FDA Review of Needle Safety and Sharps-Injury-Prevention Features on Prefilled Syringes and Pen Needles

How FDA reviews needle safety and sharps-injury-prevention features on prefilled syringes and pen needles

FDA evaluates sharps-injury-prevention features through two routes. Stand-alone needle guards and safety pen needles are cleared as devices through 510(k). Drug-device combination products (prefilled syringes with integrated guards, and pen injectors) are reviewed under the drug or biologic application, with CDRH device consults. Both routes draw on the same core expectations: simulated clinical use, activation and lockout performance, override resistance, and labeling that tells the user how to confirm activation. The combination product route adds one more question: does the feature still work on the finished, aged, drug-filled product? The sections below summarize how that has played out in cleared 510(k)s and approved combination products.

The baseline: FDA's 2005 sharps guidance

The governing framework is Medical Devices with Sharps Injury Prevention Features, issued August 9, 2005 ²⁷¹. Its main recommendations:

• Simulated clinical use. Test the feature on patient substitutes, not patients, under a device-specific protocol. Fruit may stand in for subcutaneous or intramuscular use, but not for IV use. If the same feature is already legally marketed as part of another device, the sponsor may reference it instead of repeating the testing ²⁷².
• Sample size and failure tolerance. FDA describes 500 devices as feasible for many sharps-safety devices. That sample can detect grossly defective devices at a 1% level, and a successful study at that size should show zero protection-feature failures. Scientifically justified alternative sample sizes are acceptable ²⁷³.
• Design attributes.
  • The user should readily recognize that the feature has activated.
  • Once activated, the feature should not be deactivatable and should stay protective through disposal.
  • Active features should allow one-handed activation with the hands behind the sharp.
  • Report forms should capture ease of activation, unintended activation, and whether activation can be detected ^274275276.
• Failure reporting and labeling. Sponsors should submit all data, including failed tests. Each failure needs an explanation and a correction, and simulated use should be repeated after a redesign. Labeling should state sharps-injury prevention as an intended use and explain how to confirm that an active feature has activated ²⁷⁷²⁷⁸.

Later drug-delivery guidances apply these principles to combination products:

• Essential drug delivery outputs (draft). For a prefilled syringe, needle safety activation force is an essential drug delivery output only if activating the safety feature is required to complete drug delivery ^407408409.
• Glass syringes and ISO 11040-4 (draft). The guidance lists anti-needlestick mechanism performance testing with the glass syringe, and connectivity to needles and sharps-prevention features ⁴¹¹.
• Pen, jet, and related injectors (final). The guidance references ISO 11608-2 for pen needles ⁴¹³. It calls for testing that shows injector connections are compatible with devices needed for use but not included in the submission, such as needles ⁴¹⁴.
• Emergency-use injector reliability (draft). Design-reliability examples include the force to deactivate a safety mechanism and the force to remove needle shields ⁴¹⁵⁴¹⁶.

Several of these drug-delivery guidances are drafts that are not for implementation, so they should not be cited as final requirements ⁴⁰⁵⁴⁰⁶.

510(k) review of needle guards for prefilled syringes

Passive needle guards that clip onto glass prefilled syringes have been cleared under product code MEG. In these clearances, substantial equivalence arguments rest on an unchanged activation principle, supported by bench and simulated-use data.

Two points stand out:

• Because the guard is cleared as an accessory, it can be referenced by many drug sponsors. Changes to its materials, or to the primary container it fits, have been handled through new 510(k)s.
• The public summaries often confirm only that simulated use was performed. Sample sizes and activation or override forces are usually not given.

510(k) review of safety pen needles

Safety pen needles are cleared under product code FMI, often with a non-safety pen needle in the same submission ²²⁵²²⁴. The evidence package has settled into two parts:

• ISO 11608-2 for needle performance and injector compatibility.
• ISO 23908 for the sharps-protection feature: access in safe mode, activation force, and post-activation override/unlocking force.

