Tuberculosis (TB) is a re-emerging infectious disease of international health priority. It is particularly worrisome in Africa, which informed the declaration of public health emergency by the World Health Organisation in 2005. Every year, there are about 8 million new cases and between 2 to 3 million deaths resulting from Tuberculosis despite Tuberculosis is a curable disease. In this study, we carry out a statistical analysis on deaths of the patients with tuberculosis (TB) diseases using UITH, Ilorin as case study. The factors considered for TB status (dead or alive) are sex, age and length of stay. Descriptive statistics and logistics regression was used to analyze the data. The results reveal that patients age-group (eighty-one years plus) are mostly affected by deaths while the patients with less than twenty days on treatment recorded the highest numbers of deaths. The distribution of deaths with respect to sex, age group and length of stay is positively skewed. The odds ratio of 1.16 indicates that the probability of deaths for both sexes is not the same because male patients are more likely to die than female patients. The second model with two-predictor (age and length of stay) provided a statistically significant improvement over the first model with three-predictor (age, sex, and length of stay), the Nagelkerke R2 indicated that the second model accounted for 87.6% of the total variance in response and the correct prediction rate was about 92.3%. Therefore, we suggest that the second model describes the data better. Lastly, the Hosmer–Lemeshow test with p = .999 indicates that the numbers of deaths are not significantly different from those predicted by the second model and that the overall model fit is good. Finally, the data shows that the age and length of stay on treatment can predict the death of patients with tuberculosis.
Keywords: Tuberculosis; Length of Stay; Status; Age; Sex.