In this analysis, we explored differences in the ex-manufacturer and retail prices of generic drugs across seven countries in 2013 using various price indices.
The ex-manufacturer and retail prices varied widely across countries. This is consistent with earlier studies comparing the prices of patented drugs at both levels [1, 13, 14]. More research is needed to disentangle the impact of supply- and demand-side policies, such as pricing, reimbursement, prescribing, and substitution rules, on the ex-manufacturer and retail prices of generics [26]. Prices variation is also likely due, in part, to differences in the regulation of wholesaler and pharmacy margins [1].
There are various methods for comparing drug prices across settings [25, 27], and they often produce remarkably different results. For example, the ex-manufacturer Laspeyres index (dose) in Table 1 suggests that the sample of generic drugs was about 60% more expensive in Italy than in Germany. On the other hand, the ex-manufacturer Paasche index (grams of active ingredient) indicates that the sample was about 35% cheaper in Italy than in Germany.
There were even larger differences between some of the weighted and unweighted indices. It might be especially important to use weighted indices when comparing generic drug prices, since studies suggest that these prices are closely linked to volume [28, 29]. Earlier studies have shown that the results of unweighted and weighted indices can differ sharply [4, 25], which is consistent with our findings. Extreme prices can skew the results of unweighted indices, so these indices are generally considered less reliable than weighted ones for comparing drug prices [25].
There is no consensus on which weighting method is most appropriate for comparing drug prices, as each has advantages and disadvantages [12, 25]. Academic and government studies have variously calculated unweighted [9, 10], Fisher [11], Paasche [4, 25], and Laspeyres indices [4, 17, 25, 30], often using different units of volume and/or base countries. The likely reason why Paasche results are usually lower than Laspeyres results, a finding which has been reported in previous drug price indices [4, 25], is that patients tend to consume more of the drugs that are cheaper in their countries. Therefore, when prices are weighted by local consumption, the indices show lower average prices — relative to the base country — than when prices are weighted by consumption in the base country.
The choice of unit of volume can influence the results if there are large, systematic differences between countries in the average strength per dose [25]. For example, previous studies have found that price-index results for Japan vary significantly depending on whether number of doses or grams of active ingredient serve as the unit of volume [3, 4, 17, 18, 25]. The authors of those studies attributed this finding to the tendency of Japanese clinicians to prescribe higher quantities of lower-strength products.
Despite such methodological challenges, it is still possible to glean useful information from price indices. In particular, it is important to look for consistency across indices. As an example, our results indicate that Denmark and Sweden had the lowest ex-manufacturer prices in nearly all weighted indices, regardless of whether Germany or France served as the base country. This strongly suggests that generic drugs were cheaper in Denmark and Sweden in 2013 than in the other five countries. By contrast, the French and Italian ex-manufacturer prices were among the highest in all weighted indices. Ideally, the results of price indices should be interpreted alongside other quantitative and qualitative data about the impact of individual policies on drug prices. On their own, price indices do not provide causal evidence on the effects of pricing and reimbursement rules, generic substitution laws, and other factors on the prices of generic drugs.
The findings in this study raise questions which merit further research. Both Sweden and Denmark operate tender-like systems for generic drugs,Footnote 7 which may account for the low prices observed in each country [31, 32]. Tendering refers to the bulk purchase of generic drugs from the manufacturers that offer the lowest prices [33]. More work is needed to understand the impact of tendering on drug prices, and whether any observed price reductions can be sustained over time. There is concern that relying exclusively on tendering to procure generic drugs could create product shortages, drive generic drug firms out of business, and lead to higher generic drug prices over time [33]. There is little evidence, however, on the long-term effects of tendering.
It is also important to examine why there are large differences in the prices of drugs in various therapeutic areas, both within and between countries. Such variation may, in part, reflect market factors. For example, the marketing exclusivity for a drug can expire at different times across high-income countries depending on when the drug was approved in each jurisdiction. Also, some studies have observed an inverse relationship between the number of competitors in the market and generic drug prices [34, 35]. The speed of generic entry, in turn, has been found to be correlated with how much brand-name firms record in revenue in the years leading up to patent expiry [36, 37]. In other words, generic firms tend to prioritize more lucrative drug markets.
Limitations
This study has limitations, most of which are inherent to drug price indices.
First, the data did not account for confidential discounts, which can be as high as 50% for some generic drugs in certain countries [38]. All list prices may, therefore, not have corresponded to the actual prices paid. However, if profits from discounts accrue to wholesalers or pharmacists, then list prices are more important to payers.
Second, Paasche and Laspeyres indices are underpinned by assumptions about the relationship between generic drug prices and usage which may not always hold. Specifically, the results of Laspeyres indices are valid if demand for prescription medicines is price inelastic. While empirical findings contradict this assumption [39, 40], the Paasche index instead assumes that the consumption pattern in the base country would look exactly like that of the comparator country if both had the same prices. The latter assumption might be less likely to hold true, since there are differences between countries in standards of care, disease prevalence rates, prescription drug coverage, and patient preferences — all of which can affect demand [25].
Third, by restricting the analysis to a common sample of drugs, we reduced the sample size. In some previous price indices for patented drugs, researchers instead conducted a series of comparisons between the base country and one other country at a time, looking at the drugs available in both countries. Such comparisons, which are called bilateral analyses, maximize the sample size for each country pair. We chose to instead calculate what are known as multilateral indices, which compare the prices of a sample of drugs available in all study countries. Multilateral indices provide information on how prices compare across all the countries rather than just between each pair. While a common sample might over-represent older, internationally available products [25], this is less of a concern when comparing generic drug prices. However, it is important to note that two countries with identical prices could show up as having differing price levels in a Paasche index if consumption patterns differ. Thus, multilateral price comparisons using Paasche indices should be interpreted with caution.
Fourth, we used common units of volume to aggregate data across formulations of active ingredients [25]. In using prices per dose, however, we assumed that a dose of a drug provides the same therapeutic benefit to any patient, regardless of strength-form combination. By contrast, prices per gram of active ingredient are sensitive to the selection of drugs, given that drug strengths often vary considerably between drugs [11]. The price per defined daily dose is an alternative metric. A defined daily dose is the “assumed average maintenance dose per day for a drug used for its main indication in adults.” [41] We could not identify this dose for each drug in our dataset, as we did not have information about drug indications. However, defined daily doses are not always of equal therapeutic value to all patients, and they may not accurately reflect consumption patterns [25]. For example, a defined daily dose is not adjusted for differences in the duration of treatment. They are, therefore, not necessarily a better unit of comparison than doses or grams of active ingredient [25, 41]. Also, because defined daily doses are specified in terms of grams of active ingredient per day, indices based on defined daily doses and indices based on grams should generate similar findings if the average number of treatment days are fairly consistent across countries for most drugs [4].
Fifth, the drugs were listed by active ingredient, and no information was available on the indications for which the drugs were prescribed. However, a prior study found that the results of price indices were “virtually unchanged” when products were defined by active ingredient instead of active ingredient plus indication [25].
Lastly, we had to exclude 6.7% of drugs (213/3156) due to missing volume data.