Show simple item record

dc.contributor.authorBirungi, Nalugya Jesca
dc.date.accessioned2023-11-24T09:56:59Z
dc.date.available2023-11-24T09:56:59Z
dc.date.issued2023-11
dc.identifier.citationBirungi,N.J.(2023).Comparison of the performance of ordinary logistic, log-binomial, and modified Poisson regression models in identifying determinants of teenage pregnancies in Uganda.(Unpublished master's dissertation). Makerere University, Kampala, Uganda.en_US
dc.identifier.urihttp://hdl.handle.net/10570/12572
dc.descriptionThis dissertation was submitted in partial fulfillment of the requirements for the award of Master of Biostatistics of Makerere University School of Public Health.en_US
dc.description.abstractIntroduction: Teenage pregnancy is a major public health concern in low-income countries. Uganda's rate is estimated at 25%, one of the highest in sub-Saharan Africa. Commonly associated factors include poverty, low education, media exposure, rape, defilement, and early sexual debut. This study compares logistic, log-binomial, and modified Poisson regression models to identify factors predicting teenage pregnancy in Uganda. The goal is to determine the most effective model for addressing this issue, which negatively impacts teens, infants, families, and communities. Methods: The study used secondary data from the 2016 Uganda Demographic and Health Survey. The bivariable and multivariable analysis employed STATA version 14 and R software. Multi-level models identified factors associated with teenage pregnancy, determining the best-fitting model using Akaike information criteria (AIC). Analysis used the model with the lowest AIC. Clustering was assessed via Intra-class Correlation Coefficient (ICC). Results: There was a minimal regional clustering effect of 2.6% observed in the data. The logistic model was the better fit to determine the factors associated with teenage pregnancy in Uganda. Sexual experience (p=0.001), sex of the household head (p=0.001), age (p=0.001), contraceptive use (p=0.001), education level (p=0.015), the total number of children ever born (p=0.008), and knowledge of the ovulation cycle (p=0.02) were statistically associated with teenage pregnancy. Conclusion and Recommendation: The logistic regression model demonstrated the most favorable fit and uncovered notable significance with age, sexual experience, household head's sex, education level, knowledge of the ovulation cycle, number of children ever born and contraceptive use. While a slight regional clustering effect was detected, these results emphasize the significance of focused interventions to enhance reproductive health awareness to combat teenage pregnancies in Uganda. These efforts would be most effective through collaborative initiatives involving the Ministry of Gender, Labor, and Social Development, the Ministry of Education and Sports, the Uganda Police, and the Ministry of Health.en_US
dc.language.isoenen_US
dc.publisherMakerere Universityen_US
dc.subjectteenage pregnancyen_US
dc.subjectcomparison of statistical modelsen_US
dc.subjectstatisticsen_US
dc.subjectMasters of biostatisticsen_US
dc.subjectSchool of Public Healthen_US
dc.titleComparison of the performance of ordinary logistic, log-binomial, and modified Poisson regression models in identifying determinants of teenage pregnancies in Uganda..en_US
dc.typeThesisen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record