Improving participation in door-to-door vector control campaigns using behavioral economics

A. Current practice (control). The control arm will consist of the current Ministry of Health campaign. Printed campaign materials are distributed to households. Health promoters request permission to spray and schedule insecticide spray for the next day. Promoters can answer any questions or concerns.

B. Advance commitment responsive scheduling. Households are approached 2-4 weeks prior to the spray date and asked to commit, in advance, to participate. Households that agree to participate are offered convenient 2-hour appointment windows on a future spray date. Weekend and evening appointments are available. Planning prompts, which encourage households to make a specific plan to prepare the home for spraying, are also offered. Households can request a paper or text message reminder prior to their scheduled appointment, and will be able to reschedule their appointment if necessary. Households are then also visited before the previously scheduled appointment by the health promoter to confirm the appointment.

C. Neighbor recruitment. One month prior to targeted spray dates, study staff approach formal community leaders (Ministry of Health-trained health promoters, block captains, and elected community officials) and informal neighborhood opinion leaders (corner store owners, day care coordinators), explain the campaign in depth, and ask leaders to serve as neighbor recruiters. Each recruiter is given promotional t-shirts, a clipboard, and the same educational materials used by campaign staff. Recruiters are asked to promote the campaign on the block, wear the campaign t-shirts regularly, and distribute additional Ministry of Health materials.

D. Contingent group sweepstakes. Contiguous households are assigned to sweepstakes groups of 6 households. Sweepstakes groups are randomly assigned a lottery number. Households whose lottery group is selected and who participated in spraying will win a small prize such as a gift card for $15. If 100% (6/6) households in the selected group participated in the spraying, each participating household also wins a $50 gift card. Lottery groups will be put into pools of 100 households based on sampling sectors in the intervention area. Study staff will inform households about their lottery group and lottery number, and explain the rules and procedures during recruitment. In addition, flyers explaining the lottery (including lottery group membership and lottery number) will be left at each household. After spray is completed in each lottery pool (N=14 pools), an event will be scheduled to coincide with a broadcasting of Peru's Tinka Megabol© lottery drawing. Community members will be invited to a central neighborhood location, such as a health center, to listen to the lottery drawing broadcast and to see prizes awarded to households that participated in the spray campaign. Attendance at the meeting is not necessary to win a prize; winners will be notified and prizes distributed to their homes within a week of the drawing.

2015-02-16

2015-12-15

A. The target behavior: insecticide application. The typical Ministry of Health insecticide campaign requires a second application of insecticide six months following the first treatment. We will monitor and measure participation during the second round of insecticide application among households in the study site who were exposed to the intervention trial. We will not implement any lotteries or other participation interventions. Our field team will accompany the Ministry of Health campaign workers and record participation data as in the first round of insecticide application. Motivation for participating in the second round is generally decreased due to the absence of insects, which will highlight any differences in participation due to intervention exposure.

B. A related behavior: ongoing vector surveillance. The surveillance phase of the vector control campaign relies primarily on community-based reports of vector re-infestation. This passive surveillance is supplemented by annual household visits by Ministry of Health personnel who actively search households for vectors to ensure that the passive surveillance system is working. Our study team will accompany surveillance teams 24 months following the second insecticide application and assess participation in the active surveillance campaign. Participation is defined as allowing Ministry of Health workers to enter the home and check for insects; no preparation of the home is required, nor is any insecticide applied.

C. An unrelated behavior: canine rabies vaccination. The Ministry of Health's zoonotic disease division conducts annual rabies vaccination campaigns for dogs throughout the city. Vaccination takes place in publicly accessible locations such as markets and parks. When a dog is vaccinated, the home address of the owner is recorded and a colored plastic collar signifying the year of vaccination is placed on the animal. Following vaccination, door-to-door sweep-up surveys are conducted to ascertain vaccination coverage and to vaccinate any dogs without a collar. Our study staff will monitor vaccination booths and compile addresses of vaccinated dogs to add to the campaign database from the interventions. We will also accompany Ministry of Health personnel in the sweep-up surveys to gather final vaccination rates.

1. Preliminary survey of participation correlates: During the preliminary survey phase of the vector control campaign, field staff will accompany the Ministry of Health personnel as they collect baseline household data for all households in the targeted district. The campaign routinely collects household socioeconomic and demographic characteristics, including education of household head, household building materials, asset ownership, animal ownership, and employment. Vector infestation is also assessed through a rapid (20 minute) search of the household by trained personnel, who use a can of Raid© (Tetramethrin) to flush insects out of their hiding places. Data from the preliminary survey will be used to better assess the effectiveness of the interventions by controlling for the effects of known correlates of participation and to assess effect modifiers.

2. Randomized field trial: The second phase of the study is a randomized field trial of three novel interventions and the current campaign as the control. Geographic clusters of approximately 100 households will be identified using mapping data and then randomly assigned to one of the four arms of the trial. All households in sampled sectors will be included in the trial. Non-participation in the preliminary survey does not remove the households from the sampling frame.

3. Assessment of the impact of interventions on participation in subsequent health campaigns: We will collect data on participation in three subsequent public health campaigns in the same neighborhoods where the field trial was conducted. The campaigns lie along a gradient of relatedness to the initial incentivized behavior of participation in insecticide application.

4. Evaluation of attitudes toward the vector and the campaign: From a smaller subsample (n=1,400) of the randomized households, we will employ three surveys during different stages of the study to evaluate participants’ responses to the vector and to the spray campaign interventions. Households will receive three different visits to complete the three time-dependent surveys.

Clusters will be randomized in the Arequipa office by a computer.

Clusters will be randomized.

Yes

56 clusters of 100 households/cluster

5,600 households

1,400 households control, 1,400 households contingent group lottery, 1,400 responsive scheduling with advanced commitment, and 1,400 neighbor recruitment.

We use the method of Blitstein et al.132 to calculate power and sample size for the cluster randomized trial. Our cluster-randomized trial is powered for a head-to-head comparison of the four arms of the trial. All calculations use the Bonferonni correction to account for the six possible comparisons between all pairs of arms; we derived the expected intracluster correlation coefficient parameter (0.02) from our existing data from the most recent campaign in Mariano Melgar district. We selected our sample size based on the lower bound of participation. The power of our trial depends on the participation achieved under the alternative arms. We are powered at 88.5% to detect an increase of 11% over the level of participation achieved in Mariano Melgar (66%); anything less would not justify the effort and costs of the interventions. Our trial has significantly higher power to compare alternative successful interventions.