Natural Product Extracts for Ischemic Stroke: A methodological evaluation and meta-epidemiological analysis

Sijin Wang¹Minghui Zhang²Yao Lu³Peipei Du⁴Ziwen Xu¹Xu Pang¹Li Li^3*Chi Zhang^1,5*

• ¹Beijing University of Chinese Medicine Affiliated Dongzhimen Hospital, Beijing, China
• ²Beijing Hospital of Integrated Traditional Chinese and Western Medicine, Beijing, China
• ³China Academy of Chinese Medical Sciences Institute of Basic Theory of Traditional Chinese Medicine, Beijing, China
• ⁴China Academy of Chinese Medical Sciences, Beijing, China
• ⁵Institute for Brain Disorders, Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing, China

Background: Despite numerous systematic reviews (SRs) and meta-analyses (MAs) on natural product extracts (NPEs) for ischemic stroke, their methodological limitations and redundancy are unquantified, and the impact of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement remains unclear. This meta-epidemiological study aims to evaluate the methodological quality of these SRs/MAs, examine PRISMA's impact on the methodological quality, and quantify the frequency of redundant publications. Methods: We systematically searched for SRs/MAs of NPEs for ischemic stroke. Methodological quality was assessed with the Assessment of Multiple Systematic Reviews 2 (AMSTAR-2). Logistic regression was used to analyze associated factors. Redundancy was classified using PIC (population, intervention and comparison) frameworks as original, conceptual replication, or excessive replication. Seventy-nine SRs/MAs of NPEs were included, of which 68 focused specifically on Ginkgo biloba L. extracts (GBEs) and Panax notoginseng (Burkill) F.H.Chen extracts (PNEs) and were further analyzed for redundancy. Results: 86.1% (68/79) of included SRs/MAs were rated "critically low," primarily due to unregistered protocols (79.7%, 63/79) and absent lists of excluded studies (94.9%, 75/79). Publications following PRISMA 2020 statement showed improved protocol registration (OR=10.04[2.56-39.33]; p <0.001), risk of bias assessment (OR=22.96[2.86-182.63]; p=0.003), and appropriate methods for statistical combination (OR=3.23[1.27-8.23]; p=0.014). Among 55 non-benchmark SRs/MAs focusing on GBEs and PNEs, 85.5% (47/55) were redundant publications, comprising 34 conceptual and 13 excessive replications. Conclusion: SRs/MAs on NPEs for ischemic stroke exhibit suboptimal methodological quality and high redundancy. Although PRISMA statement improved the methodological quality, prospective protocol registration and transparent search process are imperative to enhance evidence synthesis.