Claim Construction as a Constraint on Obviousness and the Comptroller’s Role on Appeal: Abbott v Dexcom [2025] EWCA Civ 1633

1. Introduction

This commentary examines the Court of Appeal decision in Abbott Diabetes Care Inc & Ors v Dexcom Incorporated & Ors [2025] EWCA Civ 1633, a significant judgment at the intersection of patent law, medical-device technology, and appellate procedure.

The appeal concerned European Patent (UK) No. 3 730 044, entitled “Compact on-body physiological monitoring device”, owned by Abbott. The patent relates to integrated continuous glucose monitoring (“CGM”) systems with a combined sensor and electronics assembly and a specific form of insertion device.

At first instance, Mellor J had revoked the patent for obviousness over a prior art publication known as “Heller” (US 2003/01000821). Abbott and Dexcom settled their commercial dispute, but Abbott nevertheless pursued an appeal solely on the question of validity, seeking to overturn the revocation order. Dexcom did not participate; instead, the Comptroller-General of Patents intervened to assist the Court.

The Court of Appeal (Arnold, Cobb and Lewison LJJ) allowed Abbott’s appeal and restored the patent, holding that the trial judge’s obviousness reasoning was inconsistent with his own construction of the claims and that there was no evidential basis for a finding of obviousness on the claim as properly construed.

The judgment is important for at least three reasons:

• Doctrinally, it emphasises that a finding of obviousness must be coherent with claim construction: a judge cannot base obviousness on a configuration (here, automatic insertion) that the claim does not in fact cover (here, manually driven insertion).
• Practically, it illustrates the evidential rigor demanded when modifying mechanical prior art (altering modes of operation, e.g. automatic to manual) to reach an allegedly obvious configuration.
• Procedurally, it clarifies the limits of the Comptroller’s role in unopposed appeals from patent revocations, holding that the Comptroller cannot serve a respondent’s notice unless made a party to the appeal.

2. Summary of the Judgment

2.1 The Patent and the Prior Art

The patent concerns an integrated analyte monitoring assembly comprising:

• a sensor electronics assembly including:
  • an analyte sensor (notably a glucose sensor);
  • sensor electronics, including a power supply, controller, and activation switch; and
• an insertion device including:
  • a housing;
  • an introducer needle “coupled to the housing”, configured to move between first and second positions; and
  • a bias mechanism “operatively coupled to the housing”, configured to automatically retract the needle to a safe, fully internal position.

Claim 1 is limited to an arrangement where the insertion movement is manual (force applied to the housing by the user moves the needle), while the retraction is automatic (spring or bias mechanism-driven), as a result of how the trial judge construed “coupled to the housing” versus “operatively coupled to the housing”.

The principal prior art, Heller, discloses:

• a mainstream two-part CGM architecture (sensor + on-skin sensor control unit) with insertion guns and spring-loaded insertion and retraction (Figure 26, Figure 33); and
• an integrated disposable sensor/transmitter/battery on a substrate (“Fig. 32 Configuration”).

The key factual issue was whether it was obvious for the notional skilled team to combine the Figure 32 integrated sensor-electronics configuration with the Figure 33 insertion gun and thereby arrive at the claimed integrated CGM assembly.

2.2 First-Instance Decision

Mellor J:

• identified the skilled team as an electronic engineer and a mechanical design engineer working within a broader CGM development team;
• described substantial common general knowledge about CGM systems, insertion devices, sterilisation and electronics; and
• construed claim 1 so that:
  • the phrase “coupled to the housing” in integer 1.9 meant the needle and housing move together for insertion, i.e. manual insertion by pushing on the housing; and
  • the phrase “operatively coupled to the housing” in integer 1.10 described a bias mechanism (spring) which automatically retracts the needle, involving relative movement between needle and housing.

On that basis, he held that all claims were obvious over Heller. In particular, he concluded that a modified version of Heller’s integrated Fig. 32 unit, inserted using the spring-driven Figure 33 insertion gun (with automatic insertion and retraction) and suitably housed, would satisfy integers 1.7–1.10 of claim 1. He rejected other invalidity grounds and found Dexcom’s G7 device did not infringe on his narrow construction.

2.3 The Appeal and Main Holdings

Abbott appealed on seven grounds. The Court of Appeal:

• Dismissed grounds 1 and 2:
  • No error in treating Fig. 32 as the starting integrated configuration.
  • No impermissible “hybrid” starting point; the judge was entitled to regard Heller as offering a menu of compatible options, including combining Fig. 32 with Figure 33.
• Dismissed ground 6 (improper adoption of Dexcom’s submissions) and ground 7 (delay) as free-standing grounds, though both issues helped explain certain errors.
• Allowed grounds 3 and 4:
  • The judge had mis-identified the “housing” relevant to integers 1.7–1.10, conflating the housing of the sensor electronics assembly with the housing of the insertion device.
  • He had found obviousness based on a configuration involving automatic insertion, inconsistent with his own construction that the claim is limited to manual insertion and with his earlier dismissal of an insufficiency squeeze premised on automatic insertion.
  • There was no evidential basis to show that a manual-insertion / automatic-retraction variant of Heller’s Figure 33 insertion gun would be obvious.
• Found it unnecessary to deal with ground 5 (claims 2–3) because obviousness of claim 1 failed.

