In recent years, many commercial and non-commercial websites reporting medicine price have become available on the internet. However, both the quality of these websites, and the availability of descriptive information about the prices to allow appropriate interpretation, can limit their usefulness. I conducted a qualitative analysis of 71 non-commercial, publicly available websites from both national (n=50) and international (n=21) price sources. Of all websites the quality of the site in general and the quality of the price data reported were assessed, using general and price specific criteria. Most of the websites were fast and user-friendly, but many received a low score for important criteria such as publication of the date of the last update of site and price information, and search ability within the price information. Various types of medicine prices were reported (e.g. ex-factory, procurement, wholesale, reimbursement, retail/patient prices) therefore a specific checklist for every price type was designed to assess the comprehensiveness of the price information. It was found that the websites often published price information that lacked key information to perform valid international price comparisons. For example, to improve the interpretability of price data it is recommended that websites publish information on the components of each price (e.g. if VAT/GST is included or not, and if so the amount of the tax). Other recommendations based on the results included the listing of medicines by INN name (rather than by brand name) and the exchange rate to US $ applicable at the date of price information posting.

Secondly, a quantitative analysis of medicine prices found on national websites was conducted to assess the feasibility of comparing web-based medicine price information from different sources. The analysis was performed on the prices of 7 medicines selected for their wide range of therapeutic effects and the expected high availability of price data (e.g. amitriptyline 25mg, atenolol 50mg, ciprofloxacin 500mg, glibenclamide 5mg, omeprazole 20mg, salbutamol 100mcg/dose and simvastatin 20mg). Actual price information can be compared across various price sources but a lack of comparability of price data components across sites limits the usefulness of these price comparisons. Therefore data was adjusted (e.g. taxes and other fees excluded) where possible to improve the comparability of price data reported from different sources; the seven medicine comparisons conducted in this study showed however that this is very demanding work.

More extensive comparisons using all national websites would be even more demanding; therefore, the use of international websites such as the Hungarian CEDD website should be considered for price comparisons. In addition, although a comparison of prices is possible using a small selection of national websites, a reasonable outcome can only be expected when using multiple sites. However, large variations were still found in the prices of individual medicines; this variation was often greatest for procurement, reimbursement and retail/patient prices. While some price variation would be expected due to comparability issues, these and other national issues alone would not account for the large scale price variation found in the 7 medicines comparison study. With the exception of Tamil Nadu procurement prices (reporting 5 out of the 7 medicines), there was no consistency as to which website reported the lowest prices. In addition, there was little correlation between national GDP (Gross Domestic Product) per capita and prices. Thus, it can be concluded that prices for the same products vary widely across countries, even within similar countries (e.g. similar WHO region and country income class).

Looking at the price comparisons in general as well as their results it is surprising that although medicine price information appears to be widely available on the internet, great variations in prices still exist.

As a last step, two brand medicines were compared to their generic therapeutic equivalent substance to investigate the potential cost savings of generic therapeutic substitution. Lipitor 20mg (atorvastatin) was compared to simvastatin 20mg, while Nexium 20mg (esomeprazole) was compared to omeprazole 20mg. it was seen that all of the prices of Lipitor and Nexium were far higher. In fact, differences up to 214-fold for Lipitor compared to simvastatin and differences up to 41.1 for Nexium compared to omeprazole were reported. While Lipitor and Nexium may be marginally more effective when the same doses are compared these differences are in the range of 20 to 50 % and not many fold differences as were found for the prices. These substantial price variations clearly showed the potential cost savings of generic therapeutic substitution of simvastatin for Lipitor and omeprazole for Nexium.