This study confirms the success of the RPI as an innovative, not-for-profit option for promoting medicine price competition in Kyrgyzstan, and ultimately increasing access to medicines. The RPI not only addressed geographic access by enabling rural residents to buy medicines in their own villages, it also spurred dramatic price competition in private pharmacies located in the district center. Thus, the RPI's impact was far greater than anticipated as the new pharmacies managed also to increase access to medicines in other villages through the competitive price reductions they engendered. The ultimate result was more affordable medicine for both villagers and residents of the district center. It is worth noting, however, that for some medicines, when the RPI introduced prices higher than competitor prices, the competitor increased their prices to match those of the RPI. In this regard, the RPI demonstrates power to drive markets both downward and upward in price.
We do not believe the RPI resulted in the institution of a rural market. Instead, we believe the market already existed in Jumgal prior to the RPI and that the entry of the not-for-profit RPI spurred a more competitive market. Indeed, demand for medicines was already documented before the RPI when villagers identified access to medicines as their number one health concern. On the supply side, medicines were available in the rayon center, but not in the villages themselves. Prior to the establishment of the RPI, villagers either hired a taxi to deliver medicines from existing pharmacies to their homes or they secured some means of transport to travel outside the village to the nearest pharmacy. The establishment of the RPI, therefore, took business away from the existing pharmacies despite being located far away. The emergence of the RPI pharmacies also seems to have stimulated expansion of the private sector into new villages. Owners of pharmacies in the rayon center opened two branch pharmacies in two of the larger villages, perhaps in an attempt to minimize loss of business in the rayon center.
While we have no means to assess price collusion, we suspect that some degree of collusion existed among the private pharmacies that were established prior to the RPI. We therefore suspect the introduction of the RPI disrupted any existing price collusion in the region. Given that our study tracks prices for three years after the establishment of the RPI, we suspect the competitors' price reactions are sustained and not a one-off reaction to the RPI.
When the competitors change prices to mimic RPI medicine prices, their prices are often near identical to the RPI prices. We were unable to fully ascertain how the competitors gained market intelligence on RPI medicine prices. Upon establishment of the RPIs, the RPI nurses were instructed to display all medicines with price tags affixed to their packages. When interviewed a few months later, the nurses told investigators that they no longer displayed medicine prices because employees from competitor pharmacies would visit RPI pharmacies and record medicine prices. The RPI nurses, management, and others were greatly concerned about predatory pricing by competitor pharmacies. The RPI was quite fragile when first established and many feared the existing private sector pharmacies would intentionally undercut RPI prices in order to drive them out of business. Our results suggest that competitors are still obtaining price information from RPI pharmacies even though price tags are no longer affixed.
While our study provides compelling results, it has limitations. First, we were only able to access prices for those medicines covered by the Health Insurance Fund. The sale of all medicines covered by the Health Insurance Fund, however, accounted for 51% of total RPI revenue in the fourth quarter of 2007 . Because these pharmacies serve sparsely-populated rural regions, small sample sizes limited us to analyzing only 19 of the 30 top-selling medicines with multivariate methods. Some medicines that passed our sample size criteria for multivariate analysis had limited insurance claims prior to or after RPI introduction. We utilized a linear model that assumes a relatively constant quarter-on-quarter price change. Given that we only observed seventeen quarters of data, we believe the linear assumption is reasonable. Model checking diagnostics, such as test for residual autocorrelation, showed that our model is adequate for our purposes.
Furthermore, we note that to make optimum use of space, we provided price trends in graphic form for only four of the thirty medicines studied, however a full set of figures is provided in Additional File 1. Figures provided in this paper depict the three types of competitor price changes observed after the introduction of the RPI-price decreases, price increases, and no price changes-and are representative of the medicines in each price change category. We had no means to determine why some medicines exhibited dramatic price changes and others remained unchanged. While we had no data on the quality of medicines, we do not believe that quality confounded our price findings. For branded generic medicines, we assume the quality of medicines to be identical in RPI and non-RPI pharmacies. Because the RPI and non-RPI pharmacies purchase medicines from the same wholesalers in Bishkek, we assume quality of non-branded generic medicines is comparable. We have no evidence of pharmacies over-charging insurers, however we expect it happens to some degree. Over-charging is less likely to happen within the RPI because it is supervised by staff from the Mandatory Health Insurance Fund.
