A Whitepaper by Nancy J Stark, PhD
©2015 Clinical Device Group Inc, all rights reserved.
Not everyone agrees that there should be a quality management system for clinical research; that is, the organizational structure, procedures, processes and resources needed to implement quality control, quality assurance, and quality improvement for clinical research. They argue that there should be one quality management system for a company, and quality from all the various functions should flow naturally from the one system.
I disagree. I see a corporate quality management system as having many sub-systems, one for each critical function. Clinical research is one of those critical functions—correctly viewed as a type of manufacturing validation process—but requiring its own unique set of procedures, processes, and resources; operating under a unique mission statement; and set apart from the mainstream of the company because of externally imposed ethical principles. [1. FDA] [2. ISO] [3. OHRP]
Clinical investigations are the study of medical devices in human subjects, where the devices are not cleared or approved for commercialization by regulatory authorities. Hence, the devices are called 'investigational'.
Clinical research is a broader term, used to encompass clinical investigations, registry studies, IVD studies, feasibility studies on new technologies, or any application of a device or technology that can treat or test a human subject or specimen.
A study is a detailed examination and analysis of a subject. [4. dictionary.com] It is not a regulatory word, which makes it useful because it doesn't imply anything beyond its literal meaning. I use the word study to refer to a clinical project with a beginning, middle, and end; regardless of whether the project is a clinical investigation or other clinical research.
Crafting a Quality System for Clinical Research
I view the clinical research function as a business within itself. Your "customers" are top management, the product development team, marketing, regulatory, FDA, Notified Bodies, and foreign regulators. Your "products" are data and reports about data. A quality management system provides the written guidance, how-to instructions, forms, tools, and references to implement clinical research from project initiation to completion and to produce reports that are accurate and give management the information they need to make decisions.
You need to know the tasks required to complete a clinical research project, a standard of quality for each task, and a way to measure the quality of each task. The quality management system will lay out the tasks, provide templates and tools for each task, and provide methods for measuring quality as you progress through to project completion.
But on a broader scale, you need to understand the contribution the clinical department is expected to make to the company, what projects are under the departments purview, and how the clinical department interacts with other departments within the firm. In other words, what are the clinical department's vision and mission?
I find it useful to divide a clinical research quality manual (the manual is a binder which houses the quality system) into sections which mimic the progress of a clinical project:
- The Quality System
- Prestudy Tasks
- Study Tasks
- Study Close-out Tasks
- Database Design
- Data Entry & Analysis
- Other Tasks
Section One: The Quality System
The Quality System section describes the management system itself. Here, policy decisions about the scope of the department's commitment, responsibility, and authority are set forth. It begins with a series of statements. First is a statement of commitment issued by the chief executive officer. The statement is a promise that the company stands firmly on the side of quality, and is made by the CEO to the employees.
The second statement sets forth the vision, mission, and quality for the department. A vision is a dream or imagining of the department's function—for example, the function might be to assure that safety, efficacy, and performance claims are supported by adequate human data whenever non-human data are insufficient.
A mission is a purpose, a reason for existence—for example, the mission might be to assure all human data are collected in compliance with good clinical practice.
Quality is when a thing does what it's supposed to do—for example, quality data and reports reports about data, should accurately and clearly present the data and should meet the reader's need for information.
An operations statement describing when clinical investigations are required and how they fit into the product development cycle is very useful. I use a cross-sectional flowchart to describe the clinical department's role in the concept, prototype, prepilot, pilot, and production phases of product development.
Importantly, section one includes a procedure for the style of a written procedure, and a procedure for the rules of communication with team members located in different geographic areas.
Section Two: Prestudy Tasks
Eighty percent of your project budget will be spent on prestudy tasks, eighty percent of the project budget will be spent before the first device is applied to the first subject. It's really important that upper management understand this. A prestudy process, executed with care, typically requires six to eight months. It hurts to see time and money being spent and no data coming in, but over and over again I have seen studies fail because of poor prestudy work or too aggressive a timeline.
There are two ways to set up operating procedures: wide and shallow or deep and narrow. A wide and shallow procedure might group all the prestudy tasks into one procedure; with numerous plans, reports, and work products resulting from it.
A deep and narrow procedure is focused on one task and results in one work product. For example, there are procedures for writing a protocol, designing consent forms, designing case report forms, addressing adverse events, and every other pre-study task that results in a work product. We've identified 26 distinct tasks to be executed in the prestudy phase.