Consequently, the Court set aside the revocation order and restored the patent, explicitly noting that the outcome turned on:

1. the narrow construction of integer 1.9 (“coupled to the housing”); and
2. the absence of evidence that the claimed manually-inserted configuration was obvious over Heller.

2.4 The Comptroller’s Role

Critically, the Court examined the role of the Comptroller in appeals from revocation orders where the original defendant no longer contests the appeal. Drawing on Halliburton and Aerotel, and applying CPR Part 52 and Practice Direction 52D, the Court held:

• The Comptroller may attend, assist and oppose in an objective, non-partisan manner, but is not thereby a “respondent” to the appeal.
• Under CPR 52.13(1) and the definition of “respondent” in CPR 52.1(3)(e), the Comptroller cannot serve a respondent’s notice unless joined as a party.
• Nevertheless, the Comptroller can – and should – flag any “respondent’s notice-type” points in advance via his skeleton argument or other communication with the appellant.

This is a clear procedural clarification of the Comptroller’s status and powers on appeal.

3. Precedents Cited and Their Influence

3.1 Obviousness Framework

• Actavis Group PTC EHF v ICOS Corp [2019] UKSC 15, [2019] Bus LR 1318
  • Lord Hodge’s judgment is the main modern authority on the fact-specific, multi-factorial nature of obviousness under s 3 of the Patents Act 1977.
  • Key aspects invoked:
    • The statutory definition: an invention involves an inventive step if it is not obvious to the skilled person having regard to the state of the art (judgment [89]).
    • Endorsement of the Pozzoli structured approach to obviousness (judgment [90]).
    • Recognition that courts must weigh factors such as motive, number of research avenues, and expectation of success (judgment [91]).
    • The prohibition on hindsight and the danger of stepwise reconstructions of the inventor’s path (judgment [92]).
  • The Court of Appeal treated Mellor J’s assessment as a multi-factorial evaluative judgment; thus, under Actavis v ICOS, they would not intervene unless he erred in law or principle (judgment [136]).
• Pozzoli SPA v BDMO SA [2007] EWCA Civ 588, [2007] FSR 37
  • Jacob LJ’s four-step test for obviousness (judgment [90]) was recognised as the usual, though not mandatory, structure:
    1. Identify the skilled person and common general knowledge.
    2. Identify the inventive concept or construe the claim.
    3. Identify differences between prior art and claim.
    4. Ask whether these differences would be obvious without knowledge of the invention.
  • Arnold LJ notes that Mellor J intended to follow this scheme ([298]) but did not explicitly map the Heller/Fig. 32–33 combination onto each integer of claim 1 – a gap that likely contributed to error.
• Conor Medsystems Inc v Angiotech Pharmaceuticals Inc [2008] 4 All ER 621 and Generics (UK) Ltd v H Lundbeck A/S [2007] RPC 32
  • Lord Hoffmann in Conor endorsed Kitchin J’s non-exhaustive list of factors relevant to obviousness in Lundbeck (judgment [91]).
  • This underpins the Court’s adherence to a contextual, rather than mechanistic, approach to obviousness.

3.2 Standard of Appellate Review

• Actavis v ICOS (above)
  • Reaffirmed that appellate intervention in obviousness is limited to errors of principle or law ([136]).
• Lifestyle Equities CV v Amazon UK Services Ltd [2024] UKSC 8, [2024] Bus LR 532
• Iconix Luxembourg Holdings SARL v Dream Pairs Europe Inc [2025] UKSC 25, [2025] Bus LR 1391
  • Both Supreme Court cases confirm a cautious approach to overturning primary evaluative findings in IP disputes, emphasising judicial restraint unless clear errors are shown (judgment [136]).

3.3 Delay and Judgment-Writing

• Bank St Petersburg PJSC v Arkhangelsky [2020] EWCA Civ 408, [2020] 4 WLR 55
  • Affirms that, even in complex cases, judgments should generally be handed down within three months (judgment [137]).
• Phones 4u Ltd v EE Ltd [2025] EWCA Civ 869
  • Recognises that the three-month rule is not absolute (judgment [137]).
• NatWest Markets plc v Bilta (UK) Ltd [2021] EWCA Civ 680
  • Holds that serious delay requires the appellate court to exercise special care in reviewing the evidence and reasoning (judgment [137]).
  • The Court applied this in reviewing Mellor J’s reasoning, especially where internal inconsistencies emerged.

