Log linear modelling of the cases of Diabetes Mellitus and high blood pressure on gender in Nigeria

Abstract

This research examined cases of Diabetes mellitus (DM) and High blood pressure (HBP) on gender (Sex), using descriptive statistics, trend components, 2-Way Log Linear Model (2×2) and 3-Way Log Linear Model (2×2×2). The results revealed that as the year’s goes, the cases of DM and HBP increases, where DM cases are more in female than male and HBP cases are more in male than female. Furthermore, the trend components analysis (i.e. Linear and Quadratic) done for male and female over the years (1985-2019) with respect to the cases of DM and HBP. It was observed that the quadratic trend curve best fits the data sets, these results was confirmed by the values of R-square and accuracy measures. The DM cases R-square values for male and female are 99.88% and 99.92%, while HBP cases R-square values are 99.99% and 99.96%. The quadratic trend accuracy measures are smaller than the linear trend in both male and female cases in term of DM and HBP. The 2-Way Log Linear Model (2×2) odds ratio was 0.914 and P(odds ratio) = 1 – 0.914 = 0.086 indicated that male having diabetes are 8.6% less likely to have HBP than female. The Pearson chi square X² = 119749.712 (large value) and p-value = 0.000 <∝= 0.05, showed that the independence model fits the data in ascertain whether or not HBP is related with DM cases HBP in term of gender. Hence, the null hypothesis was accepted indicating that there is no relationship between the numbers of cases with DM & HBP in term of gender. The fact that males with DM are at high risk of having HBP does not suggest that females will also have. In addition, the 3-Way Log Linear Model (2×2×2) p-value is 0.005 which is significant at 5%, and its R-Square value of 90.68% explains that this model also fits the data sets fairly well. From this results, it is observed that gender show no association in the number of cases with DM and HBP, while there is association in term of the number cases below and above the average DM and HBP cases. The 3-Way Log Linear Model (2×2×2) result is similar to the 2-Way Log Linear Model (2×2) above. It is therefore recommended that a study on forecasting the cases of DM and HBP be carried out in subsequent researches.