A total of 949 transactions for essential cancer medication were recorded and analyzed from 2010 to 2014. The median price paid for a package of essential cancer medication was $12.63, with the lowest recorded price $0.03 and the highest recorded price $5250.00. This indicates a high degree of variation in overall essential cancer medication pricing.
Comparisons across geographies
Prices were obtained for 19 buyers representing a total of 29 different countries. Results from a Kruskal Wallis
H test suggest statistically significant differences in prices paid by different countries (
χ
2 = 148.330;
p < 0.001). The median price paid for essential cancer medication by each country is available in Table
2. The highest median price was for transactions from Namibia (
n = 119;
M = $27.75) and the lowest median price was for transactions from Ghana (
n = 5;
M = $1.10). While it is plausible that comparisons between certain countries may be biased due to the specific cancer medications purchased, the statistically significant difference in pricing detected in this analysis is supported by specific examination of a smaller sample of medications that had insufficient transactional data for statistical testing. For example, the Dominican Republic paid $4.56, on average, for 1 vial of cyclophosphamide in 2014 whereas Peru paid $9.52 for the same amount of the same drug in the same year. Similarly, Costa Rica paid $2.00 for 100 tablets of dexamethasone in 2010 whereas Namibia paid $7.96 for the same amount of the same drug in the same year. Many additional examples of the pricing disparity between countries for essential cancer medication are available from raw data in the MSH
Guide [
17].
Table 2
Median package prices of cancer medication for all included buyers, with other select characteristics
Barbados | Latin America | $14.66 | $4498 | 263.1 |
Bolivia | Latin America | $4.94 | $33,983 | 143.9 |
Botswana | Africa | $18.40 | $12,701 | 107.6 |
Costa Rica | Latin America | $21.40 | $56,908 | 179.3 |
Dominican Republic | Latin America | $6.15 | $71,433 | 153.4 |
DR Congo | Africa | $9.60 | $41,207 | 107.8 |
Ghana | Africa | $1.10 | $38,171 | 91.7 |
Guatemala | Latin America | $6.24 | $68,142 | 130.4 |
Lesotho | Africa | $6.28 | $1766 | 103 |
Namibia | Africa | $27.75 | $11,210 | 82.7 |
OECS | Latin America | $9.00 | - | - |
Peru | Latin America | $7.89 | $178,643 | 154.5 |
Rwanda | Africa | $11.45 | $8490 | 135.8 |
Senegal | Africa | $1.14 | $14,572 | 101.2 |
SICA | Latin America | $8.15 | - | - |
South Africa | Africa | $14.27 | $266,213 | 187.1 |
Sudan | Africa | $2.36 | $93,729 | 91.1 |
Tanzania | Africa | $3.19 | $45,899 | 123.7 |
Uganda | Africa | $12.17 | $24,995 | 169.7 |
While the assessment of prices across geographies allows for insight into potential pricing variation among essential cancer medications, low sample sizes did not allow sufficient statistical power for stratified hypothesis testing. While 949 prices were available from the MSH database in the five-year time period, the average geography provided only 50 prices, with the range being 2 prices (Lesotho, Senegal) to 200 prices (South Africa). As this limited set of prices was divided among 43 drugs, sufficient statistical power did not exist to control for other characteristics of a geography, such as the mix of medications purchased. Furthermore, there was a wide range of prices available per essential cancer medication examined in the five-year time period, with the range being 3 prices (trastuzumab, vinorelbine) to 62 prices (cyclophosphamide). As with geographic comparisons, the limited set of prices per drug did not provide enough data to control for potentially influential variables corresponding to the drug, such as the mix of package doses. Therefore, while our analyses generate the possibility that a given country may have purchased a given cancer medication at a significantly different price than another country, other explanations for the observed variation are also possible.
