ABSTRACT Aim Previous studies have explored the association between periodontitis and preterm birth (PTB), with conflicting results primarily due to variability in methodology, statistical analyses and the case definitions of both conditions. This study aimed to synthesise critically the available scientific evidence of observational studies that evaluate the risk of PTB in pregnant women with or without periodontitis. Material and Methods MEDLINE‐PubMed and Cochrane databases were searched from their inception through June 2025 to identify eligible studies evaluating the incidence of PTB in women with periodontitis compared to those without. Inclusion required that periodontitis be defined through full‐mouth clinical examination, assessing probing pocket depth in combination with clinical attachment loss and/or radiographic alveolar bone loss. The risk of bias was assessed. Descriptive analysis, and when feasible, meta‐analysis (MA) and trial sequential analysis (TSA) were performed. Sub‐analyses were conducted based on the risk of bias analysis, study design, geographical area, periodontitis case definition, smoking, health or country social economic status, history of urinary tract infections and the number of evaluated teeth. The total body of evidence was graded. Results A total of 723 unique papers were identified, and 11 eligible studies were included. The descriptive analysis showed that seven of the eleven studies present a significant association between PTB and periodontitis. Confounding variables on personal, medical and environmental aspects lowered the risk. The crude overall MA based on 11 studies resulted in a small effect, odds ratio (OR: 2.38 95% CI: 1.78; 3.18, p < 0.00001). The TSA showed that the required number of events was reached, and the type I error is ruled out. Conclusion There is moderate certainty that pregnant women with periodontitis compared to pregnant women without periodontitis have a small risk of PTB. Trial Registration PROSPERO number: CRD42022327501
Sokos et al. (2025) studied this question.