Examples:

• Clickfine AutoProtect (K152514). For a modified 5 mm needle, risk-based verification covered ISO 11608-2 and ISO 23908, and injector compatibility passed ⁸²⁶⁰.
• Dual-Safety Pen Needle (K191853). Added pen-end protection to the patient-end protection of predicate K161950 ⁸⁵⁸⁶. The reported ISO 23908 activation force was 1 to 5 N at both ends, and override force was greater than 20 N at both ends ⁸⁷⁸⁸⁸⁹.
• Promisemed Covered Safety Pen Needle (K202681). Moving the trigger shield covered the needle before use. Bench testing included ISO 23908 access in safe mode, activation, and override/unlocking force. A change from gamma to EO sterilization was separately validated ^92939495.
• Safety Pen Needle (K210864). Predicate K152514. Passed ISO 23908 activation and challenge-in-safe-mode testing, along with ISO 11608-2 needle and injector-compatibility tests ^1016810254.
• Safety Pen Needle for Single Use (K212514). Reported a mean activation force of 3.71 N, against 3.73 N for predicate K181447. This is a direct quantitative comparison with the predicate ¹⁰⁶¹⁰⁷.
• Disposable Insulin Pen Needle (K221178). A red slider indicates activation and the needle cannot be reused. Model ST met ISO 23908:2011 guard-activation and override/unlocking-force requirements ^108109110111.
• Promisemed X-Safety (K220129, K261719). The added cartridge-end shield was validated through ISO 23908 ^103104105. The 2026 follow-on clearance cites ISO 23908:2024 and ISO 11608-2:2022, which shows sponsors moving to newer standard editions ^115116220.

In the summaries reviewed, simulated clinical use under the 2005 guidance appears less consistently for pen needles than ISO 23908 bench testing. Some sponsors did report it:

• BD AutoShield Duo (K110703). Healthcare professionals and lay pen users performed safety-feature activation testing and a simulated-use IFU validation. Activation met criteria attributed to the 2005 guidance ³⁷⁴.
• DropSafe Safety Pen Needles (K211716) and DropSafe Sicura (K223353). Both assessed safety-feature function and validated the IFU in simulated clinical use. DropSafe used clinical and non-clinical users; Sicura used clinical users ³⁷⁶³⁷⁷.

How simulated-use studies are designed and reported in 510(k)s

Summaries vary widely in how much study design they disclose. The most quantitative examples come from adjacent syringe and lancet clearances:

• Safety Auto-Disable Syringe (K143497). Tested 512 devices with no failures. The summary reports 97.5% confidence that the true failure rate is no higher than 0.7%, and 99.5% confidence that it is no higher than 1.1% ³⁷³.
• Auto-Disable Sterile Safety Syringe (K121637). Ten nurses tested 500 subject syringes and 100 comparator syringes ³⁷².
• Lancets and safety syringes. Many clearances report 300, 500, or 600 samples per model, referencing both the 2005 guidance and ISO 23908 ^378379381382.

Most summaries state only that the feature "met pre-established criteria." They do not give the acceptance threshold ^387390395. For sponsors, the confidence/reliability framing in K143497 is the most defensible public precedent for setting sample size.

Combination product review: the finished product is what counts

In NDA and BLA reviews, CDRH consults have increasingly asked whether the needle guard works on the finished combination product. A component 510(k) is not enough on its own. Several reviews limited the device scope to needle-safety performance under ISO 23908:2011 and the 2005 sharps guidance, separate from primary-container functions such as glide force or needle-shield removal ²⁶²⁷²⁸.

Accessory guards with a 510(k)

• Zarxio. Used the CDRH-cleared UltraSafe Passive Needle Guard. The cited special 510(k) was a materials change that the review said did not affect safety-device function ²⁹³⁰.
• Actemra. Used a commercially available, 510(k)-cleared passive guard on a 1 mL glass prefilled syringe ³¹.
• Ryzneuta. Used a passive UltraSafe PLUS guard ⁴⁶.
• Imuldosa. The reviewer noted that the guard had no active 510(k) and was described in a master file instead ⁴³.

Reliability on the aged, conditioned product (Adbry)

The Adbry review is the clearest example. Activation and lockout had been evaluated, and the guard component had 510(k) testing. Reviewers still found two problems:

• The sample size did not support an adequate reliability claim.
• The evidence did not represent the final finished product after shelf-life aging, drops, and simulated shipping.

FDA asked for activation and lockout testing at 95% confidence and 99% reliability after sequential conditioning. The sponsor then described finished-product testing after real-time aging, drop, and shipping conditioning, including compressive override-force testing. The consult recorded no postmarketing commitment or requirement ³⁶³⁷³⁸.

500-activation simulated-use data (Cosentyx)

• Prefilled syringe. Novartis reported 500 attempted and successful simulated-use activations, plus design-verification checks for one-handed activation and prevention of premature activation ³²³³.
• Autoinjector. CDRH separately questioned missing sharps testing. The response cited simulated clinical use and a supplier test in which the needle cover resisted an 80 N override force ³⁴³⁵.

