**Background:** American Samoa underwent seven rounds of mass drug administration (MDA) for lymphatic filariasis (LF) from 2000–2006, passed TAS-1 (2010) and TAS-2 (2015), but failed TAS-3 in 2016, indicating ongoing transmission. Three further rounds of triple-drug MDA (3D-MDA) were implemented in 2018, 2019, and 2021, yet recent surveys show transmission persists. LF in American Samoa is transmitted by highly efficient Aedes vectors, and local hotspots may sustain transmission even when territory-wide antigen prevalence falls below the WHO threshold of 1%.
**Methods:** The authors used GEOFIL, a spatially explicit agent-based model incorporating household, workplace, and school locations, daily commuting networks, and spatially heterogeneous transmission risk. The model was fitted using approximate Bayesian computation to data from 2014 and 2016 community surveys, including intra-cluster correlation of microfilariae (mf) positivity. Simulated interventions starting in 2023 included: (1) territory-wide 3D-MDA at 65%, 73%, or 85% coverage for 1–5 annual rounds; (2) school-based testing of TAS-aged children (6–7 years), elementary (6–13 years), or all school-aged children (6–17 years) with treatment of antigen-positive children and their households; (3) workplace-based testing at workplaces with 5+ or 50+ employees, treating either the worker alone or the worker and household; (4) household-based strategies with 1, 3, or 5 teams offering antigen testing to 25% or 50% of households per village, with treatment offered to all households within 100m, 500m, or 1km of a positive case. All strategies assumed 73% acceptance of testing/treatment. Effectiveness was measured as 'control probability'—the proportion of 100 simulations per scenario in which mf prevalence decreased between 2030 and 2035.
**Key Results:** Without further intervention, antigen prevalence is predicted to rebound to 1.24% [95% PI 1.11–1.37] by 2035. For 3D-MDA, achieving ≥90% control probability required ≥4 rounds at 65% coverage (94% [95% CrI 88.7–97.9]), ≥3 rounds at 73% coverage (93% [87.4–97.2]), or ≥2 rounds at 85% coverage (90% [83.6–95.2]). School- and workplace-based strategies had negligible control probabilities (<5% for most). The most effective household-based strategies achieved ≥90% control probability: three teams with 50% household test aim (HTA) and 1km radius (93% [87.4–97.2]), five teams with 25% HTA and 500m radius (93% [87.4–97.2]), and five teams with 50% HTA and either 100m (98% [94.5–99.8]) or 500m radius (99% [96.2–100]). The five-team strategy with 50% HTA and 100m radius required only 12,491 treatments (vs 136,800 for four rounds of 73% MDA) but 139,804 tests, for a combined 152,295 tests+treatments vs 136,800 treatments for MDA. The 1% antigen threshold was a poor predictor: all 18 MDA scenarios and all 16 household-based strategies had ≥99% probability of reducing antigen below 1%, but only half of MDA and 5 of 16 household strategies had ≥90% control probability.
**Clinical Implications:** American Samoa requires further intervention to prevent LF resurgence. Triple-drug MDA remains a viable option if high coverage can be achieved and maintained. Household-based targeted strategies use far fewer treatments but require intensive testing, likely making them more costly than MDA given donated drugs. The finding that the 1% antigen prevalence threshold poorly predicts sustained elimination challenges current WHO guidelines and suggests that elimination targets may need to be context-specific, accounting for local vector ecology and spatial heterogeneity. Combining MDA with targeted post-MDA surveillance, or incorporating molecular xenomonitoring, may offer more efficient pathways to elimination.