Toolkit for Monitoring and Evaluation of Indoor Residual Spraying for Visceral Leishmaniasis Control in the Indian Subcontinent: Application and Results

Background. We field tested and validated a newly developed monitoring and evaluation (M&E) toolkit for indoor residual spraying to be used by the supervisors at different levels of the national kala-azar elimination programs in Bangladesh, India and Nepal. Methods. Methods included document analysis, in-depth interviews, direct observation of spraying squads, and entomological-chemical assessments (bioassay, susceptibility test, chemical analysis of insecticide residues on sprayed surfaces, vector density measurements at baseline, and three follow-up surveys). Results. We found that the documentation at district offices was fairly complete; important shortcomings included insufficient training of spraying squads and supervisors, deficient spray equipment, poor spraying performance, lack of protective clothing, limited coverage of houses resulting in low bioavailability of the insecticide on sprayed surfaces, and reduced vector susceptibility to DDT in India, which limited the impact on vector densities. Conclusion. The M&E toolkit is a useful instrument for detecting constraints in IRS operations and to trigger timely response.


Introduction
Visceral leishmaniasis (VL)/kala-azar (KA) is a public health problem in the Indian subcontinent, particularly in Bangladesh, India and Nepal [1]. The elimination of the disease is possible due to its unique epidemiological features including easy diagnosis at the field level with the rK39 rapid diagnostic test, availability of highly effective drugs, and effective vector control methods if applied correctly [2,3]. In a previous study, it was highlighted that there has been no VL vector control programme in Bangladesh for a long time [4], and that indoor residual spraying (IRS) with DDT in India and lambdacyhalothrin in Nepal was effective in reducing sandfly vectors under controlled conditions but showed many shortcomings within national programmes [5,6]. IRS is expensive and operationally a challenge; if not done properly it would be a wastage of human and financial resources. Thus systematic monitoring and evaluation of IRS activities is of paramount importance to have an impact on the vector population. A monitoring and evaluation (M&E) toolkit is highly desirable for detecting operational issues at an early stage so that they can be addressed adequately. Therefore such a toolkit has been developed in an interagency effort including the Special Programme for Tropical Disease Research and Training (TDR) at the World Health Organization (WHO) in Geneva, the German Cooperation (BMZ/GIZ), scientists from different research institutes as well as programme managers from Bangladesh, India and Nepal. The present study investigates the process and results of the application of the M&E toolkit at the supervisory level by spray supervisors and District Malaria Officers (DMOs)/District Public Health Officers (DPHOs)/Upazila Health and Family Planning Officers (UH&FPOs) with regard to identifying limiting factors of the routine national IRS activities in India and Nepal, and in two pilot areas of Bangladesh.

Monitoring and Evaluation
Toolkit. The main purpose of the M&E toolkit is to support IRS programmes through systematic monitoring and evaluation of processes and outcomes, allowing timely detection of gaps and constraints and ensuring that adequate responses are taken. Details about the toolkit can be found in the WHO website (http://apps.who.int/tdr/svc/publications/tdrresearch-publications/irs toolkit) [7]. It is independent of the spray schedule, of the insecticide used (in our case, DDT in India, deltamethrin in Bangladesh; lambdacyhalotrin in Nepal) and of the spray pumps (compressor pumps in Nepal and Bangladesh; stirrup pumps in India). Briefly, the toolkit contains sets of indicators classified as input, process, output, outcome, and impact indicators ( Figure 1). Input indicators include national guidelines and action plans for IRS, engagement and training of spray teams, availability of insecticides, spray pumps, spare parts, storage facilities, and protective materials. Process indicators include all information related to the performance of spraying and quantity of insecticide used. Output indicators include the coverage achieved by IRS through document analysis and household interviews, assessment of the bioavailability of the insecticide on sprayed surfaces through the WHO cone bioassay test, determination of the chemical concentration achieved on sprayed surfaces (filter paper method), and the level of acceptability of IRS by the community. The outcome indicator presents the effect of IRS on the sandfly density.
Finally, health impact is measured as the reduction of disease burden in the community. The toolkit is complemented by a data entry programme which facilitates the data entry at district/subdistrict levels.  Figure 2). The study sites were purposively selected based on high VL endemicity. In India, two primary health care centres (PHCs) were selected from each of the above-mentioned districts, in total six PHCs, chosen on the basis of previous VL case load. Twelve villages (two from each PHC, one close and one far away from the PHC) were selected randomly for assessing the IRS operations. Additionally, four villages were randomly selected as control area for the assessment of vector densities. In Nepal, in the six study districts, 24 village development committees (VDCs) were selected based on high VL case load according to passive surveillance data. In each district, two VDCs were near and two were distant from the district headquarter from where the IRS program was operated. The 24 VDCs were then divided in those where IRS was programmed and those where it was not. From each VDC one study village was selected randomly so that in the end 12 IRS intervention villages and 12 control villages were assessed. In Bangladesh, 3 and 4 villages were selected from Fulbaria and Trishal Upazila, respectively, according to the VL endemicity, and IRS has been carried out by the programme. Additionally, in each Upazila one village far away from the IRS village was selected randomly as a control ( Figure 3).

