JEANNE MADDEN

J. Madden

The field survey process in brief

ONCE local investigators decide to undertake a WHO/HAI medicine prices survey, the bulk of their work revolves around the collection of price data from medicine outlets. A selection of outlets is visited in different health "sectors". For example, in most studies, the study team will look at prices to patients at twenty or more government-sponsored clinics and also at twenty or more private for-profit pharmacy shops. Often another important sector exists that should be included in their study, such as charitable or mission facilities or parastatal cooperatives, or health facilities created for employees of a major industry. The survey manual provides guidance on choosing sectors and then identifying the sample of outlets to be visited.

The WHO/HAI survey targets a set of 30 specific essential medicines, such as cotrimoxazole syrup and diazepam tablets. Study teams are encouraged to add other medicines of local interest to this core list. Data collectors on the local study team visit the medicine outlets and ask to see each of the medicines in three precisely-defined versions. First is the innovator or originator brand version of the medicine. Next is the generic version of that medicine that is most commonly sold in the country or region being surveyed. (This "generic" version may in fact have a brand name of its own, but it cannot be the innovator product. Its name and manufacturer are identified by the lead investigators before field work begins). Finally, the survey gathers prices for whatever happens to be the lowest-priced generic version seen in each outlet. (In some outlets, this generic may be the same as the nationally "most sold" generic. It may have a brand name. But it is never the innovator product). In large for-profit pharmacies, there could be many different products that are the same substance and strength. So, surveys of the for-profit sector frequently find a different product fitting each of the three versions. However, in public facilities there is often just one simple generic version of a medicine. If the innovator or most sold version are not found at an outlet, the corresponding spaces on the data collection forms are left blank.

Prices gathered on paper data collection forms in the field are checked for mistakes and then entered into a computer spreadsheet. There is a Microsoft Excel Workbook of spreadsheets that have been specially designed for the WHO/HAI survey. The Workbook provides a clear structure for entering all the field data. Data are entered twice, to ensure good quality. After data entry, several key analyses are produced automatically and immediately.

Table 1
Sample data from the Excel Workbook for the Kenya survey 2001

Analysis of results for specific medicines

Table 1 shows a small portion of the Excel Workbook, after some data have been entered. These data are from the 2001 Kenya medicine prices survey. Normally, the header and the first nine columns of this section would appear with a blue background. These columns identify the surveyed products and present some automated analyses. The last two columns (headed "1" and "2") would have a white background, and this is where field data are entered by the local study team. The numbers below the header in the last two columns are prices for one unit of medicine, in the local currency. Thus, in the first outlet surveyed, a 1 gm vial of innovator brand ceftriaxone cost 2300 Kenyan shillings (KSh). The price in the second facility was considerably lower (1890 KSh). The generic forms of ceftriaxone were cheaper still in the first facility (both were 1400 KSh), but were not available in the second facility. A total of 26 for-profit facilities were surveyed in Kenya and all their data were entered in this spreadsheet page, but the remaining columns have not been shown here.

Once data are entered, the column headed "% with med." instantly presents results on medicine availability. We can see that half of the surveyed shops had the innovator brand of ceftriaxone in stock. The generic forms were less often found - about 19% of shops had the leading generic brand, and 35% had at least some generic version.

Columns 4 to 8 in Table 1 summarise price data from the field. These numbers are not in local currency, however. These are price ratios: local prices converted to US dollars and then divided by an international reference price. (See explanatory box on p.17). The most expensive ceftriaxone found was in that first facility, 2300 KSh, which in 2001 was equivalent to US$29.15. The international reference price (from bulk generic distributors) for ceftriaxone was US$1.48 per vial. (Reference prices are entered and visible elsewhere in the Workbook). The maximum price ratio for innovator ceftriaxone in the survey, seen at the top of column 8, therefore was 29.15/1.48 = 19.76. In other words, the maximum retail price found in any shop was about 20 times the international bulk generic price. The minimum price ratio seen for innovator brand ceftriaxone was 16.24 (shown in Column 7). The remaining price ratios were of course between the maximum and minimum. The Workbook also calculates the 25^th and 75^th percentile of the price ratios, which together mark off a more "usual" or central range. Finally, the median price ratio (or MPR, Column 4) is the result which is most representative among all the shops visited.

Generic versions of surveyed medicines are almost always less expensive than the innovators. We can see this pattern in Kenya's results in Table 1. Prices for the leading generic competitor to innovator brand ceftriaxone are typically only two-thirds as high (MPR 12.03 versus MPR 18.56). Finally, the presence of additional generics on the market results in a median "lowest price" for generic cef-triaxone that is less than half of the median for the innovator brand (MPR 8.59).

