Risk factor analysis and development of a nomogram prediction model for Plasma Cell Mastitis

Abstract

Background and objective The risk factors for plasma cell mastitis (PCM) remain unclear. Understanding and mitigating these factors to prevent PCM before its onset has become a significant concern. This study identifies PCM risk factors, develops a predictive nomogram, and offers insights for targeted prevention and awareness in high-risk groups. Methods We retrospectively analyzed the clinical data of 82 patients diagnosed with PCM at Hangzhou Women’s Hospital’s Breast Surgery Department from 01/01/2019 to 01/01/2022. A control group was randomly selected, consisting of 82 healthy women aged between 20–60 years who had undergone routine health check-ups during the same period. Using SPSS 26.0 software for univariate analysis, significant risk factors for PCM were identified. R software was used for multivariate logistic regression analysis, and a nomogram prediction model for the risk of developing PCM was established. Results The average age of patients in the study group was 32.37 ± 6.64 years, the control group was 29.54 ± 5.33 years, with no statistically significant difference between the groups (P = 0.176). The onset time after childbirth or miscarriage was 3.37 ± 1.91 years. Univariate analysis revealed significant differences in BMI, nipple retraction, number of pregnancies, recent trauma history, and hyperlipidemia (P < 0.05). Multivariate logistic regression analysis identified nipple retraction (OR=20.128, P = 0.000, 95% CI: 5.952–68.072), number of pregnancies (OR=0.343, P = 0.000, 95% CI: 0.189–0.624), and recent trauma history (within two weeks) (OR=11.154, P = 0.000, 95% CI: 2.936–42.382) as independent risk factors for PCM. Conclusion Nipple retraction, recent trauma history, and the number of pregnancies were identified as independent risk factors for PCM. Targeted education for high-risk groups, particularly women within 3 years postpartum/post-abortion, improves disease prevention. The nomogram model had a C-index of 0.809, indicating strong discriminatory power and high prediction accuracy.