Design Thinking in Health Care

1Health care systems require continuous innovation to meet the needs of patients and providers. However, these stakeholders are not always considered when new interventions or system processes are designed, which results in products that remain unused because they do not account for human context, need, or fallibility. This approach also likely contributes to the decades-long gaps between intervention development and implementation. Design Thinking offers a way to close that gap by helping investigators incorporate user needs and feedback throughout the development process.

2Design Thinking is an approach that prioritizes developing empathy for users, working in collaborative multidisciplinary teams, and using “action-oriented rapid prototyping” of solutions. It is an iterative process, with innovation emerging only after cycling through several rounds of ideation, prototyping, and testing, which distinguishes it from the traditional linear and often top-down approach to health intervention design (Figure 1). Design Thinking has been used across sectors to solve complex problems, including the redesign of an elementary school curriculum to enhance student engagement, and in domains such as aviation that, like health care, have high levels of risk. Design Thinking is similar to both “user-centered design” and “human-centered design,” which are both referred to as “Design Thinking” in this article.

3There is much enthusiasm for the use of Design Thinking in health care, from intervention development to large-scale organizational and systems changes. However, health care settings present different challenges than do other domains, so it is important to consider these challenges in assessing whether Design Thinking provides added benefit over traditional approaches. With this in mind, the purpose of this review was to answer the questions, “How has Design Thinking been used to design interventions in health care settings, and have these interventions been effective?”

4Studies published through March 31, 2017, were identified through searches of online databases (PubMed, Medline, Web of Science, CINAHL, and PyscINFO) using the following search terms: “health,” “health care,” or “healthcare”; and “Design Thinking,” “design science,” “design approach,” “user centered design,” or “human centered design.” Additional articles were included if they were referenced as original research articles in existing articles. To provide an overview of the range of uses of the Design Thinking approach, we did not limit our review to specific populations or conditions and included articles addressing multiple health promotion and disease prevention topics. Given the search terms, the likely target populations for inclusion were patients and health care professionals and the settings in which they work or seek care.

5We reviewed selected articles using PRISMA guidelines and entered citations into a reference manager, which removed duplicates. To be eligible for inclusion, studies had to be written in English, be published in a peer-reviewed journal, provide outcome data on a health-related intervention, and use Design Thinking in intervention development, implementation or both.

6There are multiple definitions of Design Thinking, so we focused on the key principles common to most descriptions of the approach; thus, the list of Design Thinking approaches is not exhaustive. Studies were considered to use Design Thinking if they 1) described user/needs assessment, 2) involved iterative prototyping/testing of the intervention with user feedback, and 3) tested the intervention with target users. The user/needs assessment could include contextual observation of users in the setting in which they would interact with the innovation, interviews, narrative accounts, and documentation from users, gathering extreme user/outlier stories or a review of existing literature and work. Prototyping included activities such as creating a series of low-fidelity and high-fidelity prototypes of the potential innovation and refining it multiple times through iterative cycles of feedback from end users, stakeholders, and experts. Testing the intervention with target users included implementing and testing the innovation while continuing to refine it on the basis of user feedback and data. Design Thinking is also similar to other techniques, such as plan-do-study-act cycles and formative evaluations. We considered the emphasis on empathizing with the user and the use of low-fidelity prototyping to be key distinguishing features of Design Thinking, so only articles that explicitly indicate their use of these approaches were included. Initial screening was completed for all selected abstracts, and a second round of screening was completed on eligible full-text articles.

7Data were collected on target users, health conditions, objective of the intervention, details on the Design Thinking process, study design and sample, and reported health outcomes. If information was not reported in the article, we contacted the study authors. Studies were also evaluated to determine whether the intervention improved all targeted outcomes (successful), at least one targeted outcome (mixed success), or no targeted outcomes (not successful). Data quality was assessed using the National Institutes of Health’s (NIH’s) National Heart, Lung, and Blood Institute Study Quality Assessment Tools.

