An MD&DI October 1997 Column
DESIGNUsing human factors methods in the early stages of device development enables designers to meet the needs of end-users.
Most product engineers would do a double take if they were to read David Vermette's empirical study, Product Development in the 1990s: Communication and Integration.1 Vermette argues that "QFD [quality functional deployment], design reviews, or computer-driven rapid prototyping tools don't have any observable relationship to success." Instead, organizations that encourage communication and have a high degree of departmental integration are usually those that develop products quickly and capture a significant market share. He found that measurement and control have a lot less to do with product development success than do factors like knowledge of customers and changes in the marketplace.
Vermette's analysis rings true in early product development, specifically the discovery phase. This phase is where murky thoughts and brainstorms are turned into requirements, models, and projects. Design engineers gather the assumptions everyone in the company has about potential customers and the market. Company personnel other than design engineers can also share their creativity as well as lessons they have learned. The assumptions can then be corrected or refined. The discovery phase is therefore the best time to use human factors design techniques, which are also sometimes referred to collectively as "usability engineering."
Use of these techniques in the discovery phase encourages all relevant parties to communicate their ideas and allows designers to integrate those ideas efficiently. Companies that use these techniques will decrease their time to market, have fewer failures of new products, and improve their market success.
DISCOVERY-PHASE HUMAN FACTORS METHODS
Human factors is the understanding of human characteristics and limitations that is necessary to design systems such as hardware, software, and organizations. The human factors methods described below should be integrated with the training, documentation, and quality assurance activities performed in the early stages of design. Subsequent hazard analyses can then use the data and results generated in the discovery phase. For instance, comprehensive lists of failures can be created by using failure mode and effects analysis. Of course, human factors methods can and should be blended with other approaches.
Contextual Inquiry. During contextual inquiry, which is similar to task analysis, information is gathered about end-users and their needs in the context of their work (Figure 1). A team made up of personnel from many parts of the company, such as domain experts, engineers, marketers, quality managers, regulatory staff, and project managers, chooses a particular focus. Ideally, upper management should also be involved.
Figure 1. The main phases of contextual inquiry.
Most contextual inquiry sessions focus on the flow of people, things, and information. The team tracks goals, tasks, end-user characteristics, and all aspects of the use environment. Next, it visits end-users in their environments, makes observations, and questions end-users while they work in order to better understand their need for a particular product or system. Such inquiries, however, can be problematic in the medical environment because of its hectic pace and concerns about confidentiality.2 Other clues about end-user needs come from bulletin boards, customized instructions, and other informal work records such as Post-It notes. Further observations center on the way existing devices are used and integrated.
After the inquiry sessions, data reduction and analysis involve collecting and organizing issues and observations and producing flowcharts of end-users, data, and devices. The entire team interprets the data. Flowcharts can be combined from different sites, or major types can be placed side by side for comparison. Teams can use affinity diagrams in which members write single issues and observations on Post-Its, group the notes to combine similar ideas, and then label the groups and subgroups.
These experiences and materials can be used to uncover designs and product-shaping issues that do not come to mind when users respond to simple interviews or surveys. Designers can tie ideas to field data, write preliminary usability requirements, and develop approaches for future contextual inquiry sessions. The data from these sessions can be combined with input from other sources to determine ideas or issues that will shape product decisions.
Competitive Usability Evaluations. Another useful and relatively low-cost technique is a competitive usability evaluation.3 This type of evaluation considers such variables as the amount of time end-users need to operate systems, the amount of time needed to train end-users, the number and type of end-user errors, and the level of satisfaction derived from using systems. Designers can evaluate competing products or devices similar to those in the area of interest from a user's vantage point. Human factors specialists can apply guidelines and perform usability laboratory tests or contextual inquiries on the device or group of devices to determine areas of unmet needs that make the competition vulnerable.4 These activities add to the list of issues and observations obtained from contextual inquiry. For instance, if the designers are considering creating a home drug-dispensing device, they could evaluate automated hospital dispensers, pill organizers sold in pharmacies, and even kitchen spice organizers. If several products already exist, designers can develop a table of usability and functional qualities for each one. The designers' intent should be to gather feedback in the context of real-life situations from a user's viewpoint.
Low-Fidelity Prototyping. Designers can augment both of the above-mentioned techniques by creating inexpensive prototypes and evaluating their use in as close to a real situation as possible. Using paper, animations, and foam, they can create tools to probe the needs and processes of end-users. If a device has a heavy software component, or if analogous products are rare or rarely used, this technique works well, although it may be difficult to apply because of the time involved. It also allows team members to shift into the mode of tweaking design and away from breakthrough thinking.
Uses of Findings. Session and evaluation findings on user needs and usage do more than just generate candidate product ideas. They make it easier to create the test plan to develop a product idea. For instance, designers might determine where and how to test higher-fidelity prototypes. The results of continuing development activities are also easier to interpret. When a company is searching for a system from another company to augment its product, it can use such findings as selection criteria.
OTHER SOURCES OF IDEAS
Designers can also consult other sources at any time during the discovery phase, even while performing the methods listed above. Internal sources and external forces can provide ideas, as can market studies (Figure 2).
Figure 2. The discovery phase of medical device design.
