Clinical, Human-Factors, and Labeling Evidence Supporting Patient Self-Administration Claims for Injectable Drugs and Biologics: FDA and EMA/CHMP Requirements

Evidence supporting patient self-administration claims for injectable drugs and biologics: what FDA and EMA/CHMP reviews actually require

A "self-administration" claim, whether it lands in the DOSAGE AND ADMINISTRATION section of a US label or in section 4.2 of an EU SmPC, is not granted on the strength of the molecule alone. It rests on a package of evidence that ties three strands together: clinical experience showing patients or caregivers can dose themselves outside the clinic, human-factors (usability) data showing the device and its user interface can be operated safely by representative lay users, and labeling/Instructions for Use (IFU) that translate the validated workflow into instructions a non-professional can follow. The mix shifts with the presentation, an autoinjector, a prefilled syringe or pen, or a vial-and-syringe requiring reconstitution, but the underlying logic is consistent across both agencies. This overview walks through what FDA and EMA/CHMP reviewers have looked for, with worked examples from Drugs@FDA reviews and EMA European Public Assessment Reports (EPARs).

The FDA human-factors framework for injectable combination products

FDA treats a self-injected product as a combination product and expects the human-factors work to cover the drug and the delivery device together, not the device in isolation. The agency asks sponsors to run a comprehensive use-related risk analysis that assesses risk against the intended patient population and the environment of use, and to consider the human-factors characteristics of the population that will actually use the injector ¹¹². Home-use populations in particular may require consideration of age, tissue characteristics, and other user attributes ¹¹¹. Critical tasks, those that could cause harm or compromise care if performed incorrectly or omitted, are identified through a systematic task analysis and become the focus of the study ¹².

The pivotal piece is human-factors validation: a simulated-use test of the actual user interface, under representative conditions, that determines whether the design has reduced use-related risk sufficiently ². FDA accepts that residual risk may remain and weighs the validation results in the overall benefit-risk assessment ². Where use errors occur, reviewers expect them to be analyzed in context: some errors are acceptable if further interface changes are unlikely to help and the benefit-risk balance holds, while errors caused by unrealistic test conditions may be treated as artifacts, though too many such explanations signal that retesting under more realistic conditions is needed ³. For home-use scenarios, FDA emphasizes simple tasks, representative users and settings, and running the entire workflow ⁴⁷. The Quick Reference Instructions/IFU used at home should sit no higher than a 7th-grade reading level, run to one or two pages, and use pictures or diagrams ⁴⁷.

When a sponsor wants to add a new presentation, for example moving from a prefilled syringe to an autoinjector, FDA's draft guidance on bridging for drug-device and biologic-device combination products is the governing document ¹²²¹²³¹²⁴¹³⁷. It contemplates bridging PFS-to-autoinjector, one autoinjector prototype to another, and same-device/different-drug cases ¹²²¹²⁴¹³⁴, and it identifies the supporting data typically needed: human-factors validation studies, design verification data, and bench, preclinical, or clinical testing as appropriate ¹²²¹²³¹²⁴. In the worked PFS-to-autoinjector example, the applicant planned a full HF validation study and report because bridging the prior autoinjector data would be difficult for the new population and indication ¹²³. FDA's companion Q&A, Application of Human Factors Engineering Principles for Combination Products, frames the purpose of validation results the same way, to evaluate whether the interface has been optimized so use-related risks are sufficiently reduced ¹⁴².

Human-factors validation in practice: what FDA review documents show

FDA review memos give a concrete picture of how self-administration is validated. The recurring design is a simulated-use study with representative end users, split across patient, caregiver, and healthcare-professional (HCP) groups, and often across trained and self-trained (IFU-only) arms, with critical tasks scored and use errors, close calls, and difficulties adjudicated.

