FDA Human Factors Validation for Autoinjectors and Pen Injectors: Caregivers, Pediatric Users, and Patients with Dexterity or Visual Impairment

FDA human factors validation for autoinjectors and pen injectors: caregivers, pediatric users, and patients with dexterity or visual impairment

FDA reviewers, mainly the Division of Medication Error Prevention and Analysis (DMEPA) in the reviews summarized here, do not assume that human factors (HF) validation data from adult self-injectors carry over to caregivers, children, adolescents, or users with physical limitations. Across Drugs@FDA review packages, the same themes recur. Each intended user population is treated as a distinct group needing its own representative participants, usually at least 15. Labeling is narrowed when a group was not tested. Use errors are judged by their root cause and clinical consequence rather than by a pass rate. Device or instruction changes are requested when existing mitigations fall short.

Key takeaways

• Distinct user groups, 15 per group. FDA guidance treats pediatric and adult users, and lay and professional users, as distinct populations. It treats caregivers as distinct when their role or tasks differ in ways that could affect use, and it generally expects at least 15 participants per distinct group ⁶¹⁶²⁶³.
• Untested pediatric self-use leads to restricted labeling. For Hadlima and Abrilada, DMEPA recommended labeling for adult self-administration or caregiver administration only, stating that pediatric self-administration had not been tested ³²³⁰. Actemra and Orencia labeling similarly limits pediatric use of the autoinjector presentations ²⁴¹²⁴³.
• Impairments must be represented, either with real participants or with simulation. FDA asked Novo Nordisk to include at least 15 patients with diabetic retinopathy and 15 with neuropathy in insulin pen studies ¹⁶⁷¹⁶⁸. Where visually impaired participants were not enrolled, FDA asked for simulated visual impairment during patient tasks ¹⁷⁴¹⁷⁵.
• Training decay and untrained use are scrutinized. DMEPA rejected an Otrexup actual-use study as validation because it had no interval between training and self-injection ¹³⁷. For Repatha SureClick, failure rates were 50% in untrained participants versus 21% in trained participants, which led to a training statement in labeling ¹⁴⁷.
• Residual risk depends on consequence and recognizability. FDA accepted some hold-time errors where users recognized them and the likely harm was a one-time underdose ⁹⁴. It required design changes where errors could block delivery of an emergency drug ^99100101.

The guidance framework FDA applies

FDA guidance tells sponsors to define distinct user populations based on differences in characteristics, knowledge, experience, roles, or tasks that could affect interaction with the device. Pediatric and adult users, and healthcare professionals and lay users, are named as distinct populations when both are intended users. Caregivers are evaluated as a distinct group if their role or tasks differ in a way that could affect use, and caregiver reliance can vary across pediatric users ⁶¹⁶². In general, validation should include at least 15 participants from each distinct population. Participants should reflect the range of characteristics within each group, including representative pediatric users and users with relevant functional limitations when those users are likely to use the device ⁶³⁶⁴.

On physical and sensory limitations, FDA's HF guidance lists size, strength, dexterity, coordination, vision, hearing, tactile sensitivity, cognition, and medical conditions as user characteristics to account for ⁴⁸. For pen, jet, and related injectors specifically, it says sponsors should consider users' limitations in perception, strength, mobility, and manual dexterity. It notes that injectors for users with visual or dexterity limitations might need features such as auditory alerts or a larger, easy-to-grip design, and that the effects of these factors should be evaluated during development ⁴⁹. A draft guidance on rheumatoid arthritis drug development adds that the dexterity and visual acuity a self-administration device requires should be within patients' capabilities ⁵⁰. The passage on users also expressly includes parents who use devices on their children or supervise their children's use ⁵².

The draft combination-product HF submission guidance asks sponsors to document participant training and explain how it corresponds to real-world training and any applicable training decay ⁶⁵⁶⁶. Critical tasks are identified through the use-related risk analysis. Each task needs defined success and failure criteria, and observed use errors, difficulties, and close calls must be analyzed ⁶⁷⁶⁸⁶⁹.