Active guards and long-term reliability (Zimhi)

For Zimhi's active needle guard, reviewers examined visible activation, protection through disposal, one-handed activation, and finger access. The sponsor reported finger-access testing under ISO 23908 Annex B ³⁹⁴⁰. Earlier, FDA had flagged inconsistent aging data and requested evidence that the device performed at expiry after shipping challenge ⁴¹. The reviewer recorded a postmarketing requirement or commitment for an updated, complete fault-tree analysis targeting 99.999% reliability at 95% confidence ⁴².

Pen injectors and needle-stick mechanisms

• Somatuline Depot. FDA requested anti-needlestick mechanism testing, simulated clinical-use evidence that sharps hazards were mitigated, bench data, and human factors validation. It cited guidance recommending 500 successful actuations with zero failures ³²⁷.
• Lanreotide acetate injection. A 500-device simulated-use study showed all needle shields deployed and locked. The reliability claim for the guard was deferred to CDRH ^202203204.
• Plegridy. FDA compared the pen's sharps protection with the Avonex Pen and found it apparently adequate. It still requested confirmation of a shield feature, an updated needlestick risk analysis, and usability protocols able to detect shield failures, injuries, or close calls ^324325326.

Pen-needle compatibility

For pen injectors sold without needles, reviewers treated compatibility with ISO 11608-2 needles as an essential performance requirement:

• Soliqua 100/33. Supported by a needle-compatibility verification report ^309310311.
• Semglee. Tested one needle type from each manufacturer against ISO 11608-2:2012 Clause 11 ³¹⁹³²⁰.
• Lyumjev KwikPen. FDA asked for the sponsor's BD needle reports because dose accuracy depended on compatibility. The response showed ISO 11608-1 dose accuracy was met with three needle sizes ³²².

None of the reviews located establishes that a named pen was specifically verified with safety pen needles. Compatibility evidence generally referenced standard ISO 11608-2 needles ³⁰⁹³¹⁸.

Human factors findings on needle-shield use

Human factors validation reviews often surface shield-activation errors. FDA's response depends on whether a needlestick could result:

• Actemra. Two participants did not fully depress the plunger, so the shield did not activate. The sponsor revised the IFU, and a supplemental study confirmed the change worked ¹⁹⁵¹⁹⁶.
• Hemlibra. Two participants failed to engage the shield but put the syringe straight into a sharps container. FDA judged these study artifacts and accepted the residual risk ¹⁹⁷.
• Cablivi. Seven participants did not activate the shield before disposal. Reasons included unfamiliarity and a habit of recapping. FDA found the IFU adequate and required no further mitigation ¹⁹⁸¹⁹⁹.
• Hadlima. FDA accepted the residual risk for early plunger release before the shield triggered, and for one participant who recapped a used needle without a stick ²⁰⁰²⁰¹.
• Ryzodeg 70/30. Eight participants had needlestick injuries despite existing IFU precautions. FDA said mitigation beyond IFU revisions was needed and would have to be validated ²⁰⁵²⁰⁶.
• Uzedy and Abilify Asimtufii. Excerpts note errors in locking Uzedy's safety needle shield, and premature shield activation that stopped an Abilify Asimtufii participant from completing the injection ²⁰⁷²⁰⁸.

On Adbry, the CDRH consult explicitly left human factors to DMEPA. This reflects the usual split: CDRH reviews engineering reliability, and DMEPA reviews use-related risk ⁵².

Practical takeaways for sponsors

1. Plan around the 2005 guidance's 500-device, zero-failure simulated-use baseline, or justify an alternative statistically. Precedents include confidence/failure-rate statements (K143497) and 95% confidence/99% reliability targets (Adbry) ^27337336.
2. For safety pen needles, pair ISO 11608-2 with full ISO 23908 testing. That means access in safe mode, activation force, and override/unlocking force, run on sterilized and aged product. Quantify activation force against the predicate where possible ^87106108.
3. For combination products, test the finished product after aging, drop, and shipping conditioning. Do not rely only on the guard supplier's 510(k) ³⁶³⁷. Confirm whether the guard has an active clearance or is supported by a master file ⁴³.
4. Decide early whether shield activation is required to complete delivery. Under the draft EDDO guidance, if it is, activation force becomes an essential drug delivery output with associated control-strategy expectations ⁴⁰⁸⁴⁰⁹.
5. Expect human factors scrutiny of shield activation and recapping behavior. IFU changes are often accepted for low-severity errors. Actual needlestick injuries in validation can lead FDA to require design-level mitigation ^197205206.