Home / Current Issue / Paper 1712685
Model Comparison of Logistic, Poisson, And Negative Binomial Regression Approaches for Determining Predictors of Mother-To-Child Transmission of Syphilis in Ebonyi North Senatorial District
Subject area: Science,Engineering and Technology · Area of research: Medical Statistics
DOI: https://doi.org/10.64388/IREV9I6-1712685
Abstract
This study examined factors influencing mother-to-child transmission (MTCT) of syphilis in Ebonyi North Senatorial District and compared the predictive performance of logistic, Poisson, and negative binomial regression models. Data from four healthcare facilities (2010?2024) were analyzed. Key service indicators, including drug availability, screening, and counseling uptake, were assessed. Model comparison results showed that the logistic regression model provided the strongest predictive power, with excellent fit indices (McFadden R? = 0.92; AUC = 0.998). Even after variable reduction using PCA, the logistic model maintained high performance (McFadden R? = 0.63; AUC = 0.964). In contrast, the Poisson and negative binomial models identified similar significant predictors?age and syphilis knowledge?but the negative binomial model outperformed Poisson due to better handling of overdispersion. Overall, logistic regression demonstrated superior accuracy and discriminatory ability compared to the count-based models. The findings highlight logistic regression as the most reliable model for identifying MTCT risk factors while emphasizing the need to strengthen screening, counseling, and treatment services to reduce congenital syphilis.
Keywords
Syphilis, Transmission, Influencing, Logistic, Poison and Negative Binomial Regression Models.
How to cite this paper
@article{1712685,
author = {Nwuzor Ozoemena, Okoro Chiemeka Nwankwor, S. C. Nwasuka},
title = {Model Comparison of Logistic, Poisson, And Negative Binomial Regression Approaches for Determining Predictors of Mother-To-Child Transmission of Syphilis in Ebonyi North Senatorial District},
journal = {Iconic Research And Engineering Journals},
year = {2025},
volume = {9},
number = {6},
pages = {934-940},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1712685.pdf},
abstract = {This study examined factors influencing mother-to-child transmission (MTCT) of syphilis in Ebonyi North Senatorial District and compared the predictive performance of logistic, Poisson, and negative binomial regression models. Data from four healthcare facilities (2010?2024) were analyzed. Key service indicators, including drug availability, screening, and counseling uptake, were assessed. Model comparison results showed that the logistic regression model provided the strongest predictive power, with excellent fit indices (McFadden R? = 0.92; AUC = 0.998). Even after variable reduction using PCA, the logistic model maintained high performance (McFadden R? = 0.63; AUC = 0.964). In contrast, the Poisson and negative binomial models identified similar significant predictors?age and syphilis knowledge?but the negative binomial model outperformed Poisson due to better handling of overdispersion. Overall, logistic regression demonstrated superior accuracy and discriminatory ability compared to the count-based models. The findings highlight logistic regression as the most reliable model for identifying MTCT risk factors while emphasizing the need to strengthen screening, counseling, and treatment services to reduce congenital syphilis.},
keywords = {Syphilis, Transmission, Influencing, Logistic, Poison and Negative Binomial Regression Models.},
month = {December},
doi = {https://doi.org/10.64388/IREV9I6-1712685}
}