Caregivers as a distinct user group

FDA reviews treat caregiver performance as its own evidence stream. Several examples show how caregiver findings changed the outcome:

• Dawnzera autoinjector. FDA reviewed separate adult/caregiver and adolescent studies. One caregiver had difficulty locating the concealed needle, and FDA found existing mitigations inadequate and recommended a further change. It did not require additional validation data for that change ^33343536.
• Repatha SureClick. Cap-removal use errors occurred in two adolescent patients and one caregiver, and two caregivers had a close call or difficulty. Users twisted the cap rather than pulling it. DMEPA disagreed with the applicant's view that no change was needed and recommended additional labeling mitigation without further HF data ³⁷.
• Zegalogue (dasiglucagon) autoinjector. The study enrolled 15 injection-experienced adult lay caregivers, 15 injection-naive adult lay caregivers, 16 adolescent lay caregivers aged 10 to 17, and 15 first responders ⁴¹. Errors (reported in aggregate, not by group) included injection at an incorrect site such as the forearm or lower back, two injections through clothing, one failure to remove the cap, and five early removals before full dose delivery ²⁵³. DMEPA found the residual risks acceptable after considering participant feedback, root-cause analysis, and similar approved products. It also found the final IFU acceptable ²⁵⁴. The clinical review noted that caregivers after marketing might take longer than trained study personnel, and that validation data confirmed caregivers could understand and use the device, with remaining errors mitigated through the IFU ¹¹².
• Gvoke HypoPen. The validation studies for the 0.5 mg and 1 mg autoinjector each enrolled 75 participants drawn from first responders and experienced and naive adult and pediatric caregivers ⁴²¹¹¹.
• Bynfezia Pen. The pen was tested with 15 patients, 15 caregivers, and 15 healthcare providers, all untrained and with optional IFU use. FDA said it could evaluate the results despite concerns about leading language in the moderator script ¹⁴⁵¹⁴⁶.

Including a caregiver group does not mean FDA reviewed it. For the Exdensur autoinjector, GSK included pediatric patients and caregivers among its lay-user groups. Because only healthcare professionals were the intended administrators, DMEPA focused on the HCP results and did not review the lay-user data ⁴⁰.

Pediatric and adolescent users

Pediatric self-administration requires pediatric data

The adalimumab biosimilar reviews show the clearest pattern. For the Hadlima autoinjector, pediatric patients with juvenile idiopathic arthritis (JIA) were not included in the HF study. DMEPA rejected the argument that shared injection responsibility with caregivers or healthcare providers removed the need to assess pediatric self-users. It recommended labeling for adult self-administration or caregiver administration only ¹⁵³².

For the Abrilada prefilled pen, the validation study included nine JIA patients and seven proposed JIA surrogates. Clinical reviewers did not consider the surrogates representative. DMEPA judged nine patients insufficient against the 15-per-group benchmark and recommended the same adult-or-caregiver labeling, with a statement that pediatric self-administration had not been tested ^30239240.

Approved labeling for other biologics follows the same logic. Actemra labeling permits JIA patients to self-inject with the prefilled syringe, or a caregiver to administer it, when the practitioner and parent or guardian agree, but does not extend the same pediatric self-injection allowance to the autoinjector ²⁴¹²⁴². Orencia labeling uses parallel language for the autoinjector ²⁴³²⁴⁴. In both cases the age-2 threshold is a dosing threshold, not a tested minimum age for autoinjector self-use ²⁴²²⁴⁵.

Pediatric sample size within subgroups

FDA applies the 15-per-group benchmark at the subgroup level when subgroups are meaningful. In the Semglee pen program, the studies enrolled 48 pediatric patients with diabetes and their caregivers among 168 participants. However, the untrained, pen-naive pediatric subgroup had only eight patients, and DMEPA said it would request additional pediatric-user data ³¹¹⁵¹.

When adolescent data were sufficient

For the Simlandi autoinjector interchangeability assessment, the untrained HF validation study included adolescent JIA patients, while the comparative-use study enrolled only Humira-experienced adults. Reviewers found few device-use errors, with a consistent pattern across groups, and JIA patients were not more error-prone. FDA concluded that adult comparative-use data plus adolescent validation data supported its conclusion without a separate adolescent comparative-use study ³⁸³⁹.

