FDA proposes mandatory eSTAR for PMAs and IDEs: what changes and what the MDUFA V results and FY2027 agenda mean for submission planning

Status: November 3, 2026

In just six weeks, FDA has proposed to make the electronic Submission Template and Resource (eSTAR) mandatory for two further submission types: premarket approval applications (PMAs, draft of September 18, 2026) and investigational device exemption applications (IDEs, draft of October 28, 2026). Neither draft is in force, but both are open for comment, and the PMA draft describes proposed early-review mechanics for designated PMA submissions. Read together with the MDUFA V independent assessment and CDRH’s FY2027 guidance agenda, they point in the same direction: structure and evidence consistency in submissions matter more, while calendar-time exposure has not gone away.

Dates to watch

• November 17, 2026: comments on the PMA eSTAR draft (docket FDA-2026-D-9429)
• November 24, 2026: comments on the robotically-assisted surgical devices draft (docket FDA-2026-N-9505), a separate deadline
• November 30, 2026: comments on CDRH’s FY2027 guidance priorities (docket FDA-2012-N-1021)
• December 28, 2026: comments on the IDE eSTAR draft (docket FDA-2026-D-11188)

Comments on guidance documents may be submitted at any time. [11] For the PMA, IDE and robotic-surgery drafts, the stated dates ensure consideration before FDA begins work on the respective final guidance. [2] [4] [47] November 30 is the separate deadline requested for comments on CDRH’s FY2027 priorities, which concern prioritization and guidance development. [36]

Key takeaways

• Drafts, not mandates. Both eSTAR drafts are marked “not for implementation.” Mandatory use would start only after final guidance and a transition of at least one year. No date is fixed. [1] [3]
• Guidance that can bind. Guidance is usually nonbinding, but the format standards, timetable and waiver or exemption criteria in a final eSTAR guidance can have binding effect. A covered submission in the wrong format would not be considered received. [1] [3] [9]
• PMA proposal: screening instead of RTA. For properly prepared Original PMAs and Panel-Track Supplements, FDA anticipates technical screening within 15 calendar days of receipt and intends to begin only once the applicable user fee has been paid. Under the draft, a screening failure would lead to a hold and a day-zero restart on receipt of the response. Filing review would remain. [1]
• IDE: same format logic, different mechanics. The 15-day screening, day-zero reset and 360-day provision are specific to the PMA draft. [1] [3]
• MDUFA V and FY2027. Communication metrics improved, but FDA missed two shared total-time-to-decision goals, and the FY2027 agenda is a plan, not a promise. [22] [37]

Where eSTAR stands today

For submissions to the Center for Devices and Radiological Health (CDRH) or the Center for Biologics Evaluation and Research (CBER), eSTAR is already required for non-exempt 510(k) premarket notifications submitted since October 1, 2023, and for De Novo classification requests submitted since October 1, 2025. Designated PMA submissions and IDE submissions can currently use eSTAR voluntarily. Existing electronic-copy (eCopy) obligations continue where applicable: voluntary eSTAR does not mean that sponsors can otherwise submit on paper. [5] [6] [7] [8]

The distinction between eCopy and eSTAR is substantive. An eCopy is an electronic duplicate assembled under FDA’s specifications. eSTAR is a structured, dynamic PDF template that combines responses with attached documents and adapts its questions to the device and the submission type. Uploading a folder of PDFs is not the same as preparing a submission in the template. [1] [3] [6]

The proposed transition and its legal weight

Both drafts propose a transition of at least one year from issuance of the respective final guidance, with voluntary use in the meantime. The period does not start with draft publication or the close of comments, and neither finalization nor implementation has a date. After implementation, a covered, non-exempt submission in the wrong format would not be considered received. That makes format compliance consequential for receipt-based timing, without changing the evidentiary standards underneath. [1] [3]

The legal basis deserves more precision than “guidance is nonbinding.” Section 745A(b)(3) of the Federal Food, Drug, and Cosmetic Act (FD&C Act) authorizes FDA to set electronic-format standards, an implementation timetable and waiver or exemption criteria in guidance, and those provisions can bind once the guidance is final. Ordinary recommendations remain nonbinding unless they restate an independent legal requirement. The drafts themselves are not enforceable. [1] [3] [9] [10] [11]

