The reliability toolkit: how to ensure medical device reliability

What is design for reliability?

Design for reliability is a development concept that focuses on systematically applying engineering tools and analysis to ensure a product’s reliability. There are three key things to consider when applying design for reliability to medical technology development: how circumstance affects its definition, its application across the development process, and finally how to apply the reliability toolkit to develop high-reliability devices.

Defining reliability – circumstance is key

Above all else, medical devices must be safe and effective. Regardless of the application area and user group, a crucial part of achieving this is ensuring that devices are reliable. Defining what reliability means for a given device, however, varies depending on a number of factors.

For example, in the FDA’s draft guidance for emergency-use autoinjectors, which talks to a reliability target of 99.999% (at the 95% confidence level), reliability is defined as:

“the probability that the injector will perform as intended, without failure, for a given time interval under specified conditions.”

Emergency-use autoinjectors are manufactured in high volumes, but each instance of the device is only ever tested once – at the point of application. Critically, this means the functionality of a given device cannot be assured prior to its use. Instead, the reliability burden lies entirely with the design and its manufacturing controls to ensure that, to an acceptable level, the device performance is established at the point of assembly and maintained across its lifetime.

A single-use autoinjector will spend the majority of its life (if not all, hopefully) in the unused state. Here, conditioning in the form of ageing, transportation, and thermal cycling may affect the performance of the device. Provided it is not unreasonable, the user of the device will expect their device to perform as required following this conditioning, and so the device should be sufficiently reliable to withstand any detrimental effects.

There are many other examples, however, where the entire device or elements of it are repeatably used across their lifetime. A common way of illustrating this is the bathtub curve, which shows the three different kinds of reliability failure over a product’s lifetime. In an ideal world, only the second kind, the constant failures, should be a worry, and even then we should be able to understand and predict when these will happen.

The challenge of increasing device complexity

The risk of device failure can increase where durable / multiple-use requirements introduce new and different challenges. Here, elements such as fatigue, creep and reprocessing become influential. For example, a complex device comprising a reusable instrument and a disposable consumable could be particularly challenging, as the reliability of the overall system is dependent on both the standalone reliability of the components, and the reliability of their interaction.

What use is the flawless execution of a final process if the system that supports it never reaches this final stage?

Design for reliability across the medical device development process

While the definition of reliability varies for different types of medical device, there are also common areas – the most prominent of these is the application of design for reliability across the medical device development process.

The development of a medical device follows a phased, stage-gated process that carries the project from design and development planning through to design transfer and market launch. Due to the challenges and demands of high-reliability design, it’s crucial to appreciate that design for reliability is an integral part of the development process and cannot be retrospectively implemented. Instead, medical device reliability standards must be considered in parallel to other key activities that take place as the project progresses.

Requirements definition and project planning stage

Design for reliability begins at the requirements definition and project planning phase. Here, the reliability target will be defined, and the workload required can be anticipated based on regulatory, technical, manufacturing and end-user requirements. Some examples include preconditioning requirements, the number of expected use cycles, and the production volumes. For medical devices, requirements for high reliability and mass manufacturability often coexist – at first, these requirements may be seen as competing, but through sound engineering design and considered, appropriate analysis, it’s possible for them to instead be complementary.

Concept generation stage

After the initial definition of the device requirements, concepts to address the key technical challenges are generated. Multiple designs may be generated in this phase, and their down-selection should be driven with reliability in mind. Considerations should include ease of use, performance robustness, ease of manufacture and assembly, as well as the utilisation of established principles and technologies. Here, simplicity and reliability often go hand in hand.

Proving the principle and suitability of the chosen technology can be achieved through a combination of paper-based and physical prototyping techniques – the exact method will depend on the complexity of the device, and the specific reliability requirements.

Detailed design stage

The detailed design stage of the device development builds on the foundation established during concept generation and can be considered as the workhorse of high-reliability design. Critical-to-quality attributes (CQAs) are identified, as are the limits of the design space and its sensitivities, which allows the relationship between Critical Material Attributes (CMAs), Critical Process Parameters (CPPs) and CQAs to be built, characterised, and understood.