Benlysta (belimumab) autoinjector/safety syringe. The simulated-use validation enrolled 30 participants, 15 SLE patients and 15 HCPs, to show intended users could inject safely and effectively without preventable use errors and to surface confusing device/IFU features; patients were trained with a supervised injection, while HCPs were untrained/self-trained ⁴⁰.
PegIntron Peninjector. The study enrolled at least 15 participants in each of three groups, HCPs, treatment-experienced patients, and treatment-naive patients/caregivers, with a training arm and an IFU-only arm. FDA focused on the training arm because it matched labeled use. Critical tasks included dialing the dose, removing the needle cap, pushing the needle straight down, pressing against the skin, and holding for 15 seconds; only one user failed a successful injection in round 1 and was excluded as not a self-injection candidate ⁵¹.
Amjevita (adalimumab-atto). FDA leveraged human-factors data from the modified Enbrel SureClick platform and judged the patient users representative for Amjevita's indications ¹⁵⁸. For the prefilled syringe, 58 of 61 participants performed all essential steps successfully; the 4 essential-step use errors (pushing the plunger too early, removing the needle too soon, improper disposal) all came from self-trained participants, and root-cause analysis linked them to injection experience and uncertainty about needle insertion depth ¹⁵⁸. FDA also asked for IFU changes to cut medication-error risk, including clearer site-rotation language, "do not inject through your clothes," and a clearer "Push Down" image ¹⁵⁹.
Zembrace SymTouch (sumatriptan autoinjector). The HF study aimed to show the autoinjector, labeling, and IFU could be used safely without preventable harmful use errors; it enrolled 45 injection-naive and injection-experienced participants across trained and untrained groups ¹⁵¹.
Dihydroergotamine autoinjector. Validation used 30 migraine/cluster-headache patients, half experienced and half inexperienced, simulating dosing in a realistic scenario with knowledge questions; the reviewer concluded users could safely self-inject with no patterns of use errors ¹⁵⁰.
Kesimpta/Arzerra (ofatumumab) PFS and pen. FDA summarized use errors on critical tasks (identifying the injection site, removing the needle cap/rigid shield, depressing the plunger to inject, releasing the plunger to retract the guard) and on non-critical tasks (safety checks, warming, site cleaning, skin pinch, disposal), but found residual risk acceptable after feedback, root-cause analysis, and mitigation review ⁴².
Idacio (adalimumab-aacf) autoinjector/pen. FDA noted use errors and difficulties on non-critical tasks (opening the carton, returning the pen to the refrigerator, checking for damage, holding the pen firmly until the guard was depressed) and accepted residual risk based on feedback, root-cause analysis, and labeling mitigations ⁴¹⁴³. In a single-injection clinical study a device failure was traced to twisting the cap contrary to the IFU, and FDA accepted the "Do not twist the cap" labeling warning ¹⁶⁹¹⁷⁴.
Besremi (ropeginterferon) prefilled syringe. The study defined 10 critical, 5 safety-critical, 5 essential, and 14 secondary tasks (storage, carton opening, liquid inspection, site selection, sharps disposal, tip-cap removal, needle attachment, 45-to-90-degree insertion), with errors and close calls judged acceptable after subjective feedback and mitigation review ⁴⁴.
Yuflyma (adalimumab-aaty). FDA required the IFU to explain the meaning of the autoinjector's audible clicks, showing reviewer attention to whether users understand device cues ¹⁶⁴¹⁶⁶.

Two cautionary examples are worth noting. For Somatuline Depot, FDA pushed back on study design, insisting the population include HCPs plus experienced and naive caregivers/patients because patient self-injection had appeared in adverse-event reports, objecting to unrealistic training, and recommending at least two arms (with and without required IFU review) ⁵⁸⁶⁰⁶¹⁶²⁶³. For Myalept, FDA found the HF report deficient, unclear participant descriptions, ambiguous dose-preparation steps, incomplete subjective data, and no use-related risk analysis, so it could not support a self-administration conclusion ⁵⁷. For Afrezza, FDA asked for prioritized testing of high-risk scenarios such as cartridge selection for the correct dose, with results organized by user group and paired with root causes, clinical impact, and mitigations ⁵⁴⁵⁹⁶⁴.

Clinical self-administration and PK device bridging

Alongside usability testing, sponsors generate two other kinds of clinical evidence: demonstrations that patients actually dosed themselves at home during trials, and pharmacokinetic (PK) bridging studies showing a new device delivers the same exposure as an already-approved presentation, so efficacy need not be re-established for each device.

On home-use clinical experience, the Dupixent program required supervised training before home dosing: after site-administered doses, the patient or caregiver injected under supervision, with six supervised injections required before self-injection at home, and home administration permitted from Week 10 for trained patients/caregivers ³⁶³⁸³⁹. Aimovig (erenumab) ran a dedicated home-use substudy in which patients self-administered the 140 mg dose at home using either two prefilled syringes or two autoinjectors; of 136 enrolled, 129 completed, with only two device-related problems reported and injection-site reaction rates comparable to the wider program ¹⁵². Repatha (evolocumab) studied home self-administration (study 20120356) and found that after training the majority of subjects could self-administer two full 420 mg doses at home with either the automated mini-doser/on-body system or the autoinjector/pen, with similar LDL-C reductions and adverse events across devices ⁷¹⁷³⁷⁴⁷⁶.