8Figure 2 presents study flow based on the PRISMA study guidelines. After the initial search, the authors separately screened all abstracts based on the eligibility criteria. One author reviewed all full-text articles (N = 297), and a second author reviewed roughly 15% as a reliability check. Agreement on inclusion/exclusion was more than 80%. Any abstracts or articles for which there was disagreement or uncertainty were reviewed by 2 authors and discussed until consensus was reached. A total of 26 papers representing 24 interventions were included in the analysis. Two authors reviewed all included studies.

9A summary of all included studies is provided in (Table 1). Eleven studies were successful, 12 reported mixed success, and one reported no success (Table 1 and Table 2). Sample sizes of included studies ranged from 12 to 291, but most studies were small; 14 studies had fewer than 40 participants. Eleven (45.8%) used a control group, and 4 (16.6%) compared a design-thinking intervention to an intervention designed using traditional methods. Two of the studies included were “good” quality, 13 were “fair” quality, and 9 were “poor” quality. All studies used Design Thinking methodology in intervention development, and 3 also used it for implementation

10The 24 included interventions targeted a range of conditions, including 19 related to physical health (17 unique conditions), 2 related to mental health, and 3 related to systems processes. Approximately two-thirds of the interventions were mobile telephone–based or tablet-based.

11Patient-facing interventions (n = 11). Five interventions were successful: 4 with a pre/post design and 1 pilot randomized control trial (RCT). Five reported mixed success, including one pre/post design, one pilot RCT, one RCT, one cohort study, and one unblinded, randomized crossover design. One, a pilot RCT, was not successful.

12Provider-facing interventions (n = 9). Six were successful, including 3 studies using a pre/post design, one field experiment, one using a quasi-experimental crossover design, and one cross-sectional study. Three had mixed success, including 2 studies with an experimental crossover design and one with primarily a pre/post design, one portion of which was a randomized crossover design.

13Patient-facing and provider-facing interventions (n = 2). Both reported mixed success and were pre/post designs.

14Caregiver-facing or family-facing interventions (n = 2). Both reported mixed success, one in a pilot RCT and one using a pre/post design.

15Summary of Randomized-Controlled Trials (K = 5). Of the RCTs and pilot RCTs reviewed, one demonstrated success on all outcomes, 3 showed mixed success, and one reported no enduring significant results.

16Four studies directly compared interventions created with Design Thinking to interventions created with traditional methods. In one study with a within-sample experimental crossover design, a Design Thinking–based graphical information display to improve nurses’ ability to detect changes in patients’ physiological states in an intensive care unit (ICU) was compared with a conventional display in commercial, electronic ICU charting systems. The Design Thinking intervention resulted in improved detection of changes in patient states and greater ease of use, usefulness, satisfaction, and support of understanding, but no differences in workload for nurses. Another study using an experimental crossover design compared 2 computer interfaces designed to display drug interaction alerts, one developed using Design Thinking and one using traditional software. Whereas the design of the traditional software was not described, the traditional display included only basic text information. In this study, users (ICU nurses) were more efficient and effective, and reported higher satisfaction with the Design Thinking interface. Another study using a quasi-experimental crossover design used Design Thinking to develop an application to guide clinicians in detecting and scoring the severity of graft versus host disease (GvHD). When compared with paper-based NIH guidelines, users of the application (app) signficantly improved diagnostic and scoring accuracy. A final study compared a Design Thinking–based app that provided nurses with information about antibiotic use with regular information sources (which were not described). In the randomized portion of this study, nurses using the app found information on antibiotic use more quickly; however, the app did not enhance their ability to improve antibiotic-related behaviors. (Only 7 participants were included in the randomized portion of the study.) Whereas the development of the control intervention was not fully described in these papers, based on the limited descriptions given, it is likely that it did not include key elements of Design Thinking such as user feedback and prototyping.

17The 24 interventions summarized in this review provide an overview of the breadth of Design Thinking's applicability in health care and demonstrate that it is feasible and applicable to multiple health care domains. It has been applied across a range of diverse patient populations and conditions, including chronic obstructive pulmonary disease, diabetes, caregiver stress, and posttraumatic stress disorder. It also has been applied to systems process changes, such as nursing handoffs and drug–drug interaction alerts. Results also demonstrate that, although it is often applied to electronic interventions, Design Thinking is feasible for use in other modalities (eg, on paper, in person).

