By Bryon Calawa, Design Science
Confession from a human factors consultant: I’m tired of being sprayed in the face during usability testing. It’s happened in nearly every simulated-use study I have conducted with an auto-injector, regardless of its design. The user, typically someone naïve to the injection process, removes the auto-injector from the injection site too soon, angles the device towards me, and empties the contents of the auto-injector on my face. At this point, I’m considering wearing goggles to my next study.
Although it’s probably quite funny for my colleagues, wet injections are a serious issue for medical device manufacturers that produce auto-injectors. As a combination product, the drug and device components of an auto-injector need to work together in a reliably safe and consistently effective fashion. The FDA guidance is clear: medical devices must be both safe and effective for use. It’s hard to demonstrate the latter when the medicine ends up on the user’s leg.
Unlike a syringe, the premature removal of an auto-injector is unrecoverable; the entire contents of the device are expelled, uninterrupted, once the injection begins. Users can click a button and be on their way to a dose, but this convenience has associated risks and costs.
Since it is currently impossible to measure the amount of medicine successfully injected in these cases, the user is often instructed to wait until the next scheduled dose to attempt another treatment. The obvious risk with this direction is delay of therapy, which can be significant when considering that these medications are often indicated for bi-weekly or monthly injection. The less-obvious waste is the financial one. Auto-injector prices and differences in insurance coverage can make this use error extremely costly.
Medical device manufacturers have attempted to address wet injections through product design. However, my experience with usability testing has demonstrated that many of these risk controls have potential design flaws that can make them ineffective at best, and detrimental at worst. In this article, I outline some of the most common risk controls utilized by on-market devices, as well as reveal where these solutions struggle in actual use. I’ll then describe different ways to better mitigate the risk of wet injection.
Visual feedback indicating a complete dose is common in most auto-injectors designs. Often, this is as simple as providing a window that allows the user to watch the plunger push the medicine out of the device. Although this solution may be the simplest to implement, it has some significant limitations.
This type of feedback requires the user to have a basic understanding of the device’s mechanics to comprehend what the window indicates. That is, do they know what they’re looking at? While this visual feedback may be common sense to engineers, user misinterpretation of the window’s meaning can just as easily lead to a wet injection as prevent one.
For example, I’ve observed users remove the auto-injector at the first sign of plunger movement and conclude that they had delivered the full dose successfully, based on the plunger filling the window. Furthermore, a colorless liquid can easily be mistaken for an empty chamber when viewed at a glance. Simply revealing the inner working of the device is not sufficient if the user does not understand the device mechanics. It shows, but it doesn’t always tell.
The position of the window is another crucial design limitation. The window must be located so that the medication and plunger are in view (i.e., typically on the lower half of the device body). However, this is also where most users tend to grip the device, either to accommodate a larger hand size or to increase the perceived stability of the injection. Another important technical consideration is that users perform injections in various locations on the body. The position of the window must be such that it is visible at all appropriate injection locations. Although the window approach may be straightforward to design, the necessary position of the window somewhat limits its utility.
What can you do to improve the visual feedback of your auto-injector? Consider adding additional indicators to complement the window design. The indicator should be located at the distal, non-needle end of the device to increase its overall visibility. It should be visible from both the top and the sides of the device to account for variations in grip and injection location. Also consider utilizing symbols to indicate the status of the injection to increase the likelihood that the user will successfully interpret the feedback. Increasing visibility and decreasing ambiguity are paramount to the development of an effective visual feedback system.
Along with visual feedback, most on-market auto-injectors incorporate audible feedback to indicate the beginning and/or the end of the injection sequence. Typically, this sound is generated by the spring-action that triggers the injection, or by the plunger bottoming out once the injection has completed. Similar to the window design approach, this feedback system makes good use of the internal mechanisms of the auto-injector, but may not always provide sufficient feedback for every individual user.
One major drawback to this approach is the potential for cross-device variability in feedback generated by these mechanisms. For example, the volume of the auditory feedback may differ between injectors. For a user that has experience with a device that clicks once at the end of an injection, using a device that clicks at both the beginning and the end of an injection sequence could result in premature removal from the injection site.
Another issue with this approach is reliance on the user to understand the purpose of the clicks and to interpret them accordingly during an injection. Therefore, what is intended to assist the user can actually contribute to failure to administer a complete injection, due to this subjectivity. A sudden, loud click sound at the beginning of an injection may unintentionally lead the user to remove the device prematurely, assuming that the injection is complete. Conversely, a quieter click sound, after the user becomes habituated to a louder click, may be overlooked as an anomalous sound, rather than important device feedback. The primary role of these mechanisms is to deliver the medication, but the risk of wet injections makes this feedback too important to be considered a secondary feature. Consistency is key when developing feedback systems.