3.4 Adoption of Parties’ Submissions

• Crinion v IG Markets Ltd [2013] EWCA Civ 587, [2013] CP Rep 41
  • In Crinion, the trial judge lifted almost the entire judgment from one party’s submissions; the Court of Appeal criticised the practice but did not set the judgment aside.
  • In the present case, only limited paragraphs of Mellor J’s judgment were drawn from Dexcom’s closing submissions (judgment [139]).
  • The Court held this was not comparable to Crinion and did not constitute an error in itself, although it likely contributed to the specific mistakes on housing and manual/automatic insertion ([140]).

3.5 Comptroller’s Role in Appeals

• Halliburton Energy Services Inc’s Patent [2006] EWCA Civ 185, [2006] RPC 26
  • Established that where a patentee appeals a revocation order and the original defendant does not oppose, the Comptroller should ordinarily be invited to assist the Court, with his costs borne by the patentee (judgment [126]).
  • Jacob LJ stressed that the Comptroller’s role is not to provide a “full-blooded opposition” but carefully balanced assistance in the public interest.
• Aerotel Ltd v Telco Holdings Ltd [2006] EWCA Civ 1371, [2006] RPC 7
  • In Aerotel, counsel for the Comptroller became an amicus curiae in relation to one of the appeals, defending the judgment below objectively (judgment [127]).
  • This case illustrated in practice how the Comptroller may step into a defensive role when the original respondent settles.
• Apimed Medical Honey Ltd v Brightwake Ltd [2012] EWCA Civ 5, [2012] RPC 17
  • In Apimed, the Comptroller had filed a respondent’s notice, but the Court in Abbott v Dexcom noted that the underlying power to do so was not examined there (judgment [133]).
  • The present judgment clarifies that the Comptroller cannot file a respondent’s notice unless joined as a party.

4. The Court’s Legal Reasoning

4.1 The Skilled Team and Common General Knowledge

The skilled team consisted of:

• a mechanical design engineer; and
• an electronic engineer,

both operating within a larger CGM product development environment ([4]–[5]). Neither needed prior CGM development experience; they would draw on knowledge of comparable medical and non-medical devices.

The common general knowledge (“CGK”) was extensive and included:

• the nature and management of diabetes, and the limitations of finger-prick blood glucose monitoring ([11]–[14]);
• the architecture of incumbent CGM systems (Navigator, Dexcom STS-7, Medtronic Guardian) with:
  • a subcutaneous electrochemical sensor;
  • a separate on-body electronics unit;
  • a distinct insertion device; and
  • a separate reader ([16]–[19], [25]);
• short life of sensors (3–7 days) versus longer life and higher cost of electronics, explaining the prevailing two-part, non-integrated architecture ([20], [24]–[26]);
• basic electrochemical sensor fabrication (e.g. enzyme deposition, screen printing, dip coating) and the need for sterilisation, with possible incompatibility of some methods (steam, gamma, e-beam) with sensor chemistry ([21]–[22]);
• use of on-body power supplies (various battery configurations) and off-the-shelf electronic components; ASICs were not yet economically attractive ([23]–[26]);
• insertion devices: typical components (housing, needle, insertion and retraction mechanisms) and known variants:
  • manual insertion/manual retraction;
  • automatic insertion/manual retraction; and
  • automatic insertion/automatic retraction ([27]–[30]).

A notable point, later important for the Court of Appeal, was that no existing device used manual insertion with automatic retraction ([28]). The patent’s configuration is thus a new combination of known functional elements, not a radical new component.

The judge also found that the specification assumed further CGK in areas like sensor construction, needle design and actuation mechanisms ([32]): the patent is high-level and leaves practical mechanical implementation to the skilled engineer.

4.2 The Patent’s Core Technical Idea

Despite opaque drafting, the patent essentially proposes:

1. An integrated on-body device where sensor and sensor electronics are pre-connected and supplied as a single unit to the user, with:
   • a compact housing;
   • adhesive patch for skin attachment;
   • wireless communication to a remote reader ([40]–[46]).
2. Use of an insertion device that:
   • houses the integrated assembly before use ([46], [49]);
   • enables insertion by user-applied pressure on the housing (manual insertion); and
   • employs a spring-bias mechanism for automatic retraction of the introducer needle, leaving the sensor in place ([49], [53] esp. [0151]).
3. Provision of a cap that mates with the open end of the insertion device housing to maintain sterility ([54], claims 2–3).

The relevant claims (1–3) distil these ideas. The appeals turned on:

• integer 1.8: “a housing” – of the insertion device;
• integer 1.9: an introducer needle “coupled to the housing” – central to the manual insertion limitation; and
• integer 1.10: a bias mechanism “operatively coupled to the housing” and configured to automatically retract the needle fully into the insertion device.

