A Statistical Spotlight on EPO Opposition Divisions vs the EPO Technical Boards of Appeal
Introduction
Some opponents are dissatisfied with proceedings before the opposition divisions at the EPO. A feeling of bias is in the world. Among others, the following allegations are often made:
A: “Before the opposition divisions, you lose if you attack on the basis of a lack of inventive merit.”
B: “The best chance is added matter.”
C: “Unless you have a novelty-destroying document.”
D: “It looks better before the Boards of Appeal.”
We will test these allegations by comparing them with statistical data concerning opposition proceedings at the EPO.
Data analysis
Our analysis is based on the published opposition decisions both from the Opposition Divisions (OD) and the Technical Boards of Appeal (BoA) of the EPO in the years 2013 to 2022. These decisions have been collected and analyzed by Darts-ip®Darts-ip is a comprehensive, searchable IP cases solution owned by Clarivate. It contains over 10 million intellectual property cases from more than 4,230 courts and patent offices across more than 140 countries., a company of the Clarivate® group. We interpret the data thus provided. The full raw data can be found in the annex.
To start with, we compare the outcomes of discussions of individual grounds for opposition as well as other grounds that may preclude the maintenance of a European patent upon an amendment of the patent in the course of the proceedings before OD and BoA. Fig. 1 shows the percentages of positive outcomes for the most relevant individual grounds for opposition as well as clarity. An outcome is considered positive if the underlying requirement for the maintenance of the patent is fulfilled. The percentages of positive outcomes before the OD are shown in blue, while the ones before the BoA are shown in orange. Thus, for example, the blue column related to novelty means that among all analysed opposition division decisions where the requirements as to novelty were discussed, the requirement was considered to be satisfied in 72,6% and not satisfied in 27,4%. The number of decisions analysed for Fig. 1 were 29 539 before the OD and 8 610 before the BoA.
Please note that the individual columns in Fig. 1 only relate to the outcomes of discussions of the respective ground not the outcome of the whole proceedings or other grounds. A specific ground may have been decided positively in a decision while others may have been decided negatively, resulting in a revocation of the patent. Moreover, Fig. 1 does not differentiate between main request and auxiliary request(s). One opposition proceedings can lead to more than one data point in Fig. 1 as there may be statements both on novelty and e.g. inventive step and/or two novelty discussions with different outcomes with respect, e.g., to the main and an auxiliary requests in a single decision.
![](uploads/2402/epo-opposition-divisions-01.png)
Does Fig. 1 support the four allegations A, B, C, or D?
Fig. 1 supports allegation A only in parts. There is a probability of roughly 23,5% to win an attack based on lack of inventive merit before the OD. According to allegation A, the probability of winning on lack of inventive merit should probably be much smaller.
How about allegation B: added matter? The chances of success of an attack based on added matter are indeed higher as compared to any of the other grounds (28,5%). Yet, the differences are not particularly high (about 1% and 5% higher as compared to the chances of success for an attack based on novelty and inventive step, respectively). Therefore, Fig. 1 does support allegation B also only in parts. Allegation B probably suggests a more pronounced gap in the chances of success which, however, we do not see.
The chances of success of an attack based on novelty are somewhat lower (27,4%) as compared to an attack based on added matter, yet comparable. Thus, Fig. 1 seems to also provide at least some support for allegation C. As before, full support may require more pronounced differences in order to fully meet the expectations behind allegation C.
Fig. 1 does, however, strongly support allegation D: “It looks better before the Boards of Appeal.” The success rate for the patentee was always lower before the BoA. Therefore, the chances of success are consistently better for the opponent before the BoA than before the OD.
With the relatively high number of cases analysed, the results we see in Fig. 1 are significant. The standard deviations of the results of the OD are in the range of 0.3% - 0,5%Assuming a binomial probability distribution; for the BoA, the standard deviations are on the order of 0.5% - 0.9%. Thus, the standard deviations are a lot smaller than most of the differences we see. Some of the results are significant by even more than five standard deviations.
Next, we will move from analysing individual grounds to the final results of the proceedings. There are four possible outcomes before the ODs:
- OD situation 1: the patent is revoked considering the granted claims only; or
- OD situation 2: the patent is upheld considering the granted claims only; or
- OD situation 3: the patent is revoked even though amendments to the granted claims have been taken into account; or
- OD situation 4: the patent is upheld with amendments to the granted claims.
Before the BoA, we consider also four possible outcomes:
- BoA situation 1: the patent is revoked considering the same set of claims that formed the basis of the preceding OD decision; or
- BoA situation 2: the patent is upheld considering the same set of claims that formed the basis of the preceding OD decision; or
- BoA situation 3: the patent is revoked even though amendments to the set of claims, which formed the basis of the preceding OD decision, haven been taken into account; or
- BoA situation 4: the patent is upheld with claims that have been amended as compared the claims, which formed the basis of the preceding OD decision.
