- Global Pharma News & Resources

Generic Pharmaceutical Patent Searching

Generic Pharmaceutical Patent Searching


'Drug In Focus' is written by Leighton Howard, a patent information expert with extensive generics experience. He has worked within the professional patent information industry and major generic pharmaceutical firms and now provides consultancy services to the generics industry and government organisations via his company XIP Pty Ltd. He is also the founder of GenericsWeb and remains instrumental in maintaining the high quality and accuracy of data contained in Pipeline Patent Intelligence.
Last Updated: 27-Aug-2010

As a provider of comprehensive patent intelligence to generic pharmaceutical companies in over 25 countries, GenericsWeb frequently fields questions about how we identify and select relevant patents for inclusion in our Pipeline Patent Intelligence. This article is not intended to explain exactly how we conduct our searches, but to summarise the various options and dispel any myths. In doing so, I will first discuss the importance of defining the scope of the search and then contrast the various techniques that are commonly used.

Defining the Patent Information Requirements
Firstly, as with any patent search, we need to define the scope of the search to be undertaken. Various search parameters are shown in table 1, which should always be used prior to forming a search strategy.

Table 1: Parameters to Define Patent Information Requirements

Search ParameterDetermines:
Geographical ScopeWhich patent publication country/territory codes need to be searched.
Subject Matter ScopeWhat subject matter is the search strategy intended to identify.
Inclusion PolicyWhat subject matter is required to be reported as relevant and what is to be specifically excluded.
Date RangeWhether the search covers all publication dates (Novelty), publications prior to a certain date (Validity), patents that may still be in force (Infringement) or very recent publications (State of the Art).
ComprehensivenessWhether the search must find all relevant patents, the most relevant, an exact disclosure or a general sample within the subject matter scope.
ACCURACYHow often you need to update the information to ensure it remains relevant.
BudgetWhether or not you can afford to conduct the search defined by the above parameters.
Because patent searching is a non-exhaustive task (it is not possible to physically cite every published patent), and one that should be undertaken by professionally trained searchers, a trade off always occurs between the one or more of the first five parameters and the budget.

Determining What’s Important
In considering a search trade-off, a good piece of advice is to consider the costs of not identifying the patent information that you specifically seek. This will help determine both the budget and where you can afford to cut corners on your search strategy.

“In considering a search trade-off, a good piece of advice is to consider the costs of not identifying the patent information that you specifically seek”

For generic pharmaceutical developments, the costs of not identifying the required patent information can be significant due to the following potential outcomes:

• Executives spending time filtering through numerous irrelevant patents.
• Patenting a possible invention that has been prior disclosed and therefore lacks novelty.
• Increased production cost-base to avoid technology claimed in a patent that has expired, lapsed or been withdrawn.
• Changing the development mid-way due to later identification of constraining patents.
• Failure to out-license products in export territories due to later discovery that the processes and or products would be infringe patents.
• Infringing patents on products that are heavily protected by the innovator and/or generic competitors, resulting in possible product withdrawal and/or litigation.

The costs may also be implicit, in terms of low customer perception of your company’s ability to develop non-infringing generics and resulting reduction in shareholder value.
There are therefore very few areas where the patent search for generic pharmaceuticals can be compromised in the trade-off process. Patent information needs to be comprehensive, given the subject matter and geographical scope, but contain only relevant patents. It also needs to contain patents and applications of all statuses from a large date range and be kept up to date. So, given that you need all of this and that you have a limited budget, the question of how to go about finding this information arises (keeping the information up to date will be discussed in a future article).

Selecting a Data Source
In preparing a search strategy, the first stage is to select the type of data sources. The sources selected should primarily be able to accommodate the scope of the search, but should also match the skills of the searcher. True patent searching is considered a skill, requiring a combination of knowledge drawn from information science, foreign languages, patent procedures and a technical background.

Broadly speaking, there are two alternative types of patent data available to a searcher, raw data and indexed data. These will be explored in more detail below.

We define Raw Data as being that published directly by the patent offices. This includes such fields as applicant, abstract, patent classification code and, nowadays, the full text of the description and claims1.

Raw data can be searched using a variety of database interfaces, some of them are free to access, such as the EPO’s Esp@cenet, but offer limited searching functionality. In order to construct the complex search queries required and manage the large numbers of patents to be reviewed for pharmaceutical patent searching, more professional databases are required such as Minesoft’s PatBase.

