Unmeasured Confounding in Summary Estimates of Maternal Periodontitis and Adverse Birth Outcomes: Meta-Analyses of Observational Studies

Abstract Importance Maternal periodontitis has been consistently linked to adverse birth outcomes. However, a causal relationship is not established due to potential unmeasured confounding factors in observational studies and inconclusive results from randomized controlled trials. Objective To assess the impact of unmeasured confounding factors in meta-analyses examining associations between maternal periodontal disease and preterm birth or low birth weight. Data Sources PubMed/Medline, the Cochrane database for systematic reviews, Embase, Google Scholar, and a manual search of the reported references between 2002 and 2023. Study Selection Systematic reviews with meta-analyses of observational studies were included if either preterm birth or low birth weight were the outcome and were published in English between 2002 and 2023. Reviews, editorials, letters, conferences, summaries, abstracts, books, opinions, umbrella reviews, experimental animal and in-vitro studies, and duplicates were excluded. Data Extraction and Synthesis Search results were entered into Rayyan software after removing duplicates. Two researchers independently screened titles and abstracts, with a third reviewer resolving discrepancies. This study followed the Meta-analyses Of Observational Studies in Epidemiology (MOOSE) guidelines. Main Outcomes and Measures Summary relative risks in meta-analyses evaluating maternal periodontal disease and preterm birth (gestation) and low birth weight (<2500g). Results There were 62 original studies from 9 meta-analyses meeting the inclusion criteria. In a random-effects meta-analysis of 19 cohort studies, more than 80% of the studies would have RR>2.00 even if an unmeasured confounder were three times more prevalent in the exposed group and raised the risk of preterm birth 3-fold. Similarly, in a meta-analysis of 10 cohort studies, if the association between an unmeasured confounder and the exposure and low birth weight were RR >3, 80% of the studies in that meta-analysis would have RR>2.00. Publication bias in these meta-analyses may have inflated the summary measures between maternal periodontal disease and preterm birth or low birth weight. Conclusions and Relevance Associations between maternal periodontal disease and adverse birth outcomes in observational studies appear robust to unmeasured confounding factors, but publication bias may tend to inflate them. These findings support the hypothesis that maternal periodontal disease is causally related to adverse birth outcomes.