On PK bridging, FDA reviews show the pattern clearly. For Benlysta, a randomized parallel-group study in healthy subjects compared a single 200 mg SC dose by PFS versus autoinjector across abdomen and thigh with weight stratification; geometric-mean exposures were similar and the 90% confidence intervals for PFS:AI ratios fell within bioequivalence limits (AUC0-inf 0.94 [0.887, 0.997], AUC0-t 0.95, Cmax 1.05), supporting addition of the autoinjector ²⁶²²⁶³²⁷⁰. For Trulicity (dulaglutide), a comparability study showed the single-use pen was PK-equivalent to the PFS (AUC0-inf ratio 1.02 [90% CI 0.998, 1.04], Cmax ratio 1.02), and FDA concluded the pen was equivalent from a clinical-pharmacology standpoint ²⁶⁴²⁶⁵. For Vyleesi (bremelanotide), a two-part crossover study (PT-141-56, 36 healthy women) served as the PFS-versus-autoinjector bridge ²⁷¹.

EMA EPARs rely on the same device-bridging logic:

Saphnelo (anifrolumab): a relative-bioavailability study of 120 mg SC via autoinjector versus prefilled syringe demonstrated bioequivalence, with a roughly 6% lower total protein content in the AI batch judged not to compromise the conclusion ²⁴⁰.
Tezspire (tezepelumab): the PATH-BRIDGE study (D5180C00012) compared accessorized PFS or autoinjector against vial-and-syringe and demonstrated bioequivalence for AUCt and AUC-infinity across the device/formulation comparisons ²⁴⁷.
Repatha (evolocumab): study 20120133 showed the 140 mg PFS was PK-equivalent to the reference autoinjector/pen, and study 20110168 bridged the 420 mg multi-device presentation ²⁵².
Praluent (alirocumab): single-dose autoinjector versus the SYDNEY device gave comparable exposure (treatment-ratio point estimates near 1.1 for Cmax and AUC0-tau) ²⁵³.
Trulicity (dulaglutide): transferability rested mainly on a bioequivalence study comparing the PFS and single-use pen (study GBDT), supported by usability studies for both devices ²⁵⁶.
Libmyris/Hukyndra (adalimumab biosimilar): the program included a randomized two-arm PK study comparing PFS versus autoinjector delivery ²⁴⁵²⁴⁸.
Oczyesa (octreotide): prefilled pens and syringes gave similar PK profiles in study HS-19-664 ²⁵⁰.

Presentation-by-presentation: what shifts the evidence burden

Autoinjectors and prefilled pens. The dominant evidence is HF validation on the final user interface plus, for platform devices, cross-reference to prior data. FDA has repeatedly allowed sponsors to leverage validation from an established platform, Amjevita from the Enbrel SureClick, Erelzi from the Cosentyx Sensoready, while harmonizing device steps with the reference-product IFU and justifying any differences ¹⁵⁸¹⁶⁷. Reviewers pay close attention to whether users understand device feedback, hence FDA's insistence on explaining the meaning of audible clicks in the Yuflyma IFU ¹⁶⁴¹⁶⁶. On the EMA side, Nucala (mepolizumab) supported self-administration with two real-world use studies: an autoinjector arm (104 subjects) and a safety-syringe arm (56 subjects, 100% injection success across observed and home doses), with the agency concluding the HF studies were sufficient validation that the EU package-leaflet IFU was readable, comprehensible, and usable ²⁰⁹²¹¹²¹²²¹⁸. Tremfya (guselkumab) bridged from its pivotal PFS to a prefilled pen (study CNTO1959PSO3006), with comparable efficacy, PK, and immunogenicity and satisfactory self-administration on the Self-Injection Assessment Questionnaire ²²²²²³.

Prefilled syringes. These carry a lighter device-interaction burden than reconstituted products but still require validated IFUs covering warming to room temperature, needle-cap handling, site selection and rotation, plunger technique, and disposal, as seen across the Kesimpta, Besremi, and Amjevita reviews ⁴²⁴⁴¹⁵⁸. Dupixent illustrates presentation-specific IFUs: the label documented two PFS presentations (with and without needle shield), each with its own IFU, and site guidance differing for self- versus caregiver-administration ³¹³²³³.

