This is the first study to systematically and rigorously measure the availability and retail price of AMFm-subsidized ACT across retail outlets with varying degrees of remoteness. It shows that aggregate availability of AMFm-subsidized ACT increased substantially within 6–8 months of the launch of AMFm in Tanzania, even though some regional differences in availability remained. While this paper does not explore changes in ACT use under the AMFm, it shows that ACT availability in drug shops across two remote regions of Tanzania has increased dramatically under the AMFm and that ACT availability in drug shops farther away from main towns and main roads has also increased significantly. In the first survey round, the average retail price of subsidized ACT was much higher (1,529 TZS or 1.03 USD) than the government recommended price of 1,000 TZS (0.64 USD). However, the price continued to decrease and by the fifth survey round the mean retail price was 1,272 TZS (0.81 USD), with the median price at the recommended 1000 TZS. Similar to availability, there are regional differences in retail prices, with prices in the Rukwa region higher than in Mtwara.
Although initial availability of subsidized ACT was higher in drug shops closer to suppliers or drug shops located proximally to towns, this difference decreased over time and does not appear to be a difference that would persist in the long run. The initial aim of this operational research study was to test a supply-side incentive that could improve availability of AMFm-subsidized ACT in remote shops, if availability in remote shops was lower. The study team decided not to launch an incentive, as availability in remote shops increased much more rapidly than expected.
Significant differences in availability across Rukwa and Mtwara—two very different regions of Tanzania with varying geography, transport/trade linkages and malaria ecology—persisted into the last survey round. Rukwa, which is much farther from Dar es Salaam, has much lower population density than Mtwara, and lower malaria prevalence, had lower availability of subsidized ACT. A larger fraction of the drug shops in Mtwara obtained their supplies directly from Dar es Salaam, whereas in Rukwa they rely on local wholesalers in Sumbawanga or Mbeya. This structural difference in sourcing could explain some of the temporal lag in availability in Rukwa because the local wholesalers could have taken longer to stock and promote subsidized ACT. It is worth noting that while the availability in Rukwa was very low as compared to Mtwara in the first round of analysis, it increased rapidly during the period of the study. This study also finds that prices in Rukwa were significantly higher than Mtwara in the last two survey rounds. The higher prices in Rukwa could be due to the higher cost of transport and distribution to this region compared to Mtwara. Differences in malaria prevalence between the regions could also contribute to differences in availability and prices.
Geographic parameters, particularly measures of distance to large population centers, and distance from major roads, form the basis of most published indices of remoteness. Literature in trade, poverty and medicine
 consider a purely geographic measure of remoteness, which excludes any consideration of socio-economic status and other demand side factors. The findings of this study suggest that it may be worth exploring more context-specific definitions of remoteness and that understanding socio-economic factors, sourcing patterns and demand creation activities by wholesalers and sub-wholesalers may be key to ensuring more widespread and equitable availability of ACT. Further study should also examine differences between Rukwa and Mtwara using an index like the Rural Access Index (RAI) which measures the number of rural people who live within two kilometers of an all-season road as a proportion of the total rural population
A related study by Cohen et al. suggested that existing supply chains for anti-malarials may not reach all individuals in rural regions and special interventions might be needed. The current study shows that existing supply chains are able to serve remote rural drug shops as well. The differences in the findings may stem from the fact that the Cohen et al. results were based on a small pilot study using a single wholesaler/distributor whereas the results presented here are from a national roll-out of the AMFm subsidies with multiple wholesalers serving each region. The national scale AMFm is able to take advantage of market competition and economies of scale that were absent or limited in the pilot study. Additionally, the national scale-up of AMFm which was accompanied by nationwide media campaigns to raise awareness about the availability and price of ACT which was not the case in the pilot study in
Availability of child and adolescent doses of AMFm-subsidized ACT was lower as compared to adult doses, possibly due to lower quantities of non-adult packs imported by the national importers/distributors under the AMFm program (Additional file
4: Table C1). It is unclear to what extent the quantities of each pack size of subsidized ACT imported by the national wholesalers/distributors reflected true end patient demand or was the result of other market dynamics.
There are a number of caveats to this study. First, the study was undertaken in only two purposively sampled regions of Tanzania. While the results should have validity outside the regions considered, caution should still be used in directly applying these findings to all other settings. The results may be strongly dependent on socioeconomic factors, malaria treatment-seeking behavior, the structure of the wholesale and retail markets and the structure and performance of the public sector--factors that are often very different between and within countries. Due to logistical and resource constraints, a full-fledged study to capture the stocking of ACT in the public sector, in the two regions, was not undertaken. As a result, any changes in public sector availability were not systematically recorded or analyzed. Additionally, measures of price were self-reported by the ADDO owner or shop attendant.
In summary, this research indicates that the AMFm has significantly increased availability of ACT in Tanzania including in remote drug shops. Even though disparities exist in availability of ACT across the two regions of Tanzania that were a part of the study, these disparities seem to be disappearing over time. The findings show that large inequities in availability of subsidized ACT are only short term and the long run availability of subsidized ACT becomes more equitable both across remote and non-remote drug shops, and across regions. Similar to availability, there are regional differences in retail prices but median retail prices have been decreasing over time. This study does not capture how increased availability translates into increased ACT use, which is the focus of ongoing research in the two regions of this study.