Anticipation Requires Necessity: Expert “Understanding” Cannot Substitute for Strict Implicit Disclosure in Novelty Attacks

1) Introduction

Abbott Diabetes Care Inc & Ors v Dexcom Inc & Ors ([2025] EWCA Civ 1687) is a Court of Appeal patent decision arising from a multi-jurisdictional dispute between two major continuous glucose monitoring (“CGM”) competitors: Abbott (appellant) and Dexcom (respondent). The UK proceedings were divided into technical trials. At first instance (Mellor J), four patents were revoked; this appeal concerned only two Abbott patents: EP 2 146 627 (“EP 627”) (user-interface/alerts during execution of a “predetermined routine”) and EP 2 476 223 (“EP 223”) (validation of safety-critical software hosted on an “uncontrolled data processing device” such as a smartphone).

Although the parties settled and Dexcom did not oppose the appeal, the Court proceeded with the assistance of the Comptroller-General under PD52D, para 14.1, reflecting the public interest in correct validity determinations.

Key issues on appeal:

  • For EP 627: whether a Dexcom user guide (“STS Guide”) anticipated claim 1, particularly: (i) whether a 3-day sensor-expiry countdown could be a “predetermined routine”; and (ii) whether two notifications related to the same “predefined alarm condition”.
  • For EP 223: whether a US patent application (“Gejdos”) anticipated claim 1, especially: (i) whether Gejdos disclosed a “safety critical application”; (ii) whether its database integrity check could satisfy installation and functional checks; and (iii) whether Gejdos disclosed “selective enablement” (disabling safety-critical features while leaving non-safety-critical features enabled).

2) Summary of the Judgment

The Court of Appeal (Birss LJ giving the lead judgment, with Stuart-Smith LJ and Miles LJ agreeing) allowed Abbott’s appeal in part:

  • EP 627: Appeal allowed. The STS Guide did not anticipate claim 1 because: (a) the sensor-expiry countdown was not a “predetermined routine” as properly construed; and (b) the “30-minute to expiry” and “expired” notifications did not relate to the same “detected predefined alarm condition”.
  • EP 223: Mixed outcome. Grounds (1) and (2) were dismissed: Gejdos could be a “safety critical application”, and its integrity check could satisfy the claim’s determination of proper installation and functioning. But ground (3) was allowed: Gejdos did not anticipate “selective enablement” because the finding relied on expert assertion without demonstrating the strict necessity required for implicit disclosure in novelty.

3) Analysis

3.1 Precedents Cited

The judgment is notable for synthesising established UK novelty doctrine with recent Court of Appeal guidance on “implicit disclosure” and the proper use of expert evidence.

(a) Construction of patents and documents

  • Actavis v Eli Lilly [2017] UKSC 48: Reaffirmed that claim construction is purposive through the eyes of the skilled person, balancing fair protection with legal certainty. The Court used Actavis as the modern foundation for construing Abbott’s claims (and, by analogy, reading prior art through the skilled person’s lens).
  • Saab Seaeye v Atlas Elektronik [2017] EWCA Civ 2175 and Virgin Atlantic v Premium Aircraft [2010] RPC 8: Cited for the structured principles of interpretation post-Actavis (contextual, purposive, whole-document reading).
  • Nokia v IPCom [2009] EWHC 3482 (Pat): Invoked for the warning against “reading down” general non-exhaustive claim language. The Comptroller relied on this to resist Abbott’s narrowing constructions. The Court accepted the warning, but emphasised that purposive interpretation still requires the claim to make coherent sense as drafted.
  • Fisher and Paykel Healthcare Ltd v Flexicare Medical Ltd [2020] EWHC 3282 (Pat): The counterpoint to Nokia: where a claim uses specific exhaustive language, one should not “read up” to broaden it. Together, these cases framed a “no artificial narrowing or widening” approach.