Both drafts propose not to grant waivers; FDA states that it has not identified circumstances appropriate for them. An exemption is a different tool: it removes a defined category of submission from the requirement. A transmission accommodation is different again. Where known CDRH Portal technical limitations apply, the electronic submission is mailed to the Document Control Center, which does not waive the electronic format. [1] [3] [12]

PMA eSTAR: proposed screening instead of refuse-to-accept

Scope. The draft covers Original PMAs, Panel-Track Supplements, 180-day Supplements, Real-Time Supplements, and 30-Day Notices/135-Day Supplements submitted to CDRH or CBER. PMA Modules and Modular Shells are outside the draft, though their existing eCopy obligations remain. Scope follows submission type, not technology: a software-intensive device, a combination product or an in vitro diagnostic (IVD) is covered only if the designated PMA route applies. [1] [6]

Screening instead of RTA. For properly prepared Original PMAs and Panel-Track Supplements, FDA does not anticipate conducting the existing refuse-to-accept (RTA) review. It would instead apply virus scanning and technical screening: do the responses accurately describe the device, and does each applicable attachment-type question have at least one relevant attachment? FDA’s own example is a response indicating no tissue-contacting components where that is not the case, which would fail screening. Because eSTAR is a dynamic template whose questions depend on earlier answers, a wrong answer of this kind can also hide follow-up questions. A populated field or an irrelevant attachment is therefore not enough. [1] [5]

Screening is anticipated within 15 calendar days of receipt, and FDA intends to begin only once the appropriate user fee has been paid. If screening identifies incomplete information, FDA notifies the applicant by email and places the submission on hold. An updated eSTAR responding to the deficiencies is logged as an amendment. Receipt of the new information restarts the review clock at day 0, followed by repeat screening within the first 15 days of the restarted clock. If no response is received within 360 days of the deficiency notification, FDA will consider the submission withdrawn and close it. The trigger is the absence of any response, not an insufficient one, and the period runs from the notification, not from the original submission. A timely response can still fail repeat screening. [1]

Two cautions on timing. First, the 360 days are a proposal. FDA’s general eSTAR page currently describes a technical-screening hold of up to 180 days, and the sources do not provide a reconciled current PMA-specific rule. Sponsors should not read the draft as permission to wait 360 days. [1] [5] Second, the day-zero restart belongs to this particular screening sequence. Existing FDA guidance distinguishes PMA amendments that resume the clock, extend the review or have no clock effect, and already provides a total 360-day response period for specified PMA action letters. [13]

Filing review remains. Original PMAs and Panel-Track Supplements that pass screening still undergo filing review within 45 calendar days of receipt. Filing is the threshold determination that the application is sufficiently complete for substantive review. FDA’s acceptance and filing guidance keeps separate administrative-acceptance and filing checklists, the latter addressing matters such as study design, sample size, follow-up, endpoints and statistical analysis. Replacing RTA does not remove this examination. [1] [14] [15]

Likewise, the template’s “eSTAR Complete” status and a passed screening say nothing about substantive adequacy. The binding PMA content requirements and the requirement for valid scientific evidence of a reasonable assurance of safety and effectiveness remain. A relevant attachment can satisfy a screening check and still fail to support the intended use or clinical claim. [5] [14] [16] [17]

Proposed exemptions. The draft would exempt interactive-review responses; specified amendments (appeals and requests for supervisory review, requests for substantive summaries, changes of correspondent, legal entity or applicant address, and certain post-decision amendments such as final labeling); withdrawal requests; and reports, including annual or periodic reports and post-approval study reports. This is not a blanket exemption for amendments. Responses to telephone or email interactive review go by email, but other responses to requests for additional information would need eSTAR after implementation. According to the draft’s footnotes, exempt amendments, withdrawal requests and reports remain subject to any applicable eCopy requirements; this is not stated for every exempt category. [1] [6]

What it means in practice. For PMA applicants the practical change is the link between device-characteristic answers, document placement and the first review steps. An internal check that compares routing answers against the actual device and confirms attachment relevance is sound practice, but it is not an FDA-mandated quality gate. Project plans should allow for a possible screening hold with a clock reset, without treating the template as a substitute for filing-ready clinical and technical evidence.

