Price Subsidies, Diagnostic Tests, and Targeting of Malaria Treatment: Evidence from a Randomized Controlled Trial

Artemisinin combination therapies (ACTs) are now the only effective drugs in many malaria-endemic regions. In Kenya they are available in government clinics and retail drug stores. Public health clinics are often out of stock while the drug is unaffordable for the poor at full retail price. Subsidizing the drugs at the retail level can make the drug affordable but at the risk of over-treatment. The study attempts two interventions to test the trade-off between overtreatment and underteatment with ACTs.

Sample households are assigned to one of three treatment arms. In the first treatment arm, households received vouchers to purchase ACTs at full price from their nearest drug store. In the second treatment arm, the ACTs are subsidized at 92%, 88% or 80%. In the third treatment arm, households received subsidized rapid diagnostic malaria tests (RDT) in addition to the ACT subsidies.

In order to find out what fraction of people buying ACTs were truly malaria positive—a proxy for how well an ACT subsidy would be targeted—the researchers also selected a random subset of the households in all treatment groups to receive the offer of a “surprise” free RDT after they completed their transaction at the drug shop.

2009-05-01

2009-12-31

The ultimate aim in this trial is to study how retail sector ACT subsidies impact both under-treatment (UT , the share of true malaria illnesses that are not treated with ACTs) and over-treatment ( OT , the share of non-malaria illnesses that are treated with ACTs). Since collecting the data to directly measure these outcomes was not logistically feasible, UT and OT are estimated indirectly by measuring:

1) Access: the share of all illness episodes treated with an ACT
2) Targeting: the share of ACT takers who are actually malaria positive

1) Access: A malaria positivity index was created for all illness episodes, based on a symptoms database (N = 533) collected from the study population. During unannounced home visits, trained surveyors asked if anyone was feeling ill, and if yes, they collected information about symptoms (using the same survey instrument used in the endline survey and then tested the patient for malaria with an RDT. This data is then used to impute a malaria probability to the universe of illness episodes enumerated at endline and all illnesses observed at drug shops.

2) Targeting: A subsample of those redeeming the ACT voucher are administered a “surprise RCT” unless they are part of the ACT+RDT treatment arm and had already redeemed their RDT voucher prior to getting their ACT.

Four rural drug stores, in a rural market setting, were selected. All households within a 4 km radius catchment area were sampled at baseline. At the end of the survey, the household was informed of its treatment status (full price ACT, ACT subsidy or ACT+RDT subsidy) and a voucher given accordingly. Each household was given two vouchers for ACT and two for RDT (if applicable) that could be redeemed at the study drug store. The drug dispensed was specific to the age of the patient. All ACTs and RDTs were provided by trained study officers posted at the drug shop for the duration of the study.

Additionally, to test what proportion of patients buying ACT was actually positive for malaria, a random subsample was given the option of a free RDT test after they had redeemed their ACT voucher. If the patient had not come to the drug store, the officer went to the patient’s home to administer the test. During the endline survey, administered about four months after the vouchers were distributed, households were informed the study was ending and all vouchers were collected back.

Computerized random number assignment algorithm and was stratified by drug shop, by the household’s distance to the drug shop (in quartiles), and by the presence of children in the household.

Household

No

Study not clustered

2789 households

Control: no subsidy = 180 households
Treatment: ACT+ RDT subsidy = 1,625 households (92% subsidy = 394, 88% subsidy =619, 80% subsidy = 612)
Treatment: ACT subsidy only = 984 households (92% subsidy = 328, 88% subsidy =326, 80% subsidy = 330)

PRICE SUBSIDIES, DIAGNOSTIC TESTS, AND TARGETING OF MALARIA TREATMENT: EVIDENCE FROM A RANDOMIZED CONTROLLED TRIAL

Both under- and over-treatment of communicable diseases are public bads. But efforts to decrease one run the risk of increasing the other. Using rich experimental data on household treatment-seeking behavior in Kenya, we study the implications of this trade-off for subsidizing life-saving antimalarials sold over-the-counter at retail drug outlets. We show that a very high subsidy (such as the one under consideration by the international community) dramatically increases access, but nearly one-half of subsidized pills go to patients without malaria. We study two ways to better target subsidized drugs: reducing the subsidy level, and introducing rapid malaria tests over-the-counter.

Cohen, Jessica, Pascaline Dupas, and Simone Schaner. 2015. "Price Subsidies, Diagnostic Tests, and Targeting of Malaria Treatment: Evidence from a Randomized Controlled Trial." American Economic Review 105(2): 609-645.

https://www.povertyactionlab.org/sites/default/files/publications/327_119%20Subsidies%20for%20ACT%20AER2015.pdf