18Initial results of the interventions included in this review are promising; all but one demonstrated positive effects on at least one identified outcome, and half showed positive effects on all measured outcomes. In addition, in the studies that directly compared the Design Thinking intervention with a traditional intervention, the Design Thinking intervention generally demonstrated improved outcomes and higher usability and satisfaction.

19However, none of these studies were RCTs with large sample sizes. Design Thinking interventions have been tested primarily in pre/post designs or pilot RCTs with small samples. Furthermore, most studies included were poor or fair quality, with only 2 being considered good quality. Importantly, the criteria used to assess quality were based on traditional research approaches, and many of the features of poor-quality studies were included by design; some had small sample sizes to generate insights and to test assumptions rapidly, and some were pilot studies. This feature of Design Thinking also may account for the limited use of large RCTs; however, this poses a challenge when evaluating the effectiveness of the approach. More work in this area using more rigorous methods and larger samples is critical to fully understanding the benefits of Design Thinking. Although many studies that used Design Thinking were excluded from our review because they did not include sufficient outcome data (n = 131), full-scale trials of many of these interventions are under way, results of which will provide more evidence about the effectiveness of this approach in health care. In addition, no studies measured Design Thinking directly to explain how or what components of Design Thinking lead to improved usability and effectiveness, limiting the field’s ability to disseminate the most effective components and refine the Design Thinking approach for health care.

20Design Thinking methods varied among the studies reviewed. For example, only 6 studies conducted contextual observations of users during the needs assessment phase, no studies reported a brainstorming stage, 10 studies did not use low-fidelity prototypes, and some reported a small number of iterations (eg, one mixed-success trial had 4 intervention iterations, but only 2 iterations were evaluated with target users). Using more thorough and structured Design Thinking methodology may have resulted in more consistent and enhanced outcomes. At the same time, Design Thinking is meant to be flexibly applied. Future work should balance that flexibility with the potential benefits of a more systematic approach.

21Our results suggest that one area where Design Thinking could be especially useful is in designing interventions for underserved populations whose needs may be overlooked by other approaches. For example, the study of a mobile health tool for detecting and managing cardiovascular disease in rural India required significant feedback from the end users — minimally trained health workers — to ensure that the intervention was suited to their level of technological familiarity as well as the inconsistent technical infrastructure (eg, creating a one-touch navigation system). Using Design Thinking allowed the multidisciplinary team to question assumptions and biases and develop an intervention that was successful, acceptable, and feasible to the actual users, an outcome that may not have been possible using traditional methods. Another study evaluated the impact of an education tool to enhance long-acting contraceptive use in a clinic serving mostly African American patients who were included early in the usability testing process to ensure the tool met their needs. Several changes were made as a result, such as including more peer testimonials, which likely increased the tool’s impact and relevance. In this way, Design Thinking could also pair well with other approaches that prioritize the inclusion of users in service of reducing health disparities, such as community-based participatory research.

22In their text and through our analysis, the studies included in this review show several challenges to consider when applying Design Thinking to health care. First, there is the possibility of tension between what users want and what providers and researchers believe to be beneficial based on research and expertise. Whereas in industry, where an innovation designer may prioritize customers’ preferences to maximize profits, in health care a balance must be struck between creating interventions that are effective and sufficiently palatable and feasible so that they will be used by providers and patients.

23Second, tension may exist between the needs assessment, a fundamental step of Design Thinking, and existing literature and evidence base for some conditions. That is, given the evidence, intervention developers may not be willing or see it necessary to conduct their own needs assessment using observation or interview strategies or to brainstorm creative solutions. Indeed, 7 of the studies included in this review reported literature reviews, and possibly expert consultation, as their only needs assessment steps, and none reported brainstorming. One way to overcome this tension is to view evidence as a set of design constraints in which needs assessment, brainstorming, ideation, and prototyping should occur.