IDE eSTAR: study authorization keeps its own scope and timing

Scope. The draft covers original applications under 21 CFR 812.20, supplemental applications under 21 CFR 812.35, amendments to both, and related IDE submissions to CDRH or CBER, subject to exemptions. It concerns FDA-facing submissions, not every record generated in a clinical investigation. Voluntary IDE template support currently includes originals, supplements and amendments, but not IDE reports. [3] [5]

Exemptions. The draft intends to exempt interactive-review responses; appeals and requests for supervisory review; withdrawal requests for applications currently under FDA review; and IDE reports and related amendments. A response amending an original application or supplement is therefore not interchangeable with an amendment to a report. Other additional-information responses would need eSTAR after implementation, while a response requested through telephone or email interactive review goes by email. Applicable eCopy obligations remain for categories such as reports and appeals. The draft identifies compassionate-use requests and reports, emergency-use reports and adverse-event reports of all types, including serious events and malfunctions, as already exempt from submission solely in electronic format, so it does not newly create these exceptions. [3] [6] [18]

The template. It is organized around prior investigations, the investigational plan, manufacturing information, investigator and site information, and informed-consent material. Device-specific sections, for example on software, cybersecurity or biocompatibility, are prompted only where responses indicate they apply. They are conditional content categories, not new universal testing obligations. [3, Table 1] The general-clinical-information section also asks the sponsor to confirm that the study is a significant-risk device study and is not exempt under 21 CFR 812.2(c). [3] This fits the existing study-risk distinctions. Significant-risk studies require FDA approval and institutional review board (IRB) approval before they start. Non-significant-risk studies generally follow abbreviated IDE requirements with IRB approval and do not require an IDE application to FDA. Some investigations are exempt from the requirements of Part 812, except for the investigator-disqualification provisions in 21 CFR 812.119. The electronic-format proposal changes none of this. [20] [53] Neither the template nor an “eSTAR Complete” indicator substitutes for scientifically adequate prior investigations or a justified investigation involving human subjects. [3] [19]

Timing. The draft states FDA’s intention to review and reach a decision on every IDE electronic submission within 30 days of receipt. That is not a new acceleration promise. By statute, an original application submitted in accordance with the applicable procedures is deemed approved on the thirtieth day unless FDA disapproves it and notifies the applicant by then. FDA’s decision guidance explains approvals, approvals with conditions and disapprovals, and the treatment of supplements and amendments. IRB approval and the conditions of the authorized study remain relevant, so day 30 is not a guarantee of unconditional study readiness. Investigations involving an exception from informed consent under 21 CFR 50.24 require prior written FDA authorization, for which FDA states it will provide a written determination within 30 days. [3] [18] [20] [21]

The PMA draft’s 15-day screening, day-zero reset and 360-day provision should not be imported into IDE planning; the IDE draft contains none of them. The operational implication is narrower. Once the format requirement is in force, a non-exempt application in the wrong format would not be considered received, so a format error could delay the start of the existing receipt-based period. [3]

IDE and PMA answer different questions. FDA may not disapprove an IDE merely because the investigation might not support a later marketing authorization or might require further investigation. That protection does not remove the need to protect subjects and to support a scientifically sound investigation. Shared technical evidence can serve both submissions, but the FDA-facing content and the adequacy judgments remain pathway-specific. [19] [21]

MDUFA V: process improvements are verified, calendar-time risk remains

The MDUFA V Independent Assessment, prepared by Booz Allen Hamilton and published by FDA at the end of September 2026, reviews selected review-process and program commitments under the fifth Medical Device User Fee Amendments program (MDUFA V, FY2023–FY2027). It is a performance assessment, not guidance, a rule or a new sponsor obligation. [22]

It concludes that, as of its publication date, FDA met the relevant commitments in the program areas specified for the assessment, among them premarket-review efficiencies, Q-Submissions, digital health, real-world evidence and international harmonization. That is not the same as all MDUFA V commitments having been completed, or completed on time: the report itself notes commitments with future or recurring deadlines. FDA’s FY2025 annual performance report states that 18 performance-enhancement goals were due in FY2025, of which 14 were completed on time and four were missed. [22] [26]