4.3 Claim Construction: Manual vs Automatic Insertion

The most consequential construction issue was the meaning of “coupled to the housing” and “operatively coupled to the housing” ([59]–[66]).

The trial judge drew a deliberate contrast:

• Integer 1.9 (“coupled to the housing … configured to move”) was read as:
  • requiring no relative movement between the needle and the housing during insertion; and
  • insertion being caused by movement of the housing, i.e. the user’s manual downward force on the housing pushes the needle into the skin.
• Integer 1.10 (“operatively coupled … configured to automatically retract”) implied:
  • a relative movement between needle and housing during retraction; and
  • retraction powered by stored energy in a spring or equivalent bias mechanism.

This reading was anchored in the detailed description of Figures 12A–12G and paragraph [0151] of the patent ([53]). Although the specification contemplated “automated or semi-automated” insertion in general ([0146]), that broader teaching was not claimed ([66]). On the judge’s construction, claim 1 was confined to:

• manual insertion, via direct user pressure on the insertion device housing; and
• automatic retraction, via spring/bias mechanism.

Critically, Abbott accepted this narrow construction on appeal. The appeal thus proceeded on the basis that the claimed invention is about a specific insertion modality, not all ways of inserting an integrated CGM unit.

4.4 Heller: Two Configurations, One Menu of Options

Heller discloses two broad configurations:

1. Main Configuration (two-part):
   • subcutaneously implantable sensor (Figure 2 et seq.);
   • on-skin sensor control unit (housing 45, base 74, cover 76) (Figures 14–16, [0212]–[0256]);
   • separate insertion devices:
     • Figure 12/13 – needle 120;
     • Figure 26 – insertion gun 200 using spring/compressed gas, with optional spring-based retraction ([0206]–[0208]);
     • Figure 33 – an insertion kit (insertion device 270, sensor 272, gun 274, mounting unit 276, spring 280), pre-loaded and optionally “cocked” with stored potential energy ([0255]–[0256]).
2. Figure 32 Configuration (integrated):
   • transmitter 263 (integrated circuit 264 + antenna 266), battery 262 and sensor trace 268 mounted on substrate 260 ([0230]);
   • transmitter encapsulated for protection ([0231]);
   • user inserts the implantable portion of the sensor transcutaneously and fixes the unit to the skin; the whole device is disposable ([0232]);
   • optionally, the battery could be separate ([0233]).

A central debate was whether Heller taught two “mutually incompatible” architectures (Abbott’s position) or provided a menu of interoperable options (Dexcom’s and Mellor J’s view). The Court of Appeal agreed with Mellor J that Heller presents multiple compatible permutations; a skilled team could reasonably decide to bring together:

• the integrated, disposable sensor/battery/transmitter of Figure 32; and
• the advanced insertion gun of Figure 33.

Thus there was no error in principle in taking the Fig. 32 Configuration plus the Figure 33 insertion kit as the relevant starting point for an obviousness assessment (ground 1 failed: [141]).

4.5 The First-Instance Obviousness Analysis and the “12 Steps”

Dexcom’s expert, Mr Varde, proposed that the obvious approach for a design engineer would be to:

• take Heller’s integrated Figure 32 unit;
• place it in a housing (akin to Figure 14 or a routine variation); and
• use the Figure 33 insertion gun concept, perhaps adapted in dimensions and mechanics, to insert this integrated unit safely and automatically, including automatic retraction of the needle ([93]–[97]).

Abbott countered by identifying 12 alleged “steps” required to get from Heller to claim 1. They argued that this was classic ex post facto reconstruction, requiring multiple motivated departures from Heller’s teaching. Mellor J considered these steps exaggerated; he found most to be routine design choices for a skilled mechanical engineer ([357]–[358], [391]).

Key findings:

• Step 1 – choosing the Figure 32 integrated configuration: essentially given by Heller’s teaching ([111]).
• Step 2–3 – providing a housing around the Fig. 32 unit: routine design ([362]–[363]).
• Step 4 – choosing Figure 33 as the starting insertion gun: not a “step” but a choice among explicit options in Heller ([364]).
• Step 5 – adapting the dimensions of Figure 33 to accommodate the Figure 32 unit: routine mechanical scaling and tweaking ([365]–[372]).
• Step 10 – opting for automatic retraction: again a disclosed option in Heller ([379]).
• Step 12 – adding a cap: a genuine step but an obvious packaging and safety measure ([384]–[387]).

On this basis, he concluded that a modified Figure 33 device providing automatic insertion and automatic retraction of a housed Fig. 32 unit met integers 1.7–1.10 ([374]), and thus claim 1 was obvious.

