JEANNE MADDEN

J. Madden

WHEN findings from WHO/HAI medicine prices surveys are com- pared from country to country, the results are often quite dramatic. Nine countries were involved in two rounds of pilot studies from 2001 to 2002. It is not possible to conduct all types of analyses for the pilots in a cross-national way, because of differences in country contexts or in the way data were collected as the methodology was under development. However, in Figures 1 - 3 we are able to show a number of valid comparisons.

The data presented in these graphs use the median price ratio (MPR) measure, which is explained in the box on page 17. The basic MPR is the median of the prices found for a single product category across a sample of outlets, converted to US dollars and then divided by an international reference price. For example, in Armenia, the price for a single tablet of the originator brand version of furosem-ide ranged from 12 to 16 drams in the shops surveyed. The most common price found was 16 drams, and this was also the median price (meaning that half of the prices found were lower than or equal to 16, and half of the prices were higher than or equal to 16). At the time of the survey, this median price was equivalent to US$0.0286. The international reference price (see box on p.17) for furosemide was US$0.0047. The median price ratio pulls all this information together. The MPR for brand furosem-ide was 6.1, meaning that, in general, brand furosemide in retail pharmacies was about six times as expensive as the international reference price.

To summarise MPR results for many medicines, we can take the median MPR for a group of them. There were 10 originator brand medicines found in the Armenian survey. The brand medicine with the lowest MPR - 2.2 - was the salbutamol inhaler. The highest MPR was for brand ciprofloxacin tablets, which was 95.5 times the international reference. Among all 10 of the originator brand drugs, the median MPR was 10.4.

When comparing the prices of the originator brands to the prices of generics for many substances, we can use the median MPR for the brands and the median MPR for the generics. However, it is important to drop any medicines where either the brand or the generic was not found. (This is done automatically in a section of the Excel Workbook that comes as part of the pricing survey methodology). Only eight medicines sought by the Armenian survey were widely sold in both their originator brand version and the leading national generic equivalent. We can think of these as matched pairs of equivalent medicines. The median MPRs for the eight branded medicines and their eight leading generic equivalents are presented at the top of Figure 1. The median MPR for the brand version in the group of eight pairs was 10.4, whereas the median MPR for the generic versions of the same medicines was about 3.2. "Brand premium" usually refers to how much larger brand prices are compared to generics. For essential drugs in Armenia, a "typical brand premium" is about 330% (that is, 10.4 median MPR for brands divided by 3.2 for generics).

Figure 1 shows the median brand MPRs and median most-sold generic MPRs for eight of the pilot country studies. These results are for private sector pharmacies only. For each country, as with Armenia, the results shown are only for pairs of drugs where both the originator brand version and the most sold generic version were widely found. The specific medicines that make up these matched pairs are slightly different from country to country (and are listed in a table in the synthesis report of pilot survey findings on the HAI website at http://www.haiweb.org/medicineprices). Nevertheless, because the median is the observation in the middle of all the observations, it is fairly representative of both the survey results as a whole and the private sectors in these countries taken as a whole. These were not unusual results, but rather the most typical results found.

High brand premiums can be seen in almost all of the pilot countries in Figure 1, (the darker brand bar tends to be much longer than the lighter generic bar). In cases where the two bars are similar in length (in Brazil and the Philippines, especially), brand medicines seem to be about as expensive as they are in other countries. The small brand premium in those two countries seems to be due to the fact that leading generics are also quite expensive - with median MPRs in the range of 15 to 25 times the international reference prices. Two other country cases that stand out as unusual are Peru and Sri Lanka. Peru's generic prices tend to be high compared with other countries, while Peru's brand prices are extremely high. A look at Peru's median brand MPR indicates that innovator brand drugs typically sell at more than 61 times the international reference price. Meanwhile, Sri Lanka has the smallest MPRs in both the brand and the most sold generic categories - with 1.2 and 4.7 for median MPRs, respectively.

In public sector health systems, it is often unusual to find originator brand name drugs. Even the nationally most sold generic version may not be seen. Governments usually try to purchase the best medicine values, rather than the most popular or well-known products. For this reason, the price of the "lowest-priced generic available in the outlet" is often the most relevant measure. In the WHO/HAI pilot surveys, this was the category where the most data could be collected in the public sector. Figure 2 shows some results for four countries where prices to patients were collected from public health facilities. These median MPRs are for the all the "lowest priced generic products" that were sought and found in these countries' public clinics. Again, these were not exactly the same groups of medicines in each country. However, only the median value among the results from the field is presented. Unusually high or low values that may exist for certain medicines have very little influence on the median MPR measure. Note the scaling in Figure 2 - the horizontal axis only goes up to an MPR of 10. Public sector "lowest-priced" generic prices found in these surveys tend to be lower than the "most-sold" generic prices in the retail sectors, as shown above.

Figure 1. Medians of MPRs in private pharmacy outlets for pairs of equivalent medicines, as found in field surveys

Figure 2. Median MPRs of prices to patients in public clinics for generic medicines

Some pilot study countries collected procurement prices from the public sector. Examples of median MPRs for procurement of medicines in the "lowest priced generics" category are displayed in Figure 3 scaled for easy comparison to Figure 2. The procurement prices found in the three countries in Figure 3 appear to be generally lower than prices charged to patients in the four countries in Figure 2. We expect that there are mark-ups between procurement and sale in public sectors. However, none of the pilot countries has collected both public procurement prices and public patient prices, so we cannot yet make a direct comparison. Kenya stands out in Figure 3 for having high public procurement prices. In part, this is because the Kenya team obtained procurement data from three large public facilities that purchase medicines independently on the private market, whereas South Africa's data come from a centralized procurement system, and Brazilian data are a mixture of procurement prices paid by several public agencies with responsibility for different classes of medicines and levels of care.

Figure 3. Median MPRs of public sector procurement prices for generic medicines