Malaria Diagnostic Testing and Conditional Subsidies to Target ACTs in the Retail Sector: the TESTsmART Trial Aim 2
Malaria Diagnostic Testing and Conditional Subsidies to Target ACTs in the Retail Sector: the TESTsmART Trial Aim 2
Highly subsidized first-line antimalarials (artemisinin combination therapy or ACT) are available over the counter in the private retail sector in most malaria-endemic countries. Overconsumption of ACTs purchased over the counter is rampant due to their low price, high perceived efficacy, and absence of diagnostic tools to guide drug use. The ultimate goal of the proposed work is to improve antimalarial stewardship in the retail sector, which is responsible for distributing the majority of antimalarials in sub-Saharan Africa. Through a combination of diagnosis and treatment subsidies and provider-directed incentives, this approach will align provider and customer incentives with appropriate case management and thereby improve health outcomes.
The main objective of this study (Aim 2) is to test two key interventions in a random sample of private medicine retail outlets in Nigeria. This will be a cluster-randomized controlled trial where the cluster is a private retail outlet that stocks and sells WHO quality-assured ACTs. This two-arm study will test 1) a provider-directed incentive for testing and reporting in combination with a consumer-directed intervention in the form of a diagnosis-dependent ACT subsidy against 2) a comparison arm. Outlets in both arms will offer malaria diagnostic testing to customers who wish to purchase one. Information for the primary and secondary outcomes will be collected during exit interviews with eligible customers. The primary outcome will be the proportion of ACTs sold to customers with a positive diagnostic test. The main secondary outcome will be the proportion of suspected malaria cases presenting to the retail outlet that are tested. Other secondary outcomes include adherence to the RDT result amongst those tested (defined as taking a quality-assured ACT following a positive test and refraining from taking an ACT following a negative test) and appropriate case management for all suspected malaria cases (proportion tested and adhered among all suspected cases).
RATIONALE:
The investigators hypothesize that decision-making in the retail sector is influenced primarily by price (or profit) supplemented by individual beliefs or preferences. Information from a diagnostic test, when available, plays only a minor role in the decision-making landscape. In response to this, the investigators propose to test a scalable, policy-relevant strategy that integrates testing and treatment subsidies and makes ACT subsidies available only to customers with a positive malaria test. Differential pricing based on the results of the diagnostic test will align consumer and provider incentives (price and profit) with testing and appropriate ACT use. The investigators' approach will also ensure that public subsidies are directed at confirmed malaria cases thereby enhancing the sustainability of such programs. By allocating subsidy dollars across both testing and conditional treatment (rather than universal, treatment-only subsidies), the investigators can reduce the cost of subsidizing malaria treatment and improve targeting of ACTs without compromising access.
OBJECTIVES:
This goal of the proposed work is to improve antimalarial stewardship in the retail sector, which has the largest market share of antimalarials in sub-Saharan Africa. In this study (Aim 2) the investigators will use a cluster randomized controlled trial to measure the impact of two innovative interventions - a provider-directed incentive for offering malaria rapid diagnostic tests to suspected malaria cases and a consumer-directed diagnosis-dependent ACT subsidy (conditional ACT subsidy). The investigators will test both interventions in tandem against control conditions. The unit of randomization will be the private medicine outlet (cluster).
STUDY DESIGN:
The study will be a cluster-randomized controlled trial with two arms. See Section 4.4 Statistical Design and Power for complete details.
STUDY POPULATION:
The study will be conducted in Nigeria, a country with high malaria burden. Nigeria has the programmatic goal of universal access to prompt parasitological confirmation before treatment and have endorsed the use of malaria RDTs in the public health sector at all levels of care. However, very high rates of treatment in the retail sector undermine the explicit policy that all malaria cases should be confirmed by parasitological diagnosis. In Nigeria, the government has recently permitted retail outlets to conduct RDTs in an effort to reduce overuse of antimalarials, particularly publicly subsidized ACTs.
Specific study site will be Lagos, Nigeria.
