Effect of insecticide-treated bed nets on visceral leishmaniasis incidence in Bangladesh. A retrospective cohort analysis.

Background Visceral leishmaniasis (VL) is a parasitic disease, transmitted by the sand fly species Phlebotomus argentipes in the Indian sub-continent. Effective vector control is highly desirable to reduce vector density and human and vector contact in the endemic communities with the aim to curtail disease transmission. We evaluated the effect of long lasting insecticide treated bed nets (LLIN) and bed nets impregnated with slow-release insecticide tablet K-O TAB 1-2-3 (jointly insecticide-treated nets or ITN) on VL incidence in a highly endemic sub-district (upazila) in Bangladesh. Methods Several distributions of LLIN or K-O TAB 1-2-3 for self-impregnation of bed nets at home took place in Fulbaria upazila, Mymensigh district from 2004 to 2008 under three research projects, respectively funded by CDC, Atlanta, USA (2004) and WHO-TDR, Geneva, Switzerland (2006 & 2008). We included all households (n = 8142) in the 20 villages that had benefited in the past from one of these interventions (1295 donated LLIN and 11,918 local bed nets impregnated with K-O TAB 1-2-3) in the "exposed cohort". We recruited a "non-exposed cohort" in villages with contemporaneously similar incidence rates who had not received such vector control interventions (7729 HHs from nine villages). In both cohorts, we visited all families house to house and ascertained any VL cases for the 3 year period before and after the intervention. We evaluated the incidence rate (IR) of VL in both cohorts as primary endpoint, applying the difference-in-differences method. Results The study identified 1011 VL cases (IR 140.47/10,000 per year [py]) before the intervention, of which 534 and 477 cases in the intervention and control areas respectively. The IR was 144.13/10,000 py (534/37050) and 136.59/10,000 py (477/34923) in the intervention and control areas respectively, with no significant difference (p = 0.3901) before the intervention. After the intervention, a total of 555 cases (IR 77.11/10,000 py) were identified of which 178 (IR 48.04/10,000 py) in the intervention and 377 (107.95/10,000 py) in the control area. The intervention area had a significant lower IR than the control area during follow up, rate difference = -59.91, p<0.0001. The IR during follow up was significantly reduced by 96.09/10,000 py in the intervention area (p<0.0001) and 28.63/10,000 py in control area (p<0.0001) compared to baseline. There was a strong and significant overall effect of the ITN intervention, delta = -67.45, p <0.0001. Sex (OR = 1.36, p<0.0001) and age (OR = 0.99, p<0.0001) also had a significant effect on VL incidence. Male had a higher risk of VL than female and one year increase in age decreased the likelihood of VL by about 0.92%. Two third of the VL incidence occurred in the age range 2 to 30 years (median age of VL patients was 17 years). Conclusion VL incidence rate was significantly lower in the ITN intervention cohort compared to control in Bangladesh. Some bias due to more intense screen-and-treat activities or other interventions in the intervention area cannot be ruled out. Nonetheless, given their feasibility and sustainability, ITNs should be considered for integrated vector control during the maintenance phase of the VL elimination programme. Author summary Visceral leishmaniasis (VL) is a deadly parasitic disease, transmitted by the sand fly species Phlebotomus argentipes in the Indian sub-continent. Humans are the only proven reservoir of the parasite, Leishmania donovani. Effective vector control is highly desirable to reduce vector density and human and vector contact in the endemic communities to stop the disease transmission. We evaluated the effect of long lasting insecticide treated bed nets (LLIN) and bed nets impregnated with slow-release insecticide tablet K-O TAB 1-2-3 (jointly insecticide-treated nets or ITN) on VL incidence in a highly endemic sub-district (upazila) in Bangladesh. The nets were either donated or impregnated between 2004 to 2008 under three studies and defined as "exposed cohort" comparing their effect on VL incidence with "non-exposed cohort" (no donation of impregnated nets) for a 3 year period before and after the intervention. The study identified 1011 VL cases (IR 140.47/10,000 per year [py]) before the intervention, of which 534 and 477 cases in the intervention and control areas respectively. There was a strong and significant overall effect of the ITN intervention, delta = -67.45, p <0.0001. The VL incidence rate was significantly lower in the ITN intervention cohort compared to control in Bangladesh, though some bias cannot be totally ruled out.