We included UK-based adult patients who were actively using treatment for asthma [37] and excluded anyone who needed carers’ support to manage their asthma. We recruited patients from volunteer databases and asthma charities social media. We identified patients in the Scottish Health Research Register (SHARE) [38], Register for Asthma Research (REACH), and Asthma UK volunteer database [39]. People registered in these databases have given consent to be contacted about asthma research. Eligible patients were invited with emails sent by SHARE, REACH, and Asthma UK, which included a recruitment link on behalf of the research team. We also posted advertisements on Asthma UK and Asthma UK Centre for Applied Research (AUKCAR) Facebook and Twitter that included a recruitment link. People who were interested in the study used the link to register their interest. They were asked to read the information leaflet, confirm their eligibility, provide basic demographics, and give consent to be contacted (via the contact details they provided) in order complete the registration.

We purposively recruited a maximum variation sample of 12 patients to take part in individual interviews. Sampling was based on age (16-25 years, 26-45 years, 16-65 years, 65 years or older); ownership of an action plan (or not); duration of asthma (diagnosed <6 months, 6-12 months, 1-10 years, >10 years); hospital admission in the previous 12 months (or not); confidence with using technology (ability to download apps by themselves, need help, never tried).

We recruited health care professionals from primary and secondary care through newsletters and social media (the NHS Research Scotland Primary Care Network and professional bodies such as the Primary Care Respiratory Society and the NHS Lothian Respiratory Managed Clinical Network). We also approached individual professionals who were actively involved in clinical care or research through personal networks.

We adopted a think-aloud approach [40] to explore patients’ and professionals’ preferences on the future design and utility of IoT systems to support asthma self-management.

We asked patients to complete 2 tasks in the interview. For task 1, we provided a list of the most wanted app features (Multimedia Appendix 1) identified in our previous study [7] and asked patients to think about their previous usage of asthma apps and their previous asthma self-management experience to decide the features that they would want to be included in a future IoT system, and how often they think they would use them. For task 2, we provided images of current and future potential smart devices in the market (smart inhaler, smart peak flow meter, smart fabric, smart watch, voice assistant, smart purifier, human robot, and robot pet) and available data and asked patients to discuss their future potential. Specifically we asked them to use the images to create a IoT system that they would want to support their daily self-management, and if, how, and when they would use each device for self-monitoring and what data (if any) they would want to be able to send to their health care professionals and their health carers, for example, parents or spouse. (Figure 1 is an example of a task completed by a participant.)

At the end of each task, participants were asked to add any features, data, or smart devices that they thought would be useful but that were not included in the list or images (or were not yet available).

Professionals were asked to think about their previous experiences providing care for people with asthma and select the features from the list [7] that could support self-management and data that they would like to receive in order to assist their consultations with patients.

Finally, we showed participants a prototype app (an interface app which included many wanted features [7]) to stimulate their thoughts on how they would like to interact with a self-management support IoT system (Multimedia Appendix 2).

The web-based questionnaire aimed to quantify the features wanted by interviewees and their perceived usefulness enabling us to triangulate the findings. Participants’ feedback in the qualitative interview informed the features listed.

The questionnaire (Multimedia Appendix 3) collected basic demographic information (age group and gender) and asked questions regarding what data that they would want an IoT system to collect and where they would want the data to be collected. The flow of questions reflected the sequence that we used in the interviews. To reduce instances of missing data, each question had to be completed before progressing to the next. First, we offered a list of features based on our previous research [7] and the interview responses. Survey participants were asked to select the top 5 types of data that they would like collected by the IoT system and to prioritize them. A free-text option was provided. Second, we asked participants to choose where it would be most useful for the technology to collect these data. The choices were informed by the interviews—“home,” “at work/office/school,” “the place where they are at leisure activities that they do regularly (eg, gym, running, etc),” and “others.” If participants chose the leisure option, we asked the participants to specify which leisure scenario they thought was relevant to that data type. We offered a variety of activities informed by the interviews and a free text option.