The value of a deep and narrow method is that each procedure focuses on a single activity, and the list of procedures—taken together—provide a schedule of prestudy activities.
Section Three: Study Tasks
Everyone is excited when study tasks begin; finally you have the first device on the first subject and the prospect of feedback about its performance. The study execution process moves away from the company and takes place primarily at the investigator's sites. Whether your clinical research associates do the monitoring or if you hire separate employees to do the monitoring, someone will have an active travel load.
Don't skimp on the monitoring process. After spending all that time and money setting up a study, make sure a qualified individual travels regularly to the site to assure the device is used per protocol and that problems are spotted early on and fixed. This is the time when data are created; the monitoring process is critical to assuring the data will meet your ultimate goals.
The most important study task procedures are monitoring trips and reports, adverse event reporting, shipping logs, and data monitoring committee meetings. We've identified nine distinct tasks for the study execution process.
Section Four: Close-out Tasks
It's tempting to skip study close-out tasks. Everyone is so tired, the focus has turned to analyzing data, and the sites can take care of themselves, right? Almost. Some things really do need oversight: make sure you've recovered all the remaining devices, wrap up financial disclosure reports, follow-up on outstanding adverse effects, and complete the study documentation checklist. We've identified four distinct close-out tasks.
Section Five: Database Design
The database is a receptacle for receiving and manipulating data. In very small studies you may be able to use Excel, but for most studies you will use a relational database application such as Microsoft SQL Server. Plan to spend $30-60,000 and six months on database design, even more if you outsource it. Unfortunately, no application comes ready-to-wear, you have to build the database de novo to suit your unique study.
The cost of database design depends on how many sites, how much data, and the duration of data storage: 1) if you want the data to be entered by the investigative site as it is collected, each investigative site needs its own logon portal and each portal costs money, 2) the quantity of data that will be stored on the server is the second cost source. You determine the quantity of data by counting the number of fields in each case report form and multiplying by the number of subjects. The more gigabytes of data the higher the cost. 3) the third cost factor is the duration of the study. The longer server space must house your data the more it will cost.
Procedures for designing a database include planning the tables and records and how you'll relate them, what standard reports you want, creating a change log, assembling a data dictionary, and making a user's manual for the data enterer. We have identified two procedures for database design.
Section Six: Data Entry & Analysis
Data may be entered into the database either at the investigative site, at an outsourced data management center, or in-house. The obvious value of having sites enter the data is faster feedback. Don't expect to save money by having investigative sites enter the data. They know that work has been off-loaded from your firm to theirs, and they expect to be paid market-value for their labor.
Entering the data in-house has some special challenges. It means that you take full responsibility for data integrity and Part 11 compliance. Think simple if you are a small firm and want to go this route. Buy a stand-alone computer, keep it off the network, put it in a locked room, and give keys only to the people who have authority to enter data, correct data, and run reports. Management does not get a key.
The procedures in this section discuss who/how/where data will be entered, how data integrity will be verified, how to manage data queries, 'locking' the database, statistically analyzing the data and writing final reports. We include six separate procedures.
Section Seven: Other Tasks
This section of the quality manual is a place for housing miscellaneous procedures. We include procedures for using healthy (employee) volunteers, maintaining your own IRB, auditing sites, auditing yourself, and reviewing label copy.
Reviewing label copy closes the loop on a clinical investigation. If the purpose of the investigation is to obtain data to support safety, efficacy, and performance; then assuring there is data on file for each claim in the labeling brings obvious closure to the project.
Section 8: Bibliography
The manual itself contains a list of significant references. These references provide the history, background, and necessity for the content of a clinical research quality management system.
I also recommend that even small companies set up a library of training and reading materials. Such books and references often provide useful examples of how to approach problems and make unusual protocol designs.
Section 9: Glossary
Every company uses clinical research terms in a manner unique to them. That's fine. The definitions provided in US regulations or international standards may certainly be refined to meet your corporate needs. A glossary of how your firm uses these terms is extremely important, however, because you want everyone in the company assigning the same meaning to the words.
 21 CFR Part 812 Investigational Device Exemptions.
 ISO 14155 Clinical investigations of medical devices in human subjects - good clinical practice (2011).
 OHRP - Office of Human Research Protections, www.hhs.gov/ohrp.