How reliable is the use of ovarian dilatations for mosquito age-grading? A blinded multi-rater validation of the Polovodova method in Anopheles mosquitoes.

Accurate estimation of mosquito age is critical for malaria surveillance, as only older female Anopheles mosquitoes can transmit the parasite. Traditional age-grading methods include the Polovodova technique in which dilatations on the ovarian pedicel to tally completed gonotrophic cycles (GC) are counted. Despite their widespread use, these techniques remain poorly validated under realistic operational conditions, with limited evidence on their precision and accuracy across different parity rates, GC, and species. In this study, the accuracy and precision of the Polovodova method were evaluated across multiple age-graders and species under both laboratory and field conditions. A blinded validation study was conducted using laboratory-reared Anopheles arabiensis and Anopheles funestus (N > 3,600) spanning 4 GC and field-collected mosquitoes from Ulanga district, Tanzania (N = 600). Three blinded researchers performed dissections and readings, while a fourth unblinded researcher served as the reference reader (R0) to obtain ground truth (for parity) or assumed truth (for GC) for laboratory mosquitoes. Accuracy and precision for parity and GC classification were assessed using pairwise inter-rater comparisons, percent agreement, Fleiss' kappa, and Cohen's kappa (κ) statistics. For field mosquitoes, which had no age-referenced group, only precision was assessed. The Polovodova method accurately classified parity in both species, correctly identifying 98% to 100% of nulliparous and 94% to 96% of parous mosquitoes compared with the reference dataset. Inter-rater reliability for parity was almost perfect in An. arabiensis and An. funestus (κ = 0.81 to 0.83), with 82% to 96% pairwise agreement. GC classification was highly accurate for early stages (GC0 to GC2: 85% to 100%) but declined at later stages GC4 (34% to 46%), reflecting systematic underestimation of mosquito age. Overall GC reliability was moderate (κ = 0.56 to 0.59). Field classifications showed almost perfect agreement (κ = 0.87 to 0.88), supporting operational applicability. The Polovodova method provides robust estimates of parity and early gonotrophic-cycle status in Anopheles mosquitoes, but late-cycle classifications are more prone to reader disagreement and systematic underestimation. Operational use should therefore rely on trained multiple-reader workflows, including adjudication of discordant or late-cycle specimens, rather than single-reader classification. These findings provide empirical benchmarks for quality-assured Polovodova age-grading in malaria vector surveillance and intervention evaluation.