On the log-linear analysis of medical conditions and blood groups of patients

Abstract

A log-linear model is used to model count data in contingency tables form, especially when the major goal is to understand the association between categorical variables, without specifying a clear dependent variable. This research work was carried out in order to model the associations and interactions among categorical variables which include age groups, blood groups and medical conditions in a contingency table using loglinear model. Total of 5500 samples were considered across six (6) medical conditions (Asthma, Diabetes, Cancer, Hypertension and Obesity), twelve age groups (18-23, 24-29, 30-35,. . ., 78-83) and eight (8) blood groups (A+, A-. AB+, AB- B+, B-, O+ and O-). The effect of age groups on the response variable were not the same (p_value of 0.0001) and that led to the rejection of null hypothesis of equal effects across all the age groups but that of medical conditions and interactions were the same (p_value are 0.9007 & 0.7267 respectively). The maximum likelihood analysis of the parameter revealed convergence at the fourth (4^th) iterations with the value of 46776.654 for -2loglikelihood and convergence criteria of near zero (4.1836E-9). The result of maximum likelihood estimates for age groups revealed five (5) age groups were significant which include 36-41, 42-47, 54-59, 72-77 and 78-83 with their respective estimates 0.1865, 0.1047, 0.0986, 0.1464 and 0.1657 whereas the rest of the age groups were not significant. No interaction between age groups and medical conditions as evidence in likelihood ratio p_value of 0.7267. Cancer is the most prevalence medical condition within the scope of this study with age group 36-41 taking the lead with 97 counts which account for 10.26 percent. All the medical conditions have its lowest cases at age 84-89; the common blood groups in this research work are B+ and A+ (719) closely followed is O+ (712). Among those living with O- blood group, Diabetic patients are the least (87) with 13.57 percentage.