Vial-and-syringe and reconstitution. When the patient must reconstitute a lyophilized product or draw from a vial, FDA has leaned heavily on clinical training with return demonstration plus detailed IFUs, and has flagged where extra training is needed. For Arcalyst (rilonacept), patients/parents were clinic-trained in aseptic reconstitution, withdrawal, and SC injection, with written instructions and practice, and the review noted some subjects needed longer training to reconstitute and self-inject correctly ⁸³. Synribo (omacetaxine), a single-dose vial requiring reconstitution, was supported by training plus a Medication Guide and IFU covering transport of reconstituted drug, storage/use times, gloves and eyewear, handwashing, injection technique, site rotation, spill procedures, and biohazard disposal ⁸⁶⁹⁵⁹⁶⁹⁸. Saizen (somatropin) required training in vial reconstitution (diluent, gentle swirling, no shaking) before home use ⁸⁷, and Enbrel (etanercept) multiple-use vials required aseptic reconstitution with a vial adapter or 25-gauge needle and detailed technique tied to the IFU and medical supervision ⁸⁸⁹⁰⁹¹⁹²⁹³⁹⁴⁹⁹. Nivestym (filgrastim-aafi) labeling makes the human-factors logic explicit: a patient/caregiver who cannot successfully demonstrate dose measurement and administration may not be an appropriate candidate, and a vial rather than the PFS should be used for doses under 0.3 mL because the needle guard can obscure the markings and raise dosing-error risk ⁸⁴⁹⁷. On the EMA side, Nplate (romiplostim) self-administration was supported by a home-administration training pack, an HFE study, an FMEA, and patient/physician education, with most adults able to administer correctly after training ²²⁵.

On-body and wearable large-volume devices. These raise the bar because the user interaction, while limited, is novel and the failure modes (dosing failures, infection) are consequential. Furoscix (furosemide) used a pre-programmed on-body infusor delivering a fixed 10 mL SC dose over 5 hours with a single activation button; FDA initially found it not approvable because the marketed device was changed after the HF study, leaving it unclear whether the data still covered the critical tasks and IFU, and it later revised confusing IFU wording that had tripped two participants ⁷⁸⁷⁹⁸⁰⁸¹⁸²²⁰⁷. The Repatha on-body delivery system with prefilled cartridge was supported by two HF summative (simulated-use) studies, though a later review of a similar device flagged that its HF study had not evaluated the final intended-to-market interface, and observed roughly 10% device-related complaints/dosing failures in a bioequivalence study ⁷⁷⁷⁵. Duopa (carbidopa/levodopa enteral suspension) underwent two HF validation tests, but the consultant observed programming failures and close calls that could cause over- or under-dosing, including confusion between Continuous Rate and Morning Dose values ²⁰⁶.

Labeling and Instructions for Use as validated evidence

Both agencies treat labeling and IFU not as an afterthought but as part of the user interface that must be tested and shown to work.

On the FDA side, the label operationalizes the validated workflow. Dupixent's labeling states the product is for use under HCP guidance and that a patient may self-inject after training in SC technique, requires training per the IFU, and restricts pediatric self-use (the pen is for ages 12 and up with adult supervision for adolescents; the PFS is caregiver-administered for children 6-11) ³⁰³¹³²³³²⁴²⁵²⁶. FDA's HF guidance also directs that important administration constraints, injection site, rotation schedule, multiple-injection instructions, depth, or duration, belong in DOSAGE AND ADMINISTRATION, and that the label should state where a pediatric subpopulation should not self-administer because self-administration was not evaluated in that group ⁶. Device labeling should describe safety features (alarms, warnings, switches) and be shown to perform as intended ¹, and home-use labeling should be clear, detailed, and identify device components ¹⁰.