Again, this is achieved through both theoretical and empirical methods, the collection of which can be considered as a ‘reliability toolkit’ that will be discussed in the final section of this article.

Furthermore, the reliability of the end product is as dependent on its manufacture as it is on its design. Open engagement with the chosen manufacturer becomes instrumental at the detailed design phase and should be carried through to when the design is transferred to manufacture.

Post-market support

Once launched, the burden of maintaining the reliability of the system lies primarily with its manufacturer. Controls are used to keep specific aspects of the manufacturing process within acceptable limits. These can be at the component or assembly levels, and can include metrology inspections, in-process force/position checks, and environmental conditioning.

The Reliability Toolkit

As already discussed, successful, high-reliability device design depends on both the definition of reliability and its careful consideration throughout the development process. Along the design journey, engineers and designers can employ a number of tools and techniques to both achieve and demonstrate their reliability targets, whether that be for a single-use ‘one opportunity’ device, or an instrument expected to execute thousands of cycles.

This ‘Reliability Toolkit’ can be divided into four areas, all of which can be used standalone, or in combination with each other:

1. Empirical tools
2. Analytical tools
3. Risk management tools
4. Control tools

Empirical tools

The use of empirical methods to test and challenge a design is a fundamental part of a product’s development. In the early stages of the process, a risk-based approach often drives the development testing strategy, where efforts are focused on establishing baseline confidence in the core system technology.

Test data may be both quantitative and qualitative and should assist with developing both the engineering and usability of the design. Both forms of data are invaluable and serve as a snapshot of the future design reliability. Test approaches may be as simple as ‘A/B’ testing, where two different designs are evaluated, or employ Design of Experiments (DoE) techniques to screen critical variables from a large pool. This approach can focus subsequent activities on the most critical interactions and features, streamlining the development work.

As the design matures, so does the rigour and scale of its testing. Critical component features may be manufactured at the extremes of tolerance to understand the design space and how reliability can vary across it. Again, a DoE can help increase the efficiency of these larger-scale activities, with output data feeding directly into efforts to either revise the design or tighten manufacturing controls.

Preconditioning and overstress testing (such as ageing, shock or thermal cycling) can be included in test programmes if deemed appropriate for the reliability requirements of the device. Finally, as test quantities increase (including as part of design verification testing), it is possible to use process capability analysis against well-established functional limits to predict system performance.

Analytical tools

In many scenarios, particularly for complex systems, it is not practical nor feasible to exclusively use empirical tools to create a high-reliability design. Instead, empirical approaches can be complemented by analytical tools to increase the breadth and depth of the analysis, ideally over a shorter timeframe.

Math modelling, for example, is a time and cost-effective way to simulate and interrogate system behaviour. The complexity of the model is directly related to two things: the physics involved and the level of fidelity required. Time or displacement-based models, based on engineering first principles and the assumptions that come with them, help identify the relationships and sensitivities between different parameters. Models can be quickly built in Excel, or with higher-end tools such as MathCAD or Python. Similarly, regression models can be derived from gathered data to interpolate or extrapolate results.

Finite Element Analysis (FEA) or Computational Fluid Dynamics (CFD) tools can offer a level of accuracy above first-principles math models and, provided the input data is representative, can be used to inform robust and reliable medical device design.

Tolerance analysis is widely used as a means of assessing the impact of variation from a manufacturing process, and to help design and manufacturing teams align on realistic and appropriate tolerances. The assessment can be both formative, to assess the variation expected from a number of concepts, or summative, to demonstrate that manufacturing controls are sufficient to support the chosen design intent. Statistical approaches to tolerance analysis, such as Monte Carlo, can be used to analyse non-linear component interactions. Again, the results of such analysis may affect design and manufacturing decisions and can also be integrated with edge-case medical device reliability testing to prototype device performance at the extremes of tolerance.

Risk management tools

Risk analysis and management is pivotal in high-reliability medical device design. As confidence in the design and manufacture of a device increases, the occurrence rate of critical failures will decrease well below what can be detected using empirical methods. As such, probabilistic techniques must be used which combine an intimate knowledge of system mechanics with its known failure modes to assess and mitigate risk, and to forecast the resultant reliability of the device.

