We describe a comprehensive active surveillance of AFM events during repeated MDAs in an area of Papua New Guinea where LF is highly endemic. The incidence of AFM in this population decreased to 57% of the pre-treatment level after 2 annual MDAs in areas of moderate transmission and by 61% in areas of high transmission. This rapid decrease in AFM incidence was sustained over the entire 5-year surveillance period, thereby highlighting the potential of MDA to alleviate this feature of LF morbidity. Of concern, however, is the doubling of AFM events during the 4-week period following the first MDA ( and ). This transient increase was not, however, observed following subsequent MDAs. In fact, the multivariable models demonstrated that microfilaremia predicted AFM incidence only during the year following the first MDA, but not during the pre-treatment or any other post-MDA year. Drug-induced adverse events have been previously recorded following treatment with anti-filarial drugs, particularly among individuals with high levels of microfilaremia 
. The significantly higher incidence of AFM events among residents of communities with high versus moderate transmission intensities () is consistent with these observations. The absence of periodic increases in AFM with subsequent treatments indicates that the individual microfilarial densities may have decreased sufficiently following the first annual MDA to eliminate drug-induced AFM following subsequent MDAs. More generally, these observations suggest that treatment naïve populations are likely to be at increased risk of adverse events during the early phase of MDA programs.
With regard to the study objectives, we have demonstrated a significant decrease in AFM following mass administration of anti-filarial drugs and the resulting decrease in indices of transmission by the local mosquito vectors. Our results showed that the incidence density of AFM remained stable following the first two MDAs, though the rates remained twice that reported from other LF endemic regions such as Tanzania 
, India 
and Ghana 
. AFM events in the latter phases of our study (post MDA years 3 and 4), when village-specific microfilaria levels and mosquito transmission of infective larvae were reduced respectively by 89–98% and 84–97% relative to the pre-MDA values 
, may be attributed to active LF infections that were not cured by earlier MDAs, pre-existing chronic LF pathology, and/or continued exposure to infective larvae 
. Alternatively, the high rate of persisting AFM compared with other endemic regions (where post-MDA AFM was not examined) may be due to increased sensitivity of our follow-up protocols or low specificity of the case definition. With respect to the latter consideration, we did not attempt to determine whether self-reported fever accompanied by painful swelling of an extremity was due to an injury that promoted secondary bacterial infection of the arm or leg (shoes or other footwear were not used by most study participants), and malaria is highly endemic in this region of Papua New Guinea. We believe that malaria is unlikely to be a cause of incorrect categorization given that malaria-associated fevers in this region occur primarily in children younger than 5 years 
, whereas AFM occurs in adults. Regardless, the case definition used here is comparable to that used in other settings where these and other co-morbidities are present 
Our results strongly support the current global strategy to eliminate LF through the use of repeated annual MDAs 
. At the same time, the observations highlight important questions pertaining to the pathogenesis of AFM. Currently, acute morbidity events that involve painful swelling of the extremities, especially the legs, are classified according to whether they are due to secondary bacterial infection versus worm-associated etiologies 
. Our study design did not allow us to distinguish between bacterial and filarial etiologies of AFM. To do so would require clinical monitoring in the home, bacterial culture of affected tissues, provision of antibiotics, and an estimate of adult worm death (e.g. scrotal ultrasound) 
. Biopsy of inflamed tissue to assess adult worm viability or the presence of W. bancrofti
third or fourth-stage larvae is not justifiable clinically or ethically. Unlike studies of secondary bacterial infections of the skin, which to date have been limited to self-referred patients with pre-existing chronic lymphedema of an extremity, our study followed an entire population during repeated annual MDAs and noted that, though AFM rates were significantly greater in participants with pre-existing chronic LF pathology documented by physical examination, AFM rates also remained high for individuals with no pre-existing chronic LF pathology (incidence 0.28 per person-year of observation before MDA and 0.18 after four annual MDAs). Thus, while secondary bacterial infections associated with chronic lymphatic pathology of the leg may have played a role in the pathogenesis of some cases of AFM observed here, inflammatory reactions to adult worm or filarial larvae were also likely to have contributed. Regardless, the 48% decrease in AFM events following five rounds of MDA occurred in the absence of foot hygiene or antibiotic interventions which have been shown in other studies to independently decrease AFM events by 30 to 60% 
. This study supports current efforts to integrate hygiene and MDA for LF elimination. Finally, cases occurring in individuals without detectable LF infection presumably represent individuals who experienced morbidity events that were truly unrelated to W. bancrofti
infection and/or have pre-patent LF infections that are below the limits of detection with the assays used here. It is also possible that some immune individuals have filarial antigen in their tissues from previous, cured infection without detectable antigenemia at the time of sampling.
The data reported here suggest various ways that the current strategy to eliminate LF may be modified from an operational perspective. Given the dramatic rise in AFM events that occurred immediately following the first MDA, community education regarding this possibility of experiencing such events may be important to sustain compliance over repeated annual MDAs, especially in endemic regions where heavily infected individuals and populations exist. Increased follow-up and case management of AFM during the early phase of MDA implementation may also be warranted. Finally, if indeed exposure to infective third stage or subsequent larval stages that develop in the lymphatics of recently re-infected or newly infected humans are involved in the pathogenesis of AFM, intensified efforts to reduce mosquito transmission such as by stepped-up vector control may be warranted.