While research on market impact is typically complicated due to many concurrent interventions and changing market conditions, we are confident that the market in Jumgal is truly local and has no other large-scale interventions that might be responsible for our findings. Indeed, even an annual inflation of 10%  did not seem to affect medicine prices in this region over the study period. We decided to use current medicine prices in lieu of adjusting prices for inflation after noting that most medicine prices trended downward or remained unchanged over the 4 years and did not seem to follow the national inflation rate. We are not sure why medicine price trends are inconsistent with national inflation trends. We suspect that medicine prices were already priced at the highest prices the market could bear or that national inflation rates simply do not represent the price trends in rural medicine markets. Lastly, we have no means of determining if the market existing prior to the RPI was competitive or collusive with regards to price setting.
Our study was designed only to assess whether the RPI induced regional price competition and not to evaluate rational use of medicines in RPI pharmacies. Others have shown that perverse incentives to misuse medicines may result when prescribers benefit from medicine sales . In this study, we note that ampicillin injection has the highest sales volume and the largest number of insurance claims (Table 2), suggesting overuse of both antibiotics and injections. Additional research is needed to assess the impact of the RPI on rational use of medicines.
Similarly, our study does not aim to explain why the RPI has been successful or how the RPI has been sustainable amidst other documented failures to increase access to medicines in rural regions. The RPI may offer a model that can be scaled up across many more regions and in other parts of Central Asia, but it is important to first determine the most critical elements that led to its success. While we cannot pinpoint theses critical components, we note the key characteristics of the RPI, including the role of highly trained and motivated players who genuinely wanted to develop and test new ways to increase access to essential medicines in rural areas. These staff devoted a great deal of their personal time and energy to seeing the project through, and it would be a mistake to overlook or underestimate the value of this social capital, especially in post-Soviet Central Asia. From a social and political standpoint, the climate of Kyrgyzstan is recognized as being more conducive to civil society than that of any of its neighbors in Central Asia. However, with the exception of a few very strong professional organizations, the country's health sector NGOs tend to be less evolved than the NGO managing the RPI. In addition, the NGOs tend to be rather fractured, often working from outside the government rather than in collaboration with it. We therefore see a need for donors, international organizations, and governments to assist in reorganizing and building the capacity of existing NGOs to refocus their social capital toward more concrete activities, such as establishing and overseeing the RPIs.
Understanding that the majority of people in developing countries still seek care from pharmacies rather than public sector health facilities, many donors have been eager to develop private sector interventions but have been wary of engaging directly with the private sector. At the same time, donors are eager to establish and promote community-based programs and civil society. The non-profit nature and involvement of civil society organizations in the rural pharmacy initiative model could provide donors with an opportunity to accomplish multiple goals without compromising their non-profit missions.
Lastly, the study reveals the utility of data on medicines that are routinely collected through mechanisms such as insurance schemes. These types of data sources are rich and should be used to build a solid body of evidence to guide policy on access to medicines for the poor. Research on interventions to increases access to medicines must include assessment of potential impact on both formal and informal markets. More work is needed to identify incentives for NGOs and other non-profits to engage in the establishment and management of rural pharmacies that can compete with existing private pharmacies. This should include determination of the operating costs to establish and maintain rural pharmacies and the minimum mark-ups needed to sustain these pharmacies, as well as pricing policies that promote rational use of medicines. Additional research is also needed to examine policies and programs that promote and impede competition in the pharmaceutical sector, including description of market size and structure, presence or absence of competition laws, price regulation, barriers to market entry, and marketing.