4.6 Ground 3: Misidentification of the “Housing”

On appeal, Abbott argued that Mellor J had failed to distinguish between:

• the housing of the sensor electronics assembly (an optional implementation detail for the Fig. 32 unit); and
• the housing of the insertion device required by integer 1.8 of claim 1.

This distinction is crucial: integer 1.9 requires the introducer needle to be “coupled to the housing” of the insertion device, not merely to some housing around the sensor electronics assembly.

The Court observed:

• Steps 2 and 3 in Abbott’s 12-step list – and the cross-examined evidence – concerned a housing for the integrated sensor-electronics unit ([145]).
• At [374], Mellor J concluded that a “modified version of the Fig. 33 device that would automatically insert the Fig. 32 integrated sensor/transmitter unit with a housing” satisfied integers 1.7–1.10.
• But that “housing” is not the housing in integer 1.8; it is a housing around the Fig. 32 unit.

The Court held:

[146] Abbott submit that it follows that the judge’s bald statement at the end of [374] that the device he postulated “satisfied integers 1.7-1.10 of claim 1” lacks foundation. The housing referred to by the judge in this paragraph is the wrong housing, and there is no reference to the housing of the insertion device.

The Comptroller did not contest this. The error appears to have arisen because Dexter’s closing submissions had the same conflation, and the judge adopted them almost verbatim ([147], [140]).

While this mistake might conceivably have been cured by revisiting Heller’s Figure 33 on a proper footing, it was intertwined with a second, more fundamental error – the manual/automatic insertion mismatch.

4.7 Ground 4: Manual Insertion vs Automatic Insertion

The more serious inconsistency lay in the fact that:

• Mellor J construed claim 1 as limited to manual insertion (integer 1.9: “coupled to the housing”);
• but his obviousness analysis, culminating in [374], relied on a device providing automatic insertion (spring-driven insertion gun of Figure 33).

The Court pointed out that:

[149] … the judge held in [374] that a system that automatically inserted the integrated sensor and sensor electronics “satisfied integers 1.7-1.10 of claim 1”. Abbott submit that this conclusion is flatly inconsistent with the judge’s construction of the claim. Furthermore, it is also flatly inconsistent with the judge’s resolution of the first insufficiency squeeze advanced by Dexcom at [393].

In [393], Mellor J had indicated that the “automatic insertion” insufficiency squeeze “does not arise due to my decision on construction” – i.e. because he held automatic insertion was not claimed. It was therefore contradictory to base obviousness on just such an automatic insertion configuration.

The Comptroller explored whether an appellate court could salvage the finding by positing an obvious manual-insertion variant of Figure 33. However:

• Dexcom had not run such a case at trial, nor had they adduced expert evidence to support a switch from automatic to manual insertion structure.
• Heller’s brief reference to “manual or mechanical” force in [0206] is directed at the earlier Figure 26 gun, not the more sophisticated Figure 33 device ([151]).
• Dexcom’s own expert had treated Figure 33 as superior and attractive precisely because of its automatic features.
• There was no explanation or evidence of a technical motivation for discarding automatic insertion (a desirable safety and usability feature) in favour of a “retrograde” manual insertion with automatic retraction, a configuration not present in the prior art ([151]).

As Lewison LJ observed during argument, any such case would have to confront the question: why would the skilled team be motivated to re-engineer a safe spring-loaded automatic-insertion gun into a manual-insertion device, particularly when no existing CGM device had used manual insertion with automatic retraction?

The Court concluded:

[152] I therefore conclude that the judge’s conclusion on obviousness cannot stand because it is inconsistent with his interpretation of the claim and there is no evidence to support a finding of obviousness of claim 1 on that interpretation.

This is the doctrinal core of the case: obviousness must be assessed against the claim as construed, and there must be evidence that the specific claimed configuration (here, manual insertion + automatic retraction) would be obvious to the skilled person, not merely some broader or different architecture (automatic insertion) disclosed in the prior art.

4.8 Grounds 1, 2, 5, 6, and 7

4.8.1 Ground 1 – “Hybrid” Starting Point

Abbott argued that Mellor J impermissibly started from a hybrid of two incompatible configurations in Heller (Main Configuration + Fig. 32). The Court rejected this:

• The judge’s starting point was properly the Fig. 32 integrated configuration ([141]).
• He then treated Heller as offering a series of options; choosing the Figure 33 insertion gun to work with the Fig. 32 unit was not itself an error of principle, subject to the manual/automatic point addressed above.

4.8.2 Ground 2 – Over-generalising Fig. 32

Abbott claimed the judge had treated Fig. 32 as a broad teaching of “integrated on-body units” of any configurational form. The Court found this inaccurate; Mellor J had taken Fig. 32 as a specific starting point and then considered routine modifications necessary to house it and mate it with Figure 33. There was no overarching legal error here ([143]).