We used Bristol Online Survey, a secure web-based survey platform that complies with ISO27001 information security standards and the General Data Protection Regulation, to build the web-based questionnaire and collect data. To test content validity and readability, we invited a patient volunteer and an independent researcher to try out the questionnaire independently. Their feedback was incorporated in the questionnaire before it was sent to participants. Potential participants who had not been selected for an interview and who had given permission to be contacted regarding the survey were emailed a participant information sheet and a link to the questionnaire.

Interviews were digitally recorded, transcribed, and coded in NVivo (version 12; QSR International). We used PRISMS taxonomy [3] as a framework to categorize the IoT features (themes) that emerged from the interviews. In addition, our thematic analysis explored issues of particular importance to the participants. One reviewer (CYH) coded 2 interviews (patient, professional) independently, and another reviewer (HP) reviewed the coded transcriptions to standardize the coding, which was then applied (by CYH) to all the transcriptions. CYH (reviewed by HP) coded features suggested by patients and professionals separately and extracted the features, which were then combined in tables for comparison. We categorized the individual assessments of the features as would “always use,” “often use,” “use when needed,” or “less likely to use” according to the words used in the interviews (Table 1). In our previous study [7] on self-management features wanted by patients’ and clinicians’, no new app features were generated after 15 interviews. We estimated that a sample size of 12 would be likely to achieve data saturation and was within our resources for conducting individual interviews.

We used descriptive statistics to analyze patients’ preference ratings for the data they would like a future IoT system to collect and in which circumstances the system would be most useful. We used data from the survey to quantify the features wanted by interviewees and their perceived usefulness.

The findings, data synthesis, and interpretation were discussed regularly within the multidisciplinary study team, which included a patient representative, technology developer, health care professionals, and researcher. The researcher had an engineering and technology background and clinical research experience with patients and clinicians in asthma app development.

Invitations were sent by email to 572 patients (SHARE n=211; Asthma UK n=220; REACH n=141). Advertisements were posted to the Asthma UK’s Facebook and Twitter (185,800 followers) and AUKCAR Twitter (1224 followers). Of 362 patients who expressed interest in the study, 297 (82.0%) were from Asthma UK’s social media, email, or website; 42 (11.6%) were from the AUKCAR Twitter; 10 (2.8%) were from SHARE; 3 (0.8%) were from REACH, 2 (0.6%) were recommended by their hospital or a relative, 8 (2.2%) received an e-mail invitation from Asthma UK and others responded to our advertising. Of the 362 patients, 12 were selected for an interview, and the remaining 350 were invited to complete the web-based questionnaire: 3 rejected the invitation (2 were no longer available to take part and 1 declined when they realized there was no payment), and 29 were undeliverable emails. Thus, 318 were sent the web-based questionnaire link, of whom 140 (44.0%) completed the questionnaire. Of the 140 participants, 139 participants (99.3%) completed the questionnaire within 15 minutes, and 1 person (0.7%) appeared to take 3 hours and 42 minutes to complete the questionnaire (most likely because the thank you page was not closed after completion). There were no instances of missing data.

Of 152 participants (12 interviewees and 140 web-based questionnaire respondents), most (74/152, 49%) were 46 years to 65 years of age, 69.1% (105/152) were female, 54.6% (83/152) had an action plan, 75.0% (114/152) had been diagnosed with asthma for more than 10 years, 12.5% (19/152) had been admitted to hospital in the previous 12 months, and 86.2% (131/152) were confident that they could download apps themselves (Table 2).

Twelve professionals were recruited from primary, secondary, and tertiary care in the United Kingdom. All provided care for people with asthma, and some had research experience using digital technology to monitor patients’ medication use and symptoms for respiratory patients (Table 3).

Perceptions of the 12 patients about the potential usefulness of features were mapped to the PRISMS taxonomy (Table 4; the full version can be found in Multimedia Appendix 4). Features wanted by patients reached saturation within 10 interviews, and we stopped sampling at 12 interviews.

