Effect of indoor residual spraying on sandfly abundance and incidence of visceral leishmaniasis in India, 2016-22: an interrupted time-series analysis and modelling study

Abstract

Background: Efforts to eliminate visceral leishmaniasis in India mainly consist of early detection and treatment of cases and indoor residual spraying with insecticides to kill the phlebotomine sandfly Phlebotomus argentipes that transmits the causative Leishmania protozoa. In this modelling study, we aimed to estimate the effect of indoor residual spraying (IRS) on vector abundance and transmission of visceral leishmaniasis in India.

Methods: In this time-series analysis and modelling study, we assessed the effect of IRS on vector abundance by using indoor vector-abundance data (from 2016 to 2022) and IRS quality-assurance data (from 2017-20) from 50 villages in eight endemic blocks in India where IRS was implemented programmatically. To assess a potential dose-response relation between insecticide concentrations and changes in sandfly abundance, we examined the correlation between site-level insecticide concentrations and the site-level data for monthly sandfly abundances. We used mathematical modelling to link vector data to visceral leishmaniasis case numbers from the national Kala-Azar Management Information System registry (2013-21), and to predict the effect of IRS on numbers of averted cases and deaths.

Findings: IRS was estimated to reduce indoor sandfly abundance by 27% (95% CI 20-34). Concentrations of insecticides on walls were significantly-but weakly-associated with the degree of reduction in vector abundance, with a reduction of -0·0023 (95% CI -0·0040 to -0·0007) sandflies per mg/m² insecticide (p=0·0057). Reported case numbers of visceral leishmaniasis were well explained by trends in vector abundance. Village-wide IRS in response to a newly detected case of visceral leishmaniasis was predicted to reduce disease incidence by 6-40% depending on the presumed reduction in vector abundance modelled.

Interpretation: Indoor residual spraying has substantially reduced sandfly abundance in India, which has contributed to reductions in visceral leishmaniasis and related deaths. To prevent the re-emergence of visceral leishmaniasis as a public health problem, surveillance of transmission and sandfly abundance is warranted.

Funding: Bill & Melinda Gates Foundation.

Translation: For the Hindi translation of the abstract see Supplementary Materials section.

Figure 1

Location of the eight districts in which blocks where entomological surveillance occurred and the quality of indoor residual spraying was monitored All the districts depicted in colour are in Bihar state, except for Darjeeling (which is in West Bengal) and Godda (Jharkhand).

Figure 2

Sandfly abundance in 11 villages in Bihar where IRS was started or stopped during the study Monthly sandfly counts represent the mean counts across 18 catchment sites in each village. Dashed vertical grey lines indicate the times when IRS quality-assurance data were collected. Note that the y-axis is plotted on a logarithmic scale (plus 0·5 to enable plotting of 0 values). IRS=indoor residual spraying.

Figure 3

Model-predicted vs recorded trends in annual block-level incidence of visceral leishmaniasis In each graph, the datapoints represent actual case numbers (as recorded in the Kala-Azar Management Information System database). Filled datapoints represent data that were used to calibrate our model in terms of absolute vector abundance (conditional on trends in relative vector abundance), whereas unfilled datapoints represent those that were not used to calibrate the model. Lines indicate model predictions based on observed trends in vector abundance. The goodness of fit (in terms of log-likelihood) was –157·85 for the training data and –166·50 for the data not used to calibrate the model.

Figure 4

Model-predicted number of new cases of visceral leishmaniasis (A) and visceral leishmaniasis-related deaths (B) during 3 years of village-wide IRS This model related to a previously visceral leishmaniasis-naive village of 500 people, where the disease was assumed to be introduced by a single case of post-kala-azar dermal leishmaniasis. The number of new cases included both detected and undetected cases. IRS was assumed to be implemented reactively in response to the occurrence of the first newly detected case. The reduction in vector abundance due to IRS was assumed to start in either April or July. Estimates represent the average of 10 000 repeated stochastic simulations. The model was run assuming sandfly abundance followed the same relative annual and seasonal patterns as recorded in the data (averaged across the eight blocks included in the study; appendix 2 p 10). Values for absolute sandfly abundance were allowed to vary across repeated simulations (assuming a log-normal distribution), such that the model reproduced the distribution of annual village-level visceral leishmaniasis incidence recorded in the available data for the eight blocks. IRS=indoor residual spraying.