4.8.3 Ground 5 – Claims 2–3 (Cap and Sterility)

Given that the Court overturned the obviousness finding for claim 1, it found it unnecessary to consider ground 5 relating to claims 2–3 ([153]). Mellor J had held that adding a cap and providing for sterility was obvious packaging design. That analysis stands formally untouched, but it is moot in light of claim 1’s survival.

4.8.4 Ground 6 – Adoption of Dexcom’s Closing Submissions

Abbott criticised Mellor J for incorporating passages from Dexcom’s written closing submissions almost verbatim in several parts of his judgment ([139]). The Court:

• distinguished the case from Crinion, noting that only a small portion of a long judgment was involved;
• noted that the judge had reorganised the material to match his analysis by reference to Abbott’s 12 steps; and
• accepted this practice was not of itself improper.

However, this copying probably contributed to the errors on housing and insertion mechanics, given that those passages carried over the same conflations as Dexcom’s submissions ([140]).

4.8.5 Ground 7 – Delay

The delay of nearly 14 months between trial and judgment was “inordinate” ([137]). However:

• Delay alone is not a ground for overturning a judgment.
• Rather, it necessitates “special care” in reviewing the reasoning, especially as delay increases the risk of internal inconsistency and misremembered nuances.
• Here, the Court saw the delay as contextual – it may help explain the internal contradictions – but the appeal succeeded on concrete errors of law/principle, not simply on delay.

4.9 The Comptroller’s Role and the Respondent’s Notice Issue

The Court provided important procedural guidance on the Comptroller’s participation in patent appeals from revocation orders (judgment [126]–[135]).

4.9.1 Practice Direction 52D, para 14.1

PD 52D 14.1 addresses appeals to the Court of Appeal from orders revoking patents. It provides, among other things, that:

• the appellant must serve the notice of appeal on the Comptroller in addition to other respondents (14.1(2));
• if the original respondent discontinues or decides not to oppose the appeal, they must notify the Comptroller (14.1(3)); and
• the Comptroller has 14 days to indicate whether he intends to attend and may attend and oppose the appeal (14.1(5)–(6)).

Arnold LJ commented that PD 52D 14.1(4)(a) is outdated (it refers to “petition” rather than modern originating process) and that, in practice, the Comptroller also needs transcripts, skeleton arguments and written closings to perform his role properly ([129]).

4.9.2 Is the Comptroller a “Respondent”? Can He File a Respondent’s Notice?

The key procedural question was whether the Comptroller, in this capacity, can serve a respondent’s notice under CPR 52.13.

The Court held:

• Under CPR 52.13(1), only a “respondent” may file a respondent’s notice.
• CPR 52.1(3)(e) defines a “respondent” as:
  1. a person (other than the appellant) who was a party below and is affected by the appeal; or
  2. a person permitted by the appeal court to be a party to the appeal.
• The Comptroller, when attending under PD 52D 14.1(6), is not automatically a party and so is not automatically a “respondent” ([134]–[135]).

Therefore:

[135] It follows that the Comptroller cannot serve a respondent’s notice. (It would be different if the Comptroller applied to be joined as a party.)

This is an important clarification. Apimed is reconciled by noting that the issue was not there examined; the present Court now provides authoritative guidance.

The Court recommended as a matter of practice:

• The Comptroller should notify the appellant of any “respondent’s notice-type” arguments sufficiently in advance (e.g. via skeleton argument) so that the appellant can address them ([135]).
• If the Comptroller wishes to have the fuller procedural powers of a “respondent”, he may seek to be joined as a party.

In Abbott v Dexcom, the Comptroller adopted this approach; he raised additional arguments in his skeleton, some of which (relating to grounds 3 and 4) he ultimately accepted were well-founded.

5. Complex Concepts Simplified

5.1 Continuous Glucose Monitoring (CGM) and Two-Part vs Integrated Systems

Traditional diabetes monitoring relies on blood glucose meters (BGM) using finger-prick samples. CGM systems instead place a small electrochemical sensor under the skin, which continuously measures glucose in interstitial fluid ([12]–[13], [20]–[21]).

Pre-2009 CGM systems typically were:

• “Two-part” / “non-integrated”:
  • A disposable sensor inserted via a separate insertion device.
  • A reusable sensor electronics/transmitter unit that the user then physically connects to the inserted sensor or to a mounting unit.

The Abbott patent instead describes an integrated on-body unit, where:

• sensor and electronics are permanently connected;
• the assembly is supplied pre-mounted within an insertion device; and
• the user does not carry out any post-insertion alignment or connection – both insertion and final on-body placement are achieved in one operation.

5.2 “Person Skilled in the Art” and Common General Knowledge

In patent law, the “person skilled in the art” is a legal construct: an unimaginative but competent technician (or team) with typical knowledge and abilities at the relevant date. They:

• know the common general knowledge (“CGK”) – widely accepted and used technical information in the field, not obscure or speculative ideas;
• can perform routine design and engineering tasks; but
• do not exercise inventive ingenuity.