For ciprofloxacin tablets, the contrast between innovator brand prices and generic prices is even sharper. The "lowest price" generic version in retail shops costs about 11 times the international reference price, while the innovator brand costs about 123 times the reference. (There were too few observations of the leading generic ciprofloxacin product in this survey, so summary statistics were not automatically calculated).

In another section of the Workbook (not shown), and again in an automatic analysis of field data, MPR results from different sectors are compared, product by product. In the case of Kenya, we learn that in the NGO sector, patient prices for the innovator's ceftriaxone were somewhat higher than at private retail outlets (MPR 23.33 versus 18.56), but innovator brand ciprofloxacin tablets were slightly cheaper in the NGO sector (MPR 117.79 versus 122.98).

Summarising results for many medicines within a sector

The WHO/HAI medicine prices survey can produce a fairly accurate picture of prices in an entire sector by sum-marising results for all the medicines targeted in the survey. Again, the main measures used are ratios of local prices to international reference prices. Table 2 shows a portion of the Workbook that automatically presents summary analyses for a sector. In this instance, it is the private for-profit sector in the Philippines. The energetic Philippines study team collected data from 77 retail outlets. They found ample price data there for 21 innovator brand products, 9 nationally "most sold" generic products, and 15 "lowest-priced" generic products.

The simplest way to make an overall comparison between, say, innovators and generics would be to take the median of the MPRs for all the innovator products, and compare that to the median of MPRs for all the "lowest priced" products. The Workbook does compute and display these two results. However, it is not wise to use this raw comparison. Differences in the availability of innovator and generic products usually result in different sets of products being. (Clearly, if there are 21 innovators and only 15 generics, then there are at least 6 additional medicines in the innovator group that were not found in the generic group). In place of the unwise comparison, the Workbook offers the analysis shown in Table 2.

There are three pairs of boxed columns in Table 2. In each pair of columns, summary price ratios for pairs of equivalent products are compared, for the private for-profit sector only. There were just nine surveyed medicines that were found in both their innovator brand and their nationally most sold generic versions. Statistics for these two types of medicines appear side by side in the first two boxed columns of Table 2. The median among the nine MPRs for the innovator versions of the medicines was 18.28. The median among the nine "most sold generic" MPRs was 16.21. So we can say that in the Philippines private sector, innovator brand medicines are estimated to cost about 13% more than their most popular generic equivalents.

In the 3^rd and 4^th boxed columns in Table 2 (headed "Brand" and "Lowest Price"), prices for equivalent groups of innovator brand medicines and lowest price generic versions are compared. For this comparison, 14 medicines were found to be widely available in both categories. The survey found that innovator brand medicines in the Philippines for-profit sector cost, overall, about twice as much as equivalent lowest-price-in-shop generics (median of MPRs 15.37 versus 7.69). Finally, in columns 5 and 6 (headed Most Sold and Lowest Price) a comparison is made between paired generic products. For these nine pairs the most sold generics were more than twice the price of the lowest price generics (16.1 versus 6.77). The implication of this finding is that in the Philippines at least, choosing between generics is very important.

Comparing summary results among different sectors

The Workbook also makes comparisons among different sectors, based on information summarised for many medicines. On the left side of Figure 1 is a view of some tables in the Workbook which present instant cross-sector analyses. On the right side of Figure 1, portions of these analyses have been presented graphically for the purpose of this article. (At present, there is no graphical capacity in the WHO/HAI Workbook. However, graphs such as those in Figure 1 can be produced quickly and easily in Excel by local investigators).

In a manner like that described earlier (for many medicines within a single sector), the cross-sector analyses shown in Figure 1 compare only similar groups of products. These are data from the 2002 survey in Peru. Medicine prices forpatients were collected in the public sector, the private for-profit sector, and an NGO sector. Innovator brand versions of the surveyed medicines were available only in the for-profit sector, so no cross-sector analyses could be produced for innovator brands. Only a few of the nationally most sold generic medicines were found outside the private sector, so it is preferable to compare "most sold generics" among sectors on a medicine-by-medicine basis. (Because two product examples are not sufficient for drawing solid conclusions about whole sectors).