Study
All vector control staff involved in IRS operations from central to field level including spray men were assessed in the study and a random selection of households in IRS areas were interviewed.

For Vector Density Measurement.
The main outcome variable of IRS is the reduction of vector density. Hence the assumptions for the sample size calculations for vector density measurement were as follows. An 80% reduction of vector density would be achieved at 4 weeks after IRS; the sandfly counts were expected to follow a negative binomial distribution with a dispersion parameter of k = 0.10, requiring an 80% power to detect a difference between baseline and followup at 5% significance level and assuming an intracluster coefficient at village level of ICC = 0.03. A

Bioassays for Bioavailability of Insecticide on Sprayed
Surfaces. Ten sandflies were exposed on each of 4 walls using WHO plastic cones [8], cones in 10 houses from 4 study villages. The same number was applied in unsprayed control houses. Sandflies were observed for knockdown at 1 hour and for mortality at 24 hours of recovery [9]. Corrected mortality was calculated with Abbot's formula [10] as follows: where, P = corrected mortality percentage, P i = percent observed mortality in insecticide exposed sandflies, C = percent mortality in control (nonexposed) sandflies. Bioassays were done at 2 weeks, 4 weeks, and 5 months after IRS.

Filter Paper
Technique to Determine the Insecticide Concentration on Sprayed Walls. Before spraying and without knowledge of the spray men, four filter papers (Whatman no. 1, 5 × 5 cm 2 ) were placed at different heights (from 1 foot to 6 feet from ground level) together with 9 different kinds of "fake" papers (to mislead the sprayers in case they would notice something "unusual") on the 4 walls of each room. Filter papers were collected within a week of insecticide spraying and were air dried, coded, sealed individually in aluminum foil, and stored at 4 • C until they were subjected to chemical analysis. All field-collected filter paper samples along with external standards (DDT and lambda-cyhalothrin having >95% purity) were subjected to chemical analysis by IICT (Indian Institute of Chemical Technology, Hyderabad) using standard GC-quantification methods. Fixed quantities of the test samples were injected on to high-resolution gas chromatograph in triplicate and the average values for the content of active ingredient (AI) were determined using the external standardization method.
Five DDT-WP filter papers from India and 6 lambdacyhalothrin-WP filter papers from Nepal along with standard DDT and cyhalothrin WP formulations were retested by the WHOPES (World Health Organization Pesticide Evaluation Scheme) collaborating Centre in Belgium (Centre Wallon de Recherches Agronomiques, CRA-W, Gembloux, Belgium) and by the Liverpool School of Tropical Medicine.

Insecticide
Resistance. Insecticide susceptibility was tested in one site (India) in tube bioassays, using the WHO's standard chamber method [8]. Test papers, impregnated with 4% DDT were provided by the WHOPES collaborating centre at the School of Biological Sciences, Universiti Sains Malaysia. In each assay, 15-20 unfed, non-gravid female P. argentipes (caught in the study area) were introduced into a WHO susceptibility chamber lined with the insecticideimpregnated paper and left for 1 hour. In each study area, five replicates (with 15-20 and flies each) and one control, with unimpregnated paper and 20 P. argentipes, were run. After 1 hour of exposure, the percentage knockdown was recorded before the P. argentipes were taken out of the test chamber, placed in a 150 mL paper cup that was covered with netting, and maintained for 24 h at 27 ± 20 C and 80%± 10% relative humidity, with a small cotton-wool swab soaked in 10% (w/v) sucrose solution placed on the netting top. Percentage mortality was recorded 24 h, after exposure.