Here, the skilled team includes both mechanical and electronic engineering expertise, familiar with CGM and analogues such as lancing devices ([4]–[5], [32]).

5.3 Obviousness and Hindsight

An invention is obvious if the skilled person, equipped with CGK and the prior art, would see it as an obvious step, not requiring inventive skill. Courts are careful to avoid hindsight bias – the temptation to reconstruct the invention once one knows the successful solution.

To avoid hindsight, the Court:

• uses structured tests like Pozzoli to separate:
  • what the prior art discloses; from
  • the claimed invention; and then
  • asks whether the differences would be obvious without knowledge of the patent.
• scrutinises stepwise “problem–solution” narratives that might smuggle in knowledge of the invention.

In this case, the Court accepted that Mellor J had generally approached Heller correctly, but found a specific divergence: his actual assessment relied on an automatic-insertion configuration not within the claim as construed.

5.4 “Housing” and “Coupled to the Housing” vs “Operatively Coupled”

A housing is simply the structural body or shell of a device. The patent involves two distinct housings:

• a housing for the sensor electronics assembly (optional – not explicitly claimed, but practically likely); and
• the housing of the insertion device, which:
  • is held by the user;
  • receives the integrated sensor-electronics assembly;
  • guides the introducer needle during insertion and retraction.

“Coupled to the housing” in claim 1 (integer 1.9), on the judge’s construction, means:

• the needle and the insertion device housing move together during insertion; and
• applying force to the housing directly causes insertion.

“Operatively coupled to the housing” (integer 1.10) means:

• the bias mechanism does not simply move with the housing, but is connected in such a way that it can perform its function (automatic retraction);
• on release of pressure, stored energy in the spring drives the needle back into the housing.

In short: “coupled” = moves with; “operatively coupled” = functionally connected to act upon.

5.5 Insufficiency Squeeze

An “insufficiency squeeze” is a forensic argument of the form:

• Either the patent is interpreted narrowly enough to be valid, but then it is not infringed; or
• It is interpreted broadly enough to be infringed, but then the patent cannot be sufficiently enabled across that broader scope or is obvious.

Dexcom raised three such squeezes (automatic insertion, activation switch, sterilisation). Mellor J considered that:

• automatic insertion did not arise on his narrow construction of the claim ([393]);
• activation switch and sterilisation could be implemented using CGK; the patent was not insufficient on those points, but this supported obviousness over Heller ([121]).

The Court of Appeal noted the inconsistency between the judge’s rejection of the automatic-insertion sufficiency squeeze and his reliance on automatic insertion for obviousness.

6. Impact and Significance

6.1 Impact on Obviousness Analysis in Mechanical and Medical Device Cases

The decision underscores several practical lessons for future obviousness assessments, especially in mechanical and medical-device contexts.

6.1.1 Coherence Between Claim Construction and Obviousness

A central takeaway is that the scope of the claim as construed constrains the obviousness inquiry. It is not enough to show that:

• some broader embodiment (e.g. automatic insertion + automatic retraction) is obvious over prior art; the court must assess the obviousness of the precise configuration claimed (here, manual insertion + automatic retraction).

Courts and parties must therefore:

• articulate constructions early and clearly;
• ensure that invalidity cases map integer-by-integer onto the construed claim; and
• avoid drifting into prior art configurations outside the claim’s true scope.

Judges should explicitly perform and record the Pozzoli step 3 comparison between the claim as construed and the prior art combination. Arnold LJ hints that this might have prevented the misalignment here ([141]–[142]).

6.1.2 Evidence is Needed for Mode-of-Operation Changes

When the alleged obvious development from prior art involves changing the mode of operation of a mechanical system (e.g. automatic to manual insertion), courts will require:

• a clear articulation of why a skilled person would contemplate such a change; and
• expert evidence that such a change is routine, not inventive, in the relevant context.

Here, moving from:

• Heller’s automatic-insertion + (optionally) automatic retraction devices, which emphasise repeatability and safety; to
• a manual-insertion + automatic retraction architecture not used in any known CGM device at the priority date,

was not supported by evidence or by any obvious technical motivation. That absence was fatal to the obviousness finding once the claim’s manual limitation was accepted.

6.1.3 Mechanical “Implementation Steps” vs Inventive Steps

The case also illustrates the fine line between:

• routine implementation details (choosing housings, changing dimensions, packaging components); and
• genuinely inventive configurations (e.g. choosing a previously unseen combination of manual insertion and automatic retraction for integrated CGM, in a context where two-part architectures were commercially entrenched).

While Mellor J was entitled to view many of Abbott’s 12 steps as routine, the Court of Appeal emphasised that:

• even if each step appears routine in isolation, the combination must be shown to be obvious as a whole; and
• there can be key “choice points” – such as the choice of insertion modality – that require a clear, evidence-backed rationale to be treated as obvious.