(b) Novelty/anticipation: disclosure, enablement, and strict necessity

  • Synthon BV v SmithKline Beecham plc [2005] UKHL 59, [2006] RPC 10: The central novelty test: anticipation requires (i) disclosure (clear and unmistakable directions) and (ii) enablement. The appeal focused on disclosure; enablement was not in issue. The Court also drew on Synthon’s treatment of explicit and implicit disclosure.
  • General Tire & Rubber Co v Firestone Tyre & Rubber Co Ltd [1971] FSR 417: The “plants the flag” metaphor and the idea that anticipation requires the prior art to have already staked out the invention. This underpinned the Court’s resistance to novelty attacks that depend on “could” or “would” reasoning rather than “must”.
  • ModernaTX Inc v Pfizer Ltd and ors [2025] EWCA Civ 1030: A key authority relied upon expressly for the strictness of non-explicit disclosure: it is insufficient that it is “possible or obvious” to implement the prior art in a way within the claim; the result must necessarily follow. Moderna was decisive in the Court’s reversal on EP 223 ground (3), where the first instance decision drifted toward obviousness-style reasoning.

(c) Role of expert evidence in construing technical documents

  • Technip France SA's Patent [2004] EWCA Civ 381, [2004] RPC 46: The Court “dons the mantle” of the skilled person, but interpretation remains a legal task for the judge. This was used to calibrate the permissible role of experts—informative, not substitutive.

(d) Appellate restraint and delay

  • FAGE UK Ltd v Chobani UK Ltd [2014] EWCA Civ 5, [2014] ETMR 26 and Byers v Saudi National Bank [2022] EWCA Civ 43, [2022] 4 WLR 22: Cited for deference to first-instance evaluative findings absent clear error. However, the Court noted the “inordinate delay” in producing the trial judgment required “special care” in review, tempering usual restraint.

3.2 Legal Reasoning

(A) EP 627: avoiding “contorting” the claim to fit a single integrated prior-art process

The STS Guide described a single sensor-expiry process with built-in notifications. Mellor J treated the 3-day expiry countdown as the “predetermined routine” and the notifications as the claim’s first and second indications.

The Court of Appeal rejected that mapping, not by importing extraneous limitations, but by construing the claim coherently as a whole:

  • Distinctness requirement inferred from claim purpose and structure (integer 1.6): Although integer 1.6 did not expressly say the predetermined routine must be “capable of interruption”, the Court held the claim’s logic presupposes that the “indications” arise from something distinct from the routine; otherwise it becomes meaningless to speak of an indication that could interrupt the routine but is required not to. Asking whether a routine “interrupts itself” was described as unreal.
  • No redundancy (integer 1.9): The Court treated the drafting as indicating that the “processes” interfacing with the user interface (1.9) are different from the “outputting of indications” (1.3/1.4). If the indications themselves satisfied 1.9, 1.9 would become surplusage—an interpretive warning sign that the proposed reading is wrong.
  • Single vs. multiple “conditions” (integers 1.3 and 1.4): The claim refers to a single “detected predefined alarm condition”. The Court held “30 minutes remaining” and “has expired” are distinct conditions. The trial judge’s characterisation of the condition as “sensor expiry” was criticised as not being a “condition” in the relevant sense.

In short, the Court insisted on a purposive, whole-claim construction that reflects what the claim language is doing, and refused an anticipation analysis that “contorts the claim out of shape” to fit a unitary prior art process.

(B) EP 223: novelty survives where “selective enablement” is not necessarily disclosed

EP 223 targets a well-identified safety problem: deploying safety-critical medical functionality on an uncontrolled device (e.g., a smartphone) whose configuration may be altered, risking malfunction. The solution is a check on installation and functioning; failure triggers “selective enablement” (disable safety-critical features, keep non-safety-critical features).

  • Ground 1 (dismissed): “safety critical application” is broad
    The Court refused to read in a “directness” threshold. The patent’s definition at [0001] covers systems whose failure “may result in” significant harm. The Court held Gejdos’ healthcare database—if inaccurate—could lead to serious patient consequences, and thus falls within the claim’s broad language. This part of the judgment underscores that where patentees choose broad ordinary-language terms, courts will generally not confine them to the paradigm embodiment.
  • Ground 2 (dismissed): one check can satisfy both installation and functional “determination”
    The claim required a determination that the application “is installed properly and functions properly” (integer 1.2(a)), but did not require two distinct checking processes. The Court accepted that Gejdos’ integrity check could serve as both: if integrity fails, the system is not functioning properly; and the check is performed on installing or launching, providing evidence of successful installation sufficient for the claim as construed.
  • Ground 3 (allowed): expert assertion cannot convert “obvious” into “necessarily implicit”
    This is the decision’s sharpest novelty point. Gejdos explicitly disclosed an error message on failure of integrity checks, but did not explicitly disclose disabling some functions while keeping others enabled. The trial judge accepted expert evidence that the skilled person “would understand” that functions would be disabled. The Court of Appeal held this reasoning did not meet the novelty standard:
    • An expert’s “would understand” often reflects what would be obvious, which is an inventive step concept, not novelty.
    • For implicit disclosure (per Synthon and emphasised in ModernaTX Inc v Pfizer Ltd and ors [2025] EWCA Civ 1030), the feature must necessarily and unambiguously follow from the prior art; mere plausibility or likelihood is insufficient.
    • Because the expert was not cross-examined and did not articulate necessity (and the judgment did not grapple with the legal standard), the finding of anticipation could not stand.