24A third possible tension relates to balancing the Design Thinking approach of understanding the narrative of outliers with traditional health research methods that prioritize statistics on large samples to produce generalizable results. Conclusions drawn from small user samples should be tested in broader populations to ensure their applicability. Mixed-methods approaches that use both strategies may reduce this tension. For example, a research team that uses a qualitative Design Thinking approach early in the research process (eg, user observations, focus groups, and usability tests with small groups of target users) may be able to generate insights into the key needs of the target population. This approach may also find ways to address these needs, and subsequent quantitative testing of the developed interventions in broader samples will allow the group to evaluate whether their assumptions generalize to the broader population, and the intervention will be more effective as a result.

25Fourth, there is inherent tension between a central philosophy of the prototyping process in Design Thinking — to rapidly move through low-fidelity then high-fidelity iterations to fail early and often to more quickly reach a better design — and the risk of serious negative outcomes due to health care failures (eg, death). Many of the studies did not use low-fidelity prototyping or multiple rapid iterations, perhaps because of this tension. However, although there may be some reluctance to experiment with low-fidelity prototypes in health care where morbidity and mortality are at stake, there are low-stakes approaches to low-fidelity prototyping that may minimize risk and improve the pace of innovation (eg, storyboards to illustrate a new clinic process).

26Considering the role of Design Thinking is important, not only in efficacious intervention development but also in effective implementation into practice. Only 3 of the included interventions addressed implementation, but this limited implementation provides insights. For example, in designing a new process for facilitating nurse handoffs between shifts, Lin and colleagues conducted an extensive 6-month intervention development design process that was user-focused and empathic and had rapid iteration in pilot sites. However, despite this strong preliminary work, the intervention was not readily accepted when implemented in other clinics. As a participant stated:

27After the concepts had been co-developed and field tested with our pilot units... we assumed the units were “bought in” to the idea of the change.... Surprisingly, our approach to the training resulted in criticism and created skepticism [at other clinics].... They attributed this to “not made here” sentiments from those units not involved in the original design.

28To overcome this tension, the team involved additional stakeholders to develop a more user-centered process for the implementation of their Design Thinking innovation, after which they successfully implemented the innovation across 125 nursing units in 14 hospitals over 2 years. This study highlights the importance of understanding the context of the setting and users, both when developing and implementing an intervention using a Design Thinking approach. It should also be noted that this process required significant time and energy from stakeholders. One stakeholder commented, “Don’t get me wrong. What we did was fantastic. But it took a lot out of us”. This study highlights the importance of staying true to the user-centered nature of Design Thinking throughout the process — from development to implementation — to maximize implementation success. It also highlights the challenges in using this approach. Teams using Design Thinking should be prepared for a more intensive process than traditional, less iterative and user-centered methods.

29Given the varied outcomes included in the review and the inconsistent reporting of qualitative outcomes it was difficult to make comparisons across studies. The range of study types and limited number of large scale RCTs testing intervention effects also made it difficult to draw definitive conclusions about effectiveness. At the same time, given that there was only one study with a null result, there was likely publication bias, which may have led to overestimation of the effectiveness of Design Thinking. It is also possible that investigators used methods but did not report them (eg, prototyping). In addition, we did not assess the use of Design Thinking in other health care areas where it may be beneficial, such as the design of physical spaces. Finally, Design Thinking–based health care innovations that were developed and implemented outside research contexts may exist and are thus not reported in the literature.

30Design Thinking is being used in varied health care settings and health conditions, and more studies are forthcoming. This review suggests that Design Thinking may result in more usable, acceptable, and effective interventions compared with traditional expert-driven methods. However, there is inconsistent use of the methodology and significant limitations inherent in the studies, which limits our ability to draw conclusions about this approach. Future studies may benefit from focusing on comparing interventions developed using Design Thinking methods with traditionally developed interventions, including those with RCT designs, and identifying the most useful components of Design Thinking methods.

31Overall, Design Thinking is a promising approach to intervention development, implementation, and dissemination that may increase the acceptability and effectiveness of health care interventions by actively engaging patients and providers in the design process and rapidly iterating innovation prototypes to maximize success.