On the EMA side, the product-information rules and risk-minimization materials carry much of the labeling evidence. QRD guidance requires statements on residual solution remaining in injection devices/prefilled pens, discourages over-labelling because it impairs readability, and emphasizes clear, consistent device instructions and directing patients to read the IFU before use ¹⁰⁰¹⁰¹¹⁰⁹. EMA expects that most users can benefit from the information, for example through appropriate user testing, and updated package leaflets after user testing feature in EU procedural guidance ¹¹⁹¹²⁰¹²¹. For self-injected and especially emergency products, EPARs document a layered set of educational materials as additional risk-minimization measures: Ogluo and Baqsimi (glucagon) relied on administration leaflets, instructional videos, and demonstration kits with trainer devices, with instructions not to test or unpack the device in advance ²¹⁴²¹⁵²¹⁷²²⁴²²⁸²³¹²³³; Zegalogue (dasiglucagon) used a leaflet and video for its single-dose PFS/pen designed for emergency use without delay ²¹⁹²²⁹; and Eurneffy (epinephrine) supported correct use with an exact-replica training device and step-by-step videos ²²⁶. Nemluvio (nemolizumab) had HF studies for both its dual-chamber syringe and autoinjector, and the IFU and carton were revised to add warnings such as "CAREFULLY read and follow Instructions for Use. This pen requires specific steps before injection" after mixing/reconstitution-step errors were observed ²¹⁰.

The EMA/CHMP framework: drug-device combinations, Article 117, and usability

EMA assesses self-injected products through its drug-device combination (DDC) framework. The guideline distinguishes integral devices (single-dose prefilled syringes, pens and injectors; multi-dose pens/injectors with non-replaceable cartridges) from non-integral ones (co-packaged or referenced, obtained separately, such as injection or filter needles) ²⁸⁶²⁸⁸. For integral DDCs, Article 117 of the MDR requires the marketing-authorisation dossier to include evidence that the device part conforms to the relevant General Safety and Performance Requirements, either an EU Declaration/Certificate of Conformity or, where unavailable, the applicant's GSPR-compliance confirmation or a Notified Body Opinion depending on the device class ²⁸⁹²⁹⁰³⁰⁹.

For usability specifically, EMA's expectations mirror FDA's risk-based logic. A usability study is expected where the device has not been used in the proposed patient population before, or where the setting of use is new or different, to show the DDC can be used safely to deliver the required dose to the target population ²⁹¹²⁹⁷. Published or other data for identical/similar devices may suffice, but a formal usability study is required if they do not ²⁹¹²⁹⁷. Usability/human-factors detail belongs in Module 5 with a summary in Module 3.2.R, and studies may also be triggered by device design changes or IFU changes that could increase user or medication error ²⁹¹²⁹⁷³⁰⁶. The guideline also expects demonstration of delivered-dose accuracy, attention to needle penetration force and mechanical functionality, and in-use stability/functionality under the SmPC conditions of use ²⁹¹³⁰⁸. Dupixent's EPAR shows the flip side of this test: when the intended user population was unchanged and bridging data from the identical marketed device part plus existing HF data were sufficient, EMA concluded no new usability study was needed ²⁵⁸²⁵⁹.

CHMP and PRAC proceedings also show the agency using post-authorisation evidence to reshape self-administration. For adrenaline auto-injectors, CHMP called user training "of paramount importance," recommended training devices, audio-visual materials, and a prescriber checklist, and cited a study showing some users could not use the device successfully ¹⁸⁰¹⁸¹. For darbepoetin alfa (Aranesp), PRAC drew on user-testing results to revise the user-manual part of the package leaflet after incorrect device use/malfunction ¹⁸². And for leuprorelin depot medicines, including a request to replace the Eligard device with one easier to handle, PRAC concluded that patients should not prepare or inject these products themselves, that they should be handled only by experienced HCPs, a reminder that a self-administration claim is not automatic and can be withdrawn when handling complexity outweighs the benefit of home use ¹⁸⁴¹⁸⁹.

What a persuasive self-administration dossier looks like

Read together, FDA and EMA reviews point to the same construct. A convincing self-administration claim is built from: (1) a use-related risk analysis that defines the critical tasks for the specific drug-device-population combination; (2) human-factors validation on the final user interface with representative patients, caregivers, and HCPs, ideally spanning trained and self-trained arms, with observed use errors adjudicated by clinical impact and mitigation; (3) clinical confirmation that patients or caregivers dosed themselves during development, plus PK bridging when a new device joins an approved presentation; and (4) a validated, plain-language IFU, backed where needed by videos, trainer devices, and other risk-minimization materials, with self-administration constraints (site, technique, pediatric limits, reconstitution) written into the label or SmPC. The evidence weight shifts with presentation, lightest for a platform autoinjector or prefilled syringe that can bridge to prior data, heaviest for reconstituted vial products and novel on-body devices, but the reviewer's question never changes: can the intended lay user, in the intended setting, complete the whole workflow safely with the labeling provided?