Eligible participants were over 18 years of age, diagnosed with CD, and US residents. Once eligibility was confirmed, a convenience sampling strategy was used to contact individuals to participate in the study.

Participants were recruited from the online social media platform Facebook [27]. The keyword “Crohn’s disease” was used to search for pages and groups that support individuals with CD. Administrators of both public and private groups were contacted prior to posting information about the study. Posts were directly submitted to pages for approval. Administrators from 9 CD groups and 11 CD pages agreed to the request. We also posted on our personal Facebook profiles. These posts contained a flyer that included information about the study purpose, a link to the recruitment survey on Survey Monkey, and compensation information (Figure 1). The recruitment survey contained questions about basic demographic information (ie, age and gender) and contact information. Compensation was a $20 gift card to a nation-wide retail chain. The recruitment survey was closed after 54 responses were received. Respondents who provided valid information and met eligibility criteria were contacted for an interview. If the respondent did not reply within 24 hours, a follow-up message was sent. Those who did not respond after 2 attempts were not contacted again.

Data collection took place from October 2016 to February 2018 in 2 phases. Data saturation [28,29] was reached after the second phase of interviews. Interviews were semistructured, lasted approximately 45 minutes, and took place over the phone or on a video-chat platform. We used an interview guide consisting of 15 questions and additional probes. Topics included diet and nutrition management, how diet and nutrition management relate to other aspects of CD management, and experience with existing management tools. The interview guide was not grounded in a particular framework, in order to gain direct information from participants [30]. An informed consent script was read to all participants at the beginning of the interview, and oral consent was audio recorded. Interviews were audio recorded with permission from the participant. Audio recordings were stored in a secured University of Virginia Box account and were deleted from the initial recording device once uploaded.

All interview recordings were transcribed. Data were analyzed using QSR NVIVO 11.3, through a conventional content analysis process informed by Hsieh and Shannon [30]. We gained an initial impression of the data and then iteratively drew themes. After analyzing the first 2 transcripts individually, 4 team members determined the preliminary codebook through consensus building [31,32]. The preliminary codebook was reviewed by the senior author (RSV). The next 6 transcripts were divided between the first 3 authors. Then, we came together and decided on the second version of the codebook with input from RSV. The remaining transcripts were analyzed by the first author, and any changes to the codebook were reviewed by RSV. We engaged in simultaneous coding when a participant’s statement was reflective of more than 1 code [31]. This coding framework was used to analyze the remaining data. A final codebook was created to define each theme and subtheme.

This study was approved by the University of Virginia Social and Behavioral Sciences institutional review board.

In total, 54 individuals filled out the survey, and, of those, 3 (6%) did not provide sufficient contact information, 27 (50%) did not respond to the follow-up, 3 (6%) were unable to interview due to medical complications, and 21 (39%) were successfully enrolled. Of the 21 participants, 16 (76%) were female, and the average age was 35 years (Table 1). Participants reported living with an official diagnosis of CD for an average of 12 years and felt that they had CD for 6 years, on average, before their official diagnosis (Figure 2).

Qualitative content analysis revealed 4 themes: physical, emotional, informational, and technology-enabled management. Across participants, diet and nutrition management was an integral part of both illness and everyday life work. However, the extent to which diet and nutrition management could be considered as biform work varied, not only by participant but also over time. Participants usually attempted to manage their diet during remission to prevent flare-ups. For some participants, the use of medication reduced—or, in some cases, eliminated—the need to manage diet and nutrition as illness work. Illness work was also reduced via relying on a consistent diet. During flare-ups, the need for diet and nutrition management as illness work was imperative, as each participant reported maintaining a list of “safe foods” that they rely on. Participants discussed the centrality of stress in CD management, as it tended to exacerbate symptoms, in turn, causing an increased need to manage diet and nutrition as illness work. For many participants, stress stemmed from frustrations with their limited diets and subsequent difficulties developing and maintaining social relationships. As information from professionals was often insufficient, participants relied mainly on social networks to learn about the dietary aspects of CD management. Obtaining and seeking information about diet and nutrition management was often illness work, as participants typically used this information to avoid or ease disease symptoms. Lastly, statements about articulation work were pervasive throughout the interviews. Tasks such as taking medication and eating needed to be coordinated to avoid adverse outcomes. In other cases, tasks were inextricably linked, such as managing stress along with diet and nutrition. For some participants, stress was caused by the diet-restrictive nature of CD, in turn, affecting disease severity and subsequently increasing the need to manage diet and nutrition.

Participants identified the need for complex and integrated functionalities for CHIT to support diet and nutrition management, as there are varying degrees of overlap between illness and everyday life work. Coordination is also required between and within lines of work. The need to manage diet and nutrition as illness work fluctuated over time for participants; however, CHIT rarely responded to oscillations in times of remission and flare-ups, as participants predominantly noted these using general apps designed to track physical activity and calorie expenditures. Moreover, available tools did not support information management as illness work, since participants sought information about diet and nutrition mostly from social networks. For those that used management tools, many reported paper-based food journaling as cumbersome illness work. Additionally, a majority experienced limitations related to usefulness and usability in available tools to meet these needs. Altogether, participants desired features that are more responsive to the realities of living with CD and the interconnected nature of this condition. These desired features included automated diet and nutrition tracking, facilitated social connections, a public restroom finder, analytics, information-sharing with health care professionals, and integration with other apps that are already part of their daily workflow.

