Citation Trends with European Patent Documents are Dramatically Different than in the US – Issues with Counting Citations Part 3

In part three of this series on citations between patent equivalents I took a closer look at European patent documents. What I found was that citation patterns for European documents are significantly different than what we find in the US. In fact, the importance of pre-grant applications is even more highly pronounced in Europe than it is in the US.

The previous post looked specifically at the US and found that not only did ~75% of all US patent documents from the collection studied have at least one forward citation associated with it but that in 60% of the cases where the granted patent did not have any forward citations the pre-grant application still did. The initial post on this topic started the discussion on whether patent equivalents should be counted together when considering citation counts. This post continues to generate a large number of comments in a variety of forums and has raised the issue of sources for patent citations and how they can vary. Looking more closely at citation sources is being considered for a future post.

To look at the situation with European patent documents and what the impact of patent equivalents would have on them I started with the same search strategy I used when looking at the US, all EP grants and applications with an application date between January 1st, 2004 and February 14th, 2004 and a corresponding publication year between 2004 and 2010. This generated 20,465 documents with 4,323 of them being granted patents. This is less than what I found for the same period of time in the US and the relatively small percentage of grants in the total set can be attributed to the additional pre-grant kinds associated with EP pre-grant applications.

Of the 4,323 granted patents only 273 or 6% of the documents had a forward citation associated with them. When compared to the 76% I found when looking at the US documents this was a significant difference. An argument can be made that this difference reflects the practice of the two offices when it comes to reporting citations from the examiners vs. the applicants since there is no requirement to provide known references when applying for a patent in Europe as opposed to the US.

What was interesting though is that when you look at the pre-grant applications associated with the European granted patents you find that 1,542 or 36% of these documents have at least one forward citation. While this is half as many as the corresponding situation in the US it appears that there is a propensity to cite European pre-grant applications as opposed to the granted patents. In the US the percentage of documents between grants and pre-grant applications was essentially equal but in Europe there is a much higher likelihood of finding forward citations associated with the pre-grant application than there is with the granted patent.

In the previous post I looked at the distribution of publication years for the total collection of grants versus the grants without any forward citations associated with it. In that case I wanted to make sure there was not a significant time since grant bias built into the collection. In the case of the European grants, since the percentage of the total with zero forward citations is so high, I decided to look instead at a comparison between the total grants and the grants with at least one forward citation. The chart below shows this comparison:

A comparison of publication year distribution of all European granted patents in this collection versus the distribution for patents which have at least one forward citation

What we find in this case is that age does seem to have an impact since the publication year distribution for the total number of grants decreases less quickly and then levels off as opposed to the situation with the grants with at least one forward citation which peaks in 2006 and then falls off.

To make sure I wasn’t missing something due to a timing bias I created a second collection of European granted patents all of which were published between January 1st, 2006 and March 20th, 2006 (March 20th is my birthday by the way if any of you would like to send me a card). This produced 11,167 granted European patents from the period.

Looking again at the number of granted patents with at least one forward citation from this collection I found only 952 or 8% of the cases had them. While higher than the 6% I found for the initial collection with a four year publication year spread it is still significantly lower than what is found in the US.

Meanwhile, when looking at the corresponding pre-grant applications I found that 4,665 or 42% of the cases had at least one forward citation associated with them. Again this compares favorably to the 36% I found when looking at the earlier collection from a collection of publication years.

Finally, I looked at the number of cases where both the granted patent and the pre-grant application had forward citations associated with them. This, after all, was the question we started with, do pre-grant applications significantly impact the forward citation count of the corresponding granted patent when they are considered as equivalents. Of the 952 cases where the granted patent has at least one forward citation I find in this collection that 60% of the pre-grant applications also have forward citations associated with them.

There are three primary lessons that this study seems to provide:

  1. Even more so than in the US, it is critical to consider pre-grant applications in Europe when talking about forward citations with patent equivalents.
  2. Unlike the US, where the likelihood of finding citations associated with granted patents is about the same as the likelihood of finding a forward citation associated with a pre-grant application in Europe the likelihood of finding forward citations associated with pre-grant applications is four-times higher than finding them with the granted patent.
  3. When considering forward citations between documents from the two countries you are about half as likely to find citations associated with European documents as you are with ones from the US.