Substantive interaction improves, but it is not authorization

The assessment reports that the share of 510(k)s meeting substantive-interaction goals rose from 87% in FY2021 to 96% in FY2024. For Original PMAs and Panel-Track Supplements, it rose from 79% to 100%. FDA’s FY2025 annual report gives the FY2024 counts: 3,332 of 3,476 510(k)s and 67 of 67 Original PMAs and Panel-Track Supplements met the goal. The denominators exclude submissions still pending within goal, so they are not the full receipt cohorts. [22] [26]

A substantive interaction can be an additional-information or major-deficiency communication, a communication that no substantive deficiencies were identified, or an early clearance or approval. It is not synonymous with a positive decision. The commitment letter sets substantive-interaction goals of 60 calendar days of receipt for 95% of 510(k)s and 90 calendar days of the filing date for 95% of Original PMAs and Panel-Track Supplements. These milestones are distinct from the decision goals measured in FDA review days. Better performance here indicates more timely substantive communication, not that an individual sponsor can expect clearance or approval by that point. [25] [26]

The assessment also reports that first-cycle administrative acceptance, meaning an RTA acceptance or a passed technical screening in the first review cycle, increased meaningfully for 510(k)s and De Novo requests between FY2021 and FY2024. It suggests that effective use of templates may partly explain this, without establishing causality. [22]

The assessment further states that PMA first-cycle acceptance remained above 90% since FY2022, but FDA’s November 2025 quarterly report does not reproduce that series for the whole period. For Original PMAs and Panel-Track Supplements it shows CDRH first-cycle acceptance of 87.7% in FY2023 (64 of 73 submissions received) and 91.3% in FY2024 (63 of 69, counting two submissions that had no first RTA review more than 15 days after receipt). The discrepancy is unresolved, so the assessment’s PMA acceptance figure should not serve as a planning premise. First-cycle acceptance is in any case a completeness milestone, not a clearance, approval or grant, and nothing in the assessment supports treating eSTAR as a guarantee of acceptance or of shorter substantive review. [22] [54]

PCCP authorizations grow, but published rates are not success probabilities

A predetermined change control plan (PCCP) identifies planned device modifications that FDA can review as part of a marketing submission. Section 515C of the FD&C Act provides a legal basis for certain changes consistent with an authorized PCCP to proceed without another PMA supplement or 510(k). Authorization is pathway-specific: approval for PMA and clearance for 510(k), the pathways the statute addresses. FDA also authorizes PCCPs in De Novo grants, which is FDA practice rather than a statutory provision. Neither the statute nor the assessment makes a PCCP obligatory for every device or algorithm change. [31] [32]

The assessment reports 118 authorized PCCPs across FY2023–FY2025 for devices under the 510(k), PMA and De Novo pathways. This is a count of authorized plans, not of unique manufacturers or independent development programs. The assessment also compares FY2025 PCCP authorizations with annual MDUFA workload, reporting approximately 1.3% for 510(k), 34% for PMA and 1.4% for De Novo. These are not cohort-based uptake rates. The assessment does not state how numerator and denominator are composed. The PMA numerator may include supplement authorizations, whereas the MDUFA PMA decision-goal workload covers Original PMAs, PDPs, Panel-Track Supplements and Premarket Reports, not all PMA supplements. [22] [26]

The report gives FY2025 authorization rates of 66.7% for 510(k), 88.6% for PMA and 100% for De Novo, compared with 60.0%, 0% and 0% in FY2023. The report text does not explain how these rates are calculated, and the FY2025 De Novo result accompanies only one reported authorization. The FY2023 zero values should therefore not be read as demonstrated rejection rates, and none of these percentages is a general probability of obtaining PCCP authorization. [22]

The useful signal is increased use of authorized change plans. It does not justify assuming that a proposed plan will be accepted, that all future modifications fit it, or that the same approach works across devices. The already-final AI-specific PCCP guidance (issued December 2024, updated August 2025) applies to suitable AI-enabled device software functions and is separate from the device-wide PCCP draft of August 2024 that CDRH intends to finalize. [22] [32] [43]

