New Framework Demands Proof That Recalled Medical Devices Are Actually Closed

When a medical device is recalled, the notice that hospitals receive is often treated as the end of the story. A new conceptual paper argues that this assumption leaves a dangerous gap between what regulators announce and what actually happens at the bedside, and it proposes a governance framework designed to close that gap by demanding auditable proof that every affected unit has been located, corrected and verified. The framework, published in Discover Health Systems, was developed by Wissam Mazraani as a specification for hospitals and health systems rather than as an empirically validated instrument, and its author is explicit that prospective testing in real-world service environments is still needed.

The central proposition is simple but consequential: a recall notice is not equivalent to recall closure. A recall may be announced, acknowledged and administratively tracked without ever establishing that each affected unit in clinical use has been identified, acted upon and verified. The paper describes this as a measurable execution gap between regulatory communication and actual risk mitigation, one that persists because most systems monitor whether a recall was issued or communicated rather than whether closure can be evidenced at the level of the individual device.

The gap arises, according to the framework, from incomplete installed-base visibility, redistribution of devices between facilities, weak documentation of local action, limited exception governance and the absence of verification procedures matched to the relevant hazard. These weaknesses are amplified where asset registers are unreliable, biomedical engineering capacity is thin, or equipment arrives through donation channels and secondary markets with incomplete service histories. The paper stresses that the problem is not confined to lower-capacity settings, but that it is distributed unevenly: facilities with robust digital infrastructure and mature quality-management systems may execute recalls efficiently, while others leave affected devices in service for longer or miss them entirely, producing unequal patient protection within and across health systems.

High-severity recalls are not rare events. The paper notes that empirical studies of United States Food and Drug Administration device data have shown that products associated with serious Class I recalls frequently enter patient use through mainstream regulatory pathways, and that repeat recalls within device families remain common, which the author cites as evidence of persistent systemic weaknesses in post-market control and field correction. In lower-capacity environments, weak asset registers, uneven engineering capacity and dependence on imported equipment with incomplete service histories may amplify these weaknesses further.

To address the problem, the framework comprises three linked components. The first is the Recall Closure Evidence Set, or RCES, which defines the minimum auditable evidence needed to verify unit-level recall completion or documented exception handling. It is organised into four evidence classes: population evidence establishes the denominator of affected units; action evidence records the corrective or interim step applied to each unit, by whom and when; verification evidence confirms that the action resolved the hazard-relevant status rather than merely being logged; and exception and communication evidence governs units that cannot be closed immediately and the communications required for patient-safety accountability.

Notably, RCES is deliberately implementation-agnostic. It does not prescribe vendor-specific systems, unique device identification infrastructure or advanced electronic health record integration. In lower-resource settings, paper-based work orders, handwritten verification logs and manual reconciliation statements can satisfy the requirements provided they meet the minimum evidence criteria for each class. The standard, the paper insists, is one of evidential completeness rather than technological sophistication, a design choice intended to make the framework usable in settings without national digital platforms.

The second component is the Recall Oversight Quality maturity model, which stages organisational recall-governance capability across five levels, following capability maturity model conventions. Level 1 is notice-driven, where organisations receive and forward notices but infer closure without unit-level evidence. Level 2 is inventory-controlled, with local identification of affected units but inconsistent action logging. Level 3 is evidence-based closure, where the evidence set is operationalised for unit-level verification with governed exceptions. Level 4 adds interoperable traceability across procurement, service and traceability systems, and Level 5 represents assured closure and learning, with continuous monitoring and lessons embedded into procurement, maintenance and governance. Equity of closure is embedded as an explicit expectation from Level 3 onwards, on the grounds that disparities in closure performance constitute a patient-safety concern at any level of maturity.

The third component is a compact metric suite of six auditable indicators: time-to-identification, time-to-action, verified closure rate, exception burden, residual risk duration and equity of closure. Each metric is intended to be derivable from routine operational artefacts such as asset registers, work orders, verification logs and exception registers. The paper is careful to state that the indicative numerical thresholds attached to these metrics, including a five working-day identification target, an 80 percent action target and a ten percentage point equity-disparity trigger, are explicitly provisional. They were selected as pragmatic illustrative starting points for operational testing, are not empirically validated benchmarks, and were not derived from outcome data. Where a site has no basis for adopting a numerical target, the paper suggests that the direction of travel, such as a decreasing residual-risk-duration trend across successive recalls, is the minimum governance expectation.

Illustrative applications show how the framework distinguishes public notification from local evidence of closure. In a publicly documented FDA Class I recall of a Puritan Bennett 980 Series ventilator, the public record confirms that a recall was issued and that affected quantities were documented at the regulatory level, but it does not disclose whether each affected device was identified locally, what corrective action was taken at specific sites, how actions were verified, or how unresolved units are managed. Applying the evidence set, the public record satisfies none of the four evidence classes, and an organisation relying solely on it would sit at Level 1, with the verified closure rate uncomputable. A comparable European field safety corrective action notice, examined through the European database on medical devices, likewise discloses the hazard and intended remedy but no facility-level evidence, leading the author to conclude that the execution gap is a structural feature of recall governance across reporting architectures rather than an artefact of any single jurisdiction.

The paper is candid about its limitations. The framework is conceptual and has not been prospectively validated across institutions or jurisdictions; the five-step development process was conducted by a single author without structured stakeholder consultation, expert-panel review or pilot implementation, so feasibility, acceptability and completeness remain unknown. The illustrative cases rely on public recall metadata and cannot establish the prevalence or severity of the execution gap, and no empirical data currently link recall execution gaps to measurable patient-safety outcomes, meaning the safety case remains inferential. The author also warns that a narrow focus on auditable evidence could incentivise paper compliance without genuine corrective action, and that implementation burden was not measured. A staged validation pathway is proposed: expert consensus testing of face validity, retrospective testing against hospital recall datasets, multicentre pilot implementation, and prospective evaluation of patient-safety outcomes. Until those stages are complete, the framework stands as a testable governance specification, one that reframes recall closure as a measurable patient-safety process and gives governance bodies a concrete basis for asking whether all patients, regardless of where they are treated, receive the same protection from recalled devices.

Subject of Research: A governance framework for verifying medical device recall completion in health systems

Article Title: A health systems governance framework for verifying medical device recall completion

Article References: Mazraani, W. (2026). A health systems governance framework for verifying medical device recall completion. Discover Health Systems, 5(1), Article 80. https://doi.org/10.1007/s44250-026-00420-0

Image Credits: AI Generated

DOI: 10.1007/s44250-026-00420-0

Keywords: medical devices, recalls, patient safety, health systems governance, recall closure, maturity model, auditable evidence, post-market surveillance, health equity, lower-capacity settings, quality management, framework development