Guest Column | July 19, 2017

Best Practices For Measuring System Quality For Drug-Device Combination Products

By Bikash Chatterjee, chief operating and science officer, Pharmatech Associates

bikash chatterjee headshot

Combination product development is a complex matter governed by different areas of regulatory oversight. In my previous article, we discussed which current good manufacturing practice (cGMP) requirements apply when drugs, devices, and biological products are combined, and the typical pitfalls to avoid when doing so.

In this article, we will examine some common difficulties that drug companies have in measuring overall system quality, and discuss the complexities associated with investigating device returns at the end of a clinical trial.

Regulations and Guidance

To streamline the way forward and promote public health, the FDA’s Title 21 CFR 4 rule released in 2013 clarified what elements of all applicable regulations must be included for drug-device single entity and co-packaged combination products. The guidance states that companies pursuing a combination product need only adopt specific sections of the corresponding CFR. In other words, a drug company that relies on CFR 210 and 211 has the option of fully adopting both CFR 210/211 and CFR 820 (device) regulations, or only adopting specific sections of CFR 820, and vice-versa for device companies.

A comparison of some of the key system activities that are unique to combination products is shown in Figure 1.

Figure 1: Applicable CFR regulations for combination products as defined by 21 CFR 4

Let’s explore some key product development activities that take on a new level of complexity because of the combined drug/device component.

Tolerance Stack-Up Analysis

Standardized methods and specifications exist for generic pharmaceutical raw materials and finished products, but not for delivery mechanisms. Unlike drugs and drug ingredients, there are no compendial monographs that establish and define the characteristics of a medical device.

In the previous article, we talked about the impact of a tolerance stack-up. Extending the notion that the drug and the device contribute to the overall system performance, it is important to understand what the cumulative sources of error are, in order to determine what represents compliant or noncompliant performance and what constitutes simply system noise.

As the device moves through its development process, there may be intermediate performance or subcomponent performance metrics that are predictive of the final product performance. For a device manufacturer, the device will have tolerances associated with each subcomponent fabrication and assembly process.  These stack-up tolerances have a corresponding performance component. In some cases it may be a pass/fail function.

Today, it is not unusual to divide the device development from the drug development, and many companies outsource either or both. In a recent product launch, a drug manufacturer in the midst of launching a prefilled syringe analgesic product had to stop the production line completely when the plunger did not fit and function properly within the syringe. The result was an 80 percent yield loss from the manufacturing run. In this case, the tolerances for each piece of the system did not consider the assembly capabilities and the overall systems tolerances. A one-sigma shift in either direction can halt production in the absence of a clear understanding of the impact of component variation.

Error Allocation Analysis

Beyond any cumulative error stemming from design and assembly, the same error concerns extend to the system performance measurements. This second area of concern is the performance specification setting. Specifications for performance will be established for both the drug and the device in terms of their contribution to final accuracy and precision. An error allocation analysis looks at the contribution to variation of each component in the total system. In the prefilled syringe example cited earlier, the contribution to error would be the sum of the uncertainty associated with drug manufacturing, analytical method precision, device design, device assembly, drug device assembly, and final testing methodology error.

The capability of a method to discern “good” from “bad” can be determined via a gauge reproducibility and repeatability analysis that measures the contribution to variability from the equipment, the analyst, and the product. If the cumulative variability of the device, drug, and method are greater than the specification range, then it is highly likely that the test is no longer measuring system performance — it is simply measuring system noise.

Integrating both the tolerance stack-up analysis and the error allocation analysis will ensure the assembly and performance measurement are capable as the system moves through the development process. The FDA’s Center for Devices and Radiological Health (CDRH), in particular, does not look favorably on design changes as the product moves through Phase 2 to Phase 3. A Monte Carlo analysis, performed on actual development component data, is a good way to confirm the likely commercial performance based upon the proposed specifications for the device, drug, and system.

Product Release

Depending upon the sophistication of the device, one of the potentially largest challenges in manufacturing a combination product is to establish stand-alone criteria for the device, drug, and system components.  When the device, drug, and reagent have performance characteristics that are endemic to each, then the big question that emerges is how to release each component separately. For less-sophisticated devices, the release test could be dimensional or involve physical measurement, such as ejection force.

For stand-alone device systems, the use of a reference system is a common way to evaluate and release devices. A similar philosophy can be applied to systems that require independent controls. Initially, these elements are evaluated against an independent test bed. However, systems that exhibit tight variability — 2 sigma instead of 3 sigma, for example — can be designated as reference standards and can be used to release devices and controls in the future. The acceptance criteria are a direct byproduct of the error allocation analysis.

Having a clear understanding of how each independent component will be released is critical as the product moves toward registration lots and Phase 3 clinical trials. The ability to establish specifications and capable test methods is key to creating a viable commercial system.

Clinical Device Management and Returns

Anticipating how medical devices will be handled and analyzed at the end of each clinical trial is a challenging area for most drug companies pursuing a combination product. In the case of intelligent devices, the data will have to be extracted and analyzed on an individual device basis. A unique device identifier (UDI) framework is essential to maintain traceability between the clinical and device data from an investigation and analysis perspective.  Safe handling of biohazard material requires a thoughtful system and facilities to get accurate data that can be used to further the program while protecting analysts. The framework for receiving, extracting, and analyzing the data needs to be established before initiating clinical trials, as it may require a controlled environment as well as special handling and documentation processes. In turn, these processes may serve as the test bed for developing the commercial process for handling returned systems as part of the commercial complaint management system.


While the regulatory pathway has become clearer with the issuance of 21 CFR 4, the practical challenges faced when integrating the device and drug quality management systems (QMS) and development requirements for combination devices are tangible and require careful preparation. Many companies find themselves attempting to loop back and generate the necessary documentation to support a combination drug filing, only to realize the cost and time required to do so are prohibitive. By considering the common pitfalls laid out here, it is possible to raise the primary technical and program questions required to lower your program risk and increase the likelihood of a successful regulatory and commercial outcome.

About The Author:

Bikash Chatterjee is president and chief science officer for Pharmatech Associates. He has over 30 years’ experience in the design and development of pharmaceutical, biotech, medical device, and IVD products. His work has guided the successful approval and commercialization of over a dozen new products in the U.S. and Europe.

Mr. Chatterjee is a member of the USP National Advisory Board, and is the past-chairman of the Golden Gate Chapter of the American Society of Quality. He is the author of Applying Lean Six Sigma in the Pharmaceutical Industry (ISBN: 978-0-566-09204-6) and is a keynote speaker at international conferences. Mr. Chatterjee holds a B.A. in biochemistry and a B.S. in chemical engineering from the University of California at San Diego.