MODELLING THE DYNAMIC RELATIONSHIP BETWEEN HYPERTENSION AND DIABETES CASES IN NORTHERN REGION

Abstract

Hypertension and diabetes are among the well-known and potent risk factors for cardiovascular disease development globally and they rarely exist in isolation. This study therefore modelsthe dynamic relationship between hypertension and diabetes cases in the Northern region of Ghana. Monthly data on hypertension and diabetes cases from the Tamale Teaching Hospital database was modeled using Vector Autoregressive model. Before fitting the model, the nature of the trend characterising the hypertension and diabetes cases were investigated. The results revealed that both hypertension and diabetes cases exhibit log-linear trend. Appropriate order of the Vector Autoregressive model was determined using the AlC, BIC, FPE and HQIC. The BIC selected lag 1 and HQIC selected lag 2. Both VAR (1) and VAR (2) models were fitted to the data and the LRT used to select the best. The results from the LRT revealed that the VAR (2) model was best for modeling the dynamic relationship between the hypertension and diabetes cases. The diagnostic checks of the VAR(2) model using both the univariate and multivariate Ljung-Boxtest and the ARCH-LM test revealed that the model was free from serial correlation and conditional heteroscedasticity respectively. The JB test revealed that the normality assumption of the VAR (2) model was satisfied. An infemce with the model using the Grange causality test revealed that there was a unidirectional relationship between hypertension and diabetes whiles the instantaneous causality test revealed that there was a bilateral relationship between the cases. The FEVD revealed that both hypertension and diabetes cases explain some amount of forecast uncertainty in each other.