Data Collection for Outcome Indicators.
Sandfly densities were measured in IRS intervention villages and in control villages at 2-weeks before spraying and at 2, 4, and 20 weeks after spraying through the collection of sandflies with CDC light traps which is preferred to the aspirator method [11]. Additionally in India and Nepal sandfly densities were determined in sentinel houses (house in the IRS area which was not sprayed due to absence of the household members or refusal of household for spraying). In each site, 72 households where IRS was done were selected as well as 72 households which remained unsprayed (sentinel houses for determining the overall reduction of vector densities). Similarly, 72 control houses were randomly selected in unsprayed control villages next to the IRS villages. In each house one CDC light trap [11] was fixed from 6 PM to 6 AM in a standardized way in the corner of the main room 2.5-5.0 cm from the wall and 15.0 cm from the floor for one night. The captured sandflies were examined on the same day of collection. Test tubes were left in −20 • C for 20 minutes or chloroform-soaked cotton was used to kill the sandflies. Identification of sandflies was based on external morphological characteristics under a binocular microscope. The sandfly species was identified, sex and abdominal conditions of females were noted separately (P. argentipes, P. papatasi, Sergentomyia spp.) [12].

Data
Collection for Impact Indicators. The monitoring of VL cases over time through routine surveillance and active case detection [13] is being undertaken but is beyond the scope of this study.

Data Analysis.
A database was developed by the data management centre at ICDDR,B in Bangladesh, using EPI Info software version 3.5.1 to enter field as well as laboratory data. Data were cleaned and checked in duplicate. Descriptive analysis was performed and 95% Confidence Interval (CI) was calculated using Normal as well as a Binomial distributional approach where applicable. The percent reduction (PR) of sandfly count attributed to IRS was calculated as: The effect is negative/positive if sandfly density is decreased/increased after intervention and the effect should be zero if the sandfly density is same as baseline. All calculations were performed by STATA 10.

Ethical Approval.
The study protocol was approved by the ethical review committee (ERC) at WHO Geneva and in each research institution by their respective ethical boards. Informed written consent was obtained from vector control staff involved with IRS operations from central to field level operation and from households which have been interviewed.

National Control Programme Inputs into IRS.
Input indicators of the M&E toolkit revealed that district action plans and guidelines for IRS were in place across the all study sites. Except for the subdistricts in Bangladesh, all other study districts in India and Nepal had proper storage facilities for insecticides. However, the proportion of functional pumps was low and the availability of spare parts was inadequate in India and in half of the study districts in Nepal ( Table 1). The training of the spray men before spraying was not systematically done in India. Personal protection of the spray men was completely or partially ignored by the program in all study districts.

The IRS Process: Prospects and Limitations of Spraying.
In general process indicators of the M&E toolkit reflected nonhomogeneous spraying performance within and between countries ( Table 2). It was found in India that none of the spray squads adequately filled the pump whereas in Bangladesh and Nepal it was adequately done. In India, only 29.4% (95% CI, 16.9-41.9) of the spray squads mixed DDT and water properly while in Bangladesh and Nepal this was done properly with pyrethroids by all squads. In India during spraying, 23.5% (95% CI, 11.9-35.2) of the stirrup pumps were found to have leakages. In three study districts of Nepal and in the Trishal Upazila in Bangladesh, none of the pumps had a pressure gauze. In further three districts in Nepal, only 44.6% of pumps (95% CI, 35.4-53.9) had pressure gauzes. Most of the sites showed that proper distance of the nozzle from the surface was not maintained correctly (Table 2). In India, 41.2% (95% CI, 27.7-54.7) of spray squads did not maintain the spray swath properly. In three districts of Nepal, none of the spray squads maintained the spray swath properly whereas in the other three districts the spray squads did perfectly well. In Bangladesh, the squads in one Upazila maintained the spray swath poorly but in the other Upazila the spray squads did well. In one Upazila of Bangladesh and in India, 100% of the supervisors were present and supported the squads, but in the other Upazila of Bangladesh, less than half of the supervisors (44.2%; 95% CI, 35.3-53.1) were present. None of the supervisors at any district in Nepal were present during the spraying. The management of left over insecticides was properly done except for one Upazila in Bangladesh where most of the left over insecticide (90%; 95% CI, 84.6-95.4) was thrown in a nearby water body. The marking of sprayed houses (stencils) was partially done in all sites except in one Nepali site. Instructions had been given to households in all study sites to prepare the rooms before spraying and not to enter the house during spraying.   (Table 3).

Direct Results
(b) Acceptability of IRS. IRS was highly accepted by the community people in all sites and negligible side effects were reported after IRS. However, in all study sites (except one site in Nepal) instructions from spray men were rarely received about how to prepare their houses for the spraying or to leave the house while the spraying was done or not to mud plaster or paint the wall after spraying (Table 4).

