE
Zakari and Abdullahi [7] examined economic impact of HIV/AIDS and stigmatization on women in Nigeria as a challenge for the actualization of Millennium Development Goals (MDGs). The study was carried out using primary data among groups of Nigerian women. The data were analyzed using simple percentage method. The results of the study revealed that negative presentation by some medical personnel and the sensational captions by the Nigerian mass media on the so-called dead sentence nature of HIV/AIDS epidemic made it so scary that people found it difficult to accept its presence and so stigmatize people especially women with the disease. However, the study recommends that religious organizations, government and nongovernmental organizations intensify sensitization efforts towards combating the epidemic.
Obansa et al. [8] examined the burden of HIV/AIDS on income groups (upper and lower income earners) in Nigeria, its impacts on human capital development and economic growth. Income differential, the relative difference in income per capita of the quintile group, life expectancy, outof-pocket health expenditure, direct health expenditure, gross per capita formation was estimated, using data for the period 1986-2010. The study employed a panel data analysis procedure in order to capture the relative incidence (burden) of the epidemic between these income groups in Nigeria. Stationarity test was conducted on the variables used in the estimation. It was found that all the variables with an exception of health expenditure were stationary at first difference. Similarly, the long-run variability test of the incidence (burden) of HIV/AIDS on differential income earners and economic growth in Nigerian was also carried out and the residuals were found parsimonious over the period. Findings showed that the epidemic is already putting pressure on the income earners in Nigeria, especially those in the lower income group.
Sunday et al. [9] in their study evaluated the impact of HIV/AIDS on the Performance of the Nigerian Economy using annual time series data sourced from the World Bank Database, and Central Bank of Nigeria statistical bulletin. The variables considered in their study includes: gross domestic product which was used as proxy for economic growth, hence the dependent variable while HIV/AIDS and government expenditure on health were considered as independent variables respectively. The findings of their study indicated that all the variables defined in the model were stationary and there exists a unique long run relationship between the dependent and independent variables in the model. Hence, it was revealed that HIV/AIDS had a significant negative impact on productivity and by implication economic growth. Similarly, findings showed that government spending on health had a significant positive impact on economic growth in Nigeria during the period studied.
Ekezie et al [10] researched on application of multivariate multiple linear regression model on vital signs and social characteristics of patients. The result revealed that the multivariate multiple linear regression model was adequate for the relationship between the variables: Systolic Blood Pressure, Temperature and Height of patients on one hand, and the two social characteristics: Age and Sex on the other. A test of significance revealed that Age and Sex have influence on the Vital Signs. Following the result, they recommended that researchers should carry out a similar research work, making the predictor variables up to four to compare result.
The study shall examine the comparison on impact of HIV/AIDS patient's characteristics on their blood pressure in Nigeria: a case of NAUTH, COOUTH and Onitsha general hospital, having reviewed past works.

III. METHODOLOGY
The multiple linear regression with n independent observations on Y and the associated values of Zi, is the complete model of Representing equations (2) in matrix form, we have Cov() = E() =  2 I A one in the first column of the design matrix Z is the multiplier of the constant term 0. It is customary to introduce the artificial variable Zj0 = 1 so Classical Linear Regression Model where  and  2 are unknown parameters and the design matrix Z has jth row [Zj0, Zj1, ⋯, Zjr]

IV. LEAST SQUARES ESTIMATION
One of the objectives of regression analysis is to develop an equation that will allow the investigator to predict the response for given values of the predictor variables. Thus, it is necessary to "fit" the model in (3) to the observed yj corresponding to the known values b0 + b1Zj1 + ⋯ + brZjr .That is, we must determine the values for the regression coefficients  and the error variance  2 consistent with the available data. Let b be the trial values for  . Consider the that would be expected if b were the "true" parameter vector [11]. The method of least squares selects b to minimize the sum of squared differences.
The coefficient b chosen by the least squares criterion is called least squares estimates of the regression parameters.
They will henceforth be denoted by  to emphasize their rule as estimates of .The coefficients  are consistent with the data in the sense that they produce estimated (fitted) mean responses, , whose sum of squared differences from the observed j y is as small as are called residuals. The vector of residuals  Ẑ y   contains the information about the remaining unknown parameter  2 .
To establish notation conforming to the classical linear   The multivariate linear regression model is with Cov((i)) = ii I . However, the errors for different responses on the same trial can be correlated. Given the outcomes Y and the values of the predictor variables Z with full column rank, we determine the least squares estimates ) ( i  exclusively from the observations, Y(i), on the ith response. Conforming to the single-response solution, we take Collecting these univariate least squares estimates produces For any choice of parameters say , the matrix of errors is Y -ZB. The error sum of squares and cross-products matrix is (Y -ZB)(Y -ZB) is minimized by the least squares estimates  .
Then, using the least squares estimates  , we can form the matrices of The orthogonally conditions among the residuals, predicted values, and columns of Z, which hold in classical linear regression, hold in multivariate multiple regression.
so the residuals ) i (  are perpendicular to the columns of Z.
Confirming that the predicted values The residual sum of squares and cross-products can also be written as determined under the multivariate multiple regression model (3)(4)(5)(6)(7)(8)(9) with full rank (Z) = r + 1 < n ) ( ) ) ( The mean of the ith response variable is

A. Fitting Full Multivariate Regression Model for NAUTH, COOUTH and General Hospital
This means that the fitted multivariate regression models of Systolic and Diastolic blood pressure for NAUTH, COOUTH and general hospital, Onitsha Anambra State are respectively;

B. Test of Significance for the Multivariate Regression Parameters
The test of significance for the multivariate regression parameters of Systolic and Diastolic blood pressure models for NAUTH, COOUTH and general hospital, Onitsha Anambra State are discussed. Outputs 1 & 2, Outputs 3 & 4 and Outputs 5 & 6 are for Systolic and Diastolic blood pressure models for NAUTH, COOUTH and general hospital, Onitsha Anambra State data respectively.   C. Source: R software output R 2 is the coefficient of determination, while RSE is the residual standard error. Looking at the summarized results in Table I, it can be observed that the data collected from the general hospital Onitsha has the highest coefficient of determination (0.9735) with the lowest AIC (1348.944), BIC (1374.462) and residual standard error (2.587) for systolic blood pressure model, which makes the data used in this study the most suitable. Also, it can be observed that the same data collected from the general hospital Onitsha has the highest coefficient of determination (0.5715) with the lowest AIC (1825.917), BIC (1851.435) and residual standard error (6.008) for diastolic blood pressure model, which makes the data used in this study the most suitable. It is clear from the result obtained in this study that the data set collected from general hospital, Onitsha from 2003 to 2017 is most appropriate for the multivariate multiple linear regression models.

VII. CONCLUSION
From the whole study within the data collected for this study, the following conclusions are drawn from both our preliminary results and the results from our model in achieving our objectives; only age has a significant and positive relationship with both systolic and diastolic BP at 5% level of significance for the three hospitals employed in this study, while baseline count of HIV/AIDS patients has significant and negative relationship with systolic BP for NAUTH data. Again, the data collected from general hospital, Onitsha from 2003 to 2017 is suitably adequate for the multivariate multiple regression models.