The principle of searching raw patent data is that the searcher refines the available data down to a set of patents that contains all of those that are likely to fall within the subject matter scope (comprehensiveness) using a combination of the available fields. This set of patents contains a significant amount of ‘noise’, which is inevitable in building a comprehensive strategy. This set of data is then filtered manually using the inclusion policy to identify the relevant patent information (relevance).

Preparation of indexed data involves the ‘provider’ taking a set of raw patent data has been reviewed (manually, by computer or by a combination of the two) and assigned one or more ‘labels’ according to a defined index. In the case of pharmaceutical patents, the index may be a chemical abstract number or an INN. The principle of identifying patents using indexed information is that the user retrieves it in a standardised fashion, with reference to the relevant indexes. Due to the higher costs of indexed patent information, it does not lend itself to searching using the same methods as for raw data, therefore the level of relevance and comprehensiveness is reliant on the indexation of the information by the provider.

“True patent searching is considered a skill, requiring a combination of knowledge drawn from information science, foreign languages, patent procedures and a technical background”

Weighing Up the Pros and Cons
Speaking as a professional patent searcher, it is my opinion that a patent searcher who uses indexed data as their only data source is not conducting a true patent search, they are simply retrieving the results of another person’s analysis or, at worst, that of a computer. This is not to say that this would be an inappropriate method, on condition that the provider who carried out the indexation used the same the parameters (as outlined in table 1) as are appropriate for your specific search requirements. Failure to ask the right questions of the data provider in regard to indexation could result in patent information that is totally unsuitable for your requirements. For example some indexation has only been undertaken on a particular date range of patent publications, and others only on pre-selected sub set of raw patent data – this could lead to inherent limitations on the resulting indexed data.

In contrast, the options for searching raw data are almost unlimited, resulting in the ability to customise the search to meet the specific search parameters. However, searching of raw patent without applying the correct methodology and understanding the content and limitations of the data is unlikely to provide appropriate results. To emphasise this point, GenericsWeb conducted some analysis of a commonly heard example of searching for generic pharmaceutical patents by typing the drug INN as a keyword into full-text, raw patent databases. The results were astounding in that, on average, this method showed that only 62% of patents were identified for a particular pharmaceutical product when compared with the patents identified using GenericsWeb’s comprehensive searching techniques. Furthermore, of the data retrieved by such methods, 94% of publications did not relate to subject matter considered to be relevant to development of a generic pharmaceutical, based on the GenericsWeb inclusion policy. Such patent searching results not only give false confidence, but also cost the searchers’ organisation a significant amount of time in filtering through irrelevant data, meeting neither of the previously outlined key criteria for searching generic pharmaceutical patents.

Avoiding the Patent Search Trade-Off
So, given the above discussion, how is one supposed to meet all of the criteria of the generic pharmaceutical patent search and stay within the budget? Well, it will be no great surprise that the principle of GenericsWeb’s Pipeline Developer reports is entirely based on this dilemma.

“a patent searcher who uses indexed data as their only data source is not conducting a true patent search, they are simply retrieving the results of another person’s analysis or, at worst, that of a computer”

Because of our expertise working within the generic pharmaceutical industry we understand exactly what is important to our customers in meeting their patent information needs. We adopt the same search parameters that we believe meet the requirements of any generic pharmaceutical developer and index it under one INN. The search is undertaken using a combination of professional search techniques and data sources, and information is then filtered and sub-divided into development categories to assist in understanding and interpreting the patent landscape.

And how can we afford to do all of this and keep it within your budget? The answer is simple – economies of scale. There are many generic pharmaceutical developers all of whom seek patent information with the same search parameters. GenericsWeb can therefore afford to charge less than it costs to conduct an individual search, based on our ability to sell it numerous times.

The result is that GenericsWeb customers get the best patent information at the best price. And if, after reading this article, you feel that you should not start a development with the same patent information as your competitors, ask yourself whether you should really be starting with less.

GenericsWeb Pipeline Patent Intelligence is available for any pharmaceutical product on a per-drug basis. Please visit or contact us by email at for further information.

For questions and comments about this article please contact the author at

1 Although it could be argued that patent classification is a form of indexing found in raw data, the history of development of classification systems, the varying processes by which they are applied and the size of the population of pharmaceutical classes means that they are useful, but not reliable.