Internal sources include company personnel such as subject area experts, who can monitor scientific and technological advances; domain experts, who can contribute a basic understanding of the problems facing health care in general; anyone in the company who interacts with buyers and end-users; and engineering, safety, and manufacturing personnel. External forces include regulatory agencies such as FDA, which provides constraints like the quality system regulation; legislators and the judiciary, which create and interpret laws governing product liability; human factors professionals, who often act as expert witnesses in product liability suits; entities that pay for health care, such as the Health Care Financing Administration and insurance companies; group-purchasing organizations and other buyers; scientific societies, such as the American Medical Association, that sponsor and publish technology assessment studies; and health-care delivery organizations, which have formed cooperatives and laboratories to review and develop guidelines for purchasing devices as well as for placing restrictions on their use.
Market research is another way to determine the information needs and wants of end-users. But market research techniques, like focus groups, can come up short. Many device company marketing groups know this and use data sources like success stories about similar products or information about troublesome products from sources such as medical device reports. Using this kind of information can complement human factors methods in early device design phases.
Even though much of the discovery-phase work on medical products is proprietary, a few designers have written about their successes. The following case studies from a medical ultrasound imaging company, a pharmaceutical company, and a consumer products company reveal that contextual inquiry can be used successfully in the discovery phase of design. (For more examples, see Field Methods Casebook for Software Design.5)
Diane Brown of ATL Ultrasound (Bothell, WA), a medical ultrasound imaging company, has found contextual inquiry helpful in improving current products and developing new ones.6 One of her goals was to modify ATL's current ultrasound devices to accommodate work practices from around the world. She discovered that none of the currently manufactured ultrasound systems was easy to use, and that the majority of an end-user's job was not physical but mental, and therefore not directly observable. She also learned that summarizing contextual inquiry data could best be done in a hierarchical outline. She used a graphical tool that resembled a musical score to show the sonographer's work patterns over time. ATL presently uses many of the discovery-phase techniques for product development. According to Brown, "Not only does the data serve as a solid foundation for design, but also the data provide a crucial understanding of when and how to make the necessary design trade-offs."
Glaxo-Wellcome (Research Triangle Park, NC) uses contextual inquiry and low-fidelity prototyping in the early design of software created for internal use. According to one of Glaxo-Wellcome's developers, Kurt Morehouse, "Usability needs to be built in, not added on."7 Morehouse said this because their group had abandoned usability testing late in the process. He also found that the use of interviews or focus groups alone isn't enough. "If I ask a mechanic over dinner how he changes brakes, he may give me four steps. But if I were to go and watch him, I might see that there are 30 to 40 steps. And some of the really key tricks he has developed to make the job easy, he doesn't talk about at all."
Discovery-phase human factors techniques can also help companies develop breakthrough products. When designing its Snakelight, Black & Decker found through the use of focus groups, internal discussions, and market trends analysis that users wanted a smaller, lighter, longer-lasting, and waterproof flashlight.8 But when the designers observed people using flashlights in different scenarios, conventional models did not meet observed needs. Instead, designers found that more than half of the flashlight users wanted hands-free operation, so the idea for the Snakelightrather large and relatively expensive, but flexible and hands free, contrary to what the focus groups had indicatedwas born.
Hugh Beyer, of InContext Enterprises, Inc. (Harvard, MA), has used contextual inquiry at several companies, including clinical laboratories. He remarks that the high-level notions of what the market needs, such as more efficient processes, are usually correct. However, the power of early design methods is to get the next level of issues and implementation of ideas delineated correctly. In these cases, clients' assumptions can be wrong up to 80% of the time.
Discovery-phase human factors techniques are essential for the successful development of any medical device. The process can be as simple as finding a small project with willing personnel, reading some of the references in this article, and trying it. On the other hand, it could also involve significantly reorganizing a product development department and hiring and training new types of personnel. Thomson Consumer Electronics (Indianapolis), for instance, developed a new product division made up of cross-trained personnel with human factors, art, anthropology, business, and technology backgrounds. The Center for Applied Medical Informatics at Michigan State University Kalamazoo Center for Medical Studies, along with other consultants and academic groups, have also helped various types of companies.
Designers should also monitor research into approaches for using data from human factors activities and integrate such research findings into their organizations' own discovery phases.9 Companies interested in a better design process, faster rates of new product success, and meeting customer needs in novel ways should start using these techniques today.
1. Vermette D, Product Development in the 1990s: Communication and Integration, Waltham, MA, Management Roundtable, 1997.
2. Gosbee JW, and Ritchie EM, "Human-Computer Interaction and Medical Software Development," interactions, 4(4):1318, 1997.
3. Nielsen J, Usability Engineering, Cambridge, MA, AP Professional, 1993.
4. "Human Factors Engineering Guidelines and Preferred Practices for the Design of Medical Devices," ANSI/AAMI HE-48, developed by the Association for the Advancement of Medical Instrumentation (AAMI) and approved by the American National Standards Institute (ANSI), Arlington, VA, AAMI, 1993.
5. Wixon D, and Ramey J, Field Methods Casebook for Software Design, New York, Wiley, 1996.
6. Brown D, "The Challenges of User-Based Design in a Medical Equipment Market," in Wixon D, and Ramey J (eds), Field Methods Casebook for Software Design, New York, Wiley, pp 157176, 1996.
7. Anthes GH, "Clear Vision," Computerworld, May 19, p 81, 1997.
8. Friedland J, "Shoppers Talk, Black & Decker Listens, Profits," Wall Street Journal, January 9, p 1, 1995.
9. Neilsen J, "As They May Work," interactions, (10)2:1924, 1994.
John Gosbee, MD, is director of the Center for Applied Medical Informatics at Michigan State University Kalamazoo Center for Medical Studies. He is also an assistant professor at the Michigan State University College of Human Medicine and an adjunct assistant professor in Western Michigan University's industrial engineering department.