However, the Peru study produced plentiful and interesting results in terms of the category of lowest priced generics found at surveyed outlets. In the first "lowest price" row on the left side of Figure 1, we see a summary comparison of public sector and NGO sector prices for lowest-priced generics. Thirteen surveyed medicines were found in generic versions in both sectors. (If there were several generic versions of a substance available in a single outlet, then the lowest price among these was recorded as "lowest price"). The median of the MPRs for the 13 medicines in the public sector was 3.37. That is to say, in general in the public sector, patients pay a little more than three times the international reference price for essential medicines. For the same medicines in the NGO sector, patients pay a little more than double the international reference price (median MPR 2.18). Therefore, for matching groups of equivalent medicines, the Peru study found that the NGO sector was less expensive for patients than the public sector. The Workbook also automatically expresses this comparison as a ratio of NGO sector prices to public sector prices: 64.8%. On the right side of Figure 1, this cross-sector analysis has been presented graphically.

Table 2
Private Sector Medicines Outlets (n=77 in survey)

Includes Both Core and Non-Core Medicines (n=29 on list)

Analysis Includes Only Medicines With Prices Found for Both Types in Pair

 

Reference Price Data Used = MSH

Figure 1
Comparisons of Median MPRs for Medicines With Prices in Both Sectors

Includes Both Core and Non-Core Medicines (n=25 on list)

 

Reference Price Data Used = MSH

Graphical Comparison: Median MPRs for Matching Groups of Lowest-Price Generic Medicines, Sector vs. Sector

The second "lowest price" row on the left of Figure 1 presents a cross-sector comparison of the public sector to the private for-profit sector. There were 14 medicines widely available in generic form in both sectors. Because the group of medicines used in this comparison is different and slightly larger than in the first cross-sector comparison, the median of MPRs for the public sector has changed slightly, to 3.66. For the same 14 generic substances in the for-profit sector, the median of MPRs was 7.92. Thus, generics in Peru's for-profit sector cost more than twice as much as they do in the public sector. (Or, in other words, retail medicine prices are estimated to be 216.3% of government prices). Again, this contrast is shown graphically on the right.

Finally, along the bottom of Figure 1, we find a comparison of generic medicine prices for the private for-profit sector versus the NGO sector. For this analysis, there were 15 surveyed substances that were widely available in both sectors. Again the median MPRs shift slightly from the earlier analyses, because there are more data included, but the overall results are consistent. The private sector has the most expensive generics, followed by the public sector. The least expensive medicines in Peru are apparently offered in the NGO sector.

Other automated analyses in WHO/HAI medicine price surveys

In addition to patient price data collected in the field at health facilities and pharmacies, the WHO/HAI survey aims to collect procurement price data at the central level. Procurement prices are usually obtained from large recent government orders. These are entered into the survey Workbook in the same way as field outlet prices are. Subsequently, procurement prices are automatically compared to international reference prices (as above, in the form of price ratios). This is an excellent tool for evaluating the efficiency of public bidding processes for medicines. Other interesting analyses of procurement prices include medicine-to-medicine comparisons, innovator brands versus generics, and examinations of how much variation there is among prices found for the same substance (e.g., from one order to the next). Procurement prices (which are at the start of the supply chain) can also be compared to patient prices (at the end), to quickly estimate the size of price mark-ups.

Data subsetting is a useful tool for all types of price data entered into the survey Workbook. Before using any automated analysis, the local investigator can decide whether to include data from all the outlets (or procurement orders) in a sector, or whether to select a limited subset. Subsetting is done by using a simple "switch" that appears in the Workbook above each facility ID. For example, within a private for-profit sector, medicine prices to patients are often collected from both hospitals and pharmacies. An investigator may wish to compare results for these two different types of private facilities. To do so, the investigator first includes only hospital outlets, then prints out the resulting price ratios and sector summary statistics (or copies them into a separate spreadsheet). These results are only for hospitals. Next, the investigator un-selects the hospitals, and selects the remaining outlets, which are the pharmacy shops. The second set of results is printed, or copied into the separate spreadsheet. Finally, the two sets of results are compared, either manually or with the help of a spreadsheet. This sort of subsetting is also useful for comparing results among different provinces, for rural facilities versus urban facilities, or for simply focusing on specific outlet types.

Having the data collected, analysed and produced in a standard manner allows policy-makers and programme managers to ask the questions that can lead to improved prices. For example, based on the three country examples in this article, Kenyan policy-makers can ask "Why should brand name products be so much more expensive than generic products? Are the taxes, duties or markups different?". For the Philippines a key issue would appear to be the high price of the most sold generics, which are closer to brand name prices than to the lowest price generics. For Peru, the question could be why the NGO sector is selling the lowest price generics cheaper than the public sector. All of the analyses are designed to provide decision-makers with information that they can use to intervene to make essential medicines more accessible by reducing their price.