Total time to decision depends on sponsor holds and cohort methods

Total time to decision (TTD) is the shared FDA and industry outcome measure. It combines FDA review time with the time industry needs to respond to deficiencies while the review clock is stopped. The goals were 128 calendar days for FY2023 510(k)s, 124 for FY2024 510(k)s, and 290 for Original PMAs and Panel-Track Supplements in FY2023 and FY2024. In its FY2025 annual report, FDA records that FY2023 510(k)s averaged 127 days against the 128-day goal, meeting it, and that FY2024 510(k)s averaged 139 against 124 days, missing it. The FY2023 PMA result was 302 against 290 days, also a miss. The PMA measure is a three-year average over FY2021, FY2022 and FY2023, and FDA points to the pandemic-affected FY2021 and FY2022 data within it. [25] [26]

FDA’s explanation of the FY2024 510(k) miss is instructive: it met both the substantive-interaction and the decision goals for that cohort, but submissions placed on hold lengthened the total time. [26] FDA review days exclude clock-stopping periods awaiting sponsor responses. Percentage-based decision goals, such as 95% within 90 FDA days for 510(k)s and 90% within 180 FDA days for PMAs without advisory committee input, are not calendar deadlines for each application, and TTD does not represent the path from development to launch. A MDUFA decision is also not necessarily an approval: for PMAs the recorded decisions include not-approvable letters, denials and withdrawals. [25] [26]

The FY2023 and FY2024 510(k) values were calculated with different methodologies under Section VIII.G of the commitment letter. FDA states that the FY2024 value is comparable to the values recorded for FY2020 to FY2022 under the same methodology, 139, 141 and 147 days. A cohort counts as closed when 99% of 510(k)s or 95% of PMAs have reached a decision. As of September 30, 2025, the FY2024 and FY2025 PMA cohorts and the FY2025 510(k) cohort had not reached that threshold, so no TTD was reported for them. [25] [26]

For planning, the credible signal is better communication and process efficiency alongside continuing calendar-time exposure, not a general basis for reducing launch contingencies. Sponsor response time is part of the measure.

FY2027 guidance priorities: planned outputs, not newly issued requirements

CDRH announced its FY2027 proposed guidance priorities on October 1, 2026. FY2027 runs from October 1, 2026, to September 30, 2027, not the calendar year. Comments are requested by November 30, 2026, in docket FDA-2012-N-1021. The agenda is a nonbinding workplan: it neither issues the listed guidances nor fixes their content. [36] [37] [39]

The three lists carry different meanings. The A-list names documents CDRH intends to publish during FY2027. The B-list names documents it intends to publish as resources permit. Under Construction lists topics intended for development without a publication commitment. FDA may change priorities, omit listed documents or issue unlisted ones, and guidance development cannot displace statutory obligations or review timelines. [36] [39] [25]

A-list: AI lifecycle, device-wide PCCPs and defined product or review topics

The A-list has seven final-guidance topics. For software and change control, the prominent entries are marketing-submission and lifecycle-management recommendations for AI-enabled device software functions, and Predetermined Change Control Plans for Medical Devices. Both are planned final outputs, not guidance issued by the agenda. The AI-lifecycle document remains a draft from January 2025 and the device-wide PCCP document a draft from August 2024. By contrast, the separate AI-specific PCCP guidance is already final. [37] [42] [43] [32]

The quality-management entry is Quality Management System Information for Certain Premarket Application Reviews. The existing draft, issued on October 27, 2025, under the title “Quality Management System Information for Certain Premarket Submission Reviews,” addresses PMA and humanitarian device exemption (HDE) applications; it should not be read as a blanket new documentation requirement for every 510(k). Its finalization is distinct from the Quality Management System Regulation (QMSR), a binding rule effective February 2, 2026. [37] [44] [45]

Two further final topics concern robotically-assisted surgical devices and the Accreditation Scheme for Conformity Assessment (ASCA) program. The robotic-surgery draft is dated September 25, 2026; the Federal Register notice was filed on September 24 and published on September 25. It has its own comment deadline of November 24, 2026 (docket FDA-2026-N-9505), which the annual-priorities consultation does not extend. The ASCA entry is a planned final guidance on an existing program; the agenda does not create a new conformity-assessment program. [37] [46] [47]

The remaining final topics must be kept apart: menstrual products (performance testing and labeling recommendations) and a compliance policy for certain NIOSH-approved air-purifying respirators. NIOSH is the National Institute for Occupational Safety and Health. “Performance testing and labeling” describes the menstrual-products topic, not the respirator entry, and the respirator proposal concerns a compliance policy for specified respirators, not all respiratory protection. [37]

