A Three-Level Meta-Analysis On Household Food Processing, Phytate Reduction and In Vitro Iron and Zinc Bioaccessibility in Cereals, Legumes and Composite Matrices.

ABSTRACT Phytic acid (myo-inositol 1,2,3,4,5,6-hexakisphosphate, InsP6) chelates iron (Fe) and zinc (Zn) in the intestinal lumen, severely constraining mineral bioaccessibility from cereal and legume staples. This quantitative three-level meta-analysis evaluated the pooled impact of traditional household processing techniques on in vitro Fe and Zn bioaccessibility, partitioned variance across hierarchical levels, and assessed potential moderation by mineral species, processing method, assay construct, and percentage phytate degradation. A total of 108 effect sizes (k = 108) nested within 54 pairwise comparisons (m = 54) across 6 independent primary studies (s = 6) were analyzed using log response ratios (LRR) fitted via restricted maximum likelihood in a three-level random-effects framework, incorporating small-sample containment degrees of freedom, CR2 cluster-robust variance estimation, and shared-control sensitivity analyses (rho = 0.0 to 0.7). The overall pooled LRR was 0.2151 (SE = 0.0996; 95% CI: -0.0411 to 0.4712; p = 0.0833), corresponding to a response ratio (RR) of 1.2400 (95% CI: 0.9598 to 1.6020), reflecting an imprecise +24.00% positive tendency (CR2 robust p = 0.106). Multilevel variance partitioning indicated that total heterogeneity was concentrated within studies at the comparison (58.75%) and outcome (33.14%) levels, with between-study differences accounting for 7.04% and sampling error for 1.07%. Iron bioaccessibility responded more strongly (+38.43%; RR = 1.3843) than zinc (+13.50%; RR = 1.1350; contrast p = 0.0732). Among processing techniques, germination (RR = 1.4947) and combined multi-step processing (RR = 1.3827) demonstrated the largest enhancements relative to soaking (RR = 1.0981), whereas thermal cooking alone reduced bioaccessibility (RR = 0.7871). Meta-regression indicated that percentage phytate reduction did not significantly predict bioaccessibility gains (slope: 0.0471 per 10% reduction, p = 0.273; pseudo-R2 = 0.00%), while leave-one-study-out sensitivity models remained positive throughout (RR range: 1.1748 to 1.3409). In conclusion, household processing tends to enhance in vitro Fe and Zn bioaccessibility, with germination and combined schedules showing the greatest potential, but total phytate removal alone is an inadequate proxy for mineral release without direct bioaccessibility quantification. Keywords: Phytic Acid; Meta-Analysis; Three-Level Model; Iron Bioaccessibility; Zinc Bioaccessibility; Food Processing; Germination; In Vitro Digestion; Cereals; Legumes.