 |
|||
|
GfK Healthcare | October 2010 | |
|
|||
Designing Better Medical Devices  By Doug Willson, Senior Vice President, David Hamming, Associate Vice President, and Rob Wynn, Associate Vice PresidentMany market observers have lamented the prospects of drug manufacturers for the next few years. Limited pipelines, greater competition, genericization and increased regulatory scrutiny have combined to dampen expectations for growth in the short and medium term. The corresponding outlook for medical devices appears much rosier. Diagnostics, imaging, cardiovascular, surgical, orthopedic and other devices currently represent approximately half of the global pharma marketplace from a revenue perspective and are a growing share of the total. Devices have shorter development timelines than drugs and often less risk is associated with obtaining regulatory approval. The increased focus on personalized medicine has also generated growth in demand for diagnostics (e.g., the biomarker segment) and imaging. As with drugs, developing markets represent an often untapped opportunity with enormous potential. Overall, the medical device space appears poised for strong growth when traditional pharma markets for drugs appear to be struggling. Against this backdrop it is not surprising that the medical device sector has also heated up from a marketing research perspective. Product development research in the medical device space is similar to traditional product development research for drugs in some ways, but there are also some important differences. Market participants are of course the same – physicians, nurses, patients and payers. And the research methodologies also appear, at least superficially, to be similar. Choice modeling is the preferred methodology for early stage opportunity assessment with drugs as well as product design work for devices. However, when looking at opportunities for drugs in development, the big unknown is clinical uncertainty – how the drug will perform from efficacy, safety and tolerability perspectives in clinical trials for the relevant patient populations of interest. We often use choice models to investigate the sensitivity of demand estimates to varying clinical trial outcomes. But these attributes are not controlled by the manufacturer, and so this uncertainty complicates forecasting and makes early stage opportunity assessment research for drugs very different than traditional product design research. In contrast, developing devices is much closer to the traditional paradigm where a manufacturer can create new attributes and product features as long as technology will allow it. This difference often influences the research design and analyses that are conducted in medical device design research. In this paper we provide an overview of product design and opportunity assessment research for medical devices, including an outline of research issues and solutions that are specifically relevant for devices. Where relevant, we contrast the design issues in device markets with traditional opportunity assessment research for drugs, and we provide several real world examples to illustrate. The Nature of Medical Device Design The complexity of medical devices runs the gamut from relatively simple commodities like IV bags to extremely complicated cardiovascular devices such as ICDs and CRTs, which involve sophisticated software and nanotechnology. Regardless of category, almost all device manufacturers have product development teams staffed by engineers, software developers, etc., working on the next generation of devices. When doing product development research these teams are often the source of potential new product features and attributes. In many cases, the engineers on the product development teams develop large lists of detailed new product features and attributes. One goal of product development market research is to help prioritize these features, to focus future development efforts on features that resonate with customers in the marketplace. However, the ability to develop quantitative measures of attribute and feature importance in a survey is often compromised if the initial list is too long. Many drug market researchers might remark that their research also often employs long lists of attributes – on the order of 10 or 20. In device work, product development teams regularly develop attribute lists that are two, three or even four times larger. As a result, a preliminary round of qualitative research is often invaluable for winnowing down the list of new product features and attributes to a more manageable size. Preliminary Qualitative Research Depending on the category, qualitative research may investigate possible new product characteristics with customers including physicians, nurses, patients, purchasing managers, hospital administrators, etc. For each group of respondents, structured sorting exercises are often useful for identifying the most important new product features. The qualitative research is also useful for ensuring that attributes and associated levels are clearly delineated and appropriately understandable for each customer group in its own words. For device research, oftentimes new product features are truly “out there” – almost speculative or conceptual in nature. Properly conveying the nature of the concept, and what it means for the patient or physician, is crucial for measuring the importance of the feature. Quantitative Survey Research Depending on the category, quantitative survey research might be undertaken with physicians, nurses, patients, and/or hospital purchasing managers or administrators. For example, research investigating a new insulin pump would focus on physicians and perhaps nurses/diabetes educators and patients. Researchers investigating a new device for wound care to be used in a hospital might talk with surgeons who ultimately choose to use the device, nurses who influence the choice and hospital purchasing managers who decide whether the device will be stocked in the hospital in the first place. If we were investigating demand for new imaging equipment (a large-scale capital investment), the interviews might be conducted with hospital administrators responsible for purchasing in departments where the equipment would be used. In each case, the survey interview would typically investigate baseline current market behaviors (use, purchasing, etc.), perceptions of attribute importance and current brand performance on those attributes, and current unmet needs. Prior to the choice model portion of the interview, respondents would be exposed to a base or “most likely” new product profile to gauge overall interest and perceived performance of the new device on relevant product attributes. It is important to note that accurately depicting the characteristics of new devices – new features and functionality – often requires the use of graphics or video in an online setting, or even developing full-blown prototypes to be used with face-to-face personal interviewing. In the choice model portion of the interview, respondents would be shown a sequence of future market scenarios where one or more new products are assumed to be available. In each scenario the new products shown to respondents are different – the attributes and levels representing new product features vary according to an experimental design. For each scenario, respondents would be asked to reply to a question that properly reflects their decision-making context in the market:
When different respondent groups are surveyed about their interest in the same product, it is sometimes possible to develop an integrated model that properly combines choice model results from each group. So for example, patient and physicians may appreciate different aspects of the same device; since market outcomes are influenced by both, it is important to include both groups in product design work. Nowhere is the difference between product development research for drugs versus devices more apparent than in the specification of the choice tasks. With large numbers of attributes the typical “full profile” design where all new product attributes are displayed and vary across scenarios is often too cluttered for respondents to evaluate properly. Partial profile designs that focus attention on smaller groups of attributes are often used in device research for this reason. Device work also often includes “configurator” or “build your own device” tasks and menu choice designs; while these tools are more common in consumer and B-to-B product design research, they are rarely seen in product development research for drugs. Where the Real Action Is – Simulation and Optimization Results from the choice model are summarized in a user-friendly Excel simulator that estimates the impact of different configurations of product characteristics and perhaps price on demand for the new device. A screen shot of an example simulator for a new gastric band is presented below. The simulator includes attributes reflecting improved clinical performance (amount of weight loss, reduction in complications), ease of use (number and time associated with each adjustment visit), and price. The output shows new product shares of procedures for any configuration of these attributes. Importantly, the simulator can be built to incorporate information from the survey and secondary data when available. In some device markets, manufacturers will have information on their own sales and market shares and this information can be used to calibrate current market share estimates in the simulator. This step improves the accuracy of the model, promotes internal consistency with other company data sources, and ultimately establishes face validity with the product development teams and other internal research clients.  When price sensitivity is investigated in the research it is also possible to develop estimates of demand curves which display quantity demanded for any price in the tested range. A related set of output is total revenue (price x quantity) or total profit (Price –COGS x quantity) graphs. We typically display these as indexes with (100 as the max). An example is included below; ideally the maximum revenue or profit should occur somewhere in the interior of the tested range.  The simulator provides answers to the following types of questions:
In summary, while there are many similarities between product development research for drugs and devices, there are also some crucial differences. Well-designed product design research for devices addresses and capitalizes on the unique characteristics of these markets. Want to learn more on this topic? Please contact:    |