It is clear from these examinations that working with patent citations can be complicated and it is essential to consider patent equivalents when representing this data.

Comments 6

  1. Tony,

    0) I apologize for posting this on both the PIUG wiki and your website.

    1) The European pre-grant forward cites seems to be a self-fulfilling European issue. They (externally via Espacenet and I assume internally on EPOQUE) preferentially display the first published family member for a patent family. This is almost always an EP or WO A1 document.

    2) The Europeans do the same thing with US family documents and display the US PG-Pub A1 preferentially over the issued US B1 or B2.

    3) Database providers that use the EPO published data as their core data (most of them) generally have elements 1 and 2 above as their default settings and it takes a pretty sophisticated user to understand and change these settings. For novelty and validity searching this change is not required. It’s only FTO / clearance projects that require the change.

    4) For US documents, it is much easier to cite a US PG-Pub A1 document rather than an issued US B1 or B2. They have paragraph numbers which are easy to identify versus difficult to find Column / line information. Many attorneys I know would much rather have citations to US PG-Pub when not doing FTO or clearance work. (Plus they have earlier publication dates.)

    5) Some databases do not differentiate between kind codes (A1, B1, B2, etc.). Yes, yes, the fancy ones with European customers do, but not everyone uses a fancy database costing thousands of US$ per seat license. In the US, we never really cared about kind codes until very recently since we have different numbering systems for every stage of the process (much to the chagrin of everyone else). The US examiner search tools do not even have European B2 documents in their system.

    1. Hello Dominic,

      No apology necessary and wow what a tremendous and insightful response. Thank you for sharing your perspectives on the various searching methods of the examiners at the different patent offices. This certainly helps put things in perspective and helps explain the major differences between the two countries.

      From an analysis perspective it simply reinforces the idea that citation counts should be associated with equivalents as opposed to being associated with discrete documents. I have observed what you said about the way Espacenet displays documents. Even when I asked explicitly for the EP B I was taken to the page for the EP A and it took a click or two before I could get to the record for and hence the citations associated with the EP B specifically.

      Thanks again, Tony

  2. Nice blog – I haven’t always agreed with everything you have written on patent citations, but I strongly agree with your final comment:

    “It is clear from these examinations that working with patent citations can be complicated and it is essential to consider patent equivalents when representing this data”

    In our view, you should start with the US equivalent (earliest, although later patents can also sometimes help), then the WO and and the EP when performing citation analysis.

    Having said that, sometimes there is no ‘right’ or ‘wrong’ answer – they are all different views, and all add insight into what can be a complex situation.

    1. Mike,

      Also looking forward to chatting with you in the DC area in late April if you make it to the PIUG Annual. I disagree with your selected order of importance with regard to patenting authority when it comes to citations. From a relevance to the claimed subject matter point of view, I would begin with the EPO cited documents. These are all generated by an examiner of the European Patent Office and are material to the initially filed claims.

      I would then use citations from the ISR (whichever organization is generating these for the subject WO document) though these vary in quality depending upon who is actually performing the International Search. All of these will also be germane to the initially filed claims, though you probably won’t have more than one X reference.

      Lastly, I would use citations from the US.

      This is not meant as a knock on the search abilities of US examiners. This is based on the amount of noise which is introduced into the citation system by applicants submitting art to satisfy the duty to disclose which is unique to the US patent system.

      In the medical device field (as an example), it is common for US patents to have 100-400 facially cited references. While this can be useful for certain types of statistical analysis, it also loses much of the narrowness a system built on EPO citations would provide.

  3. Thank you Mike.

    If I see you at the 2013 PIUG Annual Conference perhaps we can talk about some of the items you disagreed with in more detail.

    I appreciate the insight you have added to these posts.

Leave a Reply

Your email address will not be published. Required fields are marked *