Modeling LF Elimination with MDA and Vector Control

Interventions were modeled by using the SIR-selected parameter vectors/models for simulating the impacts of both currently used as well as proposed MDA-based intervention strategies in reducing the observed baseline LF prevalence in each site to below either the global TAS (1% mf prevalence) or site-specific 95% EP thresholds. When simulating these interventions, the observed MDA regimens and coverages followed in each site were used (Supplementary Table 2), while MDA was assumed to target all residents aged 5 years and above. While the drug-induced mf kill rate and the duration of adult worm sterilization were fixed among the models (Supplementary Table 1), the worm kill rate was left as a free parameter to be estimated from the post-intervention data to account for uncertainty in this drug efficacy parameter. For making mf prevalence forecasts beyond the observations made in each site, predictions arising from the impacts of MDA simulated with and without vector control were carried out for 5 years and 20 years after the stoppage of MDA in each site. Three different MDA regimens: (i) annual MDA with ivermectin and albendazole (IVM+ALB) or diethylcarbamazine and ivermectin (IVM+DEC) as applied in each site, (ii) biannual MDA with the above regimens, and (iii) annual triple drug MDA (IDA: combined ivermectin, diethylcarbamazine and albendazole) were modeled with and without vector control in this study to provide a comparison of the effectiveness of these drug regimens for affecting LF elimination. In these simulations, MDAs are stopped after achieving either the 1% mf TAS threshold or the model-predicted 95% EP threshold in each modeled site but simulations of subsequent changes in mf prevalence with or without VC for the next 20 years were continued to evaluate the probability of LF elimination and the risk of recrudescence of the infection respectively over both the shorter 5-year TAS period and over the longer-term 20 years period (using the assessment methods for calculating the occurrence of either of these events described above). VC is modeled in terms of the impact of long-lasting insecticidal nets (LLINs) with 65% coverage following the equation given in ref. ^{9 (link)}. The specific details concerning the different MDA-based scenarios investigated are as listed in Tables 1 and 2.
Note that the estimated 95% EP mf prevalence threshold at ABR was used when carrying out simulations of durations or timelines to break transmission by MDA alone strategies while the corresponding and comparatively higher 95% EP thresholds (obtaining at TBR) for this indicator was employed when modeling the impact of including VC into MDA programs^{9 (link)}. The MDA plus MDA and vector control model formulations, parameters and functions used to carry out these simulations are as described previously and provided in Supplementary Information.