Even though we initially proposed the concept of biform work, the results from this study show that this concept is not necessarily fixed with regard to diet and nutrition management for people living with CD. There were instances where this type of management was both illness and everyday life work and also instances where it largely became everyday life work. Even if diet and nutrition management existed in a state of everyday life work, psychological factors often could trigger a flare-up for participants and, consequently, increase the illness work–related characteristics. Though not the focus of this study, there may be an opportunity to explore these psychological factors more deeply, as diet and nutrition management could also be a component of biographical work. Thus, this type of management could exist as multiform work, meaning that there could be overlap between everyday life, illness, and biographical work. The concept of diet and nutrition management as multiform work could be meaningful in other conditions such as obesity and anorexia. In such conditions, the need to manage diet and nutrition can be illness and everyday life work, in addition to biographical work, as relationships with food can contribute to self-identity [33-36]. By examining when and how diet and nutrition management can exist as multiple lines of work, a more holistic perspective can be elucidated, and, as a result, interventions can be developed to be more responsive to the multifaceted realities of diet and nutrition management of many chronic conditions.

Since the extent to which diet and nutrition management existed as biform work varied among the participants and over time, the reported need for articulation work between and among other management tasks varied as well. Articulation work was needed for task coordination, aligning with the original conceptualization of this type of work [6]. The understanding of articulation work in the context of chronic disease management has generally remained the same, focusing on organizing, delegating, scheduling, and regimenting consistent tasks [37-41]. However, for the participants in this study, articulation work was not so linear, as management needs typically were in flux and could change instantaneously. Also, management tasks, in some cases, mutually shaped each other and, by their nature, could not be regimented. For example, stress from daily life events triggered increases in symptoms and subsequent increases in management tasks. This kind of stress can be unpredictable and, for example, could stem from an impromptu invitation to a social gathering. As such, there is a need to expand on the understanding of articulation work, so that interventions can better attend to the interconnected nature of diet and nutrition management and other management tasks.

Interactions between the first 3 themes yielded several insights into how future patient-facing technologies can be developed to support diet and nutrition management as biform work, when needed, and facilitate articulation work. The fourth theme informs potential features and functionalities of these technologies.

The first theme demonstrated the articulation work needed to manage diet and nutrition and medication. The interconnected nature of these factors is supported by the scientific evidence, as both food and medications have been shown to affect the composition of the gut microbiome [42,43] and, thus, inflammation in the intestine [44]. As early research demonstrates the individualized and variable nature of diet and nutrition’s association with CD [42], general dietary guidelines may not be appropriate, and individualized guidance may need to be developed. Currently, CHIT designed specifically for CD does not support this type of articulation work, as the majority of tools do not make connections between medication and food intake [45,46]. To assist with this work, technology that leverages pattern detection through machine learning [47] could be designed to facilitate the identification of individualized diet and nutrition guidance by capturing and coordinating food and medication intake and monitoring symptoms. Although several participants discussed the difficulties of food journaling, others found that smartphone barcode scanning features, such as the one in the MyFitnessPal app [48], made it easier to track food intake. Moreover, the addition of nutrient content information in food databases within an app could help to mitigate malnutrition by tracking both macronutrients and micronutrients, which has shown to be important in CD management [14,49]. This capability is not readily available in the MyFitnessPal app [11] or other apps with barcode scanning functionality. Incorporating these features into technology could address barriers in capturing food intake [19]. Additionally, machine learning at both the individual and aggregate level could be used to identify both safe and irritating foods and automate meal planning by identifying acceptable ingredients and substitutions, reducing the work required and increasing the utility of technology that facilitates food journaling [9,10]. Lastly, customized features that adapt to individual needs over time could support the fluctuations in diet and nutrition management as biform work. For example, if medication has eliminated the need for an individual to manage diet and nutrition as illness work, a user could then hide the food journaling feature and rely on the symptom tracker to reopen this feature if a flare-up is imminent.