Fig. 2 shows the percentages of the above mentioned outcomes before the OD and BoA, namely whether the patent was revoked or upheld – including upheld in amended form or revoked considering possible amendments. Please note that the dataset underlying Fig. 2 is identical to the one of Fig. 1.
![](uploads/2402/epo-opposition-divisions-02.png)
It is interesting to note that only close to 64% and 63% of all decisions before the OD and the BoA are based on claims that are amended as compared to the claims as granted or the claims that formed the basis of the preceding OD decision, respectively.
Fig. 2 shows a striking difference between the OD and the BoA for cases where the final decision was based on amended claims (32% of patent revoked versus 56%, which makes a difference of 24%). This further seems to support allegation D. If the decision is based on amended claims, the outcome for the opponent before the BoA is significantly more positive than before the OD.
On the other hand, the difference is considerably smaller (4%) in cases which have been decided:
- upon the granted claims by the OD, or
- upon the claims, which formed the basis of the preceding OD decision, by the BoA.
![](uploads/2402/epo-opposition-divisions-03.png)
Fig. 3 basically shows the same data as Fig. 2. However, two technical fields are distinguished: electrical engineering and pharmaceuticals. The dataset underlying Fig. 3 is again identical to the dataset underlying Figs. 1 and 2.
The difference between the OD and the BoA for the cases, where the decision is based on the set of claims that also formed the basis of the preceding decision, is probably significant for the pharmaceuticals (about 16% difference; cf. the 3rd column in Fig. 3) but maybe not for electrical engineering (about 2% difference; cf. the 1st column in Fig. 3).
The picture dramatically changes if we look at those appeal cases where the decision is based on a set of claims that differs from the claims that formed the basis of the preceding decision. It seems to be a real uphill battle for opponents before the opposition division under these circumstances, no matter what the technical area is (about 27% and 22% difference; cf. the 2nd and 4th column, respectively).
Fig. 3 would again support allegation D.
![](uploads/2402/epo-opposition-divisions-04.png)
Fig. 4 shows the outcomes of novelty and inventive step assessment: criterion satisfied only (green), not satisfied only (red) and both satisfied and not satisfied (orange). A criterion can be both satisfied and not satisfied in basically two situations: Either it is satisfied for some requests and not for some other requests, or the criterion is not satisfied over prior art document D1 and satisfied over D2.
A differentiation is again made between whether the decision is based on claims that are amended as compared to the previous decision (i.e. the grant decision or the OD decision) or not.
It is logical to observe a significant orange part when the decision is based on amended claims, inventive step can, quite often, be not satisfied for the main request but satisfied for the auxiliary request.
It is also not surprising to see a very small orange part when there are no amended claims. The orange part here corresponds to a rare situation where inventive step is not satisfied over prior art document D1 and is satisfied over prior art document D2.
Fig. 4 again supports allegation D. The results for the opponent are better before the BoA in almost all categories. The difference between OD and BoA is, however, rather small in cases where the decision is based on claims which have not been amended.
Does Fig. 4 also support allegation A, namely that before the opposition division you lose if you attack on the basis of lack of inventive step? Partially yes, but not fully. To actually support allegation A, the percentage of fully upheld cases under inventive step before the opposition division should be e.g. at least 90% including cases where amended claims were filed. But we see a clear advantage for the patentee in this scenario.
![](uploads/2402/epo-opposition-divisions-05-neu.png)
Finally, Fig. 5 tries to find out whether there was any temporal development in the difference between the decisions of the OD and of the BoA. While there seems to be some development in favour of the patentee, the difference between the OD and the BoA remains significant and stable.
Summary
Table 1 summarises the results of our statistical test of the allegations A, B, C, and D.
![](uploads/2402/epo-opposition-divisions-06.png)
The data at hand simply do not allow any conclusive statement on allegation C. The data did not show whether a novelty destroying document which was newly filed during opposition, and which document had not been considered during examination, ?had any effect.
Allegation B is supported by the data at hand at least to some extent but not as clearly as suggested by allegation B. Added matter does not appear to stand out as the best ground for opposition but rather appears to be as relevant as the other ones.
As to allegation A, it is not clear that the opponent loses on inventive merit. However, we see a clear advantage in favour of the patentee.
The data, however, clearly supports that the proceedings before an opposition division are an uphill battle for the opponent. The opponent is better off before a Board of Appeal. This was allegation D.
Annex (raw data)
Raw data related to Fig. 1
![](uploads/2402/epo-annex-01.png)
![](uploads/2402/epo-annex-02.png)
Raw data related to Fig. 2
![](uploads/2402/epo-annex-03.png)
![](uploads/2402/epo-annex-04.png)
Raw data related to Fig. 3
![](uploads/2402/epo-annex-05.png)
![](uploads/2402/epo-annex-06.png)
Raw data related to Fig. 4
![](uploads/2402/epo-annex-07.png)
![](uploads/2402/epo-annex-08.png)
Raw data related to Fig. 5
Inventive step satisfied
![](uploads/2402/epo-annex-09.png)
![](uploads/2402/epo-annex-10.png)