Plant-derived complementary therapies for nausea and vomiting of pregnancy: a systematic review and network meta-analysis of randomized controlled trials

Abstract

Abstract Background Nausea and vomiting of pregnancy (NVP) are common in early gestation and can substantially impair quality of life and daily functioning. Plant-derived therapies have been evaluated for symptom relief, but comparative evidence remains uncertain because trials are small, direct head-to-head comparisons are limited, and outcome instruments differ. Methods PubMed, Embase, Scopus, CENTRAL, WHO ICTRP, Google Scholar, and reference lists were searched from inception to January 2026 for randomized controlled trials (RCTs) of plant-derived therapies for NVP up to 20 weeks of gestation. Pregnancy-Unique Quantification of Emesis and Nausea (PUQE), Rhodes-derived, and 0 to 10 visual analog scale (VAS) outcomes were synthesized in separate frequentist network meta-analyses, using random-effects models for PUQE and VAS and a fixed-effect model for the Rhodes network. Risk of bias was assessed using RoB 2, and confidence in network estimates was assessed using CINeMA. Results Twenty-eight RCTs (2,798 participants) were included, of which 23 contributed to at least one outcome-specific network. The PUQE network (8 trials, n = 655) showed marked global inconsistency (χ²(3) = 117.38, p < 0.001) and substantial heterogeneity (τ² = 4.25); its estimates could not be interpreted reliably, and confidence was very low for all comparisons. In the Rhodes network (8 trials, n = 835), placebo-referenced estimates favored lavender (SMD − 1.82, 95% CI -2.25 to -1.39), orange blossom (SMD − 1.44, 95% CI -1.85 to -1.03), and several other interventions. Confidence was low for lavender and orange blossom, which were each informed by a single trial, and very low for ginger because of local inconsistency. In the VAS network (7 trials, n = 594), ginger (MD -2.06, 95% CI -2.91 to -1.20) and vitamin B6 (MD -1.89, 95% CI -3.04 to -0.74) were associated with lower nausea scores than placebo, with low confidence, and showed little or no clinically important difference from each other. Of the 28 trials, 11 reported at least one adverse event, whereas adverse-event reporting in the remaining trials ranged from explicit reports of no events to incomplete or unavailable safety information. Conclusions The available randomized evidence does not establish a single most effective plant-derived therapy for NVP. Favorable estimates in the Rhodes and VAS networks were supported by low or very low confidence evidence, and marked inconsistency precluded reliable interpretation of the PUQE network. These findings represent comparative efficacy signals rather than a basis for clinical recommendations. Trial registration PROSPERO CRD420251116302.