6.2 Implications for Patent Drafting and Prosecution

For drafters of medical device patents, the judgment highlights:

• Importance of clear language:
  • Vague or inconsistent terminology (“housing”, “coupled”, “operatively coupled”) can create litigation risk and evidential confusion.
  • Including consistory clauses that align closely with claims may reduce interpretive uncertainty ([33]).
• Expressly considering modes of operation:
  • If multiple embodiments (manual vs automatic, fully or partially automated) are intended to be protected, claims should be drafted to capture them or include dependent claims that spell out each variant.
  • Leaving crucial operational variants to the description alone may render them unprotected.
• Addressing assumed CGK:
  • Patentees often rely on CGK for implementation detail, as Abbott did for sensor construction, sterilisation and mechanics ([32]).
  • While this can support enablement, it may also make it easier for defendants to argue obviousness if the novelty resides solely in high-level system architecture rather than specific technical subtleties.

6.3 Implications for Litigation Strategy

Several litigation-focused lessons emerge:

• Expert selection matters:
  • Abbott’s sole expert, an experienced CGM project leader, was found not to possess the granular mechanical/electronic toolkit of the skilled engineers ([6]).
  • Defendants used separate experts in electronics and mechanical engineering; this was largely accepted, despite some criticism about their separation ([7]–[9]).
  • In mechanical/device cases, courts expect evidence grounded in the perspective of the actual notional skilled engineer, not just high-level managers.
• Avoid over-elaborating “step” lists:
  • Abbott’s 12-step reconstruction was criticised as “greatly exaggerated” and “deeply impractical” ([357]–[358]).
  • While such lists can be useful to expose hindsight, they risk being dismissed if they break down routine engineering into artificial sub-steps.
• Align invalidity arguments with accepted claim construction:
  • Dexcom’s invalidity strategy hinged on an interpretation of claim 1 broad enough to encompass automatic insertion; once the court adopted the narrow, manual-only construction, Dexcom’s evidence no longer mapped onto the claim.
  • Parties should develop invalidity cases on both a primary and a “fall-back” construction where feasible, supported by distinct expert evidence, particularly where infringement is likely to depend on a broad construction.

6.4 Implications for the Comptroller and Appellate Practice

From a procedural perspective, the case provides clear guidance:

• When a patentee appeals revocation and the original opponent settles, the Court will invite the Comptroller to assist (as in Halliburton and Aerotel), usually on terms that the patentee pays the Comptroller’s costs ([126]–[132]).
• The Comptroller’s role is to defend the judgment below where appropriate, but in an objective, non-partisan way, safeguarding both the public interest in revoking invalid patents and in maintaining valid ones.
• The Comptroller is not automatically a party to the appeal; he cannot file a respondent’s notice unless joined ([134]–[135]).
• Instead, the Comptroller should raise any alternative grounds supporting the result (even if not adopted below) in his skeleton argument or other advance notice, allowing the appellant fair opportunity to respond ([135]).

Practitioners should therefore expect:

• in “unopposed” validity appeals, the Comptroller may vigorously test the patentee’s arguments, including suggesting additional legal or factual points that support the first-instance outcome;
• but formal procedural steps like respondent’s notices will not be available to the Comptroller without an order joining him as a party.

7. Conclusion

Abbott v Dexcom is an important addition to UK patent jurisprudence in several respects.

First, it reaffirms that claim construction is not an abstract exercise: the way a claim is construed constrains the scope of obviousness analysis. Courts must ensure that the configuration they deem obvious over prior art is truly the same configuration they have held to be claimed. Here, the mismatch between a manual-only claim and an automatic-insertion prior-art configuration proved fatal to the invalidity finding.

Second, the judgment emphasises the need for evidence-backed motivation when altering the operational mode of a mechanical system as part of an obviousness argument. Moving from automatic to manual insertion in a safety-critical device required more than intuition; no such evidential basis was present.

Third, on the procedural side, the Court clarifies the Comptroller’s status in appeals from revocation orders: he is a valuable, independent assistant to the Court but not a respondent unless formally joined. This clarification at the interface of CPR Part 52 and PD 52D will shape the conduct of future unopposed appeals.

Finally, the decision illustrates the Court’s willingness to scrutinise first-instance judgments with particular care where there has been prolonged delay and partial adoption of a party’s submissions. Yet the Court’s intervention remains tightly focused: it does not disturb findings on expert credibility or broader obviousness principles, but corrects discrete errors of legal coherence and evidential sufficiency.

For patent practitioners and judges alike, the case serves as a reminder that precision and internal consistency – in drafting, in claim construction, in mapping prior art to claims, and in articulating appellate arguments – are essential to robust patent adjudication.