8
Journal of Tropical Medicine (f) Reduction of Vector Populations. The main outcome indicator in the M&E toolkit explored whether the current national IRS programmes did achieve a high reduction rate of sandfly densities (Table 5). In India the reduction of sandflies in sprayed houses was remarkable. In Nepal IRS was effective to reduce sandfly density; however, the duration of the effect was only up to 4 weeks. In Bangladesh there was no effect on the vector population at all in one Upazila and a small effect in the other (Table 5).

Discussion
The study showed that the newly developed M&E toolkit successfully identified the shortfalls of IRS in the Indian subcontinent. IRS is an operationally challenging programme which includes a set of activities requiring (a) close coordination among national, state (India), district, and subdistrict authorities, (b) a number of technical ingredients such as insecticides, storage facilities, spraying pumps, spare parts, and protective clothing, and (c) training and supervision in order to achieve the necessary level of reduction of vector densities to interrupt disease transmission. Our study reconfirmed the original hypothesis that through systematic monitoring, the weak parts in the chain from the procurement of materials to the final endpoint reduction of sandfly vectors can be identified and solutions can be sought. The newly developed M&E toolkit was applied by spray supervisors, DMOs/DPHOs/UH&FPOs, and some elements by external reviewers from the research institutions (mainly observation of spraying performance, satisfaction survey and entomological tests). The toolkit was able to identify: (a) cross-cutting issues in all or most sites: poor spraying techniques, lack of protective clothing, poor marking of houses, deficient advice to household members regarding the spraying, and limited effect on the vector population.
(b) site-specific problems: lack of training (Bangladesh, India), many deficient spraying pumps (India), poor filling of pumps (India), poor disposal of left-over insecticides (Bangladesh), poor marking of sprayed houses (Nepal), low coverage of targeted houses (India), high level of insecticide resistance (India).
As a result there were serious IRS effectiveness issues: (i) bioassays indicated low standard of spraying and/or elevated insecticide resistance (as wild caught sandflies were used for the tests) or poor insecticide quality; (ii) the reduced insecticide susceptibility of sandflies for DDT in India (in contrast, in Nepal the susceptibility of sandfly vectors is still high for deltamethrin [14]; (iii) the reduction of vector densities was low, even in India where sprayed house showed a high vector reduction but unsprayed sentinel houses showed no reduction at all pointing to a lack of mass effect on the vector population by IRS. The chemical concentration of insecticides on sprayed surfaces as determined by a national laboratory was unexpectedly low compared to two European laboratories used for validation and quality control. These tests in the laboratory are time consuming, expensive, and susceptible to errors. Thus RDTs (rapid diagnostic test) for chemical residuals on sprayed walls are very important in this situation, and it is hoped that an RDT for chemical residuals on sprayed walls will soon be available [15].
Monitoring and evaluation activities only make sense if the findings are certified by the relevant authority and followed by adequate response mechanisms. However, this was beyond the scope of this particular study which assessed the performance of the M&E toolkit. The recommendations coming out of the analysis were summarized after completing the spraying cycle and were then discussed in training workshops with UH&FPOs in Bangladesh (2 UH&FPOs), DMOs in India (31 DMOs), and DPHOs in Nepal (16 DPHOs). These workshops were used for analyzing the acceptance of the toolkit and provided opportunities for the "triangulation" of the study results. It was found that DMOs/DPHOs/UH&FPOs were almost unanimously satisfied with the toolkit; they were aware of several issues detected in this study (e.g., inadequate training for spray men and supervisors, shortage of good quality pumps and spare parts, lack of proper supervision during spraying, etc.) but not of others (e.g., limited impact on the vector population; resistance issues) and, more importantly, further issues were raised such as the slow cash flow within districts to spraying squads resulting in delays of payment or the lack of travel funds for supervisors and sometimes even for the spray men. Meetings with the relevant district authorities were arranged in order to address these problems. It can be expected, however, that it will take some time until improved training and supervision together with the systematic application of the M&E toolkit by programme managers will be in place leading to improved IRS operations in our study districts and beyond. Of special concern is the stirrup pump used traditionally in India for DDT spraying as it is difficult to manage, less efficient than the compressor pumps used in Nepal and Bangladesh, and has higher health hazards [16]. After this study the toolkit also has been distributed by the health authorities to all districts in local language in India and Nepal, in Bangladesh translation into local language is under process.

Conclusion
The M&E toolkit was useful for identifying major shortcomings in IRS operations. Therefore the M&E toolkit developed with the help of WHO-TDR should be adopted and systematically applied by the programme of the three countries to achieve the expected outcome of IRS.