MULTIVARIATE TIME SERIES ANALYSIS OF PREVALENCE OF HIV/AIDS INFECTIONS IN GOMBE STATE, NIGERIA

Abstract

HIV/AIDS is a major health problem in Nigeria, it is among the greatest health challenges that
affect vast majority of people undermining all sectors of economic and social wellbeing of any
nation. The studies fits a multivariate time series model on the cases of HIV/AIDS infection
among three groups, Male, Female (non-pregnant) and pregnant women attending anti natal
clinic in Gombe State, Nigeria. The data were monthly observations collected from government
and private hospitals in Gombe State which span from 2011 – 2015 in Gombe State Agency
for the Control of Aids (SACA) office record. The Vector Autoregressive (VAR) model was
employed on the data, time series model selection criteria were performed to ensure model
parameter adequacy. The result show that a vector Error Correction (VEC) model of lags two
with at least one co integrating vector fits the data, equation of a long run equilibrium
relationship between the variables shows that for every increase in female case will amount to
an increase of at least two cases in male, on model adequacy for forecast, portmanteau test of
serial auto correlation showed no correlation among the residual, Jarque Bera test showed that
the residuals are normally distributed. The data under study shows an erratic (non-stationary)
behavior. The forecast data shows an increase in female cases due to their vulnerability for the
period of 24 month. The percentage proportion of yearly cases showed an increase in female
case with highest percentage of 57.14% in 2015, with unnoticeable cases of infection in
pregnant women attending antenatal clinic until when it became more pronounce in 2013 with
14.5% out of total case recorded.