The A-list also plans a draft titled Generative Artificial Intelligence-Enabled Conversational Devices for Mental Disorders: Evidentiary Recommendations. It is a planned output, not an issued evidentiary standard, and its subject should not be generalized to every chatbot or wellness application. [37]

B-list and Under Construction: different levels of planning confidence

B-list final topics are medical-device user-fee small-business qualification, pulse oximeters for medical purposes, and the Q-Submission Program. Planned drafts are Policy for Device Software Functions and Postmarket Management of Cybersecurity in Medical Devices. These are resource-dependent intentions. A planned Q-Submission update does not mean that the final guidance issued on May 29, 2025, has disappeared, and a planned postmarket cybersecurity draft does not mean that current cybersecurity expectations await FY2027: FDA already has final premarket cybersecurity guidance and a postmarket guidance from 2016. [37] [34] [51] [52]

Under Construction covers predicate-device selection for 510(k)s, evaluation of tissue heating or cooling effects, broader risk and evidentiary considerations for generative-AI devices, and selected updates on evaluating substantial equivalence in the 510(k) program. These topics are intended for development, not promised FY2027 publications. Merging them with B-list items would obscure the agenda’s different levels of commitment. [37] [39]

CDRH reports having published seven of eleven FY2026 A- and B-list documents, five from the A-list and two from the B-list. The figure gives context for the resource caveat; it cannot be converted into a probability that any individual FY2027 topic will be published. [36]

What this means for submission planning

Keep evidence and template answers consistent. The two eSTAR drafts and the agenda point toward a more structured interface between product characteristics and supporting evidence. A software version, device configuration, intended use and supporting test report should be consistent across the submission, otherwise a template answer can look complete while describing a different configuration from the attached evidence. This is a sponsor-level implication of the proposed screening and conditional content logic, not a new FDA documentation requirement. [1] [3]

Reuse evidence without treating the IDE as a shortened PMA. Prior-investigation evidence supports whether the proposed study is justified and subjects are protected. Marketing evidence supports the later authorization. Traceability between the investigated configuration, study endpoints and proposed marketed configuration shows whether later changes need new evidence. A single controlled evidence repository can serve both purposes, while the FDA-facing content and adequacy judgments stay pathway-specific. [16] [19] [21]

Use Q-Submissions with a defined question. The May 2025 final Q-Submission guidance recommends focusing Pre-Submissions on three or four substantial topics; the most effective Pre-Subs typically contain no more than seven to ten questions, including sub-questions. These are nonbinding recommendations, not caps introduced by the assessment. The assessment reports Pre-Submission volume rising from 3,843 in FY2023 to 4,429 in FY2025, and FDA’s annual report records the written-feedback goal as met for FY2023 and FY2024 (FY2024: 99% against an 80% goal). A focused request presents a defined regulatory question and the sponsor’s rationale rather than asking FDA to compensate for an unresolved evidence strategy. [22] [26] [34]

Govern change for AI-enabled devices now. Where a PCCP is suitable, evidence can be organized around the proposed change boundaries and their validation, preserving the link between the authorized plan, the implemented version and subsequent monitoring. The benefit depends on whether the actual modification stays within the authorized plan. The FY2027 agenda itself does not require new rollback mechanisms, authorize unrestricted model updates or finalize the device-wide PCCP recommendations. [31] [32] [37] [43]

Separate the time components. Timelines should distinguish technical preparation, administrative screening, filing where applicable, substantive review and sponsor response work. A shorter FDA-day measure does not remove the time needed to produce an adequate response. Voluntary third-party 510(k) review is available only for eligible devices, and FDA retains the final decision and can request information or re-review inadequate third-party work. [13] [25] [35]

Global programs. For global device and IVD programs, the interoperability lesson is organizational rather than legal: controlled testing, software and change-history evidence can be reused where relevant, with jurisdiction-specific assessments and submission modules. This FDA agenda does not itself amend EU obligations; it cannot establish the status or implementation dates of EU medical-device, IVD or AI legislation. A common evidence architecture should support distinct regulatory conclusions, not assume that a US PCCP or FDA template satisfies another jurisdiction’s requirements.

Image notice: The illustrative and header images were generated using AI and do not depict real people, places, or events. The portrait photographs show real people.