There are various tools available to designers to achieve this. Two of the most commonly used in high-reliability design are Failure Modes, Effects, and Criticality Analysis (FMECA), and Fault Tree Analysis (FTA), which can be used in parallel to each other.

FMECA is an example of a bottom-up approach, generating a comprehensive assessment of faults at the component level without considering their system level impacts. These failures are assessed for both severity and likelihood, with the combined score then used to drive any mitigative actions.

Conversely, FTA is a top-down approach that considers the system-level impact of combined underlying faults, though it is not effective at fault identification. Fault trees are fed data on the underlying failure probabilities – these data can be derived from empirical or analytical means (i.e., other tools in the reliability toolkit), and are combined using AND/OR gates to calculate the overall probability of success (aka reliability) for the system.

Effective fault tree analysis requires high-quality data to quantify failure probabilities. Building these datasets, either through modelling or empirical testing, can be resource and time-intensive. Therefore, it is not practical to include all foreseeable events in FTA – FMECA can be used to selectively include / exclude faults based on a predetermined and justified risk threshold. Faults requiring consideration in the fault tree can be supported as required, and faults that are excluded can be easily justified via the direct link to the FMECA.

Design and manufacturing control tools

Control tools are utilised throughout the medical device development process to ensure end-product reliability. During the design stages, formal design reviews are critical, stage-gating milestones that involve systematic evaluation of the design to identify and mitigate potential risks, and challenge how design (and manufacturing) intent will satisfy reliability targets. In addition to the core project teams, design reviews often feature other experienced technical staff that are less familiar with the device specifics, and are therefore able to offer more independent review.

One aspect of a design that is considered in a design review is its suitability for Design for Manufacture, Assembly, and Inspection (DFMAI). Optimisation for DFMAI not only facilitates the production processes but is also pivotal for achieving and demonstrating medical device reliability.

Manufacturing reliability in medical devices starts at the component level, where each part must meet stringent quality standards outlined to ensure overall device performance and safety. This involves rigorous supplier qualification processes and incoming inspection protocols to verify that components conform to required specifications.

DFMAI is also useful for streamlining the processes involved in demonstrating specification conformity. Designing components and sub-assemblies with the inspection and measurement of critical features in mind (part of DfI) reduces the complexity of generating data and increases its quality and utility. Ensuring critical features are suitable for optical inspection makes high-sample-size metrology more feasible and practical, leading to richer statistical analysis that can be used to validate tooling, and as part of demonstrating medical device reliability. This same ease-of-inspection principle also applies to the device assembly process; 100% verification of critical assembly steps (e.g., optically, via displacement or force measurement) is an effective, powerful tool for ensuring reliability in high-volume manufacturing.

Statistical process control (SPC), at both the component, assembly and batch-release testing levels, is essential for efficient high-volume, high-speed manufacture. For example, by using control charts and other SPC tools, deviations from the norm can be detected early. This allows for timely corrective actions before the variability affects the final product, ensuring that only high-quality devices reach the market. Through implementing SPC, a deep understanding of the processes involved in manufacturing are gathered, helping to reduce waste and improve overall efficiency. With data-driven decision making, this can effectively contribute to the reliability of the final product.

Life cycle management is also essential in maintaining medical device reliability and safety throughout their life (be it single-use, reusable, or a combination of both). This involves continuous monitoring and evaluation of the device’s performance post-market, with mechanisms in place for tracking and analysing field data, user feedback, and adverse event reports. Corrective and preventive actions (CAPA) are implemented as necessary to address any identified issues, ensuring ongoing compliance with regulatory requirements and maintaining the device’s reliability over time. Additionally, regular updates to manufacturing processes and design specifications based on emerging technologies and evolving standards help in enhancing device performance and extending its useful life.

The importance of design for reliability in medical device development

In conclusion, designing for reliability is a critical aspect of medical device development that ensures devices perform as expected when needed most. By systematically applying engineering tools and analysis throughout the development process, from initial concept to final production, developers can create high-reliability products that meet stringent safety and effectiveness standards. The reliability toolkit, encompassing both theoretical and empirical methods, plays a vital role in achieving this goal, ultimately leading to the reliable design of medical devices that users can trust at the point of care.