Adolescent-specific instructional failures

In the Dawnzera adolescent study, eight adolescents did not allow the device to warm to room temperature, because the IFU step title mentioned only removal from the refrigerator. Five adolescents selected an inappropriate injection site after following an illustration whose "For caregivers only" caption they missed. FDA recommended clarifying IFU Steps 1 and 3 and said these changes did not require another HF validation submission ^33343536.

Emergency-use injectors and child users

Emergency products raise the stakes because a use error can delay a life-saving dose. The Auvi-Q (Intelliject EAI) program tested untrained users including children aged 7 to 10 and 11 to 15. Children aged 7 to 10 were less likely than older users to follow instructions and use the device without error ²¹⁶²¹⁷. FDA identified residual concerns with safety guard removal, voice prompt timing, battery and switch malfunctions, and injection-site instructions. The sponsor revised the design and labeling, and a follow-up simulation in 20 pediatric and 20 adult users assessed those changes. CDRH ultimately found the evaluation of high-priority tasks acceptable ^196218219. A later Auvi-Q supplement shortened the instructed hold time from five to two seconds after reports of lacerations and embedded needles in young children who moved during caregiver injection ²²⁵.

For Teva's generic EpiPen/EpiPen Jr, FDA distinguished an exploratory head-to-head comparison from validation by representative users. It questioned a criterion under which 90% of users would deliver a dose within five minutes, noting that the remaining 10% and the reasons for their failures mattered. A later review found the results did not show that users switching from EpiPen could use Teva's device safely without provider intervention or training ²²⁰²²¹. On activation force, FDA noted a typical range of 5 to 10 lb and said maturity and judgment, rather than the ability to generate force, were the main considerations for children's self-administration ²²².

Symjepi, a prefilled syringe later approved after design and labeling changes, that FDA worried caregivers could mistake for an autoinjector, is a counterexample ¹⁰⁸. Its study had only 8 injection-naive lay users and 28 injection-experienced users. Testing also had training concerns and did not show that emergency stress was simulated ¹⁰⁹. Three of 36 participants failed to remove the cap, and others prematurely locked the safety guard or did not fully depress the plunger. In that review cycle, DMEPA concluded the study did not demonstrate safe and effective use ¹¹⁰. FDA attributed some errors to negative transfer from EpiPen experience and proposed design differentiation of the guard from the cap ^99100101. It also rejected patient preference data as a substitute for fixing failures that could cause serious harm ²²³²²⁴.

Users with dexterity impairment

Review records show FDA expects dexterity-limited users to be represented, but the reviews summarized here rarely report dexterity-specific error rates.

• Otrexup (methotrexate autoinjector). The validation study included at least 15 participants in each of three groups, one of which was patients with moderate-to-severe rheumatoid arthritis. The applicant screened so that participants met or exceeded the expected hand-function impairment of intended users. FDA found the validation study acceptable with no further HF concerns ¹⁸⁴¹⁸⁵.
• Toujeo SoloStar. FDA asked for clearer identification of disease-related limitations among validation participants ²⁵⁷. The patient sample included 9 of 60 with neuropathy, 12 of 60 with dexterity or strength issues, and 16 of 60 with arthritis in the hands or fingers ²⁵⁸.
• Ryzodeg 70/30 pen (Novo Nordisk FlexTouch/PDS290 program). In study UT54, five participants reported dexterity impairments (finger arthritis, poor hand rotation, a missing finger, mild diabetic neuropathy). Eight others performed handling tasks wearing gloves that limited sensation and movement ^186187256. FDA found the applicant's response on representation of users with neuropathy acceptable ¹⁸⁹. However, it kept separate concerns about carton and pen selection that were unrelated to dexterity ¹⁷³.

Two patterns stand out. First, FDA accepts simulation (sensation-limiting gloves) to supplement, though not fully replace, enrollment of impaired users. Second, acceptance of user representation is not the same as a finding that every impaired user can operate the device. For autoinjectors used in multiple sclerosis or psoriatic arthritis, the review excerpts did not describe dexterity-specific HF analyses ⁵⁴.

Users with visual impairment

Visual impairment is most visible in insulin pen reviews, where dose-setting and product selection are critical tasks.