The second theme shows that for the participants in this study, psychological factors also had an effect on the extent to which diet and nutrition needed to be managed as biform work. Participants described how diet and nutrition can affect social well-being and mental health, in turn, affecting disease states and requiring more articulation work. This finding supports the evidence that individuals with food restrictions have reduced social and emotional well-being [21], experience increased levels of stress and anxiety [50], and experience stress that, in turn, affects disease severity [20]. Moreover, this study provides further evidence that the processes of diet and nutrition management, not only disease severity and illness perceptions, are a source of stress [51]. Though a few apps designed to support diet and nutrition management for individuals with CD allow for mood tracking [45,46], these features are still nascent and neither include capabilities to track social engagements nor are adaptive to the extent to which these factors exist as biform work. Patient-facing technology could assist in monitoring and reducing stress and promoting social well-being by supporting biform work when needed and elucidating interactions among diet and nutrition, social engagement, and stress. Mobile sensing can identify increased stress levels [52,53] to facilitate activities [54] such as providing meditation guidance or prompting a connection to a health care provider [55]. By monitoring stress levels, these data can be analyzed along with diet and nutrition, medication, and symptom data. Machine learning could be applied to explore the relationships between these factors to further refine a food irritant profile and identify patterns in flare-ups. This profile could be used to identify restaurants and specific menu items that align with a user’s restrictions, which could reduce the psychological burdens participants expressed they experience when attending social activities. Additionally, these data can be synthesized with location and calendar data, which could help to predict stressful events based on past data related to daily activities and stress. Location and calendar data could also be cross-referenced with data available on public restrooms [56], to automatically have a restroom option available to the user, reducing stress in urgent situations. As some apps have started to integrate behavioral insights [46], these features could be expanded by using predictive analytics to identify when flare-ups are about to occur and provide recommendations for foods that are least likely to initiate or exacerbate a flare-up and preemptively suggest users cancel activities. By tracking a wide range of contextual factors beyond food intake, there is an opportunity to explore why participants experienced inconsistencies with fixed diets. Though there are apps that are providing support to track a broad range of variables, these apps aim to provide data to health care providers to improve clinical care, rather than to support individuals with CD in performing biform and articulation work [46].

As discussed in the third theme, participants sought and obtained diet and nutrition information from health care providers, text sources, and social networks. Similar to this study, one study documented that individuals with IBD wanted to know more about dietary management to reduce flare-ups and noted limitations in dietary advice offered by health care providers [57]. Additionally, seeking health information on social media is increasing in prevalence [58,59]. Available apps offer a range of education materials, and some allow connections to health care providers [60-62]. A few apps offer social networking functions [45,46], which increases the articulation work needed to find and integrate information into disease management strategies. To limit articulation work, technology could help promote online relationships among individuals with CD by suggesting connections based on similar food profiles to share recipes and general diet and nutrition information. Furthermore, those who have similar types of daily routines could be connected to share diet and nutrition management techniques to reduce the illness component of biform work. Additionally, features that connect individuals to health care providers could be expanded to facilitate appointment reminders and scheduling, to promote clinical management, as this has been shown to be a key component of long-term management of CD [63,64]. The collection and sharing of the patient-generated health data with providers could also support clinical care of patients with CD [65-68]. Lastly, the aggregation of these data across users, such as those in the IBD Partners research network [69], can be used to collect the amount of data needed to develop personal nutrition guidance through machine learning [70], deliver precision medicine [71-73], and promote citizen-science in health care [74]. However, it is important to note that there are uncertainties surrounding the amount and variety of data required to build and validate predictive models, as studies range from using the n of 1 study design [75] to including hundreds [76] or thousands of participants [77].

Although this study provided several insights about diet and nutrition management for CD, it also had a few limitations. Since we asked participants to describe CD management over time, the data could be subject to recall bias. However, the participants were not asked to recount certain details about events but, rather, asked to reflect on the most salient points of diet and nutrition management and their overall experience with CD. By recruiting on social media, the sample inherently was comfortable using technology, and, thus, this study did not allow for an understanding of the needs of those with limited technology experience. However, use of this sampling strategy could be beneficial, as the participants could identify challenges with the technology that did not necessarily stem from general inexperience or discomfort using technology. Additionally, apps that partially address some of the needs identified may have been developed since the study was conducted. However, our discussion section acknowledges these apps when they are relevant to a need that was articulated by participants. Lastly, the CHIT guidance provided centers around tracking symptoms to improve outcomes. It has been shown that monitoring biomarkers may be a more effective indicator of disease progression than tracking symptoms alone [64]. Nonetheless, symptom management is an important part of an integrated approach for managing CD and has been shown to be effective in improving the quality of life for individuals with CD [78-80].

Our interviews with individuals with CD reflected 4 main themes, illustrating the fluctuating biform nature of diet and nutrition management as well as the continuous and, in some cases, spontaneous articulation work needed to manage complex interactions between diet and nutrition and other aspects of life with CD. The participants in this study experienced numerous challenges with diet and nutrition management and use of existing technologies to support management activities. Future work should focus on other chronic conditions that require diet and nutrition management to supplement the understanding of biform work, explore the concept of multiform work, and expand on the definition of articulation work. Methods such as journaling could help to capture the nuances and complexities of daily diet and nutrition management. This study also provided descriptive and prescriptive design guidance [81] for CHIT designed to support the holistic and variable experience of diet and nutrition management for individuals with CD. Subsequent research should focus on using this guidance as a foundation for original designs or to redesign technologies for this purpose. Refinement of the proposed design guidance through an iterative participatory design process is essential for the development of interventions that will benefit individuals with CD and promote long-term engagement [82].