To partially address this limitation in data availability, a sub-analysis was carried out to more closely examine the variation in transactional prices among buyers for certain individual cancer medications. This sub-analysis was conducted on cyclophosphamide (a cytotoxic drug used since 1959 for treatment of both blood cancers and solid tumors), which included the greatest number of package prices in the dataset during the five-year period (
n = 62). Additionally, cyclophosphamide was available as a generic during this entire time frame (i.e. generic versions have been available since 1999) [
26]. Buyers with less than five transactions recorded were excluded, leaving four buyers (Barbados, Namibia, the OECS multinational buyer, and South Africa) with 46 transactions. Purchases of tablets were excluded, allowing for the remaining 31 transactions to be normalized to the price per vial. A Kruskall Wallis
H test among these 31 transactions indicated a statistically significant difference in prices (
χ
2 = 9.105;
p = 0.028), with the median price being $3.85 per vial for the OECS (
n = 5), $5.29 per vial for Barbados (
n = 5), $9.46 per vial for Namibia (
n = 12), and $10.00 per vial for South Africa (
n = 8). Further attempts to normalize to vial strength were limited due to sample size, resulting in insufficient statistical power for further testing.
To further address this data limitation, we also examined pricing data across the different non-tablet formulations of cyclophosphamide (200-mg vials, 500-mg vials, and 1-g vials) and observed appreciable variation in mean transactional prices: $0.006 per milligram for the OECS (n = 5), $0.012 per milligram for Barbados (n = 5), $0.015 per milligram for South Africa (n = 8), and $0.126 per milligram for Namibia (n = 12). Furthermore, when assessing only the price paid for the 1-g formulation (which provided the greatest number of transactions compared to the other formulations), we continued to observe noteworthy variations in transactional price: $6.39 per 1-g vial for Barbados (n = 3), $11.17 per 1-g vial for Namibia (n = 4), $3.85 per 1-g vial for the OECS (n = 5), and $13.02 per 1-g vial for South Africa (n = 4).
In regards to regional variation, transactions were recorded from 11 countries in the UN African region and 8 buyers (6 countries and 2 international organizations) in the UN Latin America region. Results from a Mann-Whitney U test suggest that transactions from African buyers exhibited a significantly higher price than transactions from Latin American buyers (z = −3.468; p = 0.001). The median price paid for essential cancer medication for an African buyer was $15.03 (n = 530), while the median price in these transactions for a Latin American buyer was $10.89 (n = 419). The significant difference in prices paid for essential cancer medication between these two regions further underscores a potential for pricing and procurement variation based on a country’s geographic region and possibly its procurement relationships.
Linear regression found no statistically significant relationship between country-level GDP (nominal) and median prices paid for essential cancer medication (β = −4.778 × 10−6; p = 0.867; R2 = 0.002). Linear regression also found no statistically significant relationship between country-level all-cancer incidence and median prices paid for essential cancer medication (β = 0.040; p = 0.347; R2 = 0.059).
Comparisons across medications
Results from a Kruskal Wallis
H test suggest statistically significant differences in prices paid for different medications (
χ
2 = 514.493;
p < 0.001). The median price paid for each essential cancer medication is available in Table
3. The lowest median price paid for an essential cancer medication was for allopurinol (
n = 17;
M = $6.40) and the highest median price paid for an essential cancer medication was for trastuzumab (
n = 3;
M = $1800.00). Rituximab (
n = 12;
M = $413.53), capecitabine (
n = 22;
M = $354.04), and tioguanine (
n = 6;
M = $102.83) also exhibited median purchase prices exceeding $100 per package. As these medications are intended to be considered essential for health systems, prices over $100 per package likely impacts access for patients and coverage of these products by national health systems. Though efforts have been taken by some national governments to ameliorate the high price of trastuzumab [
27,
28], these efforts have been geographically limited [
29], with generally insufficient options for residents of developing countries to gain access to this treatment. It is worth noting that trastuzumab is a treatment for HER2-positive breast cancer, a form of breast cancer diagnosed for several hundred thousand women each year [
30].
