Main conclusion
Researchers used a parametric g-formula in an IVF cohort to estimate outcomes under hypothetical dietary changes. The simulation suggested fewer miscarriages with a lower inflammatory potential of the diet. No dietary intervention was actually assigned, so the result must not be presented as a treatment effect.
What the study did
The prospective cohort included 1,414 women from one specialist center in Shandong, China. Diet was assessed with a simplified 25-item questionnaire. Baseline miscarriage risk was 17.35%. Under simulated moderate, high, and optimal adherence to a lower E-DII diet, estimated risks were 11.41% (95% CI 9.61–13.77), 8.66% (7.73–10.23), and 7.11% (7.02–7.39). Favorable patterns were also reported for live birth and gestational diabetes.
Key limitations
The g-formula attempts to account for measured confounding, but estimates depend on complete data, correct model specification, and the absence of unmeasured confounding. Diet was self-reported, sampling was by convenience, and the cohort came from one center. Most importantly, no real dietary program was tested.
What can be said to patients
Healthy preconception nutrition has many reasonable goals, but patients should not be promised the miscarriage reduction estimated by a simulation. A randomized trial of a specific, feasible intervention with live-birth and safety outcomes is needed.
Source
International Journal of Gynecology and Obstetrics, 2026. DOI: https://doi.org/10.1002/ijgo.71258
PubMed: https://pubmed.ncbi.nlm.nih.gov/42690291/
Educational material; it does not replace individual medical advice.