3.3 Impact

(1) Reinforcement of a bright line: novelty is not “obviousness-lite”

The judgment reinforces a strict discipline in novelty attacks based on non-explicit disclosure: courts must be satisfied that the claimed feature necessarily follows from the prior disclosure. Evidence that something would be a natural design response, common practice, or an obvious implementation choice will not “plug gaps” in anticipation unless it establishes necessity.

(2) Expert evidence: useful for technical context, insufficient as a substitute for the legal test

The decision is a practical warning for litigators: an expert’s statement that the skilled person “would understand” X may be inadequate unless tied to the correct legal category—explicit disclosure, necessarily implicit disclosure, or inevitability of result. Where a novelty case depends on implication, the evidential presentation must show why the implication is mandatory, not just sensible.

(3) Claim construction as a constraint on “over-fitting” prior art

For EP 627, the Court’s reasoning shows how purposive construction can prevent an anticipation case that only works by assigning inconsistent roles to the same feature (here: treating notifications as both the “indications” and the user-interface-interfacing “processes” within the “predetermined routine”). This has broader relevance for software/UI patents where prior art often describes integrated workflows.

(4) Broader terms mean broader exposure (and broader prior art), but also broader prior-art burdens

EP 223 demonstrates the double-edged nature of broad claim language: Abbott failed to narrow “safety critical application” to its preferred “direct, real-time patient self-treatment” context. However, the same strictness on novelty still protected Abbott from a non-explicit “selective enablement” finding not anchored in necessity.

4) Complex Concepts Simplified

  • Novelty / Anticipation: An invention is not new if a single earlier public disclosure already teaches it. In UK terms, that earlier disclosure “anticipates” the claim.
  • Disclosure vs Enablement (from Synthon BV v SmithKline Beecham plc [2005] UKHL 59, [2006] RPC 10): The prior art must (1) clearly disclose the invention and (2) enable the skilled person to perform it. This appeal was about (1), not (2).
  • Implicit disclosure (strict): A feature can count as disclosed even if not stated, but only if it is necessarily and unambiguously implied, or if following the prior-art directions inevitably produces something within the claim. “Could do it that way” is not enough.
  • Inventive step vs novelty: “Obvious” developments can invalidate for inventive step, but they do not destroy novelty unless they are actually disclosed (explicitly or strictly implicitly) by a single prior-art source.
  • Purposive construction: Interpreting patent language as a skilled person would, in context, aiming for fair protection and legal certainty, without artificially narrowing or widening the words used.
  • Selective enablement (as used in EP 223): If safety checks fail, disable the features that could harm the user, but keep benign features running.

5) Conclusion

[2025] EWCA Civ 1687 is significant for two connected propositions in UK patent validity law:

  • Anticipation by implication demands necessity: novelty cannot be defeated by expert-driven “this is what one would do” reasoning unless it proves the feature necessarily follows from the prior art (reinforcing the approach highlighted in ModernaTX Inc v Pfizer Ltd and ors [2025] EWCA Civ 1030).
  • Construction disciplines anticipation: courts will not force a claim to fit a single integrated prior-art process by making the same elements perform incompatible roles; purposive, whole-claim interpretation remains central (rooted in Actavis v Eli Lilly [2017] UKSC 48 and elaborated in Saab Seaeye v Atlas Elektronik [2017] EWCA Civ 2175).

The decision therefore strengthens doctrinal clarity between novelty and inventive step, and gives practical guidance on how implicit disclosure must be pleaded and proved, particularly where an anticipation case risks sliding into “obviousness” territory.