Patients decided the usefulness of potential IoT features based on their own past asthma self-management experiences (eg, ownership or use of action plan and smart peak flow meter, their relationship with their clinicians, medication usage, use of emergency services), their asthma triggers, their curiosity about what affected their asthma, the severity and control of their asthma condition, and what they considered (or had been told by professionals) was best practice for asthma.

Monitoring, supported by feedback advice, was the feature that most patients wanted to see in an IoT system. Information about asthma and an action plan were also priorities. Flexible access to follow-up advice with a general practitioner or asthma nurse by text or email service were “safety net” features that most patients thought would create a sense of “staying connected” with clinicians, although they stated they would only use it when needed. One patient with hearing problems who struggled to communicate during an exacerbation wanted a panic button to automatically text for emergency help. Perceptions on monitoring of control, and feedback are presented in more detail below. Other features were wanted by patients but were lower priority (Multimedia Appendix 4).

The responses to the survey are illustrated with a bubble plot (Figure 2). In keeping with the priorities of the interviewees, the largest bubbles represent monitoring of symptoms (prioritized by 25/140, 17.9%) and peak flow measurement (33/140, 23.6%). For most respondents (24/32, 75.0%), this was a task to be undertaken at home, though some considered that symptom monitoring could be useful at home, work, or leisure locations. Monitoring environmental asthma triggers (pollen, humidity, air temperature) was prioritized by 23.6% (33/140).

Some patients explained that a peak flow was “useful” and “needed” to help clinicians assess their asthma. Most professional interviewees agreed with this and thought that a log of peak flows, symptom scores, and medication use would engage patients and could inform assessment of control and management strategies.

However, in contrast to the priority attached to logging, most patients acknowledged that in reality they checked their peak flow “rarely” or “only when their asthma was getting worse.” The reasons for not measuring every day were varied. Some “forgot,” while others felt “weird doing it in front of people,” but many suggested it was unnecessary as they knew their asthma and could assess status by how they felt.

Several suggestions were made about how an IoT system could help overcome this discrepancy. A word that was used frequently was “automatic.” Most patients wanted an IoT system that could automatically log their asthma condition. They were interested in trying smart devices (eg, smart peak flow meter); some explicitly mentioned that this would enable them to capture data automatically though they were not clear how the devices could capture their asthma status without some effort on their part. A device with an automated data transfer feature still requires the user to blow the air into the meter, and patients acknowledged that to do this regularly would require motivation such as a request from researchers or their clinician. Patients had different opinions on using voice assistants, smart watches, smart fabrics, smart purifiers, human robots, and robot pets to collect additional data automatically:

How the data were used was an important motivator. Most patients said it was “good” or even “essential” that data from smart peak flow meters and inhalers were transferred automatically to the system so these logs could be correlated with other data and displayed graphically on a mobile phone. Some suggested that they would log peak flows if their clinicians asked them to do so.

Clinicians were keen to see a record of medication use to enable assessment of adherence. They favored automated logging of this information via a smart inhaler as that was perceived to be more accurate.

Patients in the interviews suggested a broad range of information and advice that could be usefully provided through an IoT system such as a timely alert when their asthma control changed; the amount of medication to be taken according to asthma control to reduce medication side effects; the numbers of doses taken and remaining in the inhaler, ideally with an option to order a repeat prescription when the medication was running low; and correct inhaler technique.

The clinician interviewees concurred with these priorities, especially the need for detecting poor inhaler technique and linking to information about correct use of an inhaler. They also saw value in feedback supporting treatment adjustment according to an action plan.

Outdoor environmental data such as pollen, humidity, and air temperature were considered “good to know” and “useful” data by many patients. Patients who were newly diagnosed with asthma wanted to learn which environmental factors affected their asthma, whereas those who already knew what triggered their asthma wanted to use daily environmental data to plan their day.

These data were of less interest to clinicians though some suggested that, together with indoor triggers, outdoor pollution, exercise intensity, weight, peak flow, and symptoms, data could provide real-time feedback and help patients understand which factors affected their asthma in order to avoid them.