STUDY PROCEDURES:
A complete sampling frame of eligible retail outlets will be made for the site. Outlets will be eligible if they regularly carry quality-assured, subsidized ACTs. Participating retail outlets will receive an initial small start-up stock of RDTs and will be given contact information for RDT wholesalers who can provide replenishment stocks on demand. In addition, all attendants will be trained to use a mobile reporting app that captures basic information about suspected malaria cases. This app will not be used to assess outcomes, but will be an important monitoring tool and serve as the conduit for all payments according to the arm assignment.
Following training, shops will be randomized to one of the two arms:
Customers who come to a participating outlet in either arm with fever or suspected malaria are eligible for an RDT at the study-recommended price and, in arm 2, a conditional ACT subsidy when the RDT is positive. Attendants will be trained to recognize danger signs and refer those individuals to a health facility for immediate care. Customers purchasing medicine on someone's behalf or those who have already had a malaria test will not be eligible for RDT testing or conditional subsidies.
Study outcome measures will be collected by interviewing a random sample of customers leaving participating shops (exit interviews). A team of field interviewers will be trained in each site and will be assigned to participating outlets on random days in a month. Both the day of the month and the interviewer assigned to the specific outlet will be randomized in order to balance data collection across interviewers and arms and to minimize the observer effect. Each outlet will be visited thrice in a month and the interviewer will enroll eligible participants in succession until they have completed 3-4 interviews. Three customers per month per outlet are needed to reach the sample size of 32 interviews per outlet over the data collection time frame. Exit interviews will capture information about customer demographics, symptoms of their current illness, whether they had a malaria diagnostic test at the outlet or elsewhere, how much they paid for the test, the results of the test, and the drug(s) they purchased. Customers will be eligible to participate in the exit interview if s/he was seeking drugs for him/herself or their minor child older than one year of age and the child is physically present, s/he had a febrile illness or suspected malaria, and s/he agrees to be interviewed.
Study monitoring will be continuous throughout the study period using the mobile app. Study staff will regularly review reporting data captured by the mobile app. In particular, pictures of RDTs will be reviewed alongside results reported by outlet attendants to ensure correct use and interpretation of the RDTs. In addition, exit interviews will be summarized monthly by outlet and compared to the data submitted via the app. This monitoring step will allow us to identify major discrepancies between actions reported by the shop and those reported by the customers such as testing rates, RDT positivity rates, dispensing of qualified ACTs and prices charged for RDTs and ACTs.
SAMPLE SIZE AND POWER:
The investigators calculated power based on a cluster randomized two-sample two-tailed t-test for the comparison of two proportions using standard formulae. We calculated power for differences in the primary outcome for the comparison of interest noted above. With a sample of 32 exit interviews per outlet (48 outlets in Nigeria), we would have >90% power to detect a minimum difference between Arms 1 (control) and 2 (combined interventions) in the primary outcome of 12 percentage points.
While interview rates at some of the higher volume shops have been higher than expected, interview rates at some of the lower volume shops are on pace to finish below the target of 32 interviews per shop. This higher than anticipated variation in cluster size could negatively impact power, therefore sampling will be continued across all shops for the duration of the study in an effort for each shop to meet or exceed the target number of interviews, or at least to come as close as possible. With a revised mean of approximately 53 interviews per shop, the above power calculations will be maintained.
The sample size estimates correspond to a total of 2524 exit interviews with clients in Nigeria. Since not everyone interviewed will have purchased an ACT, the estimates account for the fact that only a subset will enter into the analysis for the primary outcome. The investigators will have a team of data collectors (one team per site) and each team will target 3 participants per outlet per month. Exit interviews conducted up until the study design change was implemented (n = 1676) may be used for supplemental analyses.
See Section 4.4 Statistical Design and Power for comprehensive description of sample size, power and statistical analysis.
SUBJECT CONFIDENTIALITY:
Customers leaving the shop will be approached by the field interviewer to ask if they are interested in participating in a short interview. If they agree, then they will be taken to a private area and the study will be explained to them. Once they provide consent, the interviewer will record the details of their illness and treatment at the outlet. No participant identifiers will be recorded in the electronic data collection form. This limits the possibility of identifying the source of any health information (See Protection of Human Subjects Section 3.3).
All clients attending a participating outlet on the day selected for exit interviews will be eligible to be screened for inclusion into the interview sample.
Inclusion Criteria:
Exclusion Criteria: