The Divisional Game: Using Procedural Rights to Impede Generic/Biosimilar Market Entry

Despite being used frequently by practitioners in a wide array of technical fields, divisional patent applications (“divisionals”) seldom attract scholarly attention. The lack of scholarly attention is an error, particularly in the pharmaceutical field. Recent case law in the UK reveals that after generic/biosimilar manufacturers successfully revoke patents standing in the way of market entry, divisionals claiming protection for similar subject-matter as the patent that has just been revoked can crop up, hindering generic/biosimilar medicines from entering the market. Moreover, right before or after proceedings start, rightsholders de-designate countries to avoid a negative judgment that may affect a court elsewhere, meaning generic/biosimilar manufacturers initiated legal proceedings for nothing. Such practices demonstrate that divisionals deserve our attention. This article thus fills the scholarly gap by showing how divisionals are (mis)used in practice, before arguing that patent reform is necessary to prevent this from occurring.


Introduction
'''Divide and rule' ('divide et impera' in Latin) has been employed as a way of achieving political success ever since the time of Philip of Macedon, who ruled from 359 to 336 BC. For many, the expression has negative connotations. Yet in the medical use patents. 8 Each tactic deserves significant attention, but one piece in this puzzle is missing: procedural rights. This article therefore looks at how filing, albeit not necessarily obtaining, patents may be misused to achieve the same. In its 2009 Pharmaceutical Sector Inquiry Final Report, the European Commission (EC) found that divisionals were being misused to block the entry of generic drugs. 9 The originator companies interviewed disclosed that divisionals are ''filed in order to reset the acceptance deadline clock and allow more time for prosecution'' or to give ''more time for preparing the answers [to objections to parent applications] adequately''. 10 Generic companies unsurprisingly shared that divisionals create ''legal uncertainty'' as ''multiple divisional patent applications combined with abusive patent litigation and preliminary injunctions hinder the development of generic medicines''. 11 The inquiry further revealed that it was not about a few bad apples: 16 out of the 27 generic firms interviewed ''had problems with divisional filings''. 12 Albeit cited as an issue again in a EC 2019 report, 13 divisionals largely remain under the public's radar, 14 save for a couple of case notes in intellectual property journals 15 and white papers. 16 This study is one of the first attempts to investigate the use of divisionals within the pharmaceutical field, whereby its findings may be of interest to those researching pharmaceutical prices, incentives to innovate, competition law, human rights and patent law.
The effect divisionals may have within these areas can be seen in practice. After generic manufacturers initiated proceedings to revoke a patent standing in the way of market entry for a generic version of Fosamax, Merck filed additional divisionals, each effectively claiming the same subject-matter as the opposed patent, thus providing new grounds to use for patent infringement claims, delaying the entry of cheaper generics/biosimilars. 17 Abbvie employed a similar strategy for Humira, though it went a step further and started withdrawing patents once challenged, to avoid a negative court judgment, and then filed divisionals claiming the same subject-matter as the patents it just withdrew: much like a ''Hydria -for every head chopped off, a couple of heads regrow''. 18 This article argues that filing divisionals is a procedural right that can be (mis)used to deliberately delay the entry of generics/biosimilars. Section 2 explains the process, scope and requirements of filing divisionals, as well as their original justification. Using three case studies, Sect. 3 shows how divisionals are used as a fallback to keep the threats of subsequent patents claiming similar, if not exactly, the same subject-matter (i.e. ''double patenting'') as a previously rejected patent alive and, moreover, to prolong judicial proceedings by de-designating countries from patents once they commence -practices frequently referred to as ''the divisional game''. 19 The consequences of these practices not only impede generic/ biosimilar market entry, but they render judicial remedies at the national level, such as Arrow Declarations, 20 ineffective, because the courts cannot fix what is legal under European patent law. By going through the body of EPO Board decisions on divisionals, Sect. 4 shows how the interpretation of the two requirements for filing divisionals -namely, that the patent application must (1) claim similar-subjectmatter contained in a (2) earlier, pending parent -has developed in a way which enables the behaviour described above to happen. This shows that patent reform at the European level is needed to stop the procedural misuse of divisionals. The section ends by proposing several suggestions for reform.

Filing Patents for Medicines
The process of filing a patent for a medicine usually starts with an applicant drafting claims, i.e. the subject-matter (compound) it seeks to protect, as well as additional information, including a description of the invention, accompanied by drawings or figures, and an abstract. Claim construction is a delicate exercise. It requires one to find the ''sweet spot'' between competing interests. Narrow claims are, on the one hand, more likely to be quickly accepted, but may limit the scope of protection and enable third parties to make use of certain aspects of the invention without fear of infringement. Broad claims, on the other hand, meet greater resistance throughout the examination process. There is a higher chance that they overlap too much with the prior art, or are so broad that they are insufficiently disclosed, meaning they cannot be enabled by the person skilled in the art, delaying their grant. Applicants, nevertheless, usually opt for broad claims. Attorneys are even expected to do so on their behalf. The Rt. Hon. Professor Sir Robin Jacob affirms, It must be understood that any well-drafted patent application will have claims wider than those which will eventually be granted. No competent patent attorney would claim narrowly in the first instance. He goes wide to see what prior art is found and shapes his eventual widest claim around that. It would be negligent for him not to do so. There is no sanction for applying for claims which would be unduly wide and are unlikely to be granted. 21 Without consequences, applicants have nothing to lose, but much to gain by filing broad claims. The ''gain'', as commonly viewed, is that the broader the claims, the broader an invention's protection. 22 However, broad claims can carry another benefit, namely they cover more subject-matter and so give more opportunities for add-on divisionals to be filed, as is explained below. 23

Dividing One Patent into Many
A divisional is a type of patent application that claims subject-matter contained in an earlier, pending parent, as required under EPC Art. 76 and Rule 36 of Implementing Regulations to the Convention on the Grant of European Patents (''Implementing Regulations''). 24 Subject-matter is interpreted broadly and is not limited to the claims of the parent. It encompasses the parent's technical information, including its description, claims and drawings. 25 However, the subject-matter claimed may not be identical, i.e. double patenting, which is prohibited, or extend beyond the granted parent. 26 A divisional may, for example, claim only part of a parent's subject-matter, provided said part is ''directly and unambiguously derivable from the parent'' 27 by constituting a ''separate part or entity, i.e. one which can even be used outside the context of the invention of the parent application''. 28 Or it may claim its own subject-matter, in combination to that claimed in the parent. As the EPO Guidelines state: ''if the parent and divisional applications claim separate and distinct elements A and B respectively which function in combination, one of the two applications may also include a claim for A plus B''. 29 Subject-matter deleted from the parent, such as when subject-matter is removed to overcome a unity objection, can also be claimed in the divisional. 30 Subject-matter refused on appeal for the parent cannot. 31 Accordingly, what applicants pursue in the divisional need not even be claimed in the parent. The 21 Jacob (2013) Dechezleprêtre et al. (2017) p. 800. 31 EPO, Guidelines, C-IX, 1.6. amount of divisionals applicants can get out of the parent depends on how many integers (i.e. combination of features) applicants can claim to define a patent.
The divisional must be filed while the parent is pending, which it still is if it has been refused, but there is still time to file an appeal, as well as during the appeal process. Parties must submit a notice of appeal to the EPO within two months of being notified of the refusal and the statement laying out the grounds for the appeal four months after receiving notice. 32 The appeal process may take years before a decision is final, providing an extended period time where applicants may file divisionals, with further incentive to do so if the applicant notices the appeal decision is not headed in their favour. Also, as with regular patents, a divisional's full claims need not be filed when filing the application. 33 They can be filed later, even when the parent is no longer pending. 34 These rules apply when the divisional is itself divided into another divisional. 35 Conversely, albeit related, the two applications exist separately from one another: a divisional is treated as a new application, largely independent from the process concerning the parent. Any pending opposition or appeal proceeding against the parent does not give grounds for postponing the examination process of its divisional counterpart. 36 Nor does the outcome of such proceedings against the parent affect the divisional after its filing. 37 It is its own patent. A divisional must, like its parent, therefore, satisfy the three requirements for patentability in addition to the requirements under Art. 76 and Rule 36. 38 If these requirements are met, the divisional is ''deemed to have been filed on the date of filing of the earlier application and shall enjoy any right of priority''. 39 As such, a divisional should not extend an invention's period of protection. Yet it can do exactly that. Because the divisional must claim similar subject-matter to the parent but is independent from said parent and results in its own patent, it means that if the parent is revoked, a divisional protecting similar subject-matter may remain in force.
Minn observes, ''divisional applications are commonly used for strategic purposes, to obtain a comprehensive coverage equivalent to that of the excessively broad first application''. 40 This is an odd development given that divisionals are meant to be a solution to broad patents. The subject-matter link to the parent is largely because applicants can, if a patent meets a unity objection, use divisionals to bifurcate the original subject-matter of the parent into separate divisionals. As established by Art. 4(G) (1)  concerns more than one invention, ''the applicant may divide the application into a certain number of divisional applications''. The ''unity of invention'' rule, a requirement under Art. 82 of the EPC, stipulates that each patent application must only cover one invention. Or, if need be, one patent for a group of closely related inventions which form a ''single inventive concept''. 41 Unity objections may be raised a priori, before a prior-art search, or a posteriori, after a prior art search, 42 but cannot give grounds for opposition/revocation post-grant because, as the EBA confirmed in G 0001/91, unity is ''merely an administrative regulation''. 43 Requiring each patent to claim one invention or a single inventive concept ensures that the examination fee paid per application only involves one invention 44 and that each patent is properly classified. Unity is, therefore, in theory, not a mechanism which seeks to regulate the scope of protection afforded, but in practice it does. 45 Even though a unity objection gives applicants a chance to not only get it right, but to get it better, by filing divisionals, the common view is that most applicants ''limit'' their protection to one or at most a couple patents per invention and abandon patents rejected due to a lack of unity, reducing the overall scope of protection. 46 This is largely attributable to the cost of a European patent, which reaches an estimate of €32,000. 47 However, recent trends challenge this view. Applicants with more financial means, such as pharmaceutical companies, file divisionals voluntarily without any unity objections. Like mandatory filings, voluntary filings have been permitted since the Paris Convention. Article 4(G)(2) inter alia states that applicants may on their ''own initiative, divide a patent application'', albeit leaving it up to each country to decide when this option should be available. The EPC 1973 allowed applicants to file divisionals at their own discretion, 48 as does the EPC 2000.
Mandatory divisionals are allegedly ''comparatively rare'' in comparison with ''divisionals resulting from the voluntary action of the patentee''. 49 The number of divisionals filed, and why, remains unclear as the EPO no longer publishes such data, though the data previously available suggests that divisionals often filed in the pharmaceutical and biotechnology sector and that filings are increasing. 50 Hegde et al. (2007). Data however suggest that divisional filings are lower at the EPO than the US, but that divisionals are ''more frequently used at the USPTO … and at the EPO'' see Dechezleprêtre et al. (2017) p. 805. patenting in these sectors does not mean that divisionals are more likely to be misused as such, but recent case law, as will now be discussed, shows that misuse happens in practice.

The Divisional Game
Despite mandatory divisionals being permitted since 1925, and voluntary divisionals since 1958, 51 it is only in the last decade that divisionals started cropping up more readily in national patent proceedings, particularly those concerning the pharmaceutical industry. Not all divisionals are misused. They are filed for various (and legitimate) reasons. One is, as above, to overcome unity objections, thus fulfilling the original objective for this procedural avenue. Another is that if the patent office disapproves of parts of the subject-matter in the parent, then applicants can file two sets of divisionals: one set claiming the ''acceptable'' content to secure faster protection, while the other set claims the broader subject-matter likely to be subject to ''argument, evidence and possible appeal''. 52 Industry notes that continuation practices in the US, including divisionals, are needed to overcome prior art searches, particularly given the amount of literature published worldwide, and to help rectify or clarify errors in the patent examination process. 53 These patents are, according to the Biotechnology Industry Organization, vital for the biotechnology and pharmaceuticals because companies file patent applications early to attract investor interest, as well as often to ''protect both the initial concepts of their discoveries and additional practical embodiments supported by the applications''. 54 GlaxoSmithKline further provides that companies file them because they are ''trying in good faith to obtain coverage they believe they are entitled to under the patent system''. 55 Divisionals are also increasingly being misused. One unnamed member of Medicines for Europe, an organisation representing the European generic, biosimilar and valued added pharmaceutical industries, observes that, ''the abuse of the divisional patent system has indeed deteriorated over the past years''. 56 Another unnamed member likewise claims that, ''the abuse of the divisional patent system has increased over the last decade''. 57 One example of misuse is by filing multiple divisionals to serve as back-ups should the parent get revoked, or by drafting claims covering products competitors already have, or are looking to put on the market, to create uncertainty. 58 As Jacob states, 51 Bodenhausen (1968 Lemley and Moore (2004) p. 65. the game is to create a patent with narrow claims (hence easier to defend) covering the product -but which does not offend the ''no added matter'' rule. Any would-be entrant must therefore consider not only all pending applications but also divisional application, as well as potential divisionals from these. What valid claims could be obtained on the basis of the application is seldom an easy question to answer. 59 Used in this way, divisionals can, according to Minn, harm competition because competitors will not know whether the rightsholder has filed a divisional which ''can defeat its attempts to design-around the parent application''. 60 And, moreover, ''a generic company, despite a successful challenge to a parent application, suffers the risk that a patent covering substantially the same subject-matter may issue from a pending divisional application. Further, a divisional application can be divided into other divisional claims, multiplying the effects just describe''. 61 This is both true and false. The parties will likely know of the existence a divisional filing because the EPO publishes the patent application ''as soon as possible after the expiry of a period of 18 months from the date of filing, or if priority has been claimed, from the date of priority'', or earlier if requested by the applicant, 62 but they might not know exactly what the final divisional will be as applicants can file claims after the ''date of filing the divisional 63 … even if the parent is no longer pending at the date of filing the claims'', 64 and can amend the claims during EPO proceedings. 65 Although the EPO includes the new or amended claims when available, this entire process takes years. The patent grant procedure takes on average 3-5 years from the filing date, 66 but sometimes, as with the Humira case study below, it can take up to 11 years before the patent is granted. As such, there is a significant time lag between filing, prosecution and grant whereby the claims can change throughout, thus creating a prolonged period of uncertainty.
The following section will present three recent cases from the UK demonstrating how this misuse occurs in practice. These cases were selected as they are recent and have yet to attract significant scholarly attention. Additional examples in the UK and Europe are listed in the subsequent section, before Sect. 4 shows how patent law has developed in a manner allowing this behaviour.

Fosamax
Arrow v. Merck 67 is most known for ''establishing'' the Arrow Declaration, i.e. a declaration that a specific product or process was obvious at a particular date. 68 Less known are the problematic ways in which divisionals were used leading up to this case. Yet a key question in this case was: ''where one person holds over another a threat of patent infringement proceedings on the basis of divisional applications'', should this court ''necessarily refuse to allow the person so threatened to seek a determination of its rights''? 69 Although judgement was given in summary proceedings, and the parties settled before going to trial, it is nevertheless an important case to consider for this reason.
The case concerned Fosamax (alendronate), a medicine approved to treat bone wasting diseases. Merck first marketed Fosamax in 1995. A daily 10 mg oral dosage was approved in July of 1997. Merck then filed a patent for another dosage -70 mg once a week rather than 10 mg per day -in 1998. The EPO granted the patent in 2001 as EP0998292 (''292''). In order to legally pave the way for generic versions of Fosamax, Arrow and two other generic companies initiated proceedings to revoke the Great Britain designation of 292. The UK Court revoked 292, finding that it, inter alia, lacked novelty and an inventive step. The EPO Opposition Division revoked the patent on identical grounds in 2004. 70 However, during the prosecution of 292, Merck filed four divisionals, all relating to 292 and each potentially giving grounds for an infringement suit. The generic companies started marketing a generic version of Fosamax with a 70 mg dosage routine in Europe after the revocation of 292 for £1-2 per 4970 mg pack -a significant reduction from the original price of £15-18.
Although the EPO paused their examination of the divisionals during the 292 opposition proceedings, it granted one divisional shortly after as EP1175904 (''904''). This patent was the most problematic for the generic manufacturers. Albeit phrased differently, 904 and 292 claimed exactly the same substance, for exactly the same disease, with exactly the same type of administration and dosage interval. As Kitchin J ascertained, ''the key idea therefore appears to be essentially the same as that the subject of the 292 patent, namely give osteoporosis patients 70 mg of alendronate once a week''. 71 This meant that the subject-matter which generic 67 Arrow v. Merck ] EWHC 1900. 68 An Arrow Declaration mirrors the ''Gillette defence'', that the allegedly infringing process or product was obvious, rendering the infringing claim moot because one cannot infringe an invalid patent, see Gillette Safety Razor Co v. Anglo-American Trading Co ( More problems could follow if the other divisionals were granted as both 911.13 and 912.1 claimed subject-matter the generic drugs used, including alendronate, the active substance of Fosamax. The third divisional, 910.5, made no mention of alendronate, but Kitchin J recognised that, ''the claims of any of these applications can be amended during the course of prosecution so it is quite possible that the claims of any patent granted on application 910.5 could also be directed to alendronate''. 72 But the generic manufactures did not know whether this would happen. Arrow therefore sought several declarations. The most significant was the one which would establish that ''its own product was obvious at the priority date of the divisional applications''. 73 Kitchin J found that Arrow had clear interest in the declaration. 74 Of particular importance was the fact that, because Arrow already attempted to clear the way for its generic drug by revoking the original 292 patent, no doubt devoting ''significant resources to those proceedings and ultimately'' succeeding, Arrow reasonably believed no objections could be raised so it launched its product. 75 The divisionals cast doubt on this belief for two main reasons. First, Arrow faced ''the prospect of EP (UK) patents being granted on the divisional applications that will cover the very same product''. 76 Second, Merck could change the scope of the claims until grant. Reaching the grant stage also takes time. If granted and enforced, as Merck publicly announced it would do, Arrow would potentially have to pay damages for the period starting from the publication date of the patents until the case was decided, meaning Arrow was accruing liability charges for two divisionals while the case was ongoing. All in all, Arrow faced ''a considerable period of commercial uncertainty''. 77 The Arrow Declaration was a victory for generics. But not for divisionals per se. Kitchin J promulgated, ''I find it hard to conceive of any circumstances in which it would be appropriate for this court to grant a declaration that no valid patent could be granted on a divisional application which is being prosecuted before the EPO''. 78 This is understandable because neither the EPC nor national patent law provide a legal basis for such a restriction. However, confining Arrow Declarations in such a way leaves the root of the issue untouched, while also taking the sting out of the Declaration, raising questions whether it even has a deterring effect against abusive divisionals. Subsequent cases suggest otherwise. These adjudications, as will now be shown, follow Kitchin J's approach, though they raise further doubts over the effectiveness of Arrow Declarations as rightsholders have found a way to remove their ''need'' by de-designating countries from a European patent at the right time.

Humira
Humira (adalimumab) is a drug approved to treat arthritis, psoriasis and a host of other indications. 79 AbbVie created an extensive patent portfolio for Humira, protecting it with 23 patent families predominantly comprising second medical use, combination and methods of production claims. 80 Protecting Humira with so many patents is understandable from a business point of view: Humira generated over $18 billion in sales in 2017 alone, making it the world's best-selling drug. 81 It is therefore not surprising that the expiration of its primary product patent in 2018 elicited ''unprecedented patent disputes''. 82 Quite the contrary. It was expected. Richard Gonzalez, AbbVie's CEO, announced that he would ''vigorously'' enforce Humira's patents and further threatened with ''extremely large'' damages for potential infringers. 83 AbbVie sued biosimilar developers for patent infringement while the latter attacked AbbVie's patents. 84 Although biosimilar developers won the majority of cases, they raise serious concerns over why these patents were granted, as well as over AbbVie's unethical patent practices.
AbbVie employed a strategy of avoiding negative decisions at the EPO, as well as in the national courts, by withdrawing patents once challenged, and then filing divisionals with the same subject-matter, which was possible as divisionals stemming from the parent were still pending. When, in 2013, 15 companies filed opposition proceedings against EP1406656 (''656''), a mode of administration patent expiring in 2022, AbbVie withdrew the patent stating they no longer approved of the text. 85 In the time period between the opposition and assumed court proceedings AbbVie filed divisionals with similar claims as 656. This was possible as the divisional was not filed directly from the 656 patent but via another divisional based on 656. 86  who delivered in judgment in the Court of Appeal, upheld these two judgements. 89 This section will focus on the subsequent case decided by Carr J. 90 After Fujifilm and Samsung filed proceedings in the UK to revoke two dosage patents (EP1406656 and EP1944322), which stood in the way for their biosimilars (they needed to incorporate the dosage regimes), AbbVie withdrew the opposed patents, yet filed divisionals with identical claims. AbbVie was able to do so as divisionals based off of the original parent were pending at the time, meaning the divisionals filed were all second or later generation divisionals. In April 2015, AbbVie filed its fourth divisional stemming from 656. In November 2015, AbbVie told the EPO it no longer approved of 656's text. On that same day, the EPO published the fourth divisional ''which claimed essentially the same subject matter as the 656 patent''. 91 AbbVie argued that it abandoned 656 due to a late sufficiency objection. Carr J disagreed, finding that AbbVie did so to ''avoid scrutiny by the UK Court, and to prolong commercial uncertainty as to the validity of those patents''. 92 The EPO subsequently revoked 656 in November 2015 and granted the divisional in November 2016 as 044. Providing no explanation, AbbVie removed the UK designation one day prior to this grant, though proceedings between the parties had already started. In other words, AbbVie knew a UK case was coming. AbbVie filed yet another fifth divisional based on 044 before de-designating the UK from 044. AbbVie stated that it also disapproved of 322, causing it to be revoked. Although AbbVie informed the claimants that ''there were no extant members of that [322] patent family and no divisionals could be filed'', the claimants submitted that a pending EP application could arise out of 322, which in turn could give rise to a ''further psoriasis divisional''. 93 Carr J therefore found that other patents could emerge from 322 or from other families protecting the same subject-matter. Fujifilm and Samsung argued that AbbVie has a, ''well-established strategy of dragging out proceedings for as long as possible, causing maximum expense and inconvenience to its opponents, and then throwing in the towel just before its patents are scrutinised by the court or tribunal, whilst covering the same subject matter with further divisionals''. 94 Meanwhile, AbbVie refuted the need for an Arrow Declaration as it had dedesignated its UK EP patents, meaning it had no legal rights in the UK and biosimilars were free to enter the market. The Court granted the Arrow Declaration rendering the patents in question obvious at their time of filing. In his reasoning, Carr J affirmed that such a declaration was useful as, ''AbbVie has abandoned all relevant UK patent protection in order to avoid scrutiny by the UK Court, and to prolong commercial uncertainty as to the validity of those patents'', 95  furthermore, ''found that AbbVie has made threats that it will enforce its patents against biosimilar competition anywhere in the world. The declarations will serve a useful purpose of dispelling commercial uncertainty in the UK (and European) market, which those threats have created''. 96

Avastin
Avastin (bevacizumab) is a medicine produced by Genentech, authorised to treat ten medical indications, including several cancers and certain eye diseases, generating an estimated $7 billion in profits annually. 97 Although the EC approved biosimilars Mvasi and Zirabrev in 2018 and 2019 respectively, both struggled to enter the market after Avastin's main patent expired due to secondary patents 98 and divisionals. Genentech filed new patents with claims protecting dosages, bevacizumab combinations and medical uses for various cancers throughout the basic patent's lifecycle, acquiring enough patents to protect Avastin with at least three patent families: Fyfe, Dupoint and Bernasconi. Protecting Avastin with so many patents is understandable as Avastin is approved to treat ten different therapeutic indications, each constituting new sub-inventions under the EPC despite all relying on the same antibody. Each patent family contained a series of pending divisionals which could give grounds for an infringement suit if Pfizer launched its biosimilar. lacked an inventive step. Right before filing the appeal in March 2016, Roche filed another divisional (EP3064509 ''509''), but did not file the grounds for the appeal which was due in June 2016. 509 would protect similar subject-matter to the previously rejected parent patent. As with the Fyfe family, Roche de-designated the UK from the 509 divisional, though it could file further divisionals for this patent family as 509 remained pending at the time of the case.
The Bernasconi family, the third family protecting Avastin, concerned using bevacizumab combinations to treat platinum resistant ovarian cancer. Roche removed the UK designation from 558 (the main patent in this family) in November 2017. The EPO granted the patent in May 2019, though EPO opposition would likely, according to Birss J, start within the 9 month deadline. Roche disclosed of plans to file a divisional from 558, stating that it intended for it only to cover one indication; meanwhile, Pfizer and Birss J believed there was a risk claims to more indications ''could eventuate despite Roche's assurances''. 100 Accordingly, Birss J found that ''there will be a pending divisional application whose purpose is to operate as a backup to try and cover indication 9(a) even if the granted patent is revoked in opposition proceedings. There is at least a risk that the pending application may allow for coverage of indications 9(b) and 9(c)''. 101 It is against this background that Pfizer sought an Arrow Declaration. Roche disputed that an Arrow declaration could be granted as it had no UK patent rights. As a result, the Arrow Declaration failed, with Birss J observing that, ''when the action began it was not forum shopping at all. There were pending UK applications which provided a basis for considering an Arrow declaration. However now they have gone''. 102 The only prima facie unusual element in this case was, according to Birss J, that Roche dropped the appeal at the EPO, albeit noting that this was, like in the UK case, shielding in order to avoid a negative appeal decision ''which would carry more weight than a refusal by the examining division'' with the divisional later filed delaying the process even further. 103 Of particular importance is Birss J's statement that Roche, has used the ability to file divisionals to maintain pending applications of as wide a scope as possible (save possibly in relation to indications 9(b) and 9(c)). Nothing Roche has done is unlawful. Objectively this conduct gives rise to significant uncertainty for Roche's competitors. Roche knows that perfectly well. On the other hand Roche would no doubt like to get a valid patent for one of these indications if it can. It is entitled to try … None of this means that Roche does not have a bona fide belief that despite the obvious weakness of this portfolio, it might manage to get or hold on to some kind of valid patent claim relating to perhaps some of the relevant indications. Something might turn up. 104 Birss J's accepted Roche's expert witness Mr Bassett's view that ''the way in which Roche approached this [patent] portfolio has been standard industry practice and not unusual''. 105

Case Study Evaluation
When deliberate obfuscation is industry practice, this is not a few bad apples spoiling the barrel. Nor are these practices limited to the UK. In 2019, the District Court of the Hague confirmed that it has jurisdiction to grant Arrow Declarations on similar grounds as the UK case and will consider Pfizer's claims against Roche. 106 This case remains pending. Other cases where divisionals were misused include Symbicort and Nexium, 107 as well the two examples included in the EC's 2019 report. First, Xalatan, where the Italian national competition authority and Italy's highest administrative court found that Pfizer violated antitrust rules by using a ''complex legal strategy'', which included filing several divisionals and applying for additional supplementary certificates, to delay generic versions of Xalatan, a glaucoma drug, from entering the market by seven months. 108 Second, the EC investigated Boehringer Ingelheim (BI)'s patent practices for its drug used to treat chronic obstructive pulmonary disease as it filed a plethora of patents, including several divisionals that it never asserted but could have in theory, which could prolong the patent disputes beyond the basic patent. However, the EC and BI settled. In March 2021, the EC also announced that it had opened a formal antitrust investigation to assess whether Teva ''abusively blocked or delayed the market entry of competitors to its blockbuster drug Copaxone … by strategically filing and withdrawing divisional patents, repeatedly delaying entry of its generic competitor who was obliged to file a new legal challenge each time''. 109 A pending divisional creates legal uncertainty for generic/biosimilars manufacturers regardless of whether the parent is revoked. 110 This is because they do not know whether future divisionals will be granted or not. Nor do they know their scope as they may be unpublished and the scope often changes throughout the patent prosecution period. What they do know is that the divisional must contain similar subject-matter to the pending parent meaning the chances are high that it encompasses their own product. This subject-matter similarity may, on the one hand, prove advantageous as there is a higher chance that a granted divisional will get opposed and revoked if the parent was revoked. On the other hand, it may be disadvantageous. Challenging a divisional requires another round of legal proceedings and legal fees as the divisional and parent patent exists independently from one another. And if both the parent and divisional survives prosecution, the medicine enjoys overlapping protection.
The uncertainty behind this process is exacerbated by the prolonged time period of the process. Divisionals often stem from second medical use patents, particularly dosage and formulation patents, which are filed later than the patent protecting the medical compound. When medicines are protected with several patent families comprising a myriad of secondary patents, each carrying the possibility of additional divisionals, this uncertainty is multiplied. The EC found in 2009 that the number of divisionals filed per patent is anywhere from one up to 30. 111 Several cascading divisionals may thus appear several years after generic/biosimilar manufacturers think the coast is clear because the main patent expired, creating not only significant period of commercial uncertainty, but a legal minefield. Equally, actors can file divisionals during the period between rejection and appeal, so a divisional claiming subject-matter may emerge much later on, particularly as divisionals stem from other divisionals and as the patent examination process takes years, whereby the second application may succeed. If manufacturers launch generics/biosimilars during this time and a divisional is granted, the pharmaceutical company can later sue for infringement. Fear of infringement and an expectation to clear the way pre-market entry is notably why generic pharmaceuticals oppose patents at the EPO and continuously attempt to revoke secondary patents in the national courts before launch. However, they cannot revoke what does not yet exist. 112 Nor does the law permit it. The right to seek revocation only applies to granted patents. There is no such national statutory right or remedy for pending or future patents. Uncertainty of whether there are new patents on the horizon is in this sense worse than certainty that such patents do exist.
Although manufacturers could oppose patents at the EPO or seek to invalidate the patents in the national courts, the case studies above show that pharmaceutical companies either try to avert negative judgements by withdrawing the European patent or de-designating countries where national proceedings start, while filing new divisionals with near identical claims. Given the cost and time involved with these proceedings, particularly when a divisional starts this process anew, there is indeed little incentive to try when the cost of waiting is cheaper.
All these factors can dissuade generic/biosimilars from entering the market because of the significant financial risk, thus having ''a chilling effect on competition''. 113 This is not in the public's interest. It delays the market entry of cheaper, generic/biosimilar drugs, thus allowing the artificially high-priced drug to continue to enjoy an exclusive market. The effect is that the market exclusivity is extended because a pending divisional stops market entry, meaning it de facto gives a market monopoly even if a pharmaceutical company does not formally have a legal monopoly by way of a granted patent. In other words, procedural patent rights can be used in a manner affording the same scope of protection as (non-existing) substantive patent rights. Alternatively, the patent protection is actually extended because, if a secondary patent is revoked (or withdrawn) and a divisionals is granted, it extends the protection period to be the same as that which the secondary patent afforded.
Although the Courts can issue Arrow Declarations to circumvent the issues divisionals pose, granting such Declaration on an ad hoc basis cannot, as the law currently stands, remedy the procedural misuse of divisionals for three main reasons. First, actors can just remove country designations right before proceedings commence to avoid the Declaration and a negative judgement in another jurisdiction. Second, even if one country is de-designated, the issue remains for all other countries because European patents are validated in each EPC country and because actors also withdraw parents once opposed and file new divisionals right before hearings at the EPO. Third, and most importantly, it leaves the root of the issue -divisional misuse -untouched. This behaviour will just continue. However, as tempting it is to say that re-filing divisionals with the same subject-matter and scope of protection as patents which have just been revoked or opposed is clearly an abuse of the patent law system, European patent law does not deem it unlawful. It has developed in a manner that enables this behaviour.

Underlying Rules
This section describes the underlying rules, doctrines and processes which allow the individual abuses to take place. Namely, Art. 76 and Rule 36, the two main legal provisions governing divisionals. Even though Rule 36 is part of the Implementing Regulations which are used to ''to determine in more detail how the Articles should be applied'', 114 meaning if Rules conflict with EPC provisions, such as Art. 76, the latter prevails, 115 Rule 36 is discussed first because it reveals why applicants may choose to file divisionals and what type of subject-matter is included in the divisional. The aim of this analysis is to show that patent reform is needed to prevent divisional misuse. Suggestions for reform, aimed at addressing the challenges listed in Sect. 3, are proposed at the end of this section.

''Pending''
Rule 36(1) of the Implementing Regulations ascertains that applicants ''may file a divisional application relating to any pending earlier European patent application''. What constitutes as ''pending'' is not defined in the EPC. 116 It obviously covers the period of consideration of the original application; it also covers the period during which an appeal is deliberated. The EBA shed some more light on this in G 01/09, when it found that a patent is still pending even if it has been refused, and no appeal is lodged but there is still time to file an appeal. 117 More specifically, ''a pending (earlier) European patent application … was a patent application in a status in which substantive rights deriving there from under the EPC were (still) in existence''. 118 As such, one can file divisionals until the time limit to file an appeal lapses and one can do so without actually lodging an appeal. Appealing for the sake of appealing may however prove beneficial as then the clock keeps running, allowing for at least another 2-4 months to file divisionals, as is the time limit to file appeals. 119 This time period extends all the way up to right before an appeal decision is made seeing as that is the point the parent is no longer pending. Because the patent process, including the appeal process, takes years -indeed even up to 20 years 120 -a patent can be pending if a family member is challenged, allowing companies to file divisionals in hopes of maintaining patent protection should the parent be revoked.
Said appeal need not be successful or even admissible for the divisional to be filed and granted. In J 23/13, the EPO Legal Board found that ''the fact that the appeal was later rejected as inadmissible could not change the fact that, on the date when the divisional was filed, substantive rights were still in existence' '. 121 In this case, the EPO initially refused the parent so the applicant subsequently filed an appeal and then later a divisional before the time limit for filing an appeal lapsed. The appeal contained no statement of grounds for why it was filed. It was effectively blank and therefore rejected as inadmissible. Applicants can file a written statement laying out the grounds of the appeal after filing notice of appeal, meaning it is normal for appeals to be blank initially and for up to four months after the date of notification of the decision, as is the deadline to file the statement of grounds of appeal. 122 However, in J 23/13, the applicant filed the appeal on the 10 August 2010, but failed to file the statement of grounds of appeal by the 14 October 2010 deadline, causing the EPO to reject the appeal on the 20 May 2011, giving the impression that the applicant only filed the appeal in order to have more time to file a divisional. The divisional application was nonetheless permissible because it had been filed before the time limit.
As long as the parent application is pending, a divisional may be filed. Applicants can then file further divisional patents on those first divisional ones, thus further dividing divisionals into another set of divisional applications, 123 and then into further generations again, creating a whole sequence of divisionals. There is no limit to how many divisionals applicants can amass per se. Provided each divisional relates to the divisional from which it stems, one can in theory continue filing while the previous divisional or parent application is pending. And, as is shown below, an applicant's next, nearly identical divisional may very well succeed regardless of past failures. Filing a divisional each time a parent is rejected or until it is accepted is therefore a strategic way of keeping the hope (or threat) of a patent claiming almost identical subject-matter alive. As an Enlarged Board of Appeal (EBA) observed, there is a ''practice of filing a sequence of substantially identical divisionals and pending for up to 20 years for strategic reasons''. 124

Same Subject-Matter
Article 76(1) stipulates that a European divisional application ''may be filed only in respect of subject-matter which does not extend beyond the content of the earlier application as filed''. The test to assess whether a divisional complies with Art. 76(1) is the same as with Art. 123(2), namely, the disclosure test: the overall change in the content of the divisional in view of the parent, whether by way of addition, alteration or excision, may not result in the skilled person being presented with information which is not directly and unambiguously derivable from that presented in the parent, even when account is taken of matter which is implicit to a person skilled in the art. 125 If a divisional exceeds the subject-matter of the earlier application filed, it is not de lege invalid. 126 It can still be amended in proceedings post-filing, even after the parent is no longer pending. 127 The rationale for this limit is to protect ''the interests of third parties'' by preventing applicants from improving their positions ''by adding subject-matter not disclosed in the application as filed, which would give him an unwarranted advantage and could be damaging to the legal security of third parties relying on the content of the original application''. 128 Filing a divisional with subject-matter exceeding its parent is, however, not the only way to improve one's position. Filing a divisional claiming ''same subject-matter'' as a granted parent, also known as ''double patenting'', is another way. Albeit prohibited, the national cases in Sect. 3 clearly evince that this happens in practice. This can be attributed to the long-standing legal uncertainty surrounding the legal basis prohibiting double patenting, as well as its narrow application. Although the EBA, who is meant to ensure the uniform application of the EPC after a BOA refers a question necessary to this effect, 129 largely resolved this issue in 2021 case G 04/19, 130 ascertaining that Art. 125 131 provides the legal basis for prohibiting double patenting, three loopholes concerning ''same subject-matter'' remain. 124 131 Article 125 provides that the EPO ''shall take into account the principles of procedural law generally recognised in Contracting States'' when procedural provisions in the EPC are absent.

Three Loopholes
Despite the prohibition of double patenting, ''(almost) identical divisional applications are still applied for and granted''. 132 This is in large parts due to G 1/05 133 and G 1/06, two decisions where the EBA not only established that the double patenting prohibition for divisionals was based on whether applicants had a legitimate interest in filing a second patent for the same subject-matter protected by another granted patent (now overturned by G 04/19), but created three legal loopholes which makes the law susceptible to misuse. Since the EBA focused on the legal basis for the prohibition in G 04/19, not even commenting on what constitutes as a ''same invention'' 134 these three loopholes remain.
First, in G 1/05 and G 1/06 the EBA affirmed that it is established practice that EPO first-instance departments reject amendments making a parent and divisional identical ''when the amended divisional application claims the same subject-matter as the pending parent application'' 135 but did not define what ''same subject-matter'' entails. This has been left to individual TBAs to decide. TBA decisions show that double patenting occurs if the patents at issue concern the same invention, 136 while there is no double patenting if none of the claims in the patents in question contain the same features 137 or, if, in a family of patents, one concerns a product, and the other a manufacturing 138 or fabricating 139 method as the claims protect different aspects of the same invention. How different subject-matter must be to not be identical is less clear, though most TBAs apply the double patenting test narrowly.
In T 118/91, the BOA found that parts of the subject-matter claimed in the divisional can also form part of subject-matter in the parent application. 140 And, moreover, that there is generally no double patenting when an application claims ''its own subject-matter in combination with that of the other application''. 141 This was inter alia repeated in T 587/98, 142 where the examining division initially refused the divisional on the grounds that its subject-matter overlapped with its parent to the extent that ''granting the divisional would amount to granting a second patent for the same invention''. 143 The TBA reversed the decision and granted the patent. While referring to EPO Guidelines, which provide that if patents claim ''separate and distinct elements A and B respectively, one of the two applications may also include a claim for A plus B'', the TBA found that a divisional may recite all the claims of its parent and include a combination with an additional feature. 144 In other words, a divisional may claim its ''its own subject-matter in combination with that of the other application''. 145 For example, A could be the compound itself, while B would be the dosage and AB, use of said compound with said dosage. An AB combination can also obtain patent protection even if A is an obvious variant provided the patent is not pre-published. 146 In this regard, the TBA acknowledged that ''it may perhaps be thought that such a prohibition is desirable to prevent the proliferation of patents for obvious variants'' but that ''such a prohibition would require a legal basis''. 147 The TBA issued these decisions before G 1/05 and G 1/06, but they remain applicable.
A mere week after G 1/05 and G 1/06, the TBA agreed with both decisions in T 307/03, further finding that double patenting objections ''can be raised where subject-matter of the granted claim is encompassed by the subject-matter of the claim later put forward''. 148 And, furthermore, that it could not, ''regard the extent of double patenting here as something that can be ignored as de minimis, given that the subject-matter which would be double patented is stated to be the preferred way of carrying out the invention of the present application''. 149 However, later decisions have confined ''same subject-matter'' to identical subject-matter wherein partial overlaps are not per se problematic. 150 In T 1491/06, the TBA affirmed that ''claims in a parent and divisional relating to overlapping, but non-identical, subjectmatter should not be objected to ''. 151 In this case, the original divisional contained claims identical to its parent and was rejected. The TBA accepted the patent because the applicant amended its claims to be more narrow and discernible from the parent, making them only partially overlap. In T 0318/14, the TBA confirmed that ''case law has confined the term ''double patenting'' to the impending grant of two patents claiming the same subject-matter''. 152 Furthermore, it recognised that an applicant first trying to obtain a patent for a preferred embodiment and then subsequently filing a divisional for the general teaching constitutes legitimate procedural behaviour. In sum, these decisions reveal that the TBA approaches ''same subjectmatter'' strictly wherein minor deviations circumvent double patenting objections.
Second, G 1/05 and G 1/06 laid the basis for creating a distinction between claiming overlapping subject-matter and overlapping protection. If the patents do not define the same subject-matter, but confer overlapping protection, there is no double patenting according to T 1391/07. 153 The TBA found that the legitimate interest approach, as laid down in G 1/05 and 1/06, could not be invoked in such cases because there was no ''manifest objective reason to deny the legitimate interest of the applicant in obtaining a protection different from -although partially overlapping with -that of the parent patent already granted''. 154 As such, it saw no basis for extending the existing practice to ''cover claims not defining the same subject-matter but conferring … a scope of protection overlapping with each other only partially in the sense that some, but not all of the embodiments notionally encompassed by one of the claims would also be encompassed by the other one of the claims''. 155 Multiple TBA decisions confirm that the general rule is that ''the mere fact that the scope of protection … would partially overlap with that of the granted parent patent does not prejudice the grant of a patent''. 156 There is therefore no double patenting when the same medical compound is protected by different types of claims, conferring different scopes of protection but with some protection overlap. 157 In less abstract terms, there is no issue if the medicine is protected with a product patent and purpose-limited product claims (X for use in the treatment of Y). Both protect the same medicine, though the former provides a wider scope of protection. The TBA similarly found in T 2402/10 that there was no double patenting even though the applicant (Pfizer) amended the first claim so that its ''scope fully encompassed the scope of the claims granted … from the grandparent'' during opposition/appeal proceedings because the patents at issue claimed different subject-matter. 158 The claims of said patents were identical except for Claim 1 of the divisional which specified a dosage regimen and that the treatment was ''in a human being''. This was one of the patents subject to the Italian administrative court ruling for the drug Xalatan mentioned above. The EBA upheld that a distinction must be made between ''double protection (claims with overlapping scope) and double patenting'' in G 04/19. 159 As such, it appears that it is permissible to obtain broader protection in a divisional than an already granted, narrower parent.
Third, the EBA focused on filing patent applications claiming subject-matter already protected by a patent, meaning filing and acquiring a patent with identical subject-matter to a rejected patent is not prohibited. In T 441/92, the TBA reversed the decision to refuse a divisional on the basis that it claimed substantially identical 153  subject-matter to its parent by finding that double patenting could not arise when the parent ceased to exist. 160 The Board concluded that, ''there is nothing in the EPC to prevent an Applicant from repeating the parent description in a divisional application and that there is no contravention of Art. 76(1) in this respect in the present case''. 161 The TBA similarly decided that re-filing a divisional ''with a text identical to'' an earlier divisional does not provide a legal basis to revoke a divisional in T 1287/07. 162 The appellants argued that ''the re-filing of the same subject-matter [from an earlier divisional] was an abuse of procedure, which, albeit not explicitly contravening any provision of the EPC, justified the revocation of the patent in suit'' but as no EPC provision gives such grounds, the TBA held that this cannot be the case. 163 Re-filing exactly the same subject-matter as another rejected parent or divisional is therefore legally permissible.
The above decisions are representative: almost all the decisions relating to double patenting in the 2019 EPO Case Law book, as well as later ones, are laid out. There are more ''cases'' of divisionals of course. Many are abandoned, amended or granted, even though they might be problematic. As the decisions reveal, the double patenting test is applied narrowly, whereby there are several ways to circumvent double patenting objections. This is understandable seeing as any material change to claim language equals a new invention. Using ''identical'' as the benchmark for what amounts to double patenting is also easier than introducing rules stipulating that any small difference in claim language is prohibited as this might open a can of worms. However, the double patenting test is applied so narrowly that divisionals, albeit worded differently, can protect the same subject-matter as the granted parent or, indeed, use almost identical claims as parents that have ceased to exist. Because double patenting is not per se allowed and each divisional results in its own patent, this means that revoking the parent does not eliminate the divisional because they should not be identical. One must attack each patent, individually. Constructing a large family tree is thus an apt way of ensuring that if even if the root of the tree (parent) is chopped off, each individual branch (divisional) somehow still stands.

Patent Law Reform
Ways to address concerns relating to excessive pricing and access to medicines already exist, including through competition law, human rights law, and administrative law, in addition to post grant mechanisms, such as compulsory licenses. Sometimes these mechanisms are effective, but it largely depends on the context. Each mechanism also requires improvement. 164  meant to resolve such concerns. But that does not mean patent law should not address the problems it creates, particularly when these problems exacerbate preexisting issues. Patent reform at the EPO level is needed. Previous EBA decisions likewise lay the basis for legal reform, particularly G 01/16, where the EBA reaffirmed that, It appears that what applicants consider a legitimate exploitation of the procedural possibilities afforded by the EPC, others consider an abuse in relation to the law as they think it ought to be rather than as it is. The Board finds it unsatisfactory that sequences of divisional applications each containing the same broad disclosures of the original patent application, by means of at least an unamended description, should be pending for up to 20 years. If administrative measures, such as giving priority to the examination of divisional applications and bundling and speedily deciding co-pending divisional applications so as to minimize the possibility for applicants to keep alive subject-matter on which the Examining Division had already given a negative opinion in one application by means of refiling the same subjectmatter again and again, are not adequate, it would be for the legislator to consider where there are abuses and what the remedy could be. 165 The whitepaper published by Medicines for Europe in 2021 makes several recommendations to curtail divisional misuse, including reducing the patent examination time lag and stricter filing criteria. 166 These are laid out and built on, though I also propose that the EPO should: confine the meaning of ''pending''; require applicants to fill out a standardised document when filing divisionals; allow third-party observations concerning double patenting; include double patenting as grounds to both oppose and revoke patents. These measures can work in tandem to increase the efficiency and quality of patent examinations.

Reduce Time Lag
There can be a significant time lag between the filing and the grant of the patent. Given the increase in patent filings, the amount of documentation required per patent application, as well as the effort and time patent examiners allot per application, it is understandable that the examination process takes years. It is also advantageous that each application is examined in detail to ensure that it meets the patentability requirements. Conversely, when it comes to divisionals, too much of a time lag is disadvantageous because it means that divisionals may very well still be pending during the revocation proceedings of its parent, allowing for more divisionals to be filed with claims similar, if not identical to the revoked patent. The EPO allows for accelerated examinations once the application is in the hands of the examining division. 167 However, this must be requested by the applicant. As it is 165  beneficial to have pending patents in the pipeline, relevant applicants might not use this mechanism.
Medicines for Europe recommends that, during examination or opposition proceedings of a divisional, all documents pertaining to the parent should be considered ''to give continuity to the arguments and evidence used in each patent family''. 168 This information is contained in the public register. All patent examiners must do is look at it. But this will not alone expedite the process. Patent examiners would still have to do the same amount of work. The EPO should therefore assess how the patent examination period might be reduced, particularly how patents may be examined more quickly to prevent divisionals from pending for years. One suggestion is to incorporate algorithms to aid and expediate the patent examination process. Said algorithms could for instance identify patent families and detect double patenting by comparing claim language between the parent and divisional.

Stricter Filing Requirements
Medicines for Europe recommends that applicants may only file divisionals with ''claims not previously included in the parent application (or other members of the family)'' and must submit documentation explaining why these claims were not included from the start, as well as ''disclose all relevant documents from any parent application's examination or opposition file and explain the reasons for filing such a divisionals''. 169 This thus places a greater burden on applicants to prove why a divisional is necessary, as opposed to merely filing divisionals to serve as backups. It also entails more effort and higher patent attorney expenses for the applicant which may have a deterring effect.
Documentation overload may, however, slow the EPO examination process. The EPO should therefore create a standardised document with set questions and words limits that all applicants must use. One question that should be asked is how a subsequent divisional differs from the parent. This information can be verified by comparing the two applicants because, as Lai et al. state, ''the EPO will keep the file of the parent for at least the same time as the file of the divisional'' so it ''will be available for inspection to assess the extension, even if the parent has been withdrawn before publication … or abandoned later''. 170 Such a comparison may reduce instances of double patenting.
In addition, the wide meaning of ''pending'' should be confined. Imposing a deadline sounds, at first blush, like a reasonable solution. Yet when the EPO introduced a 24-month limit to file divisionals, after the EC's 2009 report, this backfired. Fears that filing divisionals would no longer be possible prompted applicants to file even more divisionals. The EPO noted that, ''due to the new regulation, the number of divisional applications increased sharply in 2010 and, despite falling again by several thousand in 2011, still remained far above the levels recorded in and before 2009''. 171 As such, the EPO removed the time limit rule in 2013. Another solution is to require that appeals for refused parents must be admissible. This is to stop the practice of applicants submitting blank appeals to extend the period in which they may file divisionals. If the appeal is not admissible, the divisional should be refused.

Third-Party Observations Concerning Double Patenting
The EPC allows for third parties to submit observations following the publication of a patent. Such observations are then shared with the applicant for comments. Observations regarding an invention's patentability (or exceptions thereof), novelty and inventive step, as well as the applications clarity, its sufficiency of disclosure and unallowable amendments, such as those extending a divisional's subject-matter beyond its parent, may be submitted. 172 When patentability is called into question, the observation ''must be taken into account in any proceedings pending before a department of the EPO until such proceedings have been terminated''. 173 Observations can notably not be directed at double patenting. This should be allowed wherein such observations should, as with patentability, be considered during EPO proceedings to reduce the instances of double patenting.

Double Patenting as Grounds to Oppose and Revoke
Although the EPC prohibits double patenting, divisionals claiming near identical subject-matter are granted in practice. Expecting patent examiners to commit no mistakes is unrealistic, but so is pretending that double patenting never occurs. The new EBA 2019 decision may reduce the number of double patenting instances but the decades long battle of finding a statutory basis to prohibit double patenting, with some TBA boards stating in 2010 that ''in the absence of such provisions, a refusal of a European patent application for double patenting is not possible irrespective of whether or not double patenting is a rare occurrence'', 174 means that several problematic patents remain in effect. Parties should be able to challenge patents on the basis in which they should be invalid and to create precedence as such. Thus, as double patenting is prohibited, it should be included as grounds to oppose the patent at the EPO or to revoke the patent in a national court.

De-Designations and the Unified Patent Court
As the cases above show, actors strategically de-designate countries to avoid negative judgements based on which market it is cheaper to lose in order to retain another. After a country has been de-designated in a parent it cannot be redesignated in a divisional. 175 This is an important safeguard. But it does not solve the issue of actors de-designating countries to avoid negative judgements in one court to avoid influencing a judgement elsewhere and forum shopping, or generic/ biosimilars manufacturers spending time and money on legal proceedings unduly. The EPC can do little to remedy this as actors are largely free to determine where to raise suit.
The proposed Unified Patent Court (''UPC'') appears the best way to assuage these concerns. When the UPC is up and running, it will have exclusive competence on Unitary Patents and European Patents, including on infringement and validity, and form part of Member States' judicial system. 176 With a ''unified'' decision on a patent and thus fewer foreign judgements to influence, there will be little point in strategically de-designating countries. In turn, there are not only fewer incentives to engage in forum shopping, but the UPC will also save generic/biosimilar manufacturers significant time and money as they must no longer raise suit on the same patent(s) in an array of countries. For the Member States who have signed the UPC agreement, the UPC thus provides an opportunity to curtail divisional misuse, while this issue will likely remain problematic for countries who have not, such as the UK and Norway. 177

Stakeholder Working Group
The issue of patents and pharmaceutical prices is not new. However, despite significant case law and academic literature, European agencies have done little to engage with this issue at the European level. This should change. A working group comprising key stakeholders, such as the EPO, European Medicines Agency and the EC, should be assembled and tasked with investigating whether and how the patent system is open to misuse to maintain high medicinal prices and how these loopholes may be closed. This includes launching an overarching review over how divisionals are used in practice, where patent practitioners can, for example, assess how claim language has evolved over the years, particularly regarding double patenting. The EPO should also publish the number of divisionals filed for each technical field in its annual report, as it did in 2011 after the time limit rule.

Conclusion
This article has shown two main things. First, that divisionals are misused in practice. They are used strategically to deter and delay generic/biosimilar market entry, notably showing that procedural rights can do so just as much as substantive rights can. This development not only demonstrates that further research into the misuse of procedural rights is needed, but that procedural rights form part of strategic patenting practices by pharmaceutical firms. Second, that patent law, as it stands, enables such misuse to occur. EPO Board Decisions reveal that there are several ways to circumvent double patenting objections, such as by making minor deviations or by using different claim language to describe what is, in practice, effectively the same invention but is considered a new invention under the EPC, allowing an applicant's next, near identical divisional to very well succeed regardless of past failures. Divisionals are meant to help applicants overcome unity objections and otherwise help applicants obtain patent protection to ensure that their inventions are protected so the R&D cycle continues. That is vital. Divisionals should not, however, be used strategically to hinder generics/biosimilars market entry once the original compound patent expires to maintain a market monopoly and high prices. Using divisionals in this way is a misuse of the procedural right. When the law yields such undesirable results, legal reform is necessary. The reforms put forward in this article -procedural improvements, stricter patentability requirements and a ''raising of the bar'' -will help prevent divisional misuse and protect the ''ratio legis'' for divisionals. Without reform, this cycle will continue.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creativecommons.org/licenses/by/4.0/.