Table 3
Cancer medications in the 19th EML with median purchase price and formulations available
Allopurinol | $1.56 | X | |
All-Trans Retinoid Acid |
Not Available
| X | |
Anastrazole | $6.40 | X | |
Asparaginase | $42.58 | | X |
Azathioprine | $14.28 | X | X |
Bendamustine |
Not Available
| | X |
Bicalutamide | $17.61 | X | |
Bleomycin | $18.24 | | X |
Calcium Folinate | $15.98 | X | X |
Capecitabine | $354.04 | X | |
Carboplatin | $26.49 | | X |
Chlorambucil | $32.83 | X | |
Ciclosporin | $88.64 | X | |
Cisplatin | $8.47 | | X |
Cyclophosphamide | $13.00 | X | X |
Cytarabine | $4.27 | | X |
Dacarbazine | $14.50 | | X |
Dactinomycin | $25.23 | | X |
Daunorubicin | $6.08 | | X |
Dexamethasone | $2.01 | | X |
Docetaxel | $44.65 | | X |
Doxorubicin | $7.53 | | X |
Etoposide | $4.57 | X | X |
Filgrastim | $52.11 | | X |
Fludarabine | $91.46 | X | X |
Fluorouracil | $1.86 | | X |
Gemcitabine | $25.85 | | X |
Hydrocortisone | $1.96 | | X |
Hydroxycarbamide | $16.67 | X | |
Ifosfamide | $25.01 | | X |
Imatinib | $75.41 | X | |
Irinotecan |
Not Available
| | X |
Leuprorelin |
Not Available
| | X |
Mercaptopurine | $55.90 | X | |
Mesna | $3.47 | X | X |
Methotrexate | $9.00 | X | X |
Methylprednisolone | $9.06 | | X |
Oxaliplatin | $44.55 | | X |
Paclitaxel | $14.85 | | X |
Prednisolone | $7.18 | X | |
Procarbazine |
Not Available
| X | |
Rituximab | $413.53 | | X |
Tamoxifen | $5.24 | X | |
Tioguanine | $102.83 | X | |
Trastuzumab | $1800.00 | | X |
Vinblastine | $14.75 | | X |
Vincristine | $3.53 | | X |
Vinorelbine | $18.10 | | X |
Transactions were recorded for 43 of the 48 medications (90%) in the “Antineoplastic and Immunosuppressives” category of the 19th EML, 81 of the 167 medications (49%) in the “Anti-Infective Medicines” category, and 18 of the 28 medications (64%) in the “Cardiovascular Medicines” category. Results from a Kruskal Wallis H test suggest statistically significant differences in prices between categories (χ
2 = 108.421; p < 0.001). The median price paid for a package of essential cancer medication in 2014 was $9.31 (n = 204), which is approximately 4 times higher than the median price paid in 2014 for a package of essential infectious disease medication (n = 457; M = $2.45) and 5 times higher than the median price paid for a package of essential cardiovascular disease medication (n = 129; M = $1.73). This analysis indicates that the median prices for acquiring essential infectious disease medication and essential cardiovascular disease medication are much lower than the median price of acquiring essential cancer medication.
Results from a Mann-Whitney U test suggest that prices paid for injectable cancer medications were not significantly different than prices paid for oral cancer medications (z = −0.205; p = 0.837). The median price for a package of injectable medication was $12.17 (n = 610) and the median price for a package of oral medication was $13.31 (n = 339). About two-thirds (64.3%) of transactions were for injectable cancer medications.
We also investigated the role of generic/biosimilar availability on transactional prices of essential cancer drugs. Among the 43 drugs with available prices, 34 had generic versions available throughout the entire 5-year period from 2010 through 2014 (median price = $10.62), 3 had no generic versions throughout this entire period (median price = $29.35), and the remaining 6 had a generic approved during the 5-year period. Drugs that became available as generic are anastrozole (generic available in 2010), gemcitabine (2010), docetaxel (2012), capecitabine (2012), rituximab (2013), and imatinib (2013). Only docetaxel and gemcitabine had greater than 5 package prices available in each of the following two categories: (1) the years from 2010 up through the generic approval year and (2) the years after the generic approval year. The year of approval was included in the “before” category, as this would allow for a short period of time for international buyers to adapt to ordering from new drug manufacturers. When using Mann-Whitney U tests to compare prices of these drugs before and after generic versions were approved, we found that docetaxel exhibited a statistically significant decrease in median price from $53.84 to $27.89 (z = −2.134; p = 0.033) and capecitabine exhibited a non-significant decrease in median price from $362.84 to $209.68 (z = −1.280; p = 0.201).
Comparison across time
Results from a Kruskal Wallis H test suggest that, between 2010 and 2014, prices paid for essential cancer medications did not significantly differ by year (χ
2 = 3.497; p = 0.478). Indeed, prices did not appear to exhibit a clear longitudinal trend, with the median price for essential cancer medication being $12.41 in 2010 (n = 216), $14.90 in 2011 (n = 150), $14.77 in 2012 (n = 131), $12.81 in 2013 (n = 248), and $9.31 in 2014 (n = 204).