Risk Factors Predicting Infectious Lactational Mastitis: Decision Tree Approach versus Logistic Regression Analysis

Area of Science:

• Obstetrics and Gynecology
• Infectious Diseases
• Public Health

Background:

• Lactational mastitis is a common condition leading to early breastfeeding cessation.
• Several risk factors contribute to mastitis development, necessitating effective identification and management strategies.

Purpose of the Study:

• To identify main risk factors for lactational mastitis using a decision tree (DT) approach.
• To compare the predictive performance of DT with stepwise logistic regression (LR) for mastitis.

Main Methods:

• Collected data from 368 mastitis cases and 148 controls via questionnaire on various risk factors.
• Utilized decision tree (DT) and logistic regression (LR) models for risk factor analysis.
• Compared model performance using receiver operating characteristic (ROC) curves, sensitivity, specificity, and accuracy.

Main Results:

• Significant risk factors identified by both models include cracked nipples, medication use, infant age, breast pumps, family history, and throat infection.
• DT model identified bottle-feeding and milk supply as relevant for specific subgroups.
• Both models showed similar predictive performance (ROC AUCs: LR 0.870, DT 0.835). LR had higher accuracy and sensitivity, while DT offered better specificity.

Conclusions:

• Decision tree and logistic regression models are valuable, complementary tools for assessing lactational infectious mastitis risk.
• The DT approach effectively identifies high-risk subpopulations, enabling targeted public health interventions and resource allocation for mastitis prevention programs.