• Retinopathy representation. For the PDS290/FlexTouch pens, FDA asked for the numbers of participants with visual and other impairments and said studies should include at least 15 patients with retinopathy and 15 with neuropathy ¹⁶⁷¹⁶⁸. In the submitted studies, UT59 included one participant reporting retinopathy and 20 who performed product-differentiation tasks wearing glasses that simulated retinopathy. In UT54, 16 participants performed handling tasks with simulated retinopathy. FDA found the explanation of user representation acceptable ¹⁶⁹¹⁷⁰.
• Simulated impairment when real participants are absent. For the Tresiba FlexTouch 200 units/mL pen, FDA initially considered 10 tested users insufficient after one made dose-setting errors. It said that if visually impaired participants were not included, patient tasks should incorporate simulated visual impairment. It also said untrained participants should be allowed, not required, to read the IFU ¹⁷⁴¹⁷⁵. After the later UT103 study, DMEPA attributed the few dose errors partly to recall bias and visual impairment. It concluded that the pen design, labeling, and support line sufficiently minimized medication-error risk ¹⁷⁶¹⁷⁷.
• Product differentiation. In one differentiation study, three of 105 participants selected the wrong insulin carton, one because they could not identify the green carton. FDA suggested further label optimization given the hypo- or hyperglycemia risk ^171172173.
• Toujeo U300. The study population included 3 of 60 patients with diagnosed retinopathy, plus patients with other eye conditions ¹⁷⁸¹⁷⁹.
• Mounjaro. FDA was concerned that patients with diabetes and visual impairment could have difficulty reading the quick reference guide's small text and illustrations, and recommended enlarging them ¹⁸⁰.
• Assisted use for severe impairment. In the PDS290 materials, the sponsor stated that people who are blind or have severe visual problems should use the pen with help from a trained person with functional eyesight ¹⁸¹.

Training, training decay, and untrained users

Caregivers and children may inject infrequently or in an emergency, so FDA looks closely at whether validation reflects realistic training. For Otrexup, DMEPA said an actual-use study could not validate the device because there was no interval between training and self-injection for a weekly product. It also asked that the training device operate like the commercial device ¹³⁷¹³⁸. Other programs built in decay periods: 5 to 7 days for the Praluent autoinjector, a one-week decay after a 45-minute session for Idacio, and at least one hour for the Trulicity pen ^139140141. FDA has recommended not specifically instructing untrained participants to read the IFU before attempting injection ¹⁴⁴¹⁴³.

Untrained performance can directly drive labeling. For Repatha SureClick, 50% of untrained participants versus 21% of trained participants had simulated-injection failures. FDA recommended a statement about needed training in the prescribing information and IFU ¹⁴⁷. FDA also tracks injection-naive users separately, as in the Humalog KwikPen review's separate reporting for 15 injection-naive patients ¹⁵⁰.

How FDA weighs residual risk

The table shows a consistent approach. Errors that users notice and correct, with limited clinical consequence, can be accepted with existing labeling. Errors that silently prevent dosing, or that stem from design confusion, lead FDA to ask for IFU revisions or device changes. This bar is higher for emergency products and for populations, such as young children and impaired users, who are less able to recover from an error.

Implications for development programs

• Define caregivers, pediatric self-injectors, adolescents, and impaired users as distinct groups in the use-related risk analysis when their tasks or capabilities differ. Enroll at least 15 per group, including meaningful subgroups such as untrained pen-naive children ⁶¹⁶³³¹.
• If pediatric self-administration is intended, test actual pediatric patients. FDA has rejected surrogates and shared-responsibility arguments, and the fallback is labeling that excludes pediatric self-use ³⁰³².
• Enroll participants with retinopathy, neuropathy, and hand impairment where the indication makes them likely users. Supplement with simulation (vision-limiting glasses, sensation-limiting gloves) as needed, and document the rationale ^168187175.
• Build realistic training-decay intervals and untrained arms into the protocol, and do not prompt untrained participants to read the IFU ¹³⁷¹⁴⁴.
• Analyze root causes of every critical-task failure, not just aggregate success rates, particularly for emergency-use products ²²⁰.

Questions that remain open for deeper research include how CBER-reviewed biologics treat these populations, how comparative-use threshold analyses are applied to generic autoinjector ANDAs for pediatric strengths, and how FDA has handled dexterity-specific findings in multiple sclerosis and psoriatic arthritis autoinjector programs.