It is also critical to consider the potential use environments of the auto-injector when developing an auditory feedback system. Utilizing the injection mechanisms of the device limits the frequency and volume range that may be necessary to accommodate noisier environments. Emergency-use products are intended to be used in a wide range of ambient noise levels. Products that are not intended for emergency use must still compete with household sounds. Audible cues must be clear and distinct from the environment in order to be effective. The click noise implemented by most on-market devices may not be sufficient in actual use.
A successful audible feedback system for auto-injectors depends on the sound being consistent, intuitive, and sufficiently distinct from other environmental noises. One example of a successful system is the Auvi-Q epinephrine auto-injector. The device gives audio feedback through small speakers that provide step-by-step instructions for the injection sequence, including an explicit indication of when to remove the device from the injection location. This level of detail is likely unnecessary for most auto-injector products, but it shows one simple way to reduce ambiguity in the user’s experience.
It is expected that a doctor or nurse will review how to use a particular auto-injector with their patient before prescribing the medication. In fact, we sometimes simulate this training experience with our participants in usability testing. The training is usually effective; trained users tend to have fewer errors than untrained users. Wet injections, in particular, become much less likely.
But this training is not a guarantee in the real world. Some users will receive insufficient training, while others may receive no training at all. And given the frequency of these injections, it is likely that most users will experience some kind of training decay. For training to be most effective, it not only needs to sufficiently depict the full injection process, but it also needs to be readily available to all users at any time.
Some on-market products have attempted to meet these criteria by developing a trainer device that is provided to the first-time user. Usually this trainer device is a needle-less, resettable version of the actual product. However, engineers face a special challenge in creating an effective trainer device: the trainer must replicate all feedback mechanisms that are present in the actual product. This point is critical, as the user must learn what to expect during the injection process and must also have access to the feedback that instructs them when to remove the device from the injection site. It may be easier to develop a trainer that, for example, does not have a working plunger. However, this type of approach does not train the user to rely on the motion of the plunger when determining whether an injection is complete, thus reinforcing bad habits and potentially contributing to wet injections.
Another solution is to include a training video for users to review prior to administering an injection. There are several effective media that can be used to provide this service: DVDs, websites, or cell phone apps would all be sufficient to reach most potential users. Having this video available on multiple platforms would also increase the likelihood that users will be able to access this content.
The best possible solution is likely a combination of the previous two: a fully functional trainer device that users can practice with while being guided through the process via a training video.
Instructions for use (IFUs) are considered the lowest form of risk control by most regulatory bodies, but good IFUs can be effective in decreasing the probability of error in auto-injector use. Mitigating wet injections through effective instructions poses a unique challenge compared to other steps of the injection process, though.
In most cases, use steps should only include one action at a time, ensuring that no actionable steps may be accidentally overlooked. However, this standard is particularly ineffective for the actual injection steps; an instruction to press the auto-injector into the skin that does not also indicate to hold the auto-injector in place runs a considerable risk of wet injection. Therefore, injection steps should incorporate both actions — “activate” and “hold” — in the text of the instruction.
The IFU must also clearly describe the indicators for a complete injection. One common bit of feedback I have received from participants using an auto-injector for the first time is that the description of the visual feedback (in most cases, the plunger in the window) and/or the audible feedback (in most cases, a click noise) does not match what they experienced in actual use. Clear and accurate descriptions of the anticipated feedback decrease the likelihood of feedback misinterpretation and, therefore, wet injection.
Finally, most auto-injector IFUs describe a specific hold time to ensure that the full dose of medication has been administered and absorbed before removal. Although this practice makes sense on paper, usability testing has clearly demonstrated that users can have very different ideas of how much time has elapsed. In an extreme example, I have had participants report that they counted out a 15-second hold time, where in reality they held for less than two seconds. Given the unreliability and subjectivity of a hold time recommendation, IFUs should focus primarily on directing users to device feedback indicators in order to confirm that an injection is completed.
While training and instructions can tell users what to do and how to do it, device feedback is the first line of defense against wet injections. Visual feedback on the device itself is a common, but limited, approach. It relies on the user’s attention, understanding, and hold technique to be successful, while ambiguity and misinterpretation are not uncommon. Even more so, audible feedback is easily ignored and misunderstood, leaving the difference between a complete and incomplete injection up to a single click. Ultimately, it may take a new perspective on auto-injector design to ensure that we all stay dry.
About The Author
Bryon Calawa is an analyst for Design Science. He has experience as a research associate, moderating usability studies and writing usability study protocols, and as a photographer